Generated 2026-11-02 · State: MO · Standards scope: Grades 6–8 mathematics + Grade 6 English Language Arts · Rebuilt from the audited crosswalk data on each revision.
Grades 6–8 went live in AcornSmarts on 15 September 2026. The other bands for Missouri are published separately: K–5 · grades 9–12 · all reports.
| Code | Verdict | Evidence (generator · forms) | Notes |
|---|---|---|---|
| 6.RP.A.1 | PARTIAL two-auditor sweep 03 Oct 2026 |
ratiosSaplingQ L5 part-to-part vs part-to-whole in a group context (a:b vs a:total distractors); equivalent-ratio recognition (which ratio equals a:b, with order-flip distractor) ▶ Play a live item from this skill |
Ratio as a comparison of two quantities is assessed by recognition at Ratios L5: 'In a group of 10, there are 7 red marbles and 3 blue marbles. What is the ratio of red marbles to blue marbles?', keyed 7:3 against 7:10 (part-to-whole) and 3:7 (order flipped), plus 'Which ratio is equivalent to 3:5?'. 'Represent' is a production verb, and here the ratio is always picked from three options, never written or typed. The only built representation is the L5 plotting item, where the child copies the pairs of a printed equivalent-ratio table onto a grid (about a third of those items cannot be completed; see 6.RP.A.3a). PARTIAL. Blocking clause: represent these comparisons |
| 6.RP.A.2 | PARTIAL two-auditor sweep 03 Oct 2026 |
ratiosSaplingQ L6 price per notebook (per = divide), train unit rate with unit-labeled answers (mph), better-buy per-ounce comparison ▶ Play a live item from this skill |
Unit-rate concept and computation solidly exercised; the 'describe the meaning' clause is production — the meaning lives in answer labels and rationales, the student never describes it. 26 Sep 2026 (round 2): with Writing Coach on — a paid plan or a live school sponsorship, district switch on — the child describes the meaning of a unit rate in writing (3 notebooks for $6), graded against a rubric requiring the rate and the division that finds it; one fixed prompt, live rounds only. Without Writing Coach this row is served by selection only, so the published verdict stays PARTIAL: no row whose clause is met only through Writing Coach is marked FULL. Blocking clause: describe the meaning of unit rate |
| 6.RP.A.3a | PARTIAL two-auditor sweep 03 Oct 2026 |
ratiosSaplingQ L5 (kind 1) equivalent-ratio table described in words (n1 lemons -> n1*per oz, n2 -> ?, n3 -> n3*per oz), find the missing value ▶ Play a live item from this skill |
Find-missing-values is exercised (a Ratios L5 table with one value missing). Plotting is exercised in live rounds. About one Ratios L5 question in six prints a table of equivalent ratios ('lemons → ounces: 1 → 2, 2 → 4, 3 → 6'), and the child plots the three pairs on the coordinate grid, then picks what the pattern shows. When the rate is 4, the pair (3, 12) is off the 0-9 grid and that item cannot be completed (about a third of these items). The child never creates a table of equivalent ratios; every table is printed. Coordinate Grid L5 also has the child plot points (a rectangle's vertices), but not ratio pairs. Blocking clause: Create tables of equivalent ratios ... and plot the pairs of values on the Cartesian coordinate plane |
| 6.RP.A.3b | FULL two-auditor sweep 03 Oct 2026 |
ratiosSaplingQ L6 total cost / count = price per one (with half-dollar prices), distance/time = unit rate, better-buy via per-ounce rates ▶ Play a live item from this skill |
Every whole-number L6 form is solve-a-unit-rate-problem; distractors encode 'more total = better' and inverted-rate errors. (The level also serves the Grade-7 fraction-rate family.) |
| 6.RP.A.3c | FULL two-auditor sweep 03 Oct 2026 |
percentsSaplingQ L5-L6 L5: p% of n (p mult of 5, n mult of 20); L6: find the whole given part and percent ('part is p% of a number'), and find the percent from part/whole in context ▶ Play a live item: level 5 · level 6 |
Forward percent-of, find-the-whole inversion, and find-the-percent are all generated — the Grade-6 percent triangle is exercised on all three corners. |
| 6.RP.A.3d | PARTIAL two-auditor sweep 03 Oct 2026 |
ratiosSaplingQ L7 inches->feet (incl .5 results), hours->minutes (the .5h=50min trap), recipe ratio scaling ▶ Play a live item from this skill |
Conversions are within-system only (customary length, time). No metric<->customary 'between two systems' item exists anywhere in the generator. Blocking clause: within and between two systems of measurement |
| 6.NS.A.1a | FULL two-auditor sweep 03 Oct 2026 |
fractionsSaplingQ L6-L7 L6: whole / unit fraction (servings), unit fraction / whole (share brownies), same-denominator counting; L7: 1/b / 1/(bk) computed, a/b / 1/d word problem computed, structure check a/b / c/d = a/b x d/c ▶ Play a live item: level 6 · level 7 |
Within fractions L6-L7 every COMPUTED division has a unit-fraction divisor (or, for 1/d ÷ w, a whole-number divisor); the general a/b ÷ c/d (c>1) appears in this skill only as invert-and-multiply structure recognition. It is computed to a value only in ratios L6 unit-rate items (for example 1/3 of a liter in 3/4 of a minute, 4/9 liters per minute), which are built for the Grade 7 unit-rate standard (MO 7.RP.A.1) rather than this one. |
| 6.NS.B.2 | FULL two-auditor sweep 03 Oct 2026 |
divisionSaplingQ L6-L7 bare 3-4 digit / 2-digit division (quotient 11-42, divisor 12-29) with off-by-one distractors; factory packing word problem (up to ~3500 / 2-digit) ▶ Play a live item: level 6 · level 7 |
Fluency is shown by computing: about two items in five are typed on a number pad, so the quotient is keyed in, not picked. The multiple-choice L6 items offer near-miss quotients (q ± 1 or 2), which a child can sort by checking last digits instead of dividing. At L7 the wrong choices start as the answer plus the box size and the total ÷ 10; in about half the deals one of them is swapped for a smaller near-miss, often the answer minus the box size. The key is never the largest choice. Dividends stay under about 3,500 and every quotient is exact (no remainders), which is the honest ceiling of the claim. |
| 6.NS.B.3 | FULL two-auditor sweep 03 Oct 2026 |
decimalsSaplingQ L6-L7 L6: add and subtract money-style hundredths (place-alignment distractors +0.10/+1.00); L7: 0.x times 0.y (place-count distractors) and tenths / 0.x (multiply-both-by-10) ▶ Play a live item: level 6 · level 7 |
All four operations are generated, all as multiple choice: adding and subtracting hundredths (1.05 to 8.49), 0.x × 0.y with one-digit tenths, and a tenths dividend up to 8.8 divided by a one-digit tenth with a whole-number quotient. In addition and subtraction the choices give the key away in every deal: the wrong choices are 0.10 and 1.00 above the key, or one 0.10 below and one 0.10 above it, so the key is the smallest choice when one choice sits about 1 away and the middle choice otherwise. In about a third of the multiplication items every choice has the same digits, so only the decimal point is tested. Multiplication and division never go past one-digit tenths. Because the addition and subtraction answers can be found without adding or subtracting, the row is not marked FULL. |
| 6.NS.B.4a | FULL two-auditor sweep 03 Oct 2026 |
factorsSaplingQ L5-L6 GCF of g*m1 and g*m2 (coprime m's, values <=100) with min-multiplier and 2g distractors; LCM of pairs like 6&8 with a*b-is-not-always-least distractor ▶ Play a live item: level 5 · level 6 |
Both GCF and LCM are found by computation; whys walk the factor/multiple structure. |
| 6.NS.B.4b | PARTIAL two-auditor sweep 03 Oct 2026 |
factorsSaplingQ L6 (distributive branch) 'Use the distributive property: a + b = ?' with answer g(m1+m2) and un-divided-addends distractor g(a+b) ▶ Play a live item from this skill |
Factors Level 6 asks, for example, 'Use the distributive property: 18 + 24 = ?' and the child chooses 6(3 + 4) over 6(18 + 24) and a wrong-factor form, 3(6 + 4). This is about half of Level 6 questions, in English and Spanish. The child never writes the rewritten sum. PARTIAL. Blocking clause: express a sum of two whole numbers with a common factor as a multiple of a sum of two whole numbers |
| 6.NS.C.5 | PARTIAL two-auditor sweep 03 Oct 2026 |
integersQ L1, L4 below-zero temperature -> -n; drone/diver above/below sea level; earned/spent with what-does-0-mean-here ▶ Play a live item: level 1 · level 4 |
The child picks the signed number for a temperature below zero and for a diver below sea level, and picks what 0 and the signs mean for money earned and spent — in English and Spanish. 'Represent' is a production verb, and the child never writes or types the signed number; every one of these is a choice among three. Blocking clause: Use positive and negative numbers to represent quantities — the signed number is only picked, never written |
| 6.NS.C.6a | PARTIAL two-auditor sweep 03 Oct 2026 |
integersQ L1 (form 2) 'On the number line, where is -n?' -> 'n units LEFT of zero' (recognition) ▶ Play a live item from this skill |
Integers L1 asks where −n sits and the child picks the words 'n units LEFT of zero'; no line is drawn at that level, and since the number is always negative the answer is always LEFT. Two number lines run into the negatives: the Grade 7 frog hop at integers L5, where the child taps where a sum lands, and the inequality graph at Inequalities L3-4, where the child taps a boundary that is always a positive whole number. On both, every number is printed on its own button, so the child finds a labelled number rather than locating one. Only integers are located at the Grade 6-8 levels (fractions are placed only in below-band levels: the 0-to-1 line in the Grade 3-5 Fractions levels and line plots in Graphs & Data); no decimal is located, and no line is vertical. Blocking clause: on a horizontal or vertical number line |
| 6.NS.C.6b | PARTIAL two-auditor sweep 03 Oct 2026 |
integersQ L2 which of two negatives is LESS; account-balance comparison (-$debt vs -$debt2, who has less money) ▶ Play a live item from this skill |
Ordering is exercised at integers L2: the child types a true comparison of two negative integers using < or > ('Write a TRUE comparison of −3 and −13 using < or >', about one L2 question in five). The debt item ('Who has LESS money?') is not counted, because its right answer is always Beto. 'Explain' is asked only through the metered Writing Coach grader ('Explain why −3 is less than −1…'), which this report does not count. Only pairs are ordered — never a list of three or more — and no negative fraction or decimal is ever ordered; positive decimals are compared in pairs at Decimals L4-5. Blocking clause: Write, interpret and explain |
| 6.NS.C.6c | FULL two-auditor sweep 03 Oct 2026 |
integersQ L1 (form 0) 'What is the opposite of v?' (v either sign) with same-number and off-by-one distractors; why states other-side-of-zero, same distance, sum to zero ▶ Play a live item from this skill |
Assessed by recognition at integers L1, in English and Spanish: the child picks the opposite of a positive or negative number (the same number and a nearby wrong number are the traps), with feedback that the opposite sits on the other side of zero at the same distance; and the child picks where −n sits on the number line (in one of three wordings — 'n units LEFT of zero', 'LEFT of zero' or 'n units to the LEFT of zero' — against a RIGHT-of-zero wording and 'between 0 and n, on the POSITIVE side'). The number to locate is always negative, and no number line is drawn. |
| 6.NS.C.7 | PARTIAL two-auditor sweep 03 Oct 2026 |
integersQ L3 |v| = ? (v either sign, negative-answer distractor); order-vs-magnitude statement item: -a < -b but |-a| > |-b| (the 6.NS.C.7-family trap) ▶ Play a live item from this skill |
The child computes the absolute value of a positive or negative integer ('|−16| = ?'), and the feedback explains it as distance to zero; but the question can be answered by dropping the sign, no number line is shown, and every value is an integer. The order-versus-magnitude item ('−35 < −30, but |−35| > |−30|') is not counted: its right answer is always the only choice joined by 'but'. A written explanation is asked only through the metered Writing Coach grader, which this report does not count. Blocking clause: the absolute value of a rational number is its distance from 0 on the number line — integers only, the distance idea is in feedback, and the comparison item gives its answer away |
| 6.NS.C.8 | PARTIAL two-auditor sweep 03 Oct 2026 |
percentsSaplingQ L7 percent -> decimal (one-place-shift distractor), fraction -> percent, decimal -> fraction (p/10 denominator trap) over a 10-entry halves/quarters/fifths/tenths table ▶ Play a live item from this skill |
Percents L7 converts in three directions over a ten-entry table (halves, quarters, fifths, tenths): 'Which decimal equals 40%?', 'What percent equals 2/5?', 'Which fraction equals 0.25?'. Every answer is picked from three options. 'Generate' is a production verb, and no item in the app has the child type or write an equivalent fraction, decimal or percent. The 1/3 and 2/3 benchmarks are absent here, which the 'and/or' wording permits; thirds appear as repeating decimals at integers L7 and irrationals L1, not credited here. PARTIAL on 'generate'. Blocking clause: generate equivalent representations |
| 6.EEI.A.1 | PARTIAL two-auditor sweep 03 Oct 2026 |
expressionsQ L1 'Which of these is an EQUATION?' — ax+b=r against two of ax+b, ax−b, ax+b>r and ax+b+r ▶ Play a live item from this skill |
The distinction itself is assessed by recognition (needs the = sign, an expression claims nothing), but the expectation's verb is 'describe' — production. Honest PARTIAL rather than a verb-swap FULL. Blocking clause: Describe |
| 6.EEI.A.2a | PARTIAL two-auditor sweep 03 Oct 2026 |
expressionsQ L1 'In cx + k, what is the COEFFICIENT?' with constant and c+k distractors ▶ Play a live item from this skill |
26 Sep 2026 (round 2): with Writing Coach on, the child describes the expression 4(x + 7) in writing — its two factors, the two terms of the sum inside, and what the 4 does — graded against a rubric requiring all three; without it only 'coefficient' is asked (multiple choice). Still PARTIAL: outside Writing Coach no other part of an expression — term, factor, sum, product, constant — is ever asked for, and no row whose clause is met only through Writing Coach is marked FULL. Blocking clause: Identify parts of an expression using mathematical terminology — only 'coefficient' is asked outside Writing Coach |
| 6.EEI.A.2b | FULL two-auditor sweep 03 Oct 2026 |
expressionsQ L2 evaluate ax + b at x = value, with the digit-concatenation trap (a then x glued = 'ax') and a(x+b) as distractors ▶ Play a live item from this skill |
Substitution-then-order-of-operations is the work. Single-variable only — the plural 'variables' is technically wider, noted for the adversary. |
| 6.EEI.A.2c | PARTIAL adversarial: NOTE-ONLY · two-auditor sweep 03 Oct 2026 |
expressionsQ L2 (exponent branch) b^e = ? (b 2-5, small e) with b*e and b+e distractors ▶ Play a live item from this skill |
Non-negative rational number expressions are evaluated across the Grade 6 levels, in English and Spanish. Examples: decimal sums, differences, products and quotients (5.94 − 5.05, 0.4 × 0.7, 3.6 ÷ 0.4); fraction quotients (1/4 ÷ 1/24); volume formulas at half-unit heights (4 × 3 × 2.5; V = Bh with h = 3.5). In Expressions, b^e is evaluated. Limit: every fraction or decimal item is a single operation or a product of factors. The only evaluated expressions that mix operations so that their order matters use whole numbers (ax + b at a given x), so no rational-number expression is evaluated across operations. |
| 6.EEI.A.2d | FULL two-auditor sweep 03 Oct 2026 |
expressionsQ L2–L3 (typed exprwrite:twice / exprwrite:tickets; numpad evaluate) Type 2n − k from words and per·b + k from a ticket-cost story (judged by equivalence); type the value of ax + b at x = v on the number pad. ▶ Play a live item: level 2 · level 3 |
26 Sep 2026 (round 2): the child types algebraic expressions from a phrase and from a ticket-cost situation, judged by equivalence, in about half of live Level 3 items, and types the value of ax + b on the number pad in about a quarter of live Level 2 items; the write bank has two templates. |
| 6.EEI.A.2e | PARTIAL two-auditor sweep 03 Oct 2026 |
equationsSaplingQ L6 (dependent-variable branch) + expressionsQ L3 'Which is the DEPENDENT variable?' over 3 contexts (cost/pounds, distance/hours, pay/lawns) with 'neither' distractor; context expressions where the variable's meaning (b = tickets) drives the answer ▶ Play a live item from this skill |
Variables are used in context: the child types expressions and equations whose letter is named in the situation (5b + 3 for the cost of b tickets in Expressions Level 3; s − 7 = 12 for Ivy's stickers in Equations Level 5). No item asks what a variable stands for, and the dependent-variable items at Equations Level 6 use words, not letters. PARTIAL. Blocking clause: Understand the meaning of the variable in the context of the situation |
| 6.EEI.A.3 | FULL two-auditor sweep 03 Oct 2026 |
expressionsQ L4 (typed exprwrite:combine / exprwrite:dist) Type ax + bx with like terms combined and a(x + b) expanded; the given, bracketed and uncombined forms are refused. ▶ Play a live item from this skill |
26 Sep 2026 (round 2): the child generates equivalent expressions by typing a(x + b) expanded (any bracket refused) and ax + bx combined into one term, with the given form refused; multiple-choice items continue to cover identifying them; factoring is asked only as multiple choice (expressions Level 5, and the factors skill's distributive rewrite). A bracketed sum such as (3 + 4)x is still accepted as combined. |
| 6.EEI.B.4 | FULL two-auditor sweep 03 Oct 2026 |
equationsSaplingQ L6 (substitution branch) + inequalitiesQ L2 which value from {x-1, x, x+2} makes ax + b = rhs true; which value from a 3-element set makes x + a > c true (boundary value as distractor) ▶ Play a live item from this skill |
Equations Level 6 asks which value from a three-number set makes an equation such as 3x + 4 = 19 true, about three Level 6 questions in ten, in English and Spanish. The right value's place in the set varies, and the explanation substitutes the right value. Inequalities Level 2 asks the same for x + a > c and x + a < c, but the right value is always the largest number for > and the smallest for <, so those items can be answered by position. The equation items carry this standard. FULL. |
| 6.EEI.B.5 | FULL two-auditor sweep 03 Oct 2026 |
inequalitiesQ L2-L3 + equationsSaplingQ L6 rationale 'How many solutions does x > c have?' (infinitely many vs one vs none); test-the-set items whose whys frame the answer set ▶ Play a live item: level 2 · level 3 |
The solution-set-as-values-that-make-it-true idea is directly asked (infinitely many, the boundary excluded), not just narrated. |
| 6.EEI.B.6 | PARTIAL two-auditor sweep 03 Oct 2026 |
equationsSaplingQ L5 solve x+a=, x-a=, ax=, x/a= (bare symbolic one-steps, both-sides whys) ▶ Play a live item from this skill |
The child types equations from situations at Equations Level 5 (s − 7 = 12 for Ivy's stickers, 4c = 72 for equal boxes of crayons, with the story's numbers required) and Level 6 (d = kt from a table), and types one-step solutions. The letter and what it stands for are always given in the prompt, and no item asks what a variable means in the situation. PARTIAL. Blocking clause: Write and solve equations using variables to represent quantities |
| 6.EEI.B.7 | FULL two-auditor sweep 03 Oct 2026 |
equationsSaplingQ L5 all four one-step forms: x + a = c, x - a = c, ax = c, x / a = c, with wrong-inverse-operation distractors ▶ Play a live item from this skill |
Equations Level 5 serves one-step equations with non-negative rational numbers, solved by choosing from three, in English and Spanish: x + p = q and x − p = q with decimals (x + 1.5 = 7.3), and (1/n)x = q (such as (1/3)x = 4). These come up in about one Level 5 question in four or five, alongside whole-number one-steps in all four operations, some typed. In the decimal and fraction items the right answer's size rank among the choices varies. In the whole-number items it is never the largest for x + a and ax, and never the smallest for x ÷ a. FULL. |
| 6.EEI.B.8a | FULL two-auditor sweep 03 Oct 2026 |
inequalitiesQ L1 choose h > c for 'must be MORE than c inches'; choose d < m for 'LESS than $m left' (= sign as distractor) ▶ Play a live item from this skill |
26 Sep 2026 (round 1, two auditors agreed FULL): inequalitiesQ L1 deals a typed inequality from a worded constraint in all four forms (h > c, p ≤ c, s ≥ c, d < c), EN+ES, ~half of live L1 items; the checker refuses a strict/inclusive swap and a wrong variable. Paper/placement/duel keep the MC form (covers > and < only). |
| 6.EEI.B.8b | FULL two-auditor sweep 03 Oct 2026 |
inequalitiesQ L3 'what does the solution set of x > c look like?' asked for x > c, x < c, x ≥ c or x ≤ c — an open or a filled circle at c, shaded right or left, going on forever, beside two of the other three circle-and-direction pictures (the wrong circle, the wrong direction, or both, each named in the explanation); 'how many solutions' companion ▶ Play a live item from this skill |
26 Sep 2026 (round 1, two auditors agreed FULL): inequalitiesQ L3 carries a ray spec for x >, ≥, <, ≤ c; RayPick scores the exact boundary + inclusive + direction (36 constructions, one correct); about a quarter of live L3 items (26–27% measured 03 Oct 2026) EN+ES, none in plain mode. |
| 6.EEI.C.9a | FULL adversarial: RECLASSIFY · two-auditor sweep 03 Oct 2026 |
equationsSaplingQ L6 (twoVarRelationQ form 0, typed d = kt or d = t + k) From a table of distance d and hours t, type d in terms of t (d = …); t = d/k is refused; the table is proportional (d = kt) or additive (d = t + k). ▶ Play a live item from this skill |
26 Sep 2026 (round 2): the child types the equation from a table of distances and hours, with d required alone on one side so t = d/k is refused; one context, in about one live Level 6 item in ten. 10 Oct 2026: two equation shapes — proportional d = kt or additive d = t + k — in equal shares. |
| 6.EEI.C.9b | PARTIAL adversarial: DOWNGRADE · re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
equationsSaplingQ L6 (twoVarRelationQ: table → equation, table → value beyond the table, shown graph → equation) 'The graph shows distance d (y-axis) for hours t (x-axis). Which equation shows the relationship?' over three plotted points at t = 1, 2, 3 — e.g. (1, 2), (2, 4), (3, 6) or (1, 3), (2, 4), (3, 5); the four options use one adjacent pair of numbers, {1, 2} (d = t, d = 2t, d = t + 1, d = t + 2) or {2, 3} (d = 2t, d = 3t, d = t + 2, d = t + 3), and any of the four can be the key; the table item offers the same four-equation set for an adjacent pair from 2 to 8 (e.g. d = 4t, d = 5t, d = t + 4, d = t + 5) beside a t | d table; 'Using the relationship, what is d when t = 6?' from the table. ▶ Play a live item from this skill |
Equations Level 6 shows a table of distance d against hours t: the child types the equation from it or uses the relationship to find d for a time beyond the table, in English and Spanish. Since 10 Oct 2026 the relationship is proportional (d = kt) or additive (d = t + k), each in half the deals, and the shown-graph item (three plotted points, at t = 1, 2, 3 in every deal) and the table item each offer FOUR equations — the key, the same form with the wrong number, the other form with the key's number and the other form with the wrong number. The two numbers are an adjacent pair drawn first ({1, 2} or {2, 3} on the graph, so d = t is a real key; any adjacent pair from 2 to 8 on the table) and the key takes either number and either form, so every option has one twin in form and one in number and every equation shown is the key in a quarter of the deals that show it — none is shown that is never the key. Measured over 600+ deals per language, 'the option with a twin in form and a twin in number' scores 25% on the four options (chance) where it scored 75% on the old three (chance 33%); 'pick a product', 'pick a sum', the smaller or larger number, the shortest or longest option and 'strike any option never seen as the key' all score 25% as well. Every wrong is named in the explanation with the point (or row) it does and does not fit, and the key is re-derived from the plotted points in saplingGrade6Gaps.test.ts, which locks these rates at the 95% bound. Each of the graph and the table is related to the equation by choosing it; no item relates the graph to the table, and no value is read off the graph. Verdict held at PARTIAL for the second re-audit. Blocking clause: using graphs ... and relate these representations to each other |
| 6.GM.A.1 | FULL two-auditor sweep 03 Oct 2026 |
geometrySaplingQ L6-L7 L6: triangle (half-parallelogram, forgot-/2 distractor), parallelogram (perimeter distractor), trapezoid (missing-/2 distractor); L7: L-shape by decompose-and-subtract, kite via half-the-diagonal-rectangle ▶ Play a live item: level 6 · level 7 |
Compose/decompose is not just the method in the why — the L-shape item is decomposition by construction, and the whys derive each formula from rectangles/triangles. |
| 6.GM.A.2a | PARTIAL two-auditor sweep 03 Oct 2026 |
volumeSaplingQ L4 compute V of an l x w x (h+0.5) prism where the l*w*hWhole distractor is exactly the missing half-LAYER ▶ Play a live item from this skill |
The layers idea is carried by the distractor structure and rationale ('The half-unit of height adds half a layer') of the L4 fractional-edge item; no item at the Grade 6 levels (volume L4-L5) asks about filling a prism with layers of the base. The filling items in the skill are K-5 (volume L1-L3, whole-unit edges): the L3 computation 'How many unit cubes fill a 6 × 5 × 9 box?' and the pack-the-prism task (interactive rounds), where each tap adds the next unit cube, bottom layer first, and the child then answers for the volume. Neither ask names layers; layers appear in two pack tasks' rationales (4 × 3 × 2: 'two layers of 4 × 3 … counts the same cubes layer by layer'; 4 × 5 × 2: 'two layers of 4 × 5') and in the skill's generic hint ('count the layers of cubes'), and neither item has a fractional edge. 2a's understanding is therefore inferred from a 2b computation item. Graded down to keep 2a and 2b from double-claiming the same question. Blocking clause: Understand that the volume of a right rectangular prism can be found by filling the prism with multiple layers of the base |
| 6.GM.A.2b | PARTIAL two-auditor sweep 03 Oct 2026 |
volumeSaplingQ L4-L5 V = l*w*h with a fractional (.5) edge; V = Bh given base AREA B and fractional height (B-already-is-the-area why), typed on the number pad in interactive rounds and picked from three on printed sheets, the placement check and Duel ▶ Play a live item: level 4 · level 5 |
Volume L4 applies V = l × w × h to a right rectangular prism with an n.5-unit height, in English and Spanish, with the answer's position among the choices varied. Volume L5 gives the base area B and an n.5 height and says 'Use V = Bh'. Until 24 Oct 2026 its choices gave the answer away (a .5 choice meant the larger whole number, otherwise the middle choice); the first repair that day added 'the base counted twice' (2Bh), which was always double the answer, so 'the smaller of a doubled pair' still found it in about 70% of deals and was replaced the same day. Since the founder decision of 24 Oct 2026, interactive rounds have the child type the volume on the number pad ('What is the volume, in cubic centimeters?') with no choices on screen, so there V = Bh must be applied; the pad shows a decimal-point key only when the volume needs one (exactly when B is odd), and a typed slip (the half dropped, B + h added, or the height rounded up) gets its own explanation. Printed sheets, the placement check and Duel cannot take typing and keep three choices: the two wrong ones are a pair drawn from seven named slips (the half dropped, the height rounded up, B + h added, both bases added on, one layer too many, one layer short, the half layer worked out as B ÷ ½); no choice is ever a whole-number multiple of another, the answer is the smallest, middle or largest choice about a third of the time each, B is odd about half the time so the answer itself can end in .5, and no single .5, doubling or 'two closest choices' rule picks it more than about 44% of the time (chance 33%). But there the choices still give the answer away to a determined guesser: how many choices end in .5, read together with which two choices sit closer (or whether all three are evenly spaced), finds the answer in about 81% of deals (chance 33%). The blocking clause records both; the verdict stays PARTIAL until the row is re-audited. Blocking clause: Apply ... V = Bh — resolved for interactive rounds on 24 Oct 2026 (founder decision: the volume is typed on the number pad, with no choices on screen); printed sheets, the placement check and Duel keep three choices, which still give the answer away in about 81% of deals (how many end in .5, read with which two sit closer). The verdict stays PARTIAL until the row is re-audited. |
| 6.GM.A.3a | PARTIAL two-auditor sweep 03 Oct 2026 |
coordinateSaplingQ L4 (quadrant branch) which quadrant holds (x, y) with nonzero signed coordinates; why maps sign-of-x/sign-of-y to left-right/up-down and names the counterclockwise numbering ▶ Play a live item from this skill |
Signs-indicate-quadrant is the whole item; all four quadrants reachable. The two wrong choices are two of the other three quadrants, picked at random (24OCT2026 note update: until then they were always the first two other quadrants, so IV was offered only when it was the answer; verdict unchanged). Blocking clause: consistency with the agreed 6.NS.6b / 6.GM.A.3b downgrades: the only generator offered quadrant IV only when it was the key (fixed 24 Oct 2026: two of the three other quadrants are now drawn at random; the verdict stays PARTIAL until the row is re-audited) |
| 6.GM.A.3b | PARTIAL two-auditor sweep 03 Oct 2026 |
coordinateSaplingQ L4 (reflection branch) reflection of (x, y) across the x-axis, with across-y (-x, y) and across-both (-x, -y) as the distractors ▶ Play a live item from this skill |
Coordinate-grid L6 asks which transformation carries (x, y) to (−x, y) or to (x, −y); either axis can be the answer, with a translation as the third choice - about one question in four at that level, in English and Spanish. That serves points that differ in one sign. The coordinate-grid L4 reflection item always reflects a quadrant-I point across the x-axis, so its answer is always the only choice with a positive first coordinate; it does not test the relationship. No Grade 6-8 item asks about two points that differ in both signs (a reflection across both axes); at these levels that pair appears only as a wrong choice and in the L4 explanation ('flipping both reflects twice'). Blocking clause: related by reflections across ... both axes (never asked); the published L4 reflection item gives its key away by form |
| 6.GM.A.3c | FULL adversarial: NOTE-ONLY · two-auditor sweep 03 Oct 2026 |
coordinateSaplingQ L5 distance between (x, y1) and (x, y2) with y1 > 0 > y2; |y1 - y2| with the subtract-instead-of-add |y1 + y2| distractor ▶ Play a live item from this skill |
Coordinate-grid L5 asks distances between points that share a coordinate. The bare two-point distance item is always the same-first-coordinate case and always crosses the x-axis; its explanation adds the two distances to zero. The same-second-coordinate, same-side case is served only as the follow-up to the plot-a-rectangle item at the same level (the side from (x1, y) to (x2, y), first quadrant, subtract the x's). Not served in this generator: a same-side vertical pair, and a horizontal pair that crosses the y-axis. |
| 6.GM.A.3d | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Coordinate-grid L5 has the child construct a rectangle on a coordinate grid by tapping its four given vertices, then read the length of one side - about half of that level's questions in interactive rounds, in English and Spanish. Limits: the only polygon ever built is an axis-aligned rectangle; the grid shows the first quadrant only; the vertices are given, so the child plots rather than designs the figure; the sides are not drawn; and the item never appears on paper or in placement. Blocking clause: Construct polygons in the Cartesian coordinate plane (only axis-aligned rectangles, first quadrant, vertices given) |
| 6.GM.A.4a | PARTIAL two-auditor sweep 03 Oct 2026 |
geometrySaplingQ L8 nets appear as given, worded descriptions ('a cube's net is 6 equal a x a squares'; 'a prism's net has lxw, lxh, wxh rectangles — two of each') feeding surface-area computation ▶ Play a live item from this skill |
About one geometry L8 question in ten on a first round at that level (about one in twenty when the level is replayed, since the item is held back for two rounds once shown) shows three drawn nets — a cross of six squares, a T of five squares and a 2 × 3 block — and asks 'Which of these nets folds into a CUBE?'. The surface-area items describe a cube's or a rectangular prism's net in words. The child never represents a figure with a net: the picker is a choice among the same three drawings every time, its answer is always the cross, and the T can be ruled out by counting its five squares. No net contains a triangle. Blocking clause: Represent |
| 6.GM.A.4b | PARTIAL two-auditor sweep 03 Oct 2026 |
geometrySaplingQ L8 cube SA = 6a^2 from its net (volume a^3 as distractor); rectangular prism SA = 2(lw+lh+wh) from its listed net rectangles (forgot-the-2 and volume distractors) ▶ Play a live item from this skill |
Surface area is found from a net described in words at geometry L8: a cube (6 equal squares; the wrong options are one face's area and, in about half of deals, the volume) and a rectangular prism (three pairs of rectangles; the forgot-the-2 total is almost always a wrong option, the volume about half the time). Not practised: no net-based item has a triangle face. The only solid with triangle faces whose surface area is asked is the square pyramid at geometry L10 ('A square pyramid has base side 4 and slant height 6. What is its surface area? (SA = B + ½Pl)'), worked from the formula printed in the prompt, not from a net. The surface-area nets are words, not drawings. Blocking clause: and triangles |
| 6.DSP.A.1 | FULL two-auditor sweep 03 Oct 2026 |
statisticsQ L1 which IS / which is NOT a statistical question, drawing the correct answer and both distractors from a 7-pair bank of variability-vs-single-answer questions ▶ Play a live item from this skill |
Recognize is the standard's own verb; the anticipates-variability rationale is the discriminator between the paired questions. |
| 6.DSP.A.2 | FULL two-auditor sweep 03 Oct 2026 |
statisticsQ L3-L4 which group is more spread out (range comparison); which value is the OUTLIER in a clustered list; whys describe distributions by center, spread, and stray values ▶ Play a live item: level 3 · level 4 |
Center and spread are assessed by recognition; 'overall shape' (peaks, symmetry, skew) is named only in rationale text — the shape corner is the attack surface on this FULL. |
| 6.DSP.A.3 | FULL two-auditor sweep 03 Oct 2026 |
statisticsQ L2-L3 compute mean/median/mode/range with concept-bearing distractors (median-as-mean, mean-as-median, unsorted-middle, range-is-not-center; the median is typed on the number pad in every interactive deal, its mean and unsorted-middle slips answered with their own notes); 'the range is not even a measure of center' discrimination in L3 ▶ Play a live item: level 2 · level 3 |
Single-number summary vs how-values-vary is carried by the computations and the L3 center-vs-spread items. The computations include the median offered as the mean; the list's mean, the middle of the list as printed and other values from the list as wrong answers for the median; and 'the range' rejected as a measure of center in the L3 house item. The median item prints its five values in a shuffled order, never already sorted, and each list has one far value whose mean is itself one of the five values. Since 24 Oct 2026 (founder decision) interactive rounds have the child type the median on the number pad in every deal, with no choices on screen; typing the mean or the middle of the list as printed gets that slip's own explanation. Printed sheets, the placement check and Duel keep three choices, all values from the list, the mean among them in about five deals in eight. There, neither the middle printed value nor any rule on the choices' order or spacing finds the answer more than about 42% of the time, and the choice nearest the mean about 37%, but the choices still give the answer away, stated openly: the mean is always the median's neighbour in the sorted list, so a child who works out the mean exactly and takes the closest other choice finds the median every time (on paper the item does not force sorting on a child who can compute the mean); with no arithmetic, 'a choice that is the list's largest or smallest value → the choice farthest from it, otherwise the middle choice' finds it about 75% of the time, and about 80% read with the list's range; 'smallest choice 6 or less → the largest choice, otherwise the smallest' about 58%; and the choice second-closest to the midpoint of the extremes about 63%. |
| 6.DSP.B.4a | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Statistics L5 serves data displays. Dot plots are built by the student (from a listed data set, the student taps to stack one ✕ per value on the plot builder) and are also read from a drawn plot (total count, mode). Histograms are read from rendered interval bars (which interval holds the most, total count). Box plots are read from a rendered figure (median, range, IQR, share of the data inside the box). The statistics L2 mean, median, mode and range items give their data as comma-separated lists. Limit: within statistics L5 the dot plot is the only display the student builds; histograms and box plots always arrive already drawn, and no item in the app has the student construct either one. Blocking clause: Use histograms and box plots to display numerical data |
| 6.DSP.B.4b | FULL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
statisticsQ L5 (display round: piebuild + circleGraphQ) Create: colour a 12-wedge circle so each category gets its count (scored on the real wedge counts). Interpret: largest category, share of the whole, percent of a count on a rendered circle graph. ▶ Play a live item from this skill |
26 Sep 2026 (round 1, two auditors agreed FULL): PieBuild (v2.217.0): the child colours a 12-wedge circle to the given counts and is scored on the real wedge counts, ✓ gated until every wedge is coloured — 'create' is now performed; circleGraphQ covers 'interpret'. |
| 6.DSP.B.5a | PARTIAL two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
The number of observations is asked directly at statistics L5 in four multiple-choice forms: "How many observations does this data set have?" over a short investigation blurb that states the count; the total shown by a drawn dot plot; the total shown by a rendered histogram; and how many values a stem-and-leaf display holds. In the statistics L2 mean items, counting the five listed values stays implicit inside the computation and is not asked on its own. Limit: all four forms are multiple choice; no statistics item has the student write a summary that states the number of observations. Blocking clause: Report the number of observations: chosen from three options, never stated by the student |
| 6.DSP.B.5b | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Statistics L5 serves an investigation family: a one-sentence blurb (six contexts, for example backpacks weighed on a scale in kilograms, or lap times taken with a stopwatch in seconds) followed by one of three multiple-choice questions — what ATTRIBUTE was under investigation, what UNIT the observations were recorded in, or how many observations there are. In English each blurb names how the measuring was done (scale, stopwatch, ruler, asking, thermometer, counting); the Spanish lap-time blurb says only that the classmates were timed. None of the three questions asks about the method, so the how-it-was-measured clause is not assessed by this family. All three forms are multiple choice over a supplied blurb; no statistics item has the student describe an investigation in their own words. Blocking clause: Describe the attribute, including how it was measured |
| 6.DSP.B.5c | PARTIAL two-auditor sweep 03 Oct 2026 |
statisticsQ L2 find the MEAN (skew-aware distractor), find the MEDIAN (sort-first), find the MODE, find the RANGE of 5-value lists ▶ Play a live item from this skill |
Mean and median are computed at Level 2: the median is typed on the number pad in every interactive deal (since 24 Oct 2026), the mean in about half of interactive deals; printed sheets, the placement check and Duel keep choices. MAD and IQR are typed at Level 4, along with range, in about a quarter of L4 deals. The IQR is also read off a box plot at Level 5. The row stays PARTIAL because describing the overall pattern and striking deviations in words has no item; the outlier is selected. Blocking clause: interquartile range and/or mean absolute deviation |
| 6.DSP.B.5d | PARTIAL two-auditor sweep 03 Oct 2026 |
statisticsQ L3-L4 which measure of center BEST describes a typical house when a $200k+ outlier is present (median vs mean-uses-every-value vs range); which measure changes MOST when the outlier is removed ▶ Play a live item: level 3 · level 4 |
The choice of a measure of CENTER is analyzed by recognition at Levels 3-4. One item asks which measure best describes a typical house when one $200k-$350k house is present: the median, against 'the mean — it uses every value' and the range. Others print a list of six values with one outlier ('Data: …') and ask which measure best describes a typical time, or which changes most when the outlier is removed. The outlier dragging the mean is the deciding idea; a choice based on shape, as opposed to an outlier, never comes up. The row is held PARTIAL because no item asks the child to choose a measure of VARIABILITY (range, IQR or MAD) for the data. Blocking clause: Analyze the choice of measures of variability based on the shape of the data distribution and/or the context |
| Code | Verdict | Evidence (generator · forms) | Notes |
|---|---|---|---|
| 7.RP.A.1 | FULL two-auditor sweep 03 Oct 2026 |
ratiosSaplingQ L6 price per notebook (per = divide, half-dollar prices), train unit rate with mph-labeled answers, better-buy per-ounce comparison ▶ Play a live item from this skill |
Forms: typed complex-fraction unit rates ('A pump moves 3/4 of a liter in 2/5 of a minute. What is the unit rate, in liters per minute?', answered with the / key) in like-unit and different-unit contexts, at least two per Ratios L6 round. There are also whole-dollar and half-dollar unit prices (price per notebook) and whole-number speeds (train, typed in about six train items in ten). Note: 26 Sep 2026 (round 1, two auditors agreed FULL): sapling78PartialToFull.test.ts re-derives (a/b)/(c/d) from each prompt and requires non-unit divisors > 50%, like units > 18%, area > 18%; 600/600 probe rounds carried at least 2 typed items. |
| 7.RP.A.2a | PARTIAL two-auditor sweep 03 Oct 2026 |
ratiosSaplingQ L8 table (x,y) triples: is-it-proportional yes/no, with a broken-last-pair variant and a does-not-start-at-(1,1) distractor ▶ Play a live item from this skill |
Ratios L8's table item ('Table: (5, 35), (7, 49), (10, 73). Is the relationship proportional?') is the only item that asks this, in English and Spanish. Its options give the answer away: 'no — the last pair breaks the constant ratio' is offered only when it is the answer, and when it is not offered the answer is always 'yes — every y is k times its x'. The other two options ('the pairs do not start at (1,1)', 'cannot be determined from three pairs') are never right. Under the published rule an item whose options reveal the key does not count. Proportionality is never judged from a graph, equation or description. PARTIAL. Blocking clause: Determine when two quantities are in a proportional relationship |
| 7.RP.A.2b | FULL two-auditor sweep 03 Oct 2026 |
ratiosSaplingQ L8 find k from a proportional table (k = y/x, confirmed on the second row), with first-y and x-value distractors ▶ Play a live item from this skill |
Identify-and-compute is exercised exactly as stated; why derives k = y ÷ x. Narrow range: k = 2-8, two-row tables, and k is never read from an equation y = kx. |
| 7.RP.A.2c | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Functions L5 has the student plot a proportional line's points, (0, 0), (1, k), (2, 2k), (3, 3k), on a coordinate grid and then asks 'What does the point (1, k) tell you?', with the unit rate credited (for example 'the unit rate: y grows by k when x grows by 1'); that item names no real-world situation. Ratios L8 shows a ticket-cost graph with (0, 0) and the points from x = 1 plotted and the line y = kx drawn from the origin (since 10 Oct 2026, commit 282dd18) and asks 'What does the point (1, 2) mean?', keyed to one ticket costing $2 (for example 'buying 1 ticket costs $2'; the '— the unit rate' tag was dropped on 04 Oct 2026). A sister form asks 'Why does the line pass through (0, 0)?', keyed to zero tickets costing zero (for example 'for zero tickets you pay zero dollars'). Limits: only the special points (1, r) and (0, 0) are interpreted; no item asks what a general point such as (3, 6) means in the situation. The graph items use one context (ticket cost) and a unit rate of 2 or 3. Every answer is selected from three options, never explained by the student. Blocking clause: Explain what a point (x, y) on the graph ... means in terms of the situation |
| 7.RP.A.2d | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Ratios L8 generates an origin item — a coordinate grid with (0, 0) and the points from x = 1 plotted and the line y = rx drawn from the origin, and the prompt "The graph shows the cost y of x tickets. Why does the line pass through (0, 0)?", keyed "zero tickets cost zero dollars", with an explanation that (0, 0) is always on a proportional line and that a flat fee lifts a graph off the origin, with the three answer options beneath the grid. Since 10 Oct 2026 (commit 282dd18) the grid plots (0, 0) with the other points and draws the line y = rx from the origin, so the picture shows the line passing through (0, 0) that the prompt names; before that only the points from x = 1 onward were plotted and no line was drawn. Functions L5 has the student plot proportional lines starting at (0, 0), but its questions are about the unit rate and the slope, not the origin. Limits: the origin item is one template in one context (ticket cost, unit rate 2 or 3); its prompt tells the student the line passes through (0, 0) and asks why — no item asks the student to decide from a graph whether a relationship is proportional by checking the origin, and no non-proportional graph is shown for contrast; the same family is credited to 7.RP.A.2c. The ratios L8 table item's distractor "no — the pairs do not start at (1,1)" concerns tables and does not mention the origin. Blocking clause: Recognize that the graph of any proportional relationship will pass through the origin |
| 7.RP.A.3 | FULL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
percentsSaplingQ L8 + ratiosSaplingQ L6-L8 sales-tax total (with one-step ×1.05-style multiplier in the why), percent change against the ORIGINAL value; unit-rate and proportional-table problems ▶ Play a live item from this skill |
Multi-step percent problems and rate and proportion problems both generate. Percents L8 serves tax then total, percent increase and percent decrease ('A price falls from $140 to $112. What is the percent decrease?'). In about one question in three it serves simple interest, tip then total, and commission. Ratios L5-L8 add missing-value tables (L5), unit rates including typed complex-fraction rates (L6), unit conversions and recipe scaling (L7), and the constant of proportionality from a table (L8). Not present: markups named as such. Percent-change values are locked to {10, 20, 25, 50}. |
| 7.NS.A.1a | PARTIAL two-auditor sweep 03 Oct 2026 |
integersG7Q L5 negative + positive climb, a − (−b) subtract-a-negative, distance between signed values ▶ Play a live item from this skill |
Add/subtract with signs is exercised at integers L5. Three of the five L5 arithmetic forms use integers; two use signed decimals ('(-8.5) + 2.5 = ?', '4.5 − (−1.5) = ?'), always ending in .5 with whole-number answers. 'Rational numbers' at Grade 7 means signed fractions too (−3/4 + 1/2), and no signed-fraction operand generates. In the add/subtract forms, non-integer operands are limited to decimals ending in .5. Blocking clause: Add and subtract rational numbers (beyond integers and .5 decimals) |
| 7.NS.A.1b | PARTIAL two-auditor sweep 03 Oct 2026 |
integersG7Q L5 number-line language in prompts (distance 'on the number line') and whys ('adding on the number line is MOVEMENT') ▶ Play a live item from this skill |
26 Sep 2026: at integers L5 the child taps where a frog's hop lands on a number line that runs into the negatives. The hops are adding a positive number from below zero, and subtracting a positive number across zero; together they are about 10% of Level 5 questions. Still PARTIAL: only integers appear on the line, adding or subtracting a negative number is never shown, and the line is drawn for the child. The line's ends also give the landing spot away: on every subtraction hop it is the third number from the left, and on about half of the addition hops it is the third from the right. Blocking clause: Represent addition and subtraction on a horizontal or vertical number line |
| 7.NS.A.1c | PARTIAL adversarial: DOWNGRADE · re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
integersG7Q L5 (form 1 rationale) subtract-a-negative item whose why runs through opposites ('taking away a debt leaves you richer') ▶ Play a live item from this skill |
Three items at integers L5-7 target this. One is '8 + (−8) = ?' (answer 0). The other two are a temperature story asked two ways: 'rose 8° in the morning and fell 8° in the afternoon. What was the total change?' (answer '0° — the opposite changes cancel'), and 'rose 8° in the morning. What afternoon change would bring it back to where it started?' (answer '−8° — the exact opposite of the rise'). Together they are about 8% of questions at L5, 5% at L6 and 7% at L7. They use whole numbers only and a single context, and the term 'additive inverses' appears only in explanations. The child always picks from given choices and never describes a situation. The one item that asks for a description ('Explain why 8 + (−8) = 0 … use a number line or a real-life situation') is graded by the Writing Coach grader, so it does not count toward FULL. Blocking clause: Describe situations and show that a number and its opposite have a sum of 0 |
| 7.NS.A.1d | PARTIAL adversarial: DOWNGRADE · re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
integersG7Q L5 (form 1) a − (−b) = ? with a−b and −(a+b) distractors; why states 'subtracting a negative adds its opposite' plus the debt interpretation ▶ Play a live item from this skill |
At integers L5, 'a − (−b) = ?' (integers or .5 decimals) puts a + b beside a − b and, in about half of draws, −(a + b), so the common a − b slip is caught. A second item asks for the rewrite directly ('Rewrite 4 − 9 as addition.' → 4 + (−9)), with the wrong rewrites drawn from p + q, −p + q, q + (−p) and −p + (−q), so a bracketed negative does not mark the key. The first number is always positive: p − q with a negative p never appears in any Grade 6-8 item, and no signed fractions appear. Held PARTIAL. Blocking clause: Understand subtraction of rational numbers (any sign, beyond .5 decimals) as adding the additive inverse |
| 7.NS.A.1e | PARTIAL adversarial: DOWNGRADE · re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
integersG7Q L5 (form 2) distance between p (negative) and q (positive) on the number line; distractors are the sum's magnitude and |p| ▶ Play a live item from this skill |
Distance between two signed numbers is directly the question at integers L5 ('What is the distance between -5 and 7 on the number line?'). The straddle form's rationale states the idea: 'Distance = |7 − (-5)| = 7 + 5 = 12. The difference of two numbers IS the distance between them — signs included.' Pairs either straddle zero (p<0<q) or are two negatives on the same side of zero, where the |a| + |b| shortcut is a distractor. Two positive points never generate, the points are integers only, and no line is rendered. Blocking clause: distance between two rational numbers ... is the absolute value of their difference |
| 7.NS.A.1f | PARTIAL two-auditor sweep 03 Oct 2026 |
integersG7Q L5 (rationales) debt-removal and below-zero-climb interpretations inside whys ▶ Play a live item from this skill |
The L7 context items interpret products and quotients (a temperature dropping each hour, a debt split among friends), not sums. Sums appear in a situation only in the temperature story at integers L5-7, which asks 'rose 8° … fell 8°, what was the total change?' or 'what afternoon change would bring it back?'. Both cover only opposite quantities summing to zero. A nonzero signed sum is interpreted in a situation only in a Writing Coach explain prompt at integers L5 ('A submarine at −40 m rises 25 m. Write the sum … and explain what the answer tells you'). That prompt needs the metered writing grader and does not count toward FULL. No item interprets a difference of signed numbers in a situation; that idea appears only in explanations ('Taking away a 5-dollar debt leaves someone 5 richer.'). Held PARTIAL. Blocking clause: Interpret sums and differences of rational numbers |
| 7.NS.A.2a | PARTIAL two-auditor sweep 03 Oct 2026 |
integersG7Q L6 (−a)×(−b), (−a)×b with sign-count whys, (−nd)÷dv with sign-inheritance check ▶ Play a live item from this skill |
At integers L6, negative × negative, negative × positive and negative ÷ positive generate; the divisor is always positive. Three of the four sign forms use integer operands; the fourth uses signed decimals ('(−2.5) × 6 = ?', a negative integer ÷ 1.5), always built on .5 values with integer results. Signed fractions never appear as operands at this level. Blocking clause: Multiply and divide rational numbers (signed fractions; negative divisors) |
| 7.NS.A.2b | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
At integers L5-7 one item asks for the reciprocal directly: 'What number do you multiply 7/9 by to get 1?' (answer 9/7; the explanation shows 7/9 × 9/7 = 63/63 = 1). The given fraction is proper or improper and positive or negative (−3/4 → −4/3), and the wrong choices are two of the fraction itself, its opposite and the reciprocal with its sign flipped, dealt so the answer is not always the largest. It is in English and Spanish and makes up about 5-8% of questions. Held PARTIAL: the reciprocal is picked, never applied, the given value is never a whole number, and the item belongs to the generator family shared with the quotient-of-integers form credited to 7.NS.A.2c. Blocking clause: Determine that a number and its reciprocal have a product of 1 (the only item's choices give the answer away; positive proper fractions only) |
| 7.NS.A.2c | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Two items at integers L5-7 test this. The first asks 'What is ALWAYS true of the quotient of two integers, like -4 ÷ 8?' and credits 'it is a rational number (as long as the divisor is not 0)' against 'always an integer' and 'irrational when it does not divide evenly'. In it, the dividend is always a negative integer from -2 to -10 and the divisor a positive integer from 2 to 8. Since 20 Sep a second item asks 'What kind of number is -36 ÷ 6?' and credits 'an integer (-6), and a rational number too'. Held PARTIAL: the first item has the same correct sentence in every draw, and the second's answer is always the only choice that shows a number. The divisor is never negative, and the non-zero-divisor condition is never asked on its own. The signed-division items at L6-L7 do not mention rational numbers. Blocking clause: Understand that every quotient of integers (with non-zero divisor) is a rational number |
| 7.NS.A.2d | FULL two-auditor sweep 03 Oct 2026 |
longDivisionQ (integers L6); percentsSaplingQ L7 Long division at integers L6: supply the next quotient digit from a quotient-so-far and a remainder, read a returning remainder (repeats) or a remainder of 0 (terminates), decide terminate-or-repeat from a lowest-terms denominator, convert generated fractions (denominators such as 7, 12, 13, 16, 24) to decimals. Also at percents L7: fraction → percent → decimal triple equivalence over a fixed halves/quarters/fifths/tenths table ▶ Play a live item from this skill |
Long division serves as an algorithm at integers L6, with at least two such items in every round. The items ask for the next quotient digit from a quotient-so-far and a remainder, what a returning remainder or a remainder of 0 means, and whether a lowest-terms fraction terminates or repeats. They also include full conversions ('Use long division: write 2/7 as a decimal.'), where the wrong choices are digit slips of the true quotient. Items are in English and Spanish. Weaknesses: every fraction converted is positive and less than 1 (no negative rationals or improper fractions), the child chooses among three decimals rather than typing one, and the full-conversion form is about 6% of L6 questions. |
| 7.NS.A.2e | PARTIAL adversarial: UPGRADE · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Integers L6 makes the decimal side concrete: the child converts generated fractions such as 3/7 to decimals by long division, supplies the next quotient digit from a remainder, decides from a lowest-terms denominator whether the decimal terminates or repeats, and reads a remainder trail that either reaches 0 (the decimal terminates) or returns (it repeats); every full round at that level carries at least two long-division items. The explanation gives the general reason - dividing by d leaves only d − 1 possible non-zero remainders, so one must come back or reach 0 - and irrationals L1 asks for that bound directly ("what is the MOST digits the repeating block of 1/12 can have?"). The fraction side is asked too: integers L7 turns 0.333…, 0.666…, 0.111…, 0.444…, 0.555… and 0.777… into fractions, irrationals L1 converts generated pure and delayed repeats, and an integers item shows that an integer quotient such as −36 ÷ 6 is "an integer (−6), and a rational number too". English and Spanish. Weaknesses: every fraction is positive and proper, all answers are chosen from three options, and the general claim is reached through worked instances and the remainder-count item rather than stated by the child. |
| 7.NS.A.2f | PARTIAL adversarial: DOWNGRADE · re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
integersG7Q L7 temperature drops rate°/hr for hrs hours (product with direction-vs-size why); friends splitting a −$debt (quotient whose sign is the interpretation) ▶ Play a live item from this skill |
There are two L7 context items: a temperature that drops a set amount each hour (whole-number rates, or 1.5°, 2.5° or 3.5° an hour), and friends splitting a −$ debt. In both, one wrong choice differs from the answer only in sign; a positive share 'would say each person GAINED money.' Held PARTIAL: in these items the child always goes from a context to a number. The only item that asks the child to describe a real situation for a given quotient is a Writing Coach prompt at fractions L6-7 ('Explain what 3 ÷ 1/2 = 6 means using a real situation'). That prompt needs the metered writing grader and does not count toward FULL. No product is ever described this way. There are only two scenarios, and the debt amounts are whole numbers. Blocking clause: Interpret products and quotients of rational numbers by describing real-world contexts |
| 7.NS.A.3 | PARTIAL two-auditor sweep 03 Oct 2026 |
integersG7Q L7 (+ L5-L6 bare ops) real-world signed multiplication and division problems; bare add/subtract/multiply/divide items ▶ Play a live item from this skill |
Signed-number problems are solved at integers L5-7. There are bare add, subtract, multiply and divide items with integers or .5 decimals (always whole-number answers), and two L7 situations: a temperature dropping 2-5° or 1.5-3.5° an hour (multiply) and a debt split among friends (divide). The only addition situation is the temperature story whose opposite changes sum to 0. No subtraction situation appears, no item adds, subtracts, multiplies or divides a signed fraction (a negative fraction appears only as the given value of the reciprocal item), and no item in the integers skill chains two operations. Held PARTIAL. Blocking clause: Solve problems involving the four arithmetic operations with rational numbers (addition and subtraction problems; non-integer/fraction operands) |
| 7.EEI.A.1 | PARTIAL adversarial: DOWNGRADE · re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
expressionsG7Q factor −ax − ab (sign-checked by expanding), expand −(x − b) (distribute-the-minus), combine ax − cx to a negative coefficient ▶ Play a live item from this skill |
Simplify and factor both appear in expressionsG7Q (expressions Level 5), all multiple choice: expand (1/2)(4x + 6) or 0.5(…), add (m.5x + b) + (k.5x − d), expand −(x − b), combine ax − cx, and factor −ax − ab. Fraction and decimal coefficients appear only in the first two. The expand items require the work, though on expand −(x − b) both wrong answers keep the − before b, so the key is the only option with a +. Factoring is held back on two counts. On the factor item (the only factoring of a linear expression in Grades 6-8) the right answer shares its outside factor with one wrong answer and its bracket with the other, so 'pick the option that shares the most with the other two' finds it in every deal. The same rule finds the combine key in every deal. The add item no longer gives its key away: since 11 Oct 2026 it deals four options, the key and three named slips (the coefficient slipped, the constant subtracted the wrong way round, and both), and the option sharing the most with the others is its key in 25% of deals over 900 deals per language, which is chance. And no factoring item has a fraction or decimal coefficient. Blocking clause: factor linear algebraic expressions — the one factor item's options give the key away |
| 7.EEI.A.2 | PARTIAL adversarial: DOWNGRADE · re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
expressionsG7Q + percentsSaplingQ L8 (rationales) equivalence manipulation items; whys that reveal meaning (tax total as one multiplication ×(100+r)/100) ▶ Play a live item from this skill |
Rewriting is exercised context-free in expressionsG7Q (expressions L5). The clarify-a-quantity payoff is asked directly by one family at equations L7 (expressionContextQ): 'A price a rises 10%. The expression a + 0.10a equals 1.10a. What does that form reveal?' (answer: raising by 10% is the same as multiplying by 1.10), with a discount companion, 'A store offers 30% off a price p. Which expression gives what you PAY?', that selects 0.7p over two of 0.3p, 1.3p and p − 30, drawn at random. Limits that remain: in the reveal item both forms are printed in the prompt and the meaning is selected from three choices, which since the 10 Oct 2026 repair all open 'raising by 10% …' and name the percent and a factor, and since 31 Oct 2026 are the same length, the two wrong ones drawn from multiplying by 0.10 (the increase alone) or by 2.00 (10% read as 1.00), adding 1.10 or 2.00 dollars, or dividing by 1.10 (which would lower the price), so neither the length nor the opening words point to the right one — before the repair it was the only choice with a percent sign; picking the choice that names the 1.10 printed in the prompt still finds it about half the time in English and a little over two times in five in Spanish (one in three by chance), the most any surface rule reaches (in English the choice sharing the most words with the others, and the only one with its operation, also land on it about half the time), and knowing only that 1.10a is a multiplication finds it in about 85% of deals; in the discount item the child selects the finished form and the unfactored form p − 0.30p appears only in the rationale; and the family's only situations are a percent increase and a percent discount on a price. Blocking clause: Understand how to use equivalent expressions to clarify quantities in a problem |
| 7.EEI.B.3a | FULL two-auditor sweep 03 Oct 2026 |
percentsSaplingQ L7 fraction ↔ decimal ↔ percent, all three directions, benchmark table ▶ Play a live item from this skill |
Converting between forms of one number is practised at several Grade 6-8 levels in English and Spanish, every answer picked from three options. Percents L7: 'Which decimal equals 40%?', 'What percent equals 2/5?', 'Which fraction equals 0.25?' (terminating benchmarks only). Integers L6: 'Use long division: write 19/32 as a decimal.', including repeating results (about one Integers L6 question in seventeen; the long-division family it belongs to is held to at least two per round). Repeating decimal to fraction is practised too: irrationals L1 asks 'Write 0.666… (the 6 repeats forever) as a fraction.' and integers L7 asks 'The repeating decimal 0.666… equals which fraction?'. Convert is not a production verb, so picking the equivalent form counts. Weakness: in most percent-to-decimal items the options include the key and ten times the key. |
| 7.EEI.B.3b | NOT COVERED re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— | Equations Level 7 has one reasonableness item, about one Level 7 question in ten: 'You buy 13 notebooks at $4 each. Your classmate estimates $150 total. Is that reasonable?' (prices $2 to $9, 11 to 29 notebooks). In about half of the deals the estimate is a rounding within a quarter of the product and in the rest it is another product from the same range, 1.7 to 12 times too big or too small, so the verdict is dealt both ways and every estimate is a multiple of 10 whether fair or off; fair and off estimates share one range of sizes (repaired 10 Oct 2026: before that a fair estimate always fell between $30 and $110 and an off one at $330 or more or $12 or less, so its size alone decided the item). The yes and no choices use the same wording and both print the computed total ('13 × $4 is around $52, close to / far from $150'), and a third choice says it cannot be judged without a calculator (sometimes naming 13 × $4); the explanation rounds 13 to 10, multiplies, and names how far off the estimate is ('about 3 times too big'). The printed total still does the mental computation for the child — she compares a printed product with the estimate rather than estimating — so assessing reasonableness is not yet genuinely practised. NOT COVERED. Blocking clause: Assess the reasonableness of answers using mental computation and estimation strategies |
| 7.EEI.B.4a | PARTIAL two-auditor sweep 03 Oct 2026 |
equationsSaplingQ L5 one-step x+a, x−a, ax, x÷a with wrong-inverse distractors ▶ Play a live item from this skill |
Equations Level 5 serves x + p = q and x − p = q with decimal p and q, and (1/n)x = q, solved by choosing from three, in English and Spanish. A write item offers x + p = q, x − p = q and px = q (decimal p) against three stories, so the shape of the right answer varies. Every full Level 5 round carries at least two. p and q are never negative, and every px = q the child solves has a whole-number or unit-fraction p. In the juice-box story the number of boxes almost never comes out whole (for example, $10.00 paid for boxes at $5.60 each). PARTIAL. Blocking clause: Write and/or solve equations of the form x+p = q and px = q in which p and q are rational numbers |
| 7.EEI.B.4b | PARTIAL two-auditor sweep 03 Oct 2026 |
equationsSaplingQ L7 solve px + q = r (negative solutions welcome, stop-after-subtract distractor); select the equation for a flat-fee-plus-hourly rental ▶ Play a live item from this skill |
Both forms serve at Equations Level 7: px + q = r and p(x ± q) = r, written (typed rental, perimeter and ticket equations, half-dollar rates included) and solved (typed px + q = r; the rest by choice), in English and Spanish. What a solution means is asked only by one Writing Coach explain prompt ('Explain what t = 6 means in this situation'), which serves only with Writing Coach on, so it does not count. PARTIAL. Blocking clause: p(x + q) = r ... and interpret the meaning of the solution in the context of the problem |
| 7.EEI.B.4c | PARTIAL two-auditor sweep 03 Oct 2026 |
inequalitiesG7Q solve px + q > r and px < c, the coefficient negative in about 3 deals in 5 (dividing by a negative FLIPS the sign); four choices cross the direction (right, or flipped wrongly / not flipped) with the boundary's sign (right or lost); the why names each step and the flip ▶ Play a live item from this skill |
The child types the solution of px + q > r (positive p) and of −ax < −b (the divide-by-a-negative flip), and types a two-step budget inequality from words (≤ form). Still PARTIAL: every typed and graphed item uses integers — rational coefficients (0.5, 0.25, 0.2) appear only in multiple-choice solve items in bracket form, p(x + q) — and the solve-and-graph number line is always centred on the answer and always open, shaded right, so the graph can be built without solving. Blocking clause: graph inequalities of the form px + q > r or px + q < r |
| 7.GM.A.1 | PARTIAL two-auditor sweep 03 Oct 2026 |
geometrySaplingQ L9 (form 2) map scale 1 cm : k km, find real distance (scale-multiplies-not-adds why) ▶ Play a live item from this skill |
Actual LENGTH from a scale is one clean form; scaled AREA (the k² trap — the expectation's hard half) and reproducing at a different scale never generate. One template, lengths only. Blocking clause: computing actual ... areas from a scale drawing and reproducing the drawing at a different scale |
| 7.GM.A.2a | NOT COVERED two-auditor sweep 03 Oct 2026 |
— | No item at the Grade 6-8 levels asks whether given measures (three sides, three angles, or two sides and an angle) make a unique triangle, more than one, or none, and none has the child construct a triangle (drawing a triangle by tapping three corners on a dot grid, or freehand, exists only at the Pre-K-5 levels, with no measures given). Triangle items in the Grade 6-8 levels compute from given measures instead: the area from base and height at geometry L6; the third angle from two angles, an exterior angle, a missing side by the Pythagorean theorem, and the Pythagorean converse at geometry L10; and a missing side of similar triangles at coordinate-grid L6. |
| 7.GM.A.2b | NOT COVERED two-auditor sweep 03 Oct 2026 |
— | No item has the child construct a quadrilateral from given conditions such as side lengths, angles, diagonals or parallel sides. The only quadrilateral a Grade 6-8 child builds is at coordinate-grid level 5, where all four vertices of a rectangle are given ("Draw the rectangle on the plane - plot its four vertices: (2, 1), (5, 1), (5, 4), (2, 4)") and the child plots them, so none of the rectangle's properties is used. The parallelogram, trapezoid and kite items at geometry levels 6-7 compute areas from given measures. |
| 7.GM.A.3 | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Volume L6 serves a cross-section family: "You slice [solid] with a plane [cut]. What shape is the cross-section?" over five rows — a rectangular prism cut parallel to its base (rectangle), a cylinder cut parallel to its base (circle) and straight down through its centre (rectangle), and a square pyramid cut parallel to its base (a smaller square) and straight down through its apex (triangle) — plus three solids-of-rotation items. Limits: none of the five rows slices a cone, which the standard names; every cut is either parallel or perpendicular to the base, so no slanted cut appears; the solids are described in words with no figure; and the student chooses a shape name from three options rather than describing the section. The family shares volume L6 with the cylinder, cone, sphere and pyramid volume items. Blocking clause: Describe two-dimensional cross sections of ... cones (no cone slice; describe served by picking) |
| 7.GM.A.4a | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
geometrySaplingQ L9 (form 0) circumference-vs-area items whose distractor is always the other formula; whys contrast 'the rim' vs 'fills the inside' ▶ Play a live item from this skill |
In the geometry L9 circle item ('A circle has radius 5. What is its circumference, in terms of π?' / 'A circle has radius 5. What is its area, in terms of π?'), the area is offered as a wrong answer to the circumference question and the circumference as a wrong answer to the area question, and answers stay exact in π. No item analyzes C/d = π or derives the area from the circumference, and no circle item gives or asks for a diameter: every circle, arc and sector item gives a radius (the word 'diameter' occurs only in the explanation of a cylinder cross-section item). Blocking clause: Analyze the relationships among the circumference, the radius, the diameter, the area and Pi in a circle |
| 7.GM.A.4b | FULL two-auditor sweep 03 Oct 2026 |
geometrySaplingQ L9 (form 0) C = 2πr and A = πr², each with the other formula and rπ as distractors, answers in terms of π ▶ Play a live item from this skill |
Both formulas are known-and-applied with the confusion pair as the discriminator. Weak-FULL flag: the radius is always GIVEN (never the diameter, never solve-backwards from C or A), r is 3-9, and 'solve problems' is bare-shape only — no context. |
| 7.GM.B.5 | FULL two-auditor sweep 03 Oct 2026 |
geometrySaplingQ L9 (form 1) + geometryG8Q (kind 1-2) vertical-angle and supplementary-pair unknowns (complementary-confusion distractor); triangle third-angle; transversal angle pairs ▶ Play a live item from this skill |
Geometry L9 has the child write and solve the equation for an unknown angle, typed: 'Two angles are supplementary. One measures 65°. Write an equation for the other angle m.' (m + 65 = 180, judged by meaning) and 'Solve m + 65 = 180 for the other angle. Write m = …' (m alone on one side), for supplementary (180°) and complementary (90°) pairs. English and Spanish; about one geometry L9 question in seven in live rounds (paper, placement and duels show the multiple-choice angle items instead). Limits: every equation is one step; vertical angles, triangle sums and transversal angles are asked only as multiple choice; no figure is drawn. |
| 7.GM.B.6a | PARTIAL adversarial: NOTE-ONLY · two-auditor sweep 03 Oct 2026 |
geometrySaplingQ L6-L7 triangle, parallelogram, trapezoid (with forgot-the-÷2 traps), L-shape by decompose-and-subtract, kite via half-diagonal-rectangle ▶ Play a live item: level 6 · level 7 |
Geometry L6-7 asks the area of a triangle, a parallelogram, a trapezoid and a kite, and of an L-shaped figure given as a rectangle with a rectangular corner missing (the same items credited to MO 6.GM.A.1). Not practised: the only 'other polygon' is that L-shape, which is made of rectangles alone; no item asks the area of a polygon composed of triangles and rectangles (such as a rectangle with a triangle on top), and no figure is drawn. Blocking clause: other polygons composed of triangles and rectangles |
| 7.GM.B.6b | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
volumeSaplingQ L4-L5 + geometrySaplingQ L8 + volumeG8Q prism V = lwh / V = Bh (fractional edges); prism/cube surface area from described nets; cylinder V = πr²h ▶ Play a live item: level 4 · level 5 |
Every solid and measure the standard names is served at Grade 6-8 levels, in English and Spanish: prism volume at volume L4 (V = lwh with an n.5 height) and L5 (V = Bh from a given base area); prism and cube surface area from described nets at geometry L8; cylinder volume at volume L6 (in terms of π); cylinder surface area at geometry L8 (a soup can) and geometry L9; pyramid volume at volume L6 (a rectangular-base pyramid as one third of its prism, no formula given); and square-pyramid surface area from base side and slant height at geometry L10. Limits: every prism in these items is a right rectangular prism or a cube - no triangular or other prism appears in any Grade 6-8 volume or surface-area item; each pyramid form is a single template with whole-number answers; the pyramid surface-area prompt prints SA = B + ½Pl; pyramid volume uses rectangular bases only and pyramid surface area square bases only; and the pyramid surface-area item is about one geometry L10 question in fourteen; and the volume L5 V = Bh item is typed on the number pad in interactive rounds (since 24 Oct 2026) but keeps three choices on printed sheets, the placement check and Duel, where they still give the answer away (how many end in .5, read with which two sit closer, finds the answer in about 81% of deals). Blocking clause: pyramids |
| 7.DSP.A.1a | PARTIAL two-auditor sweep 03 Oct 2026 |
statisticsSamplingQ sampling scenarios that all presuppose sample-within-population structure ▶ Play a live item from this skill |
The subset relationship is ambient in every item ('a sample of 40 from all students') but never itself assessed — no item asks what a sample or population IS. Graded down to avoid inheriting the concept from items that assess representativeness instead. Blocking clause: Understand that a sample is a subset of a population |
| 7.DSP.A.1b | FULL two-auditor sweep 03 Oct 2026 |
statisticsSamplingQ 'why is this sample BIASED?' with the scenario-specific mechanism as the answer (early-arrivers share habits, park users pre-approve) against size/payment distractors ▶ Play a live item from this skill |
Valid-only-if-representative is directly assessed: the correct answer is always the representativeness failure, and the why states 'bias lives in HOW the sample is chosen.' Six fixed scenarios — thin bank, flagged. |
| 7.DSP.A.1c | FULL two-auditor sweep 03 Oct 2026 |
statisticsSamplingQ 'which sample is MOST representative?' where the random-selection option beats two interest-selected samples ▶ Play a live item from this skill |
Random-produces-representative is the item's whole discrimination, and the why says it verbatim ('randomness is what erases hidden selection patterns — every member gets an equal chance'). Same six-scenario bank as 1b — the two claims share one bank. |
| 7.DSP.A.2 | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Statistics L6 serves a multiple-samples family in two forms. One form lists five random-sample means from the SAME school (mean sleep hours) and asks what the five samples together show; the credited answer is that the estimates vary a little around a centre and the true mean sits nearby, against the foils that one sample was done wrong or that the school changed between samples. The second form projects one sample's proportion onto a school population, with "about" credited over "exactly" and an explanation that another random sample would give a slightly different estimate. Limits: each form has a single context (sleep hours; pizza preference), and the five means always follow the same offset pattern around the centre (−0.2, +0.1, −0.1, +0.2, 0); the sample means arrive already computed — the student never draws samples; and the variability among the estimates is judged in words only, never computed. The team-comparison item compares two different populations' means and is credited to the 7.DSP.B rows, not here. Blocking clause: Use data from multiple samples to draw inferences about a population and investigate variability in estimates |
| 7.DSP.B.3 | FULL two-auditor sweep 03 Oct 2026 |
statisticsSamplingQ (comparison form) two teams' means with 'similar spread' stated; valid-conclusion selection ▶ Play a live item from this skill |
Statistics L6 analyzes two distributions with statistical measures. Two samples are drawn as stacked dot plots on matching number lines (both lists also printed). With their MAD of 1 stated, the child picks how many MADs apart their means are, and judges how much they overlap. Two random samples are compared through a printed center and spread: mean with MAD, median with IQR, medians (equal in half the deals, far apart in the other half) with different ranges, and equal means with different modes. Range, MAD and IQR of a single data set are typed. Limits: the displayed samples are always the same eight-value shape shifted, with a MAD of exactly 1, sitting either 1 or 6-7 apart; the spreads in the two-sample comparisons are printed, not derived. |
| 7.DSP.B.4 | FULL two-auditor sweep 03 Oct 2026 |
statisticsSamplingQ (comparison form) mean A vs mean B (+gap), similar spread → 'B typically scores about gap more' against won-every-game and cannot-compare distractors ▶ Play a live item from this skill |
Statistics L6 compares measures of center, of frequency and of variability from two random samples to infer about the populations. Mean with MAD or median with IQR are compared, and the printed spread decides the inference. A third form prints both medians and both ranges: the medians are equal in half the deals and far apart in the other half, and the options cross a center claim (a typical value is about the same, or higher for one group) with a spread claim (which group varies far more), so each measure answers its own question; a wrong option naming the narrow group as the more varied is on screen in every deal. The fourth sets equal means against different MODES: which group is lopsided and in which direction both vary, and the options cross the mode's side of the mean with the group, so each mode has to be read against the mean. In the range and mode forms two of the three wrong combinations are dealt and each is named in the explanation; over 900 deals per language, the option sharing the most words with the other two is the key in 39–44% of range deals and 31–34% of mode deals (chance 33%). There are five scenario families, and every full round carries at least two comparative items. The measures are printed rather than computed from raw data; the standard asks the child to compare them, which is done. Limit: in the mean and median forms the two spread outcomes sit at the extremes, with no near-threshold case. |
| 7.DSP.C.5a | FULL two-auditor sweep 03 Oct 2026 |
probabilityG7Q L4-L5 P(color) = favorable/total (part-to-part ratio as the trap); die P(rolling more than k) with count-the-outcomes why ▶ Play a live item: level 4 · level 5 |
Simple-event probability is computed in two independent forms with the definitional distractor (a/b part-to-part vs a/total). Sample spaces stay at marbles and one die — noted, not blocking. |
| 7.DSP.C.5b | PARTIAL adversarial: UPGRADE · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Both ideas are asked at probability L4, in English and Spanish. A probability is computed as a number between 0 and 1 - from a bag of marbles, "What is P(red)?" is the favorable count over the total, and the explanation names the part-to-part ratio as not a probability. And a number is turned into a likelihood: "The probability of an event is 7/8. Which word describes it?" (unlikely, likely, as likely as not), whose explanation states that probability lives between 0 and 1, near 0 meaning unlikely, near 1 likely and 1/2 in the middle; that form is about 3 of every 10 L4 questions and appears in about 9 rounds in 10. Weaknesses: it uses five fixed probabilities (1/8, 1/10, 7/8, 9/10, 1/2) and three words; at L4-L6 no item gives a probability of 0 or 1 or offers a value outside 0 to 1, and no item compares two events on the scale. The certain/likely/unlikely/impossible items at L2-L3 are grade 3-5 levels and are not counted here. |
| 7.DSP.C.6a | FULL two-auditor sweep 03 Oct 2026 |
probabilityG7Q L4 (expected-count form) expected = P × trials with draws a multiple of the total (exact), 'about' flagged in the why as tendency-not-guarantee ▶ Play a live item from this skill |
Predicting approximate frequency from theoretical probability is the item verbatim. Single form (marbles with replacement) — flagged. |
| 7.DSP.C.6b | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
No item has the student run trials and record outcomes. Probability L5 serves a design-selection family — "You want to simulate a 3-in-10 chance. Which design models it EXACTLY?" — in which each option describes a chance device (spinner, coin, die, bag of marbles) and the student picks the one whose probability matches the stated chance, across six stated chances. The student selects a design and does not carry it out. Experimental data arrives pre-collected in the L5 coin item (the numbers of flips and heads are given), which is credited to 7.DSP.C.6c. Blocking clause: Perform experiments |
| 7.DSP.C.6c | PARTIAL two-auditor sweep 03 Oct 2026 |
probabilityG7Q L5 (coin form) experimental P = heads/flips (heads generated off-center on purpose) with theoretical 1/2 as the named distractor; why draws the drift-toward distinction ▶ Play a live item from this skill |
The probability L5 coin item asks for the experimental probability of heads from given results ('You flip a coin 50 times and get 18 heads. What is the EXPERIMENTAL probability of heads?'), with the theoretical 1/2 always among the wrong choices, so the child must tell the two kinds of probability apart. No item asks the child to compare them: which is larger, how far apart they are, why they differ, or which to use for a prediction. That comparison appears only in the explanation ('with more flips the experimental one drifts toward it'). One context (coin flips). Held PARTIAL. Blocking clause: Compare theoretical and experimental probabilities |
| 7.DSP.C.7a | PARTIAL adversarial: DOWNGRADE · two-auditor sweep 03 Oct 2026 |
probabilityG7Q L4-L6 equal-probability marble and die models used to compute event probabilities; two-coin item whose why enumerates the 4 equally likely outcomes against the 1/3 equal-outcomes fallacy ▶ Play a live item: level 4 · level 5 · level 6 |
Uniform models are used at probability L4-L6: equally likely marbles, a fair die ('P(rolling more than 2)' — count the faces), a coin with a die, and two coins. In the two-coin item, a 1/3 wrong choice tests the equally-likely assumption, but it is shown in only about half of draws. No item asks the child to assign the equal probabilities that make up the model (for example, that each face of a die has probability 1/6); the model is always given. Held PARTIAL. Blocking clause: Develop a uniform probability model by assigning equal probability to all outcomes |
| 7.DSP.C.7b | PARTIAL two-auditor sweep 03 Oct 2026 |
probabilityG7Q L5 (coin form) experimental probability from observed frequencies of a chance process ▶ Play a live item from this skill |
Estimating one event's probability from data is the mechanism, but no item builds a full (non-uniform) model — a spinner or weighted die with observed frequencies assigned across ALL outcomes never generates; only the fair-coin case appears. Blocking clause: Develop a probability model (which may not be uniform) by observing frequencies |
| 7.DSP.C.8a | PARTIAL two-auditor sweep 03 Oct 2026 |
probabilityG7Q L6 count the sample space of (color, number) pairs by multiplying; two-coin item's why lists HH, HT, TH, TT ▶ Play a live item from this skill |
Since 26 Sep 2026 the sample space of a coin and a 6-sided die appears on screen as a 2 × 6 table, and the child taps every outcome in a stated event (for example 'heads AND a prime number'). Another L6 form counts the outcomes of a spinner and a die by multiplying, and the two-coin item lists HH, HT, TH, TT in its explanation. Held PARTIAL because the child never represents a sample space: the table is drawn for them, and no item has the child build a list, table or tree diagram. Blocking clause: Represent the sample space of a compound event |
| 7.DSP.C.8b | NOT COVERED re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— | Simulation content exists in the app only for simple events: probability L5 asks "You want to simulate a 1-in-4 chance. Which design models it EXACTLY?" across six stated chances, and the student selects a device rather than designing one. One of the six credited designs uses a compound device (three coins for a 1-in-8 chance), but the chance is given as a single stated probability. No item designs or uses a simulation for a compound event, and no item presents frequencies produced by a simulation; the compound-event items at probability L6 (coin and die, two coins, counting a sample space) are theoretical calculations. |
| Code | Verdict | Evidence (generator · forms) | Notes |
|---|---|---|---|
| 8.NS.A.1a | FULL two-auditor sweep 03 Oct 2026 |
irrationalsQ L1 which number is IRRATIONAL (√non-perfect vs √perfect vs terminating/repeating rationals); why is √perfect RATIONAL (equals an integer, general-rule and size distractors) ▶ Play a live item from this skill |
Both directions of the rational/irrational split are assessed, with the every-root-is-rational overgeneralization as a distractor. Irrational examples are square roots only — π never appears as an option. |
| 8.NS.A.1b | PARTIAL two-auditor sweep 03 Oct 2026 |
irrationalsQ L1 rejecting 0.75 / 1/3 / 0.2 as irrational; rationale states 'every terminating or repeating decimal is rational' ▶ Play a live item from this skill |
The two outcomes and the reason are all asked, in English and Spanish. At integers L6 (every full round there carries at least two long-division items) the child reads a remainder trail that either reaches 0 ("the division stops - the decimal terminates") or returns ("the same digits repeat, forever"), with "the digits keep changing with no pattern" as a wrong option, and decides terminate-or-repeat from a lowest-terms denominator. Inside irrationals L1 the child names the block that repeats from a remainder trail (when the repeat starts late, the digits before it are offered as a wrong answer) and answers how many digits a repeating block can run to when dividing by d (d − 1, because only d − 1 non-zero remainders exist), whose explanation states that this is why a rational's decimal always terminates or repeats. Weaknesses: every answer is chosen from three options, every fraction is positive and proper, the general claim appears only in explanations, and the terminating case is asked only at integers L6, not inside the irrationals skill. |
| 8.NS.A.1c | FULL re-audited ×1 · re-audited ×2 (CONFIRMED-RECORD-CORRECTED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Both directions are practised in grades 6-8 skills, in English and Spanish. Repeating decimal to fraction: about 3 in 10 irrationals L1 questions convert single-digit repeats (0.444… = 4/9), two-digit blocks (0.3636… = 4/11) and delayed repeats (0.1666… = 1/6), with every choice in lowest terms; integers L7 adds single-digit repeats. Fraction to repeating decimal: integers L6 runs long division on fractions such as 2/7, 5/12 and 4/13, asking for the next digit or the full decimal (such as 0.28571…). Irrationals L1 asks which block repeats in the long division of a fraction such as 5/7. Weakness: the child chooses rather than types. |
| 8.NS.A.1d | PARTIAL two-auditor sweep 03 Oct 2026 |
percentsSaplingQ L7 fraction ↔ decimal ↔ percent benchmark equivalences ▶ Play a live item from this skill |
Equivalent representations are selected (not generated) from a below-band generator claimed at Grade 6 — same double-claim caveat as 7.NS.A.2d, plus the production verb. Blocking clause: Generate equivalent representations of rational numbers |
| 8.NS.A.2 | PARTIAL two-auditor sweep 03 Oct 2026 |
irrationalsQ L2-L3 √n between which consecutive integers (boxing-in with perfect squares); which integer is √n CLOSER to (midpoint-squared refinement) ▶ Play a live item: level 2 · level 3 |
Estimation is served by two forms at irrationals L2-L3, with the boxing method in the explanation: which two consecutive integers √n sits between, and which integer it is closer to (checked against the half-way square). English and Spanish. On the between-integers form the two wrong pairs are the neighbouring pairs on either side of the right one, with how many sit below it dealt at random, so the right pair is the middle one only by chance (for √2 and √3 nothing sits below it). The closer-to form usually offers both neighbouring integers, but in about one item in nine one of them is replaced ("Which integer is √15 CLOSER to?" offering 4, 5 and 12), and that item needs only the boxing step. No item compares two irrational numbers with each other (√8 vs √12), and 'locations on a number line' is answered in words, never on a drawn line. Blocking clause: compare the size of irrational numbers |
| 8.EEI.A.1 | PARTIAL two-auditor sweep 03 Oct 2026 |
expressionsG8Q (kind 2) b^m × b^n = b^(m+n) with multiply-the-exponents and square-the-base distractors; why counts factors and names (b^m)^n as the trap's home ▶ Play a live item from this skill |
The product-of-powers item ('3^2 × 3^4 = ?') generates at expressions Level 6 — the only bare exponent-rule item in any Grade 6-8 skill. Since 11 Oct 2026 its two wrong answers are one of three pairs of named slips — multiplying the exponents (3^8), multiplying the base by itself (9^6), both at once (9^8), subtracting the exponents (3^2) — each named in the explanation. It used to deal the same two slips every time, so the right answer always shared its base with one wrong answer and its exponent with the other and that rule alone picked it; over 900 deals per language, the option that shares the most with the other two is now the key in 45–48% of deals (chance 33%, the guard's bar 56%), and no length, position or question-echo rule reaches the bar. The add-the-exponents rule is also applied to powers of ten in '(2 × 10^4) × (3 × 10^2) = ?', and how-many-times comparisons subtract exponents of ten, including negative ones. Quotient (b^n ÷ b^m) and power-of-a-power ((b^m)^2) items generate only in the Grade 9-12 skill oak-exponentials; no Grade 6-8 item asks about a zero exponent; outside powers of ten every exponent is a positive integer; and the child never writes an equivalent expression — every item is choose-one-of-three. Blocking clause: Know and apply the properties of integer exponents to generate equivalent expressions — only product of powers is asked, and only as multiple choice |
| 8.EEI.A.2a | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
rootsSciNotationQ (expressions L6) Solve x² = p for a square's side (p a perfect square, 3² to 25²) and x³ = p for a cube's edge (perfect cubes), typed in live rounds, 'x = r' multiple choice on paper; halving p (for x²) and stopping at the square (r² for x³) are among the wrong answers. Non-perfect cases (x² = 50, x³ = 30) are answered by writing the root symbol (√50, 5√2, ∛30). ▶ Play a live item from this skill |
x² = p and x³ = p are solved as equations (a square's area with an unknown side; a cube's volume with an unknown edge). They make up about 7% of Level 6 questions and are typed in more than half of live deals. On paper, halving p (for x²) and stopping at r² (for x³) are among the wrong answers, and for x³ the near misses r ± 1 and r ± 2 are drawn too. Non-perfect cases (x² = 50, x³ = 30) are answered by writing the root symbol. Held PARTIAL because p is always a whole number: no equation has a fraction or decimal p. Blocking clause: where p is a positive rational number — p is never a fraction or decimal |
| 8.EEI.A.2b | PARTIAL adversarial: NOTE-ONLY · re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
rootsSciNotationQ (expressions L6); irrationalsQ L1 Evaluate √p for perfect squares 3² to 25² = 625 and ∛p for perfect cubes, typed in live rounds, with halving (√) and squaring (∛) as the wrong answers. Also √perfect inside the irrationals L1 rational-or-irrational sort. ▶ Play a live item from this skill |
Square roots of perfect squares are evaluated up to 25² = 625 (√9 to √625), and cube roots of 8, 27, 64 and 125, typed on the number pad in live rounds. On paper the two wrong answers are drawn from those slips (half of p and p − r for square roots; r² and p − r for cube roots) and the near misses r ± 1 and r ± 2, each with a check note, with how many sit below the answer dealt at random. Held PARTIAL because the cube roots stop at 125: 216, 343, 512, 729 and 1000, half the range the standard names, never appear. Blocking clause: cube roots of perfect cubes less than or equal to 1000 — only 8, 27, 64 and 125 are dealt |
| 8.EEI.A.2c | FULL two-auditor sweep 03 Oct 2026 |
irrationalsQ L1 √(non-perfect) identified as irrational with why: 'not the root of a perfect square: its decimal runs forever without repeating' ▶ Play a live item from this skill |
This is the item, verbatim, over a 14-value non-perfect-square bank with a √perfect foil. The strongest claim in the 8.NS/EEI radical family. |
| 8.EEI.A.3 | PARTIAL two-auditor sweep 03 Oct 2026 |
expressionsG8Q kind 0 (large write) · sciSmallUnitsQ form 0 (small write) · sciTimesAndUnitsQ (how many times) write large numerals and small decimals in scientific notation; approximate how many times larger one quantity is than another, with unequal fronts, negative exponents and regrouping ▶ Play a live item from this skill |
Very large and very small quantities are put in scientific notation in multiple-choice items — large numbers from 10^6 up with one- or two-digit fronts (4,500,000 → 4.5 × 10^6), small decimals with negative exponents. A generated how-many-times family (about 1 in 10 Level 6 questions) uses unequal fronts that divide exactly, for large and very small quantities, with front quotients below 1 regrouped, across six contexts; re-examined 03 Oct 2026, its option sets show no shared-feature or size-rank pattern. Held PARTIAL because the child never writes the notation: every 'write in scientific notation' item is choose-one-of-three. Blocking clause: approximate how many times larger one is than the other (served; not guaranteed in every round; rebuilt option sets not yet re-audited) |
| 8.EEI.A.4a | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Multiplication in scientific notation serves at expressions Level 6 ('(2 × 10^5) × (3 × 10^4) = ?'), and division through the how-many-times family ((a × 10^m) ÷ (b × 10^n), fronts divided and exponents subtracted), multiple choice. Held PARTIAL because addition and subtraction in scientific notation never generate, and no operation mixes a decimal numeral with a scientific-notation number. Blocking clause: Perform operations … including addition and subtraction, and problems mixing decimal and scientific notation |
| 8.EEI.A.4b | FULL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Both halves are dealt at expressions L6 in English and Spanish, and every full round there carries at least two items from the small-number, unit, calculator and how-many-times families. Scientific notation is used for very large and very small quantities: writing numbers from 10^6 to 10^11 and from 10^-3 to 10^-6 in scientific notation, reading calculator E-notation in both directions, multiplying in scientific notation, dividing in context (how many times as much), and writing out 1.5 × 10^8 km for the Sun's distance. Choosing a unit of appropriate size is asked directly over seven contexts - "A bacterium is about 2 × 10^-6 meters long. Which unit reports it most sensibly?" (micrometers; likewise a virus, the Sun, a video file and a hair, plus two where the printed unit is already the sensible one) - and again for a hair 0.00007 m wide; six contexts then ask for the measurement re-expressed in the unit (2 × 10^-6 m → 2 micrometers; 5 × 10^14 bytes → 500 terabytes). The unit items are about one L6 question in five and appear in about three rounds in four; a choose-the-unit item appears in a little under half of rounds. Limits: the unit contexts are fixed (seven to choose, six to re-express) and answers are chosen from three options. |
| 8.EEI.B.5a | FULL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Generator: functionsQ L5; ratios L8 (proportionalGraphQ). Note: at functions L5 the child plots a proportional line through (0, 0), (1, k), (2, 2k), (3, 3k) and picks what the point (1, k) tells: the unit rate (for example 'the unit rate: y grows by k when x grows by 1'). The other choices read (1, k) as where the line ends or k as the number of points, and are easy to rule out. At ratios L8 a proportional cost graph is shown, and the child picks what (1, k) means (one ticket costs $k, in one of three wordings such as 'buying 1 ticket costs $k', with no '— the unit rate' tag) against the reversed rate ('$1 buys k tickets') and, since 10 Oct 2026, a starting-cost reading ('the price before buying a ticket is $k') in place of a second copy of the reversed rate, so the key is no longer the only option without '$1'. Both explanations add that this rate is the slope of the line; in these items the word slope is never in a question or a choice. With Writing Coach on, the child also explains in writing the slope of a proportional graph as a unit rate. |
| 8.EEI.B.5b | PARTIAL two-auditor sweep 03 Oct 2026 |
ratiosSaplingQ L6 (better-buy) + functionsQ L2 two rates compared per-ounce; equation-vs-table growth comparison ▶ Play a live item from this skill |
Comparisons exist, but none genuinely compares two proportional relationships. The Ratios L6 better-buy compares two unit prices ('A: 8 oz for $2.80. B: 12 oz for $4.80'), but its keyed answer is always A, so the options give it away. Functions L2 compares y = mx + b (b at least 1, not proportional) with a table through (0, 0), and Functions L5 races a plotted line against y = mx + b (b at least 1). In both, the options print each function's rate, so the child picks between two given rates rather than reading one off a table or graph. The comparing verb is practised only in that form, on linear functions. Two proportional relationships in different representations are never compared. PARTIAL. Blocking clause: Compare two different proportional relationships |
| 8.EEI.B.6a | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
At functions L5 the child plots (0, 0), (2, 2m) and (4, 4m) on a grid (m is 1 or 2) and is asked "Compare rise÷run from (0, 0) to (2, 4) and from (2, 4) to (4, 8). What do the two slope triangles show?", choosing "the SAME slope between ANY two points - the triangles are similar" over two wrong sentences; the same three sentences appear every time, the line always passes through the origin, and the x-intervals are always 0 to 2 and 2 to 4. It is about one functions L5 question in six. Another L5 form has the child read the rise per step on a line y = mx + b they plotted, with an explanation that the rise is the same between any two points. With Writing Coach on (a paid plan or a live school sponsorship, district switch on), functions L2-L4 can deal a written prompt - "On the line y = 3x, explain why the slope is the same between (1, 3) and (2, 6) as it is between (2, 6) and (5, 15)" - graded against a rubric that asks for both rise-over-run values and the constant-ratio reason; it is one fixed prompt, at most one explain item per round. Without Writing Coach the child selects the reason and never writes it, so the row stays PARTIAL: no row whose clause is met only through Writing Coach is marked FULL. Blocking clause: Explain why (met only through Writing Coach) |
| 8.EEI.B.6b | NOT COVERED two-auditor sweep 03 Oct 2026 |
— | 'Derive' asks the child to produce the two equations and show why every point on such a line satisfies them; no item does. The nearest content stops short of derivation: at ratios L8 the child picks y = 2x or y = 3x as the equation of a shown proportional graph (and t = pn from a story); at equations L6 the child picks d = kt from a table or shown graph, or types it from a table ("Write d in terms of t"); at functions L5 the child plots three points of a line and reads its rise per step and where it crosses the y-axis, and the explanations call the rise the slope and the crossing the b of y = mx + b; at functions L4 the child interprets the parts of an already-given equation. No item asks the child to build a y = mx + b equation from a graph or table. |
| 8.EEI.C.7a | PARTIAL two-auditor sweep 03 Oct 2026 |
equationsG8Q (kinds 1-2) ax + b = ax + d classified as no-solution (b≠d) or infinitely-many (b=d), with x = d as the tempting third option; why: 'when the x's cancel, the truth of what is left decides everything' ▶ Play a live item from this skill |
Equations Level 8 shows ax + b = ax + d and the child picks 'no solution' or 'infinitely many solutions', in English and Spanish. 'Exactly one solution' is offered but never right, and the both-sides items ask for x, not the number of solutions, so no item has the child identify a one-solution equation. Creating an equation with a given number of solutions is never asked. PARTIAL. Blocking clause: Create ... linear equations with one solution, infinitely many solutions or no solutions |
| 8.EEI.C.7b | FULL two-auditor sweep 03 Oct 2026 |
equationsG8Q (kind 0) + inequalitiesG7Q ax + b = cx + d with variables both sides (gather-then-divide why, negative solutions); two-step inequalities including the divide-by-negative flip ▶ Play a live item from this skill |
Equations Level 8 serves rational-coefficient equations (0.5x + 3 = 7, 1.5x + 2 = 11, (1/4)(x + 2) = 3) and equations that need distributing and combining like terms (3(x − 2) + 2x = 19, also typed; 0.25(x − 4) + x = 9). Inequalities Level 4 serves −3(x + 2) ≥ 6 (the sign flips), 2(x − 3) + x > 9 and 0.5(x + 4) < 6, with answers in all four directions; no pattern in the choices gives the answer away. Every full round at both levels carries at least two, in English and Spanish. FULL. |
| 8.EEI.C.8a | PARTIAL two-auditor sweep 03 Oct 2026 |
systemsQ count-the-solutions item whose answers are geometric ('the lines cross', 'parallel lines'); solve item's why: 'a system's solution must live on both lines at once' ▶ Play a live item from this skill |
26 Sep 2026 (round 2): the child plots three given points of each line of a system and each line extends across the grid, then names the crossing (equations Level 9); still PARTIAL because the points are dictated, so the child never derives a point from an equation, and only two fixed systems appear in this form. Blocking clause: Graph systems of linear equations |
| 8.EEI.C.8b | PARTIAL two-auditor sweep 03 Oct 2026 |
systemsQ (count-solutions form) solution counts answered WITH their geometric reason attached ('none — parallel lines', 'exactly one — the lines cross') ▶ Play a live item from this skill |
The how-many-solutions item pairs each count with its geometric reason ('none — parallel lines'), and Level 9 graph items have the child plot the lines and pick where they cross as the system's solution; both are choices. 'Explain why' is asked only by a Writing Coach prompt (why (2, 5), where two lines cross, solves the system), which serves only with Writing Coach on, so this row stays PARTIAL. Blocking clause: Explain why solution(s) ... correspond to point(s) of intersection |
| 8.EEI.C.8c | PARTIAL two-auditor sweep 03 Oct 2026 |
systemsQ (count-solutions form) same slope/different intercepts → none; identical → infinitely many; different slopes → exactly one — all three cases generated and discriminated ▶ Play a live item from this skill |
The three-case trichotomy is the item, with slope/intercept comparison as the method in every why — the strongest of the C.8 partials, capped only by the 'explain' verb. Blocking clause: Explain why systems of linear equations can have one solution, no solution or infinitely many solutions |
| 8.EEI.C.8d | FULL adversarial: NOTE-ONLY · re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
systemsQ (solve form) solve y = x + a and y = 2x by equating (answer (a, 2a), coordinate-swap and y=x distractors); check-in-BOTH-equations why ▶ Play a live item from this skill |
Equations Level 9 asks the child to solve systems in two templates (y = x + a and y = 2x; y = 3x and y = x + 2k), about half of Level 9 questions, in English and Spanish, by choosing the solution point from three. The right point is always the only choice that fits the y = x + … equation (in the first template it is the only choice that fits either equation), so checking one equation finds it without solving the system. Graph items, about one Level 9 question in ten, give the points of the lines to plot and ask where the lines cross. No elimination, no standard-form pair, no context. PARTIAL. |
| 8.GM.A.1a | PARTIAL two-auditor sweep 03 Oct 2026 |
transformationsQ (kind 2, congruent branch) reflect-then-translate → 'congruent — same size and shape', why: rigid motions 'move the figure without stretching it' ▶ Play a live item from this skill |
Preservation-under-rigid-motion is assessed as a single congruence gestalt; the four named invariants are never individually verified (no item checks that a specific angle or distance survived), and 'verify' leans production. Blocking clause: Verify that angle measure, betweeness, collinearity and distance are preserved |
| 8.GM.A.1b | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Orientation appears in one item (similarityOrientationQ, coordinate-grid L6): a triangle's vertices run clockwise before a transformation and counterclockwise after, and the child picks 'a reflection — it reverses orientation' over a translation or a rotation; the rationale states that translations and rotations keep the vertex order. It is a single fixed-wording multiple-choice item with no figure and no parameters. It is dealt in about 1 of every 10 questions at coordinate-grid L6 (measured over 1,500 six-question rounds) and never twice in a round. No item has the child carry out a transformation and then check orientation; the child only recognizes the rule. Blocking clause: Investigate whether orientation is preserved |
| 8.GM.A.2a | PARTIAL two-auditor sweep 03 Oct 2026 |
transformationsQ (kinds 1-2) identify the single transformation carrying (x,y) to (−x,y); reflect-then-translate sequence classified as congruence-preserving ▶ Play a live item from this skill |
Coordinate-grid L6 asks which single transformation carries (x, y) to (−x, y) or (x, −y), and classifies a reflect-then-translate image as congruent; both are multiple choice. With Writing Coach on (a paid plan or a live school sponsorship, district switch on), the student is given two congruent triangles by their vertices and writes a sequence of rigid transformations that carries one onto the other, graded against a rubric (one fixed prompt, live rounds only). Without Writing Coach no item asks the student to describe a sequence, so the verdict stays PARTIAL: no row whose clause is met only through Writing Coach is marked FULL. Blocking clause: Describe a possible sequence of rigid transformations between two congruent figures |
| 8.GM.A.3 | PARTIAL two-auditor sweep 03 Oct 2026 |
transformationsQ translate (x,y) by (±dx, ±dy) with sign-error distractors; reflection across the y-axis identified from coordinates; dilation-by-k classified as similar-not-congruent ▶ Play a live item from this skill |
Coordinate-grid L6 serves two coordinate items, both multiple choice: where a point lands after a translation, and which transformation carries (x, y) to (−x, y) or (x, −y) (the answer is always a reflection across one axis). In the translation item the move is right or left and up or down, and the two wrong choices are two of the three other combinations of the moves, so the shared coordinates no longer give the answer away (a child who works out only one of the two coordinates still lands on the right choice about two times in three). No Grade 6-8 item puts a rotation in coordinates, and the dilation item classifies the image as similar without coordinates. No item asks the student to describe what a transformation does to a figure's coordinates; the student picks. Blocking clause: Describe the effect of ... rotations ... using coordinates (no rotation item; translation item answerable from form; describe served by picking) |
| 8.GM.A.4a | PARTIAL two-auditor sweep 03 Oct 2026 |
transformationsQ (kind 2, dilation branch) dilation by factor k → 'similar but NOT congruent — same shape, different size'; why: angles survive, side lengths do not ▶ Play a live item from this skill |
The similarity-via-dilation concept is assessed by multiple choice (a dilated figure is similar but not congruent); outside Writing Coach no item gives two similar figures and asks for the sequence that maps one onto the other. 26 Sep 2026 (round 2): with Writing Coach on - a paid plan or a live school sponsorship, district switch on - the student describes a dilation-plus-translation sequence between two rectangles, graded against a rubric requiring the scale factor and the move; one fixed prompt, live rounds only. Without Writing Coach this row is served by selection only, so the published verdict stays PARTIAL: no row whose clause is met only through Writing Coach is marked FULL. Blocking clause: Describe a possible sequence of transformations between two similar figures |
| 8.GM.A.5a | PARTIAL two-auditor sweep 03 Oct 2026 |
geometryG8Q (kind 1) third angle of a triangle from two given (180−a distractor, 360-quadrilateral-total distractor) ▶ Play a live item from this skill |
The 180° sum is APPLIED cleanly; 'derive' (the transversal/tear-the-corners argument) is production and no item walks the derivation even by recognition. Blocking clause: Derive the sum of the interior angles of a triangle |
| 8.GM.A.5b | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Exterior angles appear in one item form (g8GeomKnowledgeQ, geometry L10): 'Two interior angles of a triangle measure a° and b°. What is the EXTERIOR angle at the third vertex?', with a rationale that derives the supplement and names the exterior angle theorem and the two remote interior angles. It is a compute-and-choose item with no figure, in one direction only (two remote interiors given, exterior asked). The expectation's 'explore' verb is not exercised — the child computes a value and never investigates or justifies the relationship. Blocking clause: Explore the relationship |
| 8.GM.A.5c | NOT COVERED two-auditor sweep 03 Oct 2026 |
— | One geometry L10 item mentions a transversal (about one L10 question in five): 'A transversal crosses two PARALLEL lines. One angle measures 75°. What is its CORRESPONDING angle?' (or its alternate interior angle), in English and Spanish, with the supplement usually among the wrong options. It only ever asks for an angle equal to the given one, so the answer is always the number printed in the question — choosing that number always scores. No Grade 6-8 item has the child construct parallel lines or a transversal (drawing a line parallel to a given one is a geometry L4-5 item, with no transversal), and co-interior (same-side) angles and non-parallel lines are never asked. Blocking clause: Construct and explore the angles created when parallel lines are cut by a transversal |
| 8.GM.A.5d | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
One missing-side form exists at the Grade 6-8 levels (similarityOrientationQ, coordinate-grid L6): two triangles are similar, two sides of the smaller and one matching side of the larger are given, and the child finds the other matching side from the scale factor (whole-number factor 2 or 3). The dilation item in transformationsQ separately classifies a dilated figure as similar but not congruent. The missing-side form is a single proportion-only problem type, multiple choice, stated in words with no figure; it has no indirect-measurement or shadow context, no non-integer scale factor, and no use of angle properties of similar figures. A map-scale item is served at geometry L9 ('On a map, 1 cm represents 5 km. Two towns sit 7 cm apart on the map. Real distance?'), but it is a one-step scale multiplication, not a similar-figures problem. Coordinate-grid L6 also serves the translation, reflection and dilation items credited to 8.GM.A.1a, 8.GM.A.2a, 8.GM.A.3 and 8.GM.A.4a. Blocking clause: Use the properties of similar figures to solve problems (one proportion-only template) |
| 8.GM.B.6 | NOT COVERED two-auditor sweep 03 Oct 2026 |
— | No Pythagorean item asks the child to build, manipulate or demonstrate a model; every Pythagorean item is multiple choice with no figure. Proof-adjacent content is limited to two forms in g8GeomKnowledgeQ (geometry L10): one fixed-wording item asks which comparison finishes the squares proof (the area of the square on the hypotenuse equals the sum of the areas on the legs), and one gives three side lengths and asks, using the converse, whether the triangle is right — half are true triples and half have one side off by one, so the squares have to be compared. Neither shows a model or figure; the rearrangement is described only in the rationale. The remaining Pythagorean items apply the formula (find a hypotenuse or a leg; distance between two points). |
| 8.GM.B.7 | PARTIAL two-auditor sweep 03 Oct 2026 |
geometryG8Q (kind 0) find the hypotenuse from two legs (walk-around-the-corner a+b distractor); find a leg from hypotenuse and leg (hypotenuse-is-longest why); 7 fixed triples ▶ Play a live item from this skill |
Both directions of the theorem are exercised on seven integer triples at geometry L10 — the hypotenuse from two legs, and a leg from the hypotenuse and a leg — and a coordinate-distance item applies it to two points. Every right triangle is described in words with no figure, none sits in a word-problem context (ladder, diagonal), and no item is three-dimensional (box diagonal, slant height). No answer is ever irrational. Blocking clause: in problems in two- and three-dimensional contexts |
| 8.GM.B.8 | FULL re-audited ×1 · re-audited ×2 (CONFIRMED-RECORD-CORRECTED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Pythagorean coordinate-distance IS live: g8GeomKnowledgeQ form 2 (content.ts:12024-12038), ~1/15 at geometry:10, pinned by saplingGrade78Gaps.test.ts:130-142 (3-4-5 instance recomputed by hand in the second audit). The earlier absence claim was 18-draw sampling (P(miss) ≈ 0.29). FULL stands on quoted source. |
| 8.GM.C.9a | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Pyramid surface area generates in one form (g8GeomKnowledgeQ, geometry L10): 'A square pyramid has base side b and slant height l. What is its surface area? (SA = B + ½Pl)', with a rationale that decomposes it into the base plus four triangular faces. The formula is printed in the prompt, so the child substitutes into a given formula; in this form the pyramid is always square, the slant height is always given, and there is no net or figure. Other surface-area items cover cubes, rectangular prisms and cylinders (geometrySaplingQ L8-L9). volumeG8Q has no surface-area item — sphere surface area is named there only in the sphere-volume rationale and, when the radius is 6, offered as a wrong choice; its pyramid content is volume (one third of the prism). Blocking clause: find surface area of pyramids — only by substituting into the formula printed in the question, for one square-pyramid template |
| 8.GM.C.9b | FULL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
volumeG8Q cone V = (1/3)πr²h with the three-cones-fill-the-cylinder why; sphere V = (4/3)πr³ with surface-area foil; (cylinder as the third form) ▶ Play a live item from this skill |
Cones and spheres are found exactly in π, and pyramids in the same generator with no formula given ('A pyramid has a 4 by 4 rectangular base and height 3. What is its volume?'), keyed at one third of the enclosing prism with the full prism always among the wrong choices. English and Spanish; each of the three is roughly one volume L6 question in five or six. Limits: the pyramid's base is always rectangular and its answer whole; the sphere item has only two versions (radius 3 or 6); the fourth kind is the cylinder, a 7.GM claim. Sphere volume is served by this generator; arcsQ (arc length and sector area) has no sphere content. Blocking clause: pyramids |
| 8.DSP.A.1 | FULL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Statistics L8 serves scatter plots on the tap grid in English and Spanish. To construct, the child plots five listed bivariate pairs, scored on the exact set of points, and then reads the association of their own plot; this is about two L8 items in five. To interpret, shown plots ask for positive, negative or no association. Plots with a straight model line ask what its slope means, or whether a candidate line fits. Limit: the clouds are only linear rising, linear falling or patternless. |
| 8.DSP.A.2 | PARTIAL two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Statistics L8 serves scatter plots rendered on a coordinate grid over twelve contexts, a form where the student plots the pairs, and - for the eight contexts that have a trend - two forms with an orange line drawn over the points: a candidate line with the question "Does the orange line fit the cloud's trend?", and a model line with the question "What does its slope tell you?". English and Spanish. Held PARTIAL: the fit of a drawn line is genuinely judged (a line running against the cloud must be rejected) and a model line's slope is interpreted in context, but the student never generates a trend line - no item has the student draw, place or choose a line for the data; the drawn line is always one of two fixed diagonals (rising from (0, 0) to (9, 9) or falling from (0, 9) to (9, 0)); the fit judgment is direction only and never compares two plausible lines or how close the points lie to the line; and no item uses a trend line to predict a value. The fit and slope questions are answered by selecting one of three options. Blocking clause: Generate a trend line for bivariate data |
| 8.DSP.A.3 | PARTIAL two-auditor sweep 03 Oct 2026 |
functionsQ L4 y = rate·x + start rental cost: what does the intercept represent (per-hour-cost and max-hours distractors); m interpreted in the why ▶ Play a live item from this skill |
At functions L4 the child picks what either number of a given linear rule means in context: the cost of each hour or the starting cost in a kayak-rental equation, or the point where a savings equation reaches $0. These rules are deterministic, never fitted to bivariate data. The data-model framing appears only at statistics L8, where a model line is drawn over a scatter plot and the child picks what its slope says in context ('each extra 1 of hours studied brings ABOUT 1 more quiz score'). That line carries no equation, its slope is always 1 or −1, its intercept is never interpreted, and it is never used to solve a problem. Blocking clause: linear model of bivariate measurement data |
| 8.DSP.A.4a | PARTIAL two-auditor sweep 03 Oct 2026 |
twoWayTableQ RENDERED 2×2 table (passage) — grand total (cells-partition why), joint 'AND' cell vs row/column-sum distractors ▶ Play a live item from this skill |
The interpret half is strong and display-backed (the table is actually shown, unlike most display claims); 'construct' — building the table from raw categorical data — is production and never exercised. Blocking clause: Construct ... a two-way table |
| 8.DSP.A.4b | PARTIAL two-auditor sweep 03 Oct 2026 |
twoWayTableQ (kind 2) row-conditional fraction: of bike owners, what fraction plays a sport — denominator pinned to the ROW against grand-total and column distractors ▶ Play a live item from this skill |
Statistics L7 shows a 2 × 2 table (printed as rows of text) and asks two conditional fractions — of the bike owners, the fraction who play a sport (denominator the row), and of the sport players, the fraction who own a bike (denominator the column) — with the other denominators as wrong options (in more than half of deals one of them is replaced by a nearby fraction, such as 3/10 beside 4/10), and asks which is more common within the bike row. Describing an association is asked only in a written explain item ('of 40 students with a pet, 30 walk daily; of 40 without a pet, 10 walk daily. Describe the possible association…'), graded by the Writing Coach check and dealt only where Writing Coach is open to the child (a paid plan or a sponsored class whose district has not switched it off). No automatically graded item compares two groups' relative frequencies to judge association. Blocking clause: describe possible association between the two variables |
| 8.F.A.1a | PARTIAL two-auditor sweep 03 Oct 2026 |
functionsQ L1 the function rule stated in the yes choice of each item ('each input has one output' / 'every input gives exactly one output') ▶ Play a live item from this skill |
Forms: the rule is stated in one choice of each item ('yes — every input has exactly one output') and in every explanation ('The function rule polices the INPUTS: each must yield exactly one output.'). Note: the assigns-exactly-one-output understanding is inferred from the L1 determination item rather than separately assessed, and that item's choices each come in one of three wordings of different lengths, with the yes reading the same whether or not it is the key and no option naming an input, in English and Spanish. So neither this row nor 8.F.A.1b, which rests on the same item, is marked FULL. Blocking clause: Understand that a function assigns to each input exactly one output |
| 8.F.A.1b | PARTIAL two-auditor sweep 03 Oct 2026 |
functionsQ L1 (input, output) pair lists, function and non-function variants; correct answer names the reason (repeated input with two outputs), shared-OUTPUT foil and cannot-determine foil ▶ Play a live item from this skill |
At functions L1 the child decides whether three (input, output) pairs form a function. Both cases generate, and when the pairs are a function one wrong option is 'no — two inputs share an output'. Each option comes in one of three wordings of different lengths, the yes wording is the same whether or not it is the key, and no option names an input, so the answer cannot be found without reading the pairs, in English and Spanish. The question is posed from three-pair lists only, never from a mapping diagram or a graph. Blocking clause: Determine if a relation is a function |
| 8.F.A.1c | PARTIAL two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Every functions L5 question has the child graph a linear function on a coordinate grid: the prompt lists three or four of its points ("Plot these points of a line: (0, 1), (1, 3), (2, 5)", "Function A (graph) - plot its points"), the child taps each one, the grid joins them into a line, and a follow-up question reads the graph (rise per step, where it crosses the y-axis, the unit rate at (1, k), whether two slope triangles show the same slope, or which of two functions grows faster). English and Spanish. Held PARTIAL because every coordinate is printed in the prompt: the child never works out a function's points from its rule or table before graphing it (the systems item at equations L9 also lists the points it asks for), and only straight-line functions are graphed. Blocking clause: Graph a function from its rule or table (points are always listed in the prompt) |
| 8.F.A.2 | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
functionsQ L2 Function A as equation y = m1x + b1 vs Function B as table through (0,0) — which grows FASTER; rate-from-table computed in the why; starts-higher foil separates slope from intercept ▶ Play a live item from this skill |
Two templates compare two functions given in different representations, and growth rate is the only property compared. At L2 an equation is compared with a table whose function is always proportional. At L5 the child plots Function A as a graph and compares it with Function B's equation. Either function can win in both. In both templates the answer choices state each function's rate (at L5 the wrong comparison even contradicts itself), so the child only has to know that the larger rate grows faster, never to read the rate from the table, graph or equation. One side is always an equation, and verbal descriptions never serve as a representation. Blocking clause: each represented in a different way (the child never reads a characteristic out of either representation) |
| 8.F.A.3a | FULL two-auditor sweep 03 Oct 2026 |
functionsQ L4 (+ L2) kayak rental y = mx + b: the b identified as the starting cost at x = 0, with the m ('cost of each hour') as a distractor; L2 reads m as per-step rate ▶ Play a live item from this skill |
At functions L4 the child is given a kayak-rental equation y = mx + b and picks what one of its numbers represents. About half the time that is the rate ('the cost of each hour out on the water'), otherwise the starting value ('the starting cost, before any hours pass'); the explanation names them as the m and the b. In about one kayak item in ten the two numbers are equal (for example y = 10x + 10), so both meanings are right but only one is accepted; this is a defect. At L5 the child finds where a plotted line crosses the y-axis, and the explanation ties it to the b. No item asks whether an equation y = mx + b defines a linear function or what its graph looks like, and the child is never asked to name m as the slope or b as the y-intercept; those words appear only in explanations. There is one context family (rentals). |
| 8.F.A.3b | PARTIAL two-auditor sweep 03 Oct 2026 |
functionsQ L3 table-based linearity test: constant per-step change vs growing jumps (x²+b rows); misses-(0,0) and only-a-graph foils rejected in the why ▶ Play a live item from this skill |
Constant-rate-of-change recognition is exercised from TABLES at L3. The graph the standard names is rendered in a minority of L3 draws (about one in ten): a line graph with the question 'Is the function LINEAR or NONLINEAR?', whose explanation ties the straight line to a constant change per step. That graph item has two fixed graphs only — the line from (0, 0) to (4, 8) and the curve through (0, 0), (1, 1), (2, 4), (3, 9). The table item's explanation says 'a table decides just as well as a graph'; the table item itself shows no graph. Blocking clause: the graph of a linear function |
| 8.F.A.3c | PARTIAL two-auditor sweep 03 Oct 2026 |
functionsQ L3 (nonlinear branch) quadratic-valued table identified as NOT linear because the per-step change grows ▶ Play a live item from this skill |
Recognizing a presented nonlinear function is exercised; 'give examples' is production. One nonlinear shape only (x² + b) — no exponential, no absolute value, and never in equation form. Blocking clause: Give examples of nonlinear functions |
| 8.F.B.4a | PARTIAL two-auditor sweep 03 Oct 2026 |
functionsQ L4 what does the b represent in the rental context — the correct explanation selected against per-hour and max-hours misreadings ▶ Play a live item from this skill |
Forms: what the rate or the starting value represents in a rental equation, with the right meaning picked against a maximum-hours misreading. Note: picking the right contextual meaning, for either the rate or the starting cost, is real recognition-level coverage, but it is capped by the 'explain' production verb per house rule. 26 Sep 2026 (round 2): with Writing Coach on (a paid plan or a live school sponsorship, with the district switch on), the child explains in writing what the 40 and the 60 in y = 40x + 60 mean. The answer is graded against a rubric requiring both the hourly rate (slope) and the fixed charge (intercept); it is one fixed prompt, in live rounds only. Without Writing Coach this row is served by selection only, so the published verdict stays PARTIAL: no row whose clause is met only through Writing Coach is marked FULL. Blocking clause: Explain the parameters of a linear function based on the context |
| 8.F.B.4b | PARTIAL two-auditor sweep 03 Oct 2026 |
functionsQ L2 the rate m determined from a table (Δy/Δx computed in the why) ▶ Play a live item from this skill |
The slope is worked out at functions L5, as the rise per step on the line the child plots, chosen from numbers. It is typed at equations L6, as k in d = kt from a table (about one deal in ten). Both numbers of a linear rule are typed at expressions L3, where the child writes the cost of b tickets from a per-ticket price and a flat fee, for example 5b + 3 (about one deal in four). The y-intercept is found at L5 as the point where the plotted line crosses the y-axis. Limits: the y-intercept is always read off (from a listed point (0, b) or a stated flat fee), never worked out from points that leave out x = 0. The L2 table comparison prints both rates in its choices and is not counted. |
| 8.F.B.4c | FULL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
The x-intercept is computed and interpreted in functionsQ L4 (xInterceptQ): 'Savings: y = b − mx (dollars after x weeks). Where does the graph cross the x-axis?' with the set-y-to-0 rationale, and a second form asking what that crossing means ('after x weeks the money hits $0'). Both forms use one context — a decreasing savings line with whole-number answers — and are multiple choice. The equation generators themselves do not name or ask for an x-intercept. |
| 8.F.B.5 | PARTIAL two-auditor sweep 03 Oct 2026 |
functionsQ L3-L4 qualitative behavior read from tables ('the per-step change keeps growing') and verbal contexts (rental cost structure) ▶ Play a live item: level 3 · level 4 |
Forms: rendered line graphs at functions L3 (story match, the flat interval, linear vs nonlinear), plus tables at L3. Note: at functions L3 about one item in three shows a drawn line graph, and the child picks the story it tells, the interval with no change, or whether it is linear. The relationship is read from a graph, but by choosing a description, never by writing one. No item gives a relationship in words and asks the child to describe it; the kayak item at L4 only asks what one number in a rental equation means. Blocking clause: from a graph |
| Code | Verdict | Evidence (generator · forms) | Notes |
|---|---|---|---|
| 6.RL.1.A | PARTIAL two-auditor sweep 03 Oct 2026 |
comprehensionSaplingQ L6 (p6-secondlap, p6-static) + WRITE_PROMPTS wp-six-analyze evidence-bound inference items (why Jonah smiles — the why quotes 'because the voice that said stop had lost'; what Priya's passing 'without looking over' suggests; what the brother's ceded turn reveals); wp-six-analyze requires defending a theme reading with two specific moments ▶ Play a live item from this skill |
Inference-from-evidence is genuinely assessed — the correct answers are the text-supported readings and every why names the supporting line. But in the selection items the student never CITES anything; the one place citation is produced (wp-six-analyze) is about a book of the student's own choosing, not a presented literary text, so the cite-evidence-for-THIS-text act is never elicited on a text the platform controls. Blocking clause: citing textual evidence |
| 6.RL.1.B | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
comprehensionSaplingQ L6 + vocabularySaplingQ L6 word-in-context ('rehearsed' most nearly means — context confirmed by the contrast that follows); figurative meaning ('like a held breath' makes Saturn feel…); SAPLING_ROOTS affix/root meaning and decode forms (14 parts, e.g. tele-, dict, -less) ▶ Play a live item from this skill |
Two of the standard's three pathways (context and affixes) are real; the or-clause spares the reference-materials pathway. Word-in-context ('rehearsed' most nearly means) and figurative meaning ('like a held breath', money that 'was shirts') are asked inside the literary passages. Connotation is assessed at vocabulary L6 (pairs such as selective/picky and confident/arrogant, in English and Spanish), but only as isolated word pairs. No item asks what a word's connotation adds to a passage, so connotative meaning as used in a text is the open clause. Blocking clause: connotative meanings |
| 6.RL.1.C | NOT COVERED two-auditor sweep 03 Oct 2026 |
— | No Grade-6 literary passage carries a visual element — they are prose text only. (At Grade 6 the one rendered visual is the bar chart on the informational Pony Express passage; illustrations that a question asks about appear in the K-5 reading levels.) Nothing at Grade 6 to grade for literary texts. |
| 6.RL.1.D | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
comprehensionSaplingQ L6 (p6-secondlap, p6-static) + WRITE_PROMPTS wp-six-analyze 'Which sentence best states the passage's theme?' with plot-level and false-generalization distractors, why teaches theme-as-idea-under-plot; ending-meaning item (why the silence is 'exactly enough'); wp-six-analyze elicits naming a theme and defending it with two moments ▶ Play a live item from this skill |
Theme determination is directly assessed ('Which sentence best states the passage's theme?'), and the summarize clause has content: 'Which is the BEST summary of the passage?', with a theme statement as a wrong option. The row is held PARTIAL on two clauses. 'Summarize' is met only by choosing a summary; no prompt asks the student to write one. Evidence of the theme's development is produced only in the Writing Coach's analyze prompt, about a book the student picks, and the published rule does not count that toward FULL. Blocking clause: summarize the text |
| 6.RL.2.A | FULL two-auditor sweep 03 Oct 2026 |
comprehensionSaplingQ L6 (p6-secondlap q5, p6-static q4) how the embedded question 'the race was already decided, wasn't it?' works in the story (stages the voice Jonah must defeat); why the closing silence is 'exactly enough' (the talk-to-quiet reversal is the point) ▶ Play a live item from this skill |
Sentence-level analysis is real here. Items ask what a particular sentence DOES for meaning: the embedded question 'the race was already decided, wasn't it?' (it stages the voice Jonah must defeat), the closing silence that is 'exactly enough', and why 'The Lantern Schedule' ends with electric bulbs 'in every room'. An image item asks what 'like a held breath' makes Saturn feel. The standard names a sentence, chapter, scene, stanza, or image, so one well-practised unit meets it. Chapters and stanzas themselves never appear: the passages are short prose scenes and no poem exists at Grade 6. Five items across four stories ask this, in English and Spanish: the four above, which come up about one reading item in fourteen, and a fifth, added on 31 Oct 2026 with the story 'Enough of Each' and not yet audited as evidence, which asks why the story ends with the line Mateo adds to his own recipe card. With it, the five come up about one reading item in eleven. |
| 6.RL.2.B | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
comprehensionSaplingQ L6 (p6-static q1) 'From whose point of view is the story told, and why does that matter?' — correct answer ties first-person narration to the annoyance-to-wonder turn; brother's-POV and scientist's-POV foils ▶ Play a live item from this skill |
The item goes beyond identify-the-narrator. The credited answer is WHY the chosen point of view powers the story, which is honest development-adjacent recognition. But 'explain how' is a production verb met only by selection, and this is a single fixed item in one passage ('Static'). The other Grade-6 literary texts ('The Second Lap', 'The Lantern Schedule', Leo's first-person journal in 'Two Views of the Storm' and, added on 31 Oct 2026, 'Enough of Each' and Inés's travel journal in 'Two Ways to Hear a Whisper') have no point-of-view item. Blocking clause: Explain how an author develops the point of view |
| 6.RL.2.C | FULL two-auditor sweep 03 Oct 2026 |
comprehensionSaplingQ L6 (p6-static q3) 'The comparison «like a held breath» makes Saturn feel…' — figurative-language-to-meaning, why teaches that a simile transfers a feeling, not just a picture ▶ Play a live item from this skill |
Three literary word-choice items deal at Grade 6, all on figurative language: 'The comparison «like a held breath» makes Saturn feel…', 'What does the detail that money «was shirts» reveal about the family's world?' and, added on 31 Oct 2026 and not yet audited as evidence, 'When Grandpa says the dough «answers,» what does he mean?' in 'Enough of Each'. Each asks what the chosen words add to meaning, in English and Spanish. The two audited items come up about one reading item in twenty-three, and all three about one in seventeen. The standard reads 'figurative language and/or repetition', so figurative analysis meets it. No question asks about repetition or sound devices: a phrase like 'stride for stride' sits in a passage unasked, and The Second Lap's 'his slowest smile of the season' goes unassessed. |
| 6.RL.2.D | PARTIAL two-auditor sweep 03 Oct 2026 |
comprehensionSaplingQ L6 (p6-secondlap q2/q4, p6-static q1/q2) character response/change items: why Jonah smiles after losing (he overcame the quit-voice), what Priya's manner shows, the narrator's annoyance turning to wonder, what the brother's generosity reveals ▶ Play a live item from this skill |
The characters-respond-or-change half is the strongest literary coverage in the bank. 'The Second Lap' and 'Static' are built around a character changing as the plot resolves, and several items assess it: why Jonah smiles after losing, what Priya's manner shows, what the brother's ceded turn reveals, and the narrator's annoyance turning to wonder. The plot-unfolding half is touched only by the 'best summary' items, on those two stories and, added on 31 Oct 2026 and not yet audited as evidence, on 'Enough of Each', where the child picks the retelling that follows the story's episodes in order. No item asks about plot structure itself. And 'describe' asks the child to put both halves into their own words, while every Grade-6 reading item is answered by choosing an option. Blocking clause: how a particular text’s plot unfolds in a series of episodes |
| 6.RL.3.A | NOT COVERED two-auditor sweep 03 Oct 2026 |
— | Comparing reading a text with listening to or watching a performed version needs a performance, and an item that asks for the comparison. Every Grade-6 passage can be read aloud by the Read-to-me button, but that is a synthesized voice reading the printed page, not a performance. No item asks the child to compare the two experiences. |
| 6.RL.3.B | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
A genuine two-text item now serves: 'How do the TONES of the two passages compare?' over 'Two Views of the Storm' (a journal beside a news brief), with further items on purpose, what appears in both, and what only one tells. Held PARTIAL at audit: one paired passage, comparing tone and approach rather than theme across genres, and the same pair is credited to cross-text informational codes. A second pair, 'Two Ways to Hear a Whisper' (a science-magazine article beside a travel journal), was added on 31 Oct 2026 and has not been audited as evidence. The row stays at its audited verdict until a refute-first two-auditor re-check counts the new items. Blocking clause: a second pair of texts — every audited item sits on 'Two Views of the Storm' |
| 6.RL.3.C | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
'The Lantern Schedule' (a 1911 tenement story) has questions that interrogate its era — 'How does the 1911 setting shape the story's central problem?', what money 'was shirts' reveals about the family's world, and why the story ends with electric bulbs 'in every room'. Held PARTIAL: one passage, and 'explain how' is met by selection. |
| 6.RL.3.D | PARTIAL two-auditor sweep 03 Oct 2026 |
comprehensionSaplingQ L6 (p6-secondlap, p6-static) two Grade-6-complexity literary prose passages, self-read by design (passageSelfRead: true — 'reading it yourself IS the task'), 9 items across theme, inference, craft, POV ▶ Play a live item from this skill |
Independent reading of grade-level literary prose is genuinely practised. The passages are meant to be read by the child (an optional Read-to-me voice is available) and carry real Grade-6 syntax and interiority: four stories ('The Second Lap', 'Static', 'The Lantern Schedule' and, added on 31 Oct 2026, 'Enough of Each') and two first-person journals, each paired with an informational text, with items on theme, inference, craft, point of view, summary and historical setting, in English and Spanish. But two of the three genres in the standard's list are absent: no drama and no poem exists in the Grade-6 reading bank, and a handful of short fixed passages is a thin basis for a proficiency claim. Blocking clause: including stories, dramas and poems |
| 6.RI.1.A | PARTIAL two-auditor sweep 03 Oct 2026 |
comprehensionSaplingQ L6 (p6-octopus) evidence-matching ('Which detail supports the claim that the disguise happens fast?' — timing detail vs true-but-irrelevant foils); bounded inference ('What can you infer about why two hearts serve the gills?' — why teaches the smallest-step rule) ▶ Play a live item from this skill |
Both halves — explicit-evidence matching and text-bounded inference — are assessed across several informational passages, with whys that teach the discipline. A writing prompt now elicits evidence from an informational text (state its main idea and back it with two specific details), but the text is one the student chose outside the app. Held PARTIAL: in the reading items the student selects rather than cites. Blocking clause: citing textual evidence |
| 6.RI.1.B | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
comprehensionSaplingQ L6 (p6-octopus q3) + vocabularySaplingQ L6 content-specific word from context ('chromatophores are…' — the definition sits beside the word, and the why teaches noticing where technical texts define their terms); SAPLING_ROOTS affix/root bank (14 parts, meaning + decode forms) ▶ Play a live item from this skill |
Content-specific meaning from context is assessed on the octopus passage ('chromatophores are…'). The affix route is real: a 14-part bank of roots and affixes at vocabulary L6. Figurative meaning is interpreted inside the informational Pony Express passage ('beaten not by faster horses, but by a wire'). Held PARTIAL on connotation: the connotation pairs at vocabulary L6 are bare word pairs, and no informational passage asks what a word's connotation does in the text. Blocking clause: figurative, connotative |
| 6.RI.1.C | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
The Pony Express passage carries a rendered bar chart, and two items draw on it: 'Using the CHART together with the text, about how many days did the Pony Express save…' and 'What does the chart's telegraph bar show that the text states in words?'. Held PARTIAL: it is one chart on one passage, and visuals from different media are never used. |
| 6.RI.1.D | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
comprehensionSaplingQ L6 (p6-octopus q1/q2) central-idea item ('remarkable, unusual abilities in a short life' vs overgeneralization and know-it-all foils — why teaches that a central idea must cover the whole text); development-evidence item (which detail supports the speed claim) ▶ Play a live item from this skill |
Central idea and its development-through-details are both assessed, and a summary is now asked on the informational octopus passage ('Which is the BEST summary of the passage?'). Held PARTIAL because the summary is selected, not written, and no writing prompt asks for a summary of a presented text. Blocking clause: summarize the text |
| 6.RI.2.A | FULL adversarial: DOWNGRADE · re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
comprehensionSaplingQ L6 (p6-octopus q4) 'Why does the author most likely mention that the blood is blue?' — credited answer places the sentence as 'the next exhibit in a case about strangeness', against warning and wrong-function foils ▶ Play a live item from this skill |
Three audited items ask what a particular part does for a text's meaning. 'Why does the author most likely mention that the blood is blue?' (to build the case that the octopus is deeply unusual). 'Why does the author mention factory shifts?' in 'The Case for Quiet Hallways' (to show the bell is a leftover habit). And, for an image, 'What does the chart's telegraph bar show that the text states in words?' on the Pony Express passage. The standard names a sentence, paragraph, section or image. The sentence and image cases are both assessed, in English and Spanish. Two more items, added on 31 Oct 2026 and not yet audited as evidence, ask why 'The Color Underneath' opens by saying the maples 'seem to repaint themselves' and why 'Books on Horseback' mentions the scrapbooks. |
| 6.RI.2.B | PARTIAL two-auditor sweep 03 Oct 2026 |
comprehensionSaplingQ L6 (p6-octopus q4) author's-purpose reading of the blue-blood sentence (building a case that the octopus is deeply unusual) ▶ Play a live item from this skill |
Author's purpose and position are identified on several informational texts: why the octopus author mentions blue blood, why the Quiet Hallways author mentions factory shifts, what that author's claim is, and how a journal and a news brief differ in purpose and tone. Held PARTIAL: the child picks the purpose or position but never explains how the text conveys it. 'Explain' is met only by selection, and no Grade-6 writing prompt asks for it. Blocking clause: Explain how an author's point of view or purpose is conveyed |
| 6.RI.2.C | PARTIAL re-audited ×1 · re-audited ×2 (UPGRADED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
One informational item now assesses figurative language as a contributor to meaning: on the Pony Express passage, 'beaten not by faster horses, but by a wire'. Held PARTIAL at audit: a single fixed item; the octopus passage's own craft ('vanish without moving', the title metaphor) is still unquestioned, and connotation and repetition are never assessed inside an informational text. A new item on 'The Color Underneath', added on 31 Oct 2026 and not yet audited as evidence, asks why the author opens by saying the maples 'seem to repaint themselves' (to name a first impression that the facts then correct). The row stays at its audited verdict until a refute-first two-auditor re-check counts the new items. |
| 6.RI.2.D | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED-RECORD-CORRECTED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
An author's argument is now identified and evaluated on 'The Case for Quiet Hallways': the claim, the evidence that supports it, and a claim that is NOT supported ('Every student would earn better grades'). Held PARTIAL at audit: one passage, and the same bank is credited to the Kansas twin. A second argument, 'Every School Needs a Garden', was added on 31 Oct 2026 with four more items, including one that asks how the author answers people who say gardens cost too much; it has not been audited as evidence. The row stays at its audited verdict until a refute-first two-auditor re-check counts the new items. Blocking clause: a second argument passage — every audited item sits on 'The Case for Quiet Hallways' |
| 6.RI.3.A | NOT COVERED two-auditor sweep 03 Oct 2026 |
— | Every passage, informational ones included, has a 'Read to me' button that reads it aloud by text-to-speech. But no item asks the child to compare reading a text with hearing or watching it, and no performed audio or video version exists whose performance could shape interpretation. |
| 6.RI.3.B | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Comparing two authors' presentations now has a surface: 'Two Views of the Storm' sets a journal (Passage A) beside a town news brief (Passage B) about the same outage, with items on purpose, tone, what appears in both and what only one tells. Held PARTIAL at audit: one passage pair carries every audited item. A second pair, 'Two Ways to Hear a Whisper' (a science-magazine article beside a travel journal), was added on 31 Oct 2026 and has not been audited as evidence. The row stays at its audited verdict until a refute-first two-auditor re-check counts the new items. Blocking clause: a second pair of texts — every audited item sits on 'Two Views of the Storm' |
| 6.RI.3.C | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
The Pony Express passage carries explicit historical framing, and 'Why did the Pony Express exist at all, given its short life? What about its TIME explains it?' asks how that context shaped the subject. Held PARTIAL: the child picks the explanation rather than writing one, and at audit it was a single item on a single passage. A similar item on 'Books on Horseback', added on 31 Oct 2026 and not yet audited as evidence, asks why the program sent librarians on horses and mules. The row stays at its audited verdict until a refute-first two-auditor re-check counts the new items. |
| 6.RI.3.D | PARTIAL two-auditor sweep 03 Oct 2026 |
comprehensionSaplingQ L6 (p6-octopus) one Grade-6-complexity informational passage (science prose with an on-the-spot technical definition), self-read (passageSelfRead: true), 5 items across central idea, evidence, vocabulary, purpose, inference ▶ Play a live item from this skill |
Grade-6 informational reading is self-read on eight texts: the octopus science passage, the Pony Express history passage with its chart, the 'Quiet Hallways' argument, and a news brief paired with a journal, plus four added on 31 Oct 2026 and not yet audited as evidence: a science passage (fall leaf color), an argument (school gardens), a science-magazine article paired with a travel journal (a whispering gallery) and a history passage (Kentucky's pack-horse librarians). Items cover central idea, evidence, word meaning, purpose, inference, summary, the chart and the argument. Held PARTIAL: each text is one short paragraph (each paired set, two) and the same eight repeat, a thin basis for a claim about proficient independent reading across the year. Blocking clause: independently and proficiently |
| 6.W.1.A.a | NOT COVERED two-auditor sweep 03 Oct 2026 |
— | Conducting research across several sources with a standard citation system is the research/inquiry wall. No writing prompt supplies or requires sources; nothing citation-shaped exists anywhere in the Writing Coach. |
| 6.W.1.A.b | NOT COVERED two-auditor sweep 03 Oct 2026 |
— | Gathering from multiple print/digital sources, assessing credibility, avoiding plagiarism, bibliographic information — all research/inquiry-wall verbs with zero substrate. Not stretched. |
| 6.W.2.A.a | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
Writing Coach L4 — WRITE_PROMPTS wp-six-narrative produce a scene with two characters, one moment of tension, ≥2 lines of correctly punctuated dialogue, pacing that slows at the key moment; rubric: real tension, dialogue punctuation, pacing, sensory/concrete detail; min 280 chars, LLM-judged vs rubric |
Genuine narrative writing, with Writing Coach on (a paid plan or a live school sponsorship). Two prompts elicit characters, correctly punctuated dialogue and concrete detail, and both are judged against a rubric. One also grades pacing and an ending that follows from the scene's events; the other grades a character's earned change of mind, carried through time by transition words. Held PARTIAL: the standard's 'including poems' has no prompt, and a clause met only through Writing Coach is not marked FULL. Blocking clause: including poems |
| 6.W.2.A.b | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
Writing Coach L4 — WRITE_PROMPTS wp-six-inform explain something you genuinely know: opening that frames the topic, ≥3 precise steps or facts in logical order, domain-specific words used correctly, closing that says why it matters; min 250 chars, EN/ES rubric, LLM-judged |
With Writing Coach on (a paid plan or a live school sponsorship), informative writing is genuinely produced and judged against a rubric across three prompts. One explains something the child knows well (three or more precise steps or facts, in a logical order). One compares two games (two similarities and two differences, connected by transition words). One states the main idea of something informational the child read, backed by two specific details. Held PARTIAL: the clause is met only through Writing Coach, which is never marked FULL, and no rubric line asks for examples. Blocking clause: The whole clause is met only through the paid, metered Writing Coach; 'examples' is never a graded line |
| 6.W.2.A.c | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
Writing Coach L4 — WRITE_PROMPTS wp-six-argument take a position (later school start): clear arguable claim, two distinct pieces of evidence or reasoning, one sentence answering someone who disagrees; rubric enforces claim/evidence/counterargument/organization; min 250 chars, LLM-judged |
With Writing Coach on (a paid plan or a live school sponsorship), the child writes an argument on one of two topics (a later school start; phones in class). Each is judged against a rubric requiring a clear, arguable claim and two distinct pieces of evidence or reasoning. The school-start prompt also grades a sentence answering an opposing view, and the phones prompt grades cohesive words linking claim to reasons. Held PARTIAL: the clause is met only through Writing Coach, which is never marked FULL, and because the rubric accepts reasoning in place of evidence, relevant evidence is never required. Blocking clause: The whole clause is met only through the paid, metered Writing Coach |
| 6.W.3.A.a | PARTIAL adversarial: NOTE-ONLY · re-audited ×1 · re-audited ×2 (CONFIRMED-RECORD-CORRECTED) · two-auditor sweep 03 Oct 2026 |
Writing Coach L4 — WRITE_PROMPTS wp-six-inform/argument/analyze rubrics + WRITE_FEEDBACK rubrics require an opening that frames the topic, one organized paragraph, and a closing that follows (inform: 'a closing that says why it matters'); grow-feedback fb-on-topic polices focus |
With Writing Coach on (a paid plan or a live school sponsorship), introducing a topic and closing with a conclusion that follows are graded rubric lines on the child's own writing, and focus is coached ('keep every sentence about the same topic'). When a first draft does not pass, the child reads the coaching and gets one chance to revise before it is scored. Held PARTIAL: a draft that passes the first time is never reviewed or revised, and a clause met only through Writing Coach is not marked FULL. Blocking clause: Review, revise, and edit writing |
| 6.W.3.A.b | PARTIAL two-auditor sweep 03 Oct 2026 |
Writing Coach L4 rubrics + WRITE_FEEDBACK (fb-complete, fb-describing-words) precise language demanded (inform: 'domain-specific words used correctly'; narrative: 'sensory or concrete detail, not labels'); complete sentences required in every L4 rubric |
With Writing Coach on (a paid plan or a live school sponsorship), precise language and a consistent style are graded on the child's own writing: domain words used correctly, concrete detail rather than labels, a formal style in one argument prompt (no slang, no chatty asides) and in one explanation prompt, and a steady point of view in the other argument prompt. Held PARTIAL: sentence structure is never graded, complete sentences are coached by feedback but are not a rubric line at Grade 6, and a clause met only through Writing Coach is not marked FULL. Blocking clause: establish and maintain an appropriate and consistent style |
| 6.W.3.A.c | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
Writing Coach L4 rubrics + WRITE_FEEDBACK (fb-capital, fb-endmark, fb-spelling-close, fb-complete) correctly punctuated dialogue is a hard rubric line (wp-six-narrative); judge can flag capitals, end marks, spelling, and sentence completeness on any submission |
With Writing Coach on (a paid plan or a live school sponsorship), two conventions are graded on the child's own writing: correctly punctuated dialogue, and a point of view whose pronouns do not drift. Capitals, end marks, spelling and pronoun case get coaching feedback but do not decide the grade, and the grader is told not to fail a strong answer for minor spelling or punctuation slips. Grade-6 grammar also assesses nonrestrictive punctuation, pronoun case, intensive pronouns and a vague pronoun, by recognition. Held PARTIAL. Blocking clause: conventions of standard English grammar and usage |
| 6.W.3.A.d | PARTIAL re-audited ×1 · re-audited ×2 (CONFIRMED) · two-auditor sweep 03 Oct 2026 |
— No generator is named on this row yet; the note says what is practised. |
Grade-6 vocabulary asks which transition fits the logic between two sentences (contrast, cause and effect, example, sequence, addition, alternative), in English and Spanish. With Writing Coach on (a paid plan or a live school sponsorship), transitions are graded rubric lines on the child's own writing in three genres: connecting compare-and-contrast points, moving a narrative through time, and linking a claim to its reasons. Held PARTIAL: using transitions in one's own writing is met only through Writing Coach, which is never marked FULL. |
| 6.W.3.A.e | NOT COVERED two-auditor sweep 03 Oct 2026 |
— | Tempted and rejected: typing into the Writing Coach is literally 'technology to produce writing', but crediting the delivery medium would credit every digital product for this clause. The standard's substance — publishing writing and interacting/collaborating with others — is the collaboration wall, and neither exists (fb-private actively steers away from sharing personal content; nothing is published or shared). |
| 6.SL.1.A | NOT COVERED two-auditor sweep 03 Oct 2026 |
— | Collegial discussion rules, group goals, and individual roles are the collaborative-discussion wall. A solo adaptive app has no discussion to conduct; not stretched. |
| 6.SL.1.B | NOT COVERED two-auditor sweep 03 Oct 2026 |
— | Delineating a live speaker's argument and posing or responding to questions in discussion is speaking-and-listening interaction. At Grade 6 the app reads prompts aloud, but no Grade-6 item presents a speaker's argument to listen to or asks the child to speak. The app's few speaking items (reading a word aloud, counting aloud, Reading Buddy) are early-grade practice. |
| 6.SL.1.C | NOT COVERED two-auditor sweep 03 Oct 2026 |
— | Reviewing a speaker's key ideas across diverse media and paraphrasing multiple perspectives — speaking/listening plus multimedia walls, no substrate. |
| 6.SL.2.A | NOT COVERED two-auditor sweep 03 Oct 2026 |
— | Speaking clearly and audibly when presenting is oral production for an audience. The app does listen in a few early-grade places (reading a word aloud, counting aloud, Reading Buddy's oral reading), but no Grade-6 item asks the child to speak or present, and nothing judges clarity or volume. |
| 6.SL.2.B | NOT COVERED two-auditor sweep 03 Oct 2026 |
— | Body position, eye contact, and gesture are physical presentation behaviors — nothing in a screen-based item can assess them. |
| 6.SL.2.C | NOT COVERED two-auditor sweep 03 Oct 2026 |
— | Planning and delivering multimedia presentations is the presentation + multimedia wall; no presentation task of any kind exists. |
These are the rulings the adversarial passes changed, kept in the open: a report that never shows its corrections is asking to be trusted on faith.
| Code | Change | Reason |
|---|---|---|
| 6.EEI.A.2c | PARTIAL → PARTIAL | Citation gap, verdict unchanged: the draft supports 2c ('Evaluate non-negative rational number expressions') only with the whole-number b^e branch, but volumeSaplingQ L4-L5 evaluates V = l×w×h and V = Bh at x.5 fractional values — genuine rational-number expression evaluation that the KS twin (6.EE.2c) does cite cross-skill. Add volumeSaplingQ L4-L5 to the generator citation; still PARTIAL (the only non-whole values are exactly-.5 heights, and no fraction/decimal ever enters expressionsQ itself). |
| 6.EEI.C.9a | PARTIAL → PARTIAL | Verdict survives but its cited basis is wrong and must be replaced. The draft cites equationsSaplingQ L6 dependent-variable items as 'the prerequisite' — but 9a's verbatim statement is solely 'Write an equation to express one quantity, the dependent variable, in terms of the other quantity': dep/ind identification exercises no clause of it (identification lives in the parentStatement and in 2e/KS 6.EE.8a). Prerequisite content is not standard content — on the cited basis alone this falls to NOT COVERED, matching KS 6.EE.8b's original grade for identical content. What actually sustains PARTIAL is expressionsQ L3 branch 1 (per·b + k selected as 'the cost of b tickets'): the dependent quantity expressed in terms of the independent variable, by selection, minus the equation form. Replace the generator citation with expressionsQ L3 branch 1 (+ equationsSaplingQ L6 as context only); blockingClause 'Write an equation' stands. |
| 6.EEI.C.9b | PARTIAL → NOT COVERED | Cross-state contradiction the draft leaves standing: KS 6.EE.8c ('Analyze the relationship … using graphs and tables, and relate these to the equation') is NOT COVERED; MO 6.EEI.C.9b ('Analyze the relationship … using graphs, tables and equations and relate these representations to each other') is PARTIAL — same generator, near-identical clause. One is wrong, and it is the MO grade. 9b's statement binds the analysis entirely to representations: no graph, table, or equation of any two-variable relationship exists anywhere in content.ts (verified — equationsSaplingQ L6 contexts stay in words; the draft's own note admits this). The verbal 'changes IN RESPONSE to' reasoning belongs to the parentStatement/2e, not to 9b's text; a PARTIAL with no content behind any clause of its own statement must fall. The mapper's note already calls it 'one step above NOT COVERED' — it is not even that. |
| 6.GM.A.3c | FULL → FULL | Held on the 'or' reading of 'same first coordinate OR the same second coordinate', but verified narrowness: coordinateSaplingQ L5 fixes x and varies y ALWAYS (const x = nz(); y1 = 1+rnd(7); y2 = −(1+rnd(7))) — the horizontal (same-y) case never generates, and every pair crosses the axis (no same-side pair, so subtraction-of-like-signs is never exercised). One added same-second-coordinate form would make this FULL unassailable; the KS twin clause inside 6.NS.8 stays PARTIAL for independent reasons (graphing clause), so no cross-state contradiction. |
| 6.DSP.B.4a | NOT COVERED → PARTIAL | L5: dot plots are both DISPLAYED (construct register) and INTERPRETED (total/mode reading); histogram interpretation via interval bars. BLOCKING: "histograms and box plots" display halves. SELF-ASSESSED after the adversarial pass — re-adversarial required at go-live. |
| 6.DSP.B.5a | NOT COVERED → PARTIAL | L5 reading items report n from a display. BLOCKING: free production of the report. SELF-ASSESSED after the adversarial pass — re-adversarial required at go-live. |
| 6.EEI.B.8a | PARTIAL → FULL | Two separate adversarial auditors run by this project (a verb-fidelity auditor and a refutation auditor, round 1, wf_001cb4de-a84) both marked this FULL after the write/type/draw builds: 26 Sep 2026 (round 1, two auditors agreed FULL): inequalitiesQ L1 deals a typed inequality from a worded constraint in all four forms (h > c, p ≤ c, s ≥ c, d < c), EN+ES, ~half of live L1 items; the checker refuses a strict/inclusive swap and a wrong variable. Paper/placement/duel keep the MC form (covers > and < only). |
| 6.EEI.B.8b | PARTIAL → FULL | Two separate adversarial auditors run by this project (a verb-fidelity auditor and a refutation auditor, round 1, wf_001cb4de-a84) both marked this FULL after the write/type/draw builds: 26 Sep 2026 (round 1, two auditors agreed FULL): inequalitiesQ L3 carries a ray spec for x >, ≥, <, ≤ c; RayPick scores the exact boundary + inclusive + direction (36 constructions, one correct); ~30–40% of live L3 items EN+ES, none in plain mode. |
| 6.DSP.B.4b | PARTIAL → FULL | Two separate adversarial auditors run by this project (a verb-fidelity auditor and a refutation auditor, round 1, wf_001cb4de-a84) both marked this FULL after the write/type/draw builds: 26 Sep 2026 (round 1, two auditors agreed FULL): PieBuild (v2.217.0): the child colours a 12-wedge circle to the given counts and is scored on the real wedge counts, ✓ gated until every wedge is coloured — 'create' is now performed; circleGraphQ covers 'interpret'. |
| 6.EEI.A.3 | PARTIAL → FULL | Two separate adversarial auditors run by this project (a verb-fidelity auditor and a refutation auditor, round 2, wf_44c82963-87a) both marked this FULL after the write/type/draw builds: 26 Sep 2026 (round 2): the child generates equivalent expressions by typing a(x + b) expanded and ax + bx combined, with the given, bracketed and uncombined forms refused; multiple-choice items continue to cover identifying them; factoring is not asked. |
| 6.EEI.C.9a | PARTIAL → FULL | Two separate adversarial auditors run by this project (a verb-fidelity auditor and a refutation auditor, round 2, wf_44c82963-87a) both marked this FULL after the write/type/draw builds: 26 Sep 2026 (round 2): the child types d = kt from a table of distances and hours, with d required alone on one side so t = d/k is refused; one context and one equation shape, in about one live Level 6 item in ten. |
| 6.EEI.A.2d | PARTIAL → FULL | Two separate adversarial auditors run by this project (a verb-fidelity auditor and a refutation auditor, round 2, wf_44c82963-87a) both marked this FULL after the write/type/draw builds: 26 Sep 2026 (round 2): the child types algebraic expressions from a phrase and from a ticket-cost situation, judged by equivalence, in about half of live Level 3 items, and types the value of ax + b on the number pad in about a quarter of live Level 2 items; the write bank has two templates. |
| 6.NS.C.5 | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: Use positive and negative numbers to represent quantities — the signed number is only picked, never written |
| 6.NS.C.7 | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: the absolute value of a rational number is its distance from 0 on the number line — integers only, the distance idea is in feedback, and the comparison item gives its answer away |
| 6.DSP.B.5d | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: Analyze the choice of measures of variability based on the shape of the data distribution and/or the context |
| 6.EEI.A.2e | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: Understand the meaning of the variable in the context of the situation |
| 6.EEI.B.7 | PARTIAL → FULL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached FULL. |
| 6.EEI.C.9b | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: using graphs ... and relate these representations to each other |
| 6.NS.B.4b | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: express a sum of two whole numbers with a common factor as a multiple of a sum of two whole numbers |
| 6.RP.A.1 | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: represent these comparisons |
| 6.NS.C.8 | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: generate equivalent representations |
| 6.GM.A.3b | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: related by reflections across ... both axes (never asked); the published L4 reflection item gives its key away by form |
| 6.GM.A.2b | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: Apply ... V = Bh (the only V = Bh prism item gives its answer away by the position of its choices) |
| 6.GM.A.3a | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: consistency with the agreed 6.NS.6b / 6.GM.A.3b downgrades: the only generator offers quadrant IV only when it is the key |
| 6.EEI.C.9b | NOTE-ONLY | Note refresh, 10 Oct 2026 (after commit 9c765be): the shown-graph item's slope slip is dealt on either side of k instead of always (k + 2)t; the note names the shipped slips and keeps the remaining shares-form-and-number tell. Verdict unchanged. |
| 6.EEI.C.9b | NOTE-ONLY | Note refresh, 10 Oct 2026 (r6/twovar): the table and shown-graph multiple choice now deal four equations (both forms × both numbers) around a proportional OR additive relationship, so the shares-form-and-number tell is gone — the pairing rule fell from 75% to 25% (= chance) over 600+ deals per language, locked at the 95% bound in saplingGrade6Gaps.test.ts; generator and forms filled in. The note now says what the item still does not do (graph ↔ table; values read off the graph). Verdict unchanged — held at PARTIAL for the second re-audit. |
| 6.EEI.C.9a | NOTE-ONLY | Note refresh, 10 Oct 2026 (r6/twovar): the typed-equation table is proportional (d = kt) or additive (d = t + k) in equal shares since this round; note, generator and forms name both shapes. Verdict unchanged. |
| 6.EEI.C.9b | NOTE-ONLY | Note refresh, 10 Oct 2026 (r6/twovar, after the separate review of round 6): the forms text listed one option set (d = 2t, d = 3t, d = t + 2, d = t + 3) while the item also dealt d = t, d = 4t and other equations that were never the key — d = t as a wrong in about a quarter of graph deals, so striking it was free ('strike any option never seen as the key' scored 40% on the graph, 29% on the table, chance 25%), and only d = 3t was plotted with three points. The visible numbers are now an adjacent pair drawn inside the key range ({1, 2} or {2, 3} on the graph, with d = t a real key; 2 to 8 on the table), every graph plots t = 1, 2, 3, and the never-key rule scores 25%; saplingGrade6Gaps.test.ts bounds every equation and number dealt as also the key. Forms and note describe HEAD. Verdict unchanged (PARTIAL). |
| 6.EEI.B.8b | NOTE-ONLY | Note refresh, 11 Oct 2026 (after the separate review of widget/r7 at 949b7a3): the forms text described the multiple-choice solution-set item for x > c only (open circle, shaded right, with a filled-circle and a wrong-direction distractor); since r7 it asks x > c, x < c, x ≥ c or x ≤ c, the key sits at each circle-and-direction corner in turn, and two of the other three pictures are dealt, each named. Verdict unchanged (FULL). |
| 6.DSP.A.3 | NOTE-ONLY | Note refresh, 24 Oct 2026 (keyfirst-a, measured over 450–600 real deals per language in English and Spanish; locked in app/src/lib/keyfirstGiveaway.test.ts): the median item no longer prints the median in the middle position (shuffled, never-sorted order) and its wrong choices are values from the list, the middle of the list as printed among them; the first listed value and the .5 midpoint of the extremes are no longer offered. Same-day repair: the list's mean is offered as a wrong choice (each list has one far value, and its mean is one of the five values; about five deals in eight since round 3), because with neighbours alone 'the choice nearest the mean' found the median in every choice deal. Round 3 (re-verification, same day): that list let a spacing rule ('one choice far from the other two → the opposite end, else the middle') win 60–66%, so the list's gaps and the wrong pair are now dealt together; no rule on the choices' order or spacing tops about 42% and nearest the mean is about 37%. The note no longer says sorting is required: working out the mean exactly and taking the closest other choice still finds the median, and the note says so. Verdict unchanged. |
| 6.GM.A.2b | NOTE-ONLY | Note refresh, 24 Oct 2026 (keyfirst-a, measured over 450–600 real deals per language in English and Spanish; locked in app/src/lib/keyfirstGiveaway.test.ts): the V = Bh item's single-rule giveaways are closed (see KS 6.G.2, including the same-day repair of the doubled-slip and closest-pair tells), but the multiple-choice item is still answerable from its choices: how many end in .5, read with which two sit closer, finds the key in about 81% of deals. The blocking clause therefore stays, re-worded to name that pattern, and the PARTIAL verdict is unchanged. |
| 6.GM.A.2b | NOTE-ONLY | Note refresh, 24 Oct 2026 (keyfirst-a, founder decision of 24 Oct 2026 recorded in GAPS.md 24OCT2026-b; measured over 3,000 real deals per language and mode in English and Spanish; locked in app/src/lib/keyfirstGiveaway.test.ts): V = Bh is typed on the number pad in every interactive deal, so in practice the formula must be applied and no choices are shown. Printed sheets, the placement check and Duel keep three choices, still answerable from them in about 81% of deals. The blocking clause now says the V = Bh clause is resolved for interactive rounds and names the paper residual; following the MO 6.GM.A.3a precedent the verdict stays PARTIAL until the row is re-audited. |
| 6.DSP.A.3 | NOTE-ONLY | Note refresh, 24 Oct 2026 (keyfirst-a, founder decision of 24 Oct 2026 recorded in GAPS.md 24OCT2026-b; measured over 3,000 real deals per language and mode in English and Spanish; locked in app/src/lib/keyfirstGiveaway.test.ts): the median is typed in every interactive deal; the paper/placement/Duel choices and every open rule there (mean neighbour, list-extreme rule about 75% / about 80% with the range, smallest-choice rule about 58%, second-closest to the midrange about 63%) are named in the note. Verdict unchanged. |
| 6.DSP.B.5c | NOTE-ONLY | Note refresh, 24 Oct 2026 (keyfirst-a, founder decision of 24 Oct 2026 recorded in GAPS.md 24OCT2026-b; measured over 3,000 real deals per language and mode in English and Spanish; locked in app/src/lib/keyfirstGiveaway.test.ts): the median is typed in every interactive deal and the mean in about half; the note no longer says both are typed in about half. Verdict unchanged (PARTIAL). |
| 7.RP.A.1 | PARTIAL → FULL | Two separate adversarial auditors run by this project (a verb-fidelity auditor and a refutation auditor, round 1, wf_001cb4de-a84) both marked this FULL after the write/type/draw builds: 26 Sep 2026 (round 1, two auditors agreed FULL): Same generator; sapling78PartialToFull.test.ts re-derives (a/b)/(c/d) from each prompt and requires non-unit divisors > 50%, like units > 18%, area > 18%; 600/600 probe rounds carried ≥2 typed items. |
| 7.GM.B.5 | PARTIAL → FULL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached FULL. |
| 7.GM.B.6a | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: other polygons composed of triangles and rectangles |
| 8.GM.A.5c | PARTIAL → NOT COVERED | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached NOT COVERED. Blocking clause: Construct and explore the angles created when parallel lines are cut by a transversal |
| 7.DSP.B.3 | PARTIAL → FULL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached FULL. |
| 7.DSP.B.4 | PARTIAL → FULL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached FULL. |
| 7.EEI.B.3b | PARTIAL → NOT COVERED | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached NOT COVERED. Blocking clause: Assess the reasonableness of answers using mental computation and estimation strategies |
| 8.EEI.C.7b | PARTIAL → FULL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached FULL. |
| 7.EEI.A.1 | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: factor linear algebraic expressions — the one factor item's options give the key away |
| 8.EEI.A.2a | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: where p is a positive rational number — p is never a fraction or decimal |
| 8.EEI.A.2b | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: cube roots of perfect cubes less than or equal to 1000 — only 8, 27, 64 and 125 are dealt |
| 8.EEI.A.4b | PARTIAL → FULL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached FULL. |
| 8.DSP.A.2 | NOT COVERED → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: Generate a trend line for bivariate data |
| 8.F.A.1c | NOT COVERED → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: Graph a function from its rule or table (points are always listed in the prompt) |
| 7.RP.A.2a | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: Determine when two quantities are in a proportional relationship |
| 7.EEI.B.3a | PARTIAL → FULL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached FULL. |
| 8.F.A.1b | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: Determine if a relation is a function |
| 8.F.A.2 | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: each represented in a different way (the child never reads a characteristic out of either representation) |
| 7.DSP.C.6c | FULL → PARTIAL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached PARTIAL. Blocking clause: Compare theoretical and experimental probabilities |
| 7.NS.A.2d | PARTIAL → FULL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached FULL. |
| 8.NS.A.1c | PARTIAL → FULL | Two-auditor audit of every grades 6-8 math row, 03OCT2026: two separate auditors run by this project re-read the row against the live code and both reached FULL. |
| 7.EEI.A.2 | NOTE-ONLY | Note refresh, 10 Oct 2026 (after commit ae4ee1a): the reveal item's three choices now all open 'raising by N% …' and name the percent and the factor; the note no longer says the key alone repeats the prompt's percent. Verdict unchanged. |
| 7.EEI.B.3b | NOTE-ONLY | Note refresh, 10 Oct 2026 (after commit ae4ee1a): the reasonableness item's off estimates are now products from the same range as the fair ones (1.7–12× off either way, every estimate a multiple of 10), so the note no longer says the estimate's size alone decides the item; the printed product remains the reason for the verdict. Verdict unchanged. |
| 8.EEI.B.5a | NOTE-ONLY | Note refresh, 10 Oct 2026 (after commit 9c765be): the ratios L8 point-(1, k) item's second wrong option is a starting-cost reading rather than a second copy of the reversed rate, and the key carries no '— the unit rate' tag; the note quotes the shipped options. Verdict unchanged. |
| 7.RP.A.2d | NOTE-ONLY | Note refresh, 10 Oct 2026 (after commit 282dd18, merged at fc8983c): the origin item's grid now plots (0, 0) with the other points and draws the line y = rx from the origin, so the picture shows the line through (0, 0) the prompt names; the note no longer says the origin is not plotted and no line is drawn. Verdict unchanged. |
| 7.RP.A.2c | NOTE-ONLY | Note refresh, 10 Oct 2026 (after commit 282dd18, merged at fc8983c): the ratios L8 ticket-cost graph now plots (0, 0) with the points from x = 1 and draws the line y = kx from the origin; the note no longer describes only pre-plotted points from x = 1. Verdict unchanged. |
| 7.EEI.A.1 | NOTE-ONLY | Note refresh, 11 Oct 2026 (after the separate review of widget/r7 at 949b7a3; measured on r7/batch1-guard with everyAskGuard's sign-aware word overlap and middle length rank): the note said the add key always shares one feature with each wrong answer; since r7 the add item deals one of three pairs of named slips and its key is the option sharing the most with the other two in 72–77% of deals (chance 33%); the factor and combine keys are that option in every deal, and the expand −(x − b) key is the only option with a +. Verdict unchanged (PARTIAL). |
| 7.DSP.B.3 | NOTE-ONLY | Note refresh, 11 Oct 2026 (after the separate review of widget/r7 at 949b7a3; measured on r7/batch1-guard with everyAskGuard's sign-aware word overlap and middle length rank): the note listed equal medians with different ranges; since r7 the two medians are equal in half the deals and far apart in the other half. Verdict unchanged (FULL). |
| 7.DSP.B.4 | NOTE-ONLY | Note refresh, 11 Oct 2026 (after the separate review of widget/r7 at 949b7a3; measured on r7/batch1-guard with everyAskGuard's sign-aware word overlap and middle length rank): the note said the range form holds the medians equal; since r7 the medians are equal in half the deals and far apart in the other half, and the range and mode forms cross two claims and deal two of the three wrong combinations, each named; the overlap-centre key rate is 39–44% (range) and 31–34% (mode), chance 33%. Verdict unchanged (FULL). |
| 8.EEI.A.1 | NOTE-ONLY | Note refresh, 11 Oct 2026 (after the separate review of widget/r7 at 949b7a3; measured on r7/batch1-guard with everyAskGuard's sign-aware word overlap and middle length rank): the note and blocking clause said the product-of-powers options always give the key away; since r7 the wrong answers are one of three pairs of named slips and the shared-feature rule finds the key in 45–48% of deals (chance 33%, bar 56%). Still only product of powers, only multiple choice. Verdict unchanged (PARTIAL). |
| 8.EEI.A.2b | NOTE-ONLY | Note refresh, 11 Oct 2026 (after the separate review of widget/r7 at 949b7a3; measured on r7/batch1-guard with everyAskGuard's sign-aware word overlap and middle length rank): the adversarial note (also data/grade7_8_adversarial_findings.json) still said the direct evaluate question is never asked, the perfect squares stop at 36 and cube roots have no content; the 15 Sep 2026 hardening deals √9 to √625 and ∛8 to ∛125 at expressions L6, and the irrationals L1 key now reads 'because its value is exactly N'. The note says so; the cube roots still stop at 125. Verdict unchanged (PARTIAL). |
| 7.EEI.A.1 | NOTE-ONLY | Note refresh, 11 Oct 2026 (widget/r7 after merging r7/batch1-content, r7/batch1-guard and main; measured with everyAskGuard's sign-aware word overlap and middle length rank, 900 deals per language): the add item is now a full 2×2 and the option sharing the most with the others is its key in 25% of deals (chance 25%); the note no longer says that rule finds the add key in three deals in four. Verdict unchanged (PARTIAL). |
| 8.EEI.A.2b | NOTE-ONLY | Note refresh, 11 Oct 2026 (widget/r7 after merging r7/batch1-content, r7/batch1-guard and main; measured with everyAskGuard's sign-aware word overlap and middle length rank, 900 deals per language): the irrationals L1 why-is-it-rational key now reads 'because its value is precisely N' (one length and one word count with its wrong answers); the adversarial note (also data/grade7_8_adversarial_findings.json) quotes the live wording. Verdict unchanged (PARTIAL). |
| 7.EEI.B.4c | NOTE-ONLY | Note refresh, 11 Oct 2026 (widget/r7, ineqSolve2x2): the inequality multiple choice now deals a 2x2 — direction right or turned wrongly, crossed with the boundary's sign right or lost — with a negative coefficient in about 3 deals in 5; the why names each step and whether dividing flips the sign. The earlier 'tested by substitution in the why' wording no longer described the item. Verdict unchanged. |
| 7.GM.B.6b | NOTE-ONLY | Note refresh, 24 Oct 2026 (keyfirst-a, measured over 450–600 real deals per language in English and Spanish; locked in app/src/lib/keyfirstGiveaway.test.ts): the volume L5 V = Bh item's wrong choices are now a pair drawn from seven named slips (never one choice a whole-number multiple of another, the key the smallest, middle or largest choice a third of the time each; the first repair's doubled slip, 2Bh, was replaced the same day), so its old single-rule giveaways are gone; but the multiple-choice item is still answerable from its choices (how many end in .5, read with which two sit closer, finds the key in about 81% of deals), so the limit says so and prism volume rests mainly on volume L4. Verdict unchanged. |
| 7.GM.B.6b | NOTE-ONLY | Note refresh, 24 Oct 2026 (keyfirst-a, founder decision of 24 Oct 2026 recorded in GAPS.md 24OCT2026-b; measured over 3,000 real deals per language and mode; locked in app/src/lib/keyfirstGiveaway.test.ts): the volume L5 V = Bh item is typed on the number pad in every interactive deal, with no choices on screen; printed sheets, the placement check and Duel keep three choices, still answerable from them in about 81% of deals. The limit now says so, and no longer says prism volume rests mainly on the L4 item. Verdict unchanged (PARTIAL). |
| 7.EEI.A.2 | NOTE-ONLY | Note refresh, 31 Oct 2026 (answer-giveaway round 8 review, revised in its third pass): the reveal item writes every decimal to hundredths (1.10a, not 1.1a), and its two wrong choices are drawn from multiplying by 0.10 or 2.00, adding 1.10 or 2.00 dollars, and dividing by 1.10; 'nearly doubling the price' is gone. The note names them and the surface leans left (picking the choice that names the printed 1.10: about half of English deals, a little over two in five Spanish ones). Verdict unchanged. |
| 8.GM.A.3 | NOTE-ONLY | Note refresh, 31 Oct 2026 (answer-giveaway round 8 review): the translation item's move varies (right or left, up or down) and its wrong choices are two of the three other combinations of the moves, so the note no longer says the shared coordinates give the answer away; the one-coordinate shortcut that remains is named. Verdict unchanged. |
| Code | Attack | Why the claim held |
|---|---|---|
| 6.NS.B.3 | Cross-state contradiction attempt: same decimalsSaplingQ generator is PARTIAL for KS 6.NS.3 but FULL for MO — multiplication is only 0.x × 0.y (single tenths digits, verified at content.ts 9849-9857) and division only tenths ÷ 0.d. | The standards differ in text, not just numbering: KS demands 'multi-digit decimals' (mult/div operands are not multi-digit → PARTIAL correct); MO's expectation says only 'fluency with addition, subtraction, multiplication and division of DECIMALS' — no size term. All four operations generate on decimals; the verbatim MO statement is satisfied. The split grade is the correct resolution, not an inconsistency. |
| 6.NS.A.1a | Cross-state pressure: KS 6.NS.1 is PARTIAL on the same generator; and every COMPUTED division has a unit-fraction divisor (verified: L7 form 0 is 1/b ÷ 1/(bk), form 1 is a/b ÷ 1/d; general a/b ÷ c/d with c>1 appears only as structure recognition, never computed). | The grade split tracks the texts: KS demands 'Interpret and COMPUTE quotients of fractions' generally plus 'multiple exposures connecting various concrete and abstract models' — both blocked, hence PARTIAL. MO 1a's whole demand is 'Solve problems involving division of fractions by fractions', a coarse solve verb: L7 forms 0-1 are solved fraction-÷-fraction problems (a unit fraction is a fraction), including a word problem. Weakest surviving MO FULL — a c>1 computed divisor would end the argument — but the verbatim statement is satisfied. |
| 6.DSP.A.2 | 'Overall shape' with no display: MO's text lacks KS 6.SP.2's '(… and/or outlier)' escape hatch inside the shape list, data always arrives as a 5-6 value raw list, and shape words (peak, symmetry, skew) appear nowhere in any stem. | Narrowly. It is an understand-standard, and the L4 items make the child read shape features off the raw list — identifying the value that 'sits far from the pack' requires seeing the cluster, and the why states the distribution-described-by-center/spread/stray-values frame verbatim. Cluster + outlier are informal shape descriptors, which is all a no-display engine can honestly reach. Flagged as the thinnest surviving FULL in the MO set; the first histogram widget shipped should revisit it. |
| 6.RP.A.1 | 'Represent these comparisons' as a production verb — the same wall that blocked 'write' codes everywhere else. | House-consistent: selection of the correct representation against structurally meaningful wrong ones (a:total part-to-whole, b:a order-flip) is how this crosswalk treats 'represent'/'use … to represent' throughout (KS 6.NS.5a FULL on identical grounds), as distinct from 'write'/'describe'/'explain' which stay capped. The distractors force the actual part-to-part comparison, and the equivalent-ratio form covers ratio-as-comparison from a second angle. |
| 6.NS.C.8 | The equivalence table (SAPLING_EQUIV) contains no 1/3 or 2/3 — the benchmark fractions the standard names — and pct→fr direction uses only terminating decimals. | The parenthetical is a ceiling, not a floor: '(LIMITED TO terminating decimals and/or benchmark fractions of 1/3 and 2/3)' restricts what may appear; the and/or means terminating-only content is inside the limit. All three representation pairs generate (pct→dec, fr→pct, dec→fr) with the one-place-shift and p/10-denominator misconceptions targeted. FULL stands. |
Some verbs in the standards cannot be honestly assessed by an app, and we do not pretend otherwise: conducting real surveys, experiments, or simulations (we assess the design-judgment half); freehand drawing and construction; and speaking & listening as spoken practice (taking part in a discussion, presenting, listening to a speaker). Codes built on those verbs appear above as NOT COVERED or with the limitation named in their notes.