The tests were built against the NGSS grade ladder. Our Grade 3 course moved two things later, for good reasons — and the tests didn't move with them:
| What the test asks at Grade 3 | Questions per form | Where we teach it |
|---|---|---|
| Engineering design — state a problem with success criteria and limits (3-5-ETS1-1), compare possible solutions (3-5-ETS1-2), test designs and improve them (3-5-ETS1-3) | ≈ 4 | Grade 4 (success criteria, limits, design a solution, test it) and Grade 5 (compare designs, decide what to improve) |
| Variation helps survival — some individuals of a kind have traits that give them an advantage (3-LS4-2) | 2 | Grade 5 — a deliberate ruling: it's the seed of natural selection, and it leans on the full heredity chain (traits in G3, inherited-vs-learned in G4, offspring patterns in G5). Florida also places it at Grade 5; NGSS places it at 3. |
Everything else on the Grade 3 form is taught by our Grade 3 course — the weather tables, the climates, the fossils, the life cycles, the balanced-forces truck trials, all of it.
Why move the questions rather than the teaching: the Grade 5 test is already cumulative (it re-asks Grade 3 and 4 material), so late-taught content still gets tested — nothing is lost. Moving the teaching instead would unpick two architecture decisions for the sake of six questions.
If you'd rather the courses matched the tests, here is what the swap looks like. Nothing below is built — it's the proposal, with the lesson-hours accounted for.
| Move | What moves | Where it lands |
|---|---|---|
| Design comes down to Grade 3 | Grade 4's design topic ("What a design must do, and what it must stay inside" — success criteria, limits, design a solution; 3 lessons), the test-it lesson from the pinwheel build, and Grade 5's compare-and-improve pair (which design did better; what to improve next). Six lessons in all. | One new Grade 3 topic, "Design it, test it, make it better" (5–6 lessons), placed at the end of the Earth strand — right after Dangerous Weather, which already has students judging a flood barrier. Grade 4 keeps its pinwheel build as the first real build, now a re-entry; Grade 5's design topic folds its one remaining lesson into the circuits topic. |
| Variation-helps-survival comes down to Grade 3 | Three lessons from Grade 5's "Who pulls through when a place changes" (same place, different structures; predict who pulls through; say why). | A new small Grade 3 topic straight after "Life cycles and family likeness", which already teaches that offspring differ — the moved lessons add "…and the difference can decide who survives". Kept to observable structures (the grade-3 framing); the deeper inherited-vs- acquired reasoning stays where it is in Grade 4. |
| Rebalance the hours | Grade 3 gains ~9 lessons (183 → ~192) while 4 and 5 shed them. To pull Grade 3 back: move its "Using and conserving natural resources" topic (4 lessons) up to Grade 4 — the tests only ask about resources at Grade 4 anyway. | End state ≈ Grade 3: 188 · Grade 4: 159 · Grade 5: 142 (today: 183 · 158 · 147). Grade 3 stays the biggest year, as it already is. |
Cost and catches: about a dozen atoms re-registered, two new Grade 3 topics with fresh PP100 banks, and ~9 lessons re-authored at Grade 3 reading level (Grade 4/5 scripts can't be reused verbatim — the language rules differ). Two or three days of course-chat work plus your reviews. Two things to re-wire: the moved test-a-design lesson loses its fair-test foundation (fair tests are Grade 4), so the Grade 3 version teaches "same test for both" informally; and Florida and Arizona both place variation-helps-survival at grade 5, which the standing cross-grade citation convention handles (their grade-5 codes sit on the grade-3 atoms — content covered, timing differs). Note this reverses your earlier placement ruling on variation, which followed Florida; these tests are genuine new evidence for that re-ruling.
No clashes. Every standard on the Grade 4 forms is taught by the end of our Grade 4, and the Grade 5 form — which deliberately re-samples Grades 3 and 4 — matches our courses fully. The Grade 5 test's cumulative design is exactly the shape we already planned for the end-of-grade-5 exit test, so these forms can double as practice for it.
Content first: every standard on all forty-five middle-school forms is taught in our 6–8 courses. That sentence is only true as of two weeks ago — each form carries about four engineering-design questions, and engineering design was absent from our middle-school band until the extension you approved on 29 September. Those atoms now carry roughly 180 questions' worth of test demand.
The structural problem: the test-makers dealt the middle-school standards out across three year-group tests — Grade 6 gets cells, ecosystems, space, weather, particles and forces; Grade 7 gets plate tectonics, body systems, heredity, reactions and fields; Grade 8 re-tests everything and adds evolution, genes, human impact and a chemistry extension. Each grade's test samples all five of our subject courses at once. A self-paced student who starts with, say, all of MS Physics would meet a Grade 6 test where most questions come from courses they haven't opened.
Two workable shapes:
| Option | What changes | Fits self-pacing? |
|---|---|---|
| A — subject tests (recommended) | Re-sort the Grade 6–7 question bank into five end-of-course tests (MS Physics, MS Chemistry, MS Life Science, MS Earth & Space, MS Scientific Reasoning). Grade 8's cumulative test stays as the final anchor — it already covers every standard, so it works whenever the courses finish. | Yes — a test fires when its course ends, whichever year that is. |
| B — keep year tests, fix the pacing | Keep the Grade 6/7/8 forms, and publish a recommended cross-course schedule so that by the end of each year a student has met that year's tested standards. We can derive this mapping from the courses' own topic orders. | Only if students roughly follow the schedule — it reintroduces the year-group clock self-pacing removed. |
Across all 3,420 questions, this is the complete list:
| What the questions ask | How many | Where it comes from | Status |
|---|---|---|---|
| Balance a chemical equation — choose the missing coefficient ("Aluminum Oxide Coefficient", "Engine Smog Equation Balance"). We taught students to check a given balanced equation, but never to balance one. | ≈ 1 per Grade 8 form | Standards-linked. Texas grade 8 (8.6E: "relate conservation of mass to the rearrangement of atoms using chemical equations") and some of the bank's own balancing items are tagged to the NGSS conservation standard (MS-PS1-5). | Atom added (15 Oct): choose the whole-number coefficient; coefficient multiplies, subscript never changes. Lands in MS Chemistry's scope. |
| An atom's mass from its particle counts — adding protons and neutrons to get the mass number. | ≈ 1 per Grade 8 form | Not in any middle-school standard we track. The bank tags it "EXT: subatomic structure" — a legacy-course extension. "Masses of protons and neutrons" first appears in high-school Chemistry (TEKS CHEM-6B); the 2021 middle-school TEKS only ask students to use the periodic table. | Your call: one tiny atom (it's a one-lesson fact that feeds high school), or re-aim those questions at periodic-table reading, which IS the grade 8 standard. |
| Transparent, translucent, opaque ("Femi's Lantern Covers") — sorting materials by how much light passes through. | 2 questions in all 15 Grade 4 forms combined | No standard anywhere — not Texas, NGSS, Florida or Arizona at any grade we track. The bank's own tag says where it came from: "extension … taught [in the original course's] Unit 3 light lesson". It exists because the old course taught it. | Recommend retiring the two questions. |
The tests are built on the NGSS ladder, and our courses deliberately teach more than NGSS — Texas (the TEKS) demands a lot of physical science at these ages that NGSS leaves out. Because the forms sample by NGSS standard, that extra content is structurally invisible to them:
| Grade | Share of our atoms these forms can't sample | The biggest invisible clusters — all STAAR staples |
|---|---|---|
| Grade 3 | 43% | How pushes and pulls change motion; heating, cooling and changes of state; rapid changes to Earth's surface; measurable properties of matter; how light behaves |
| Grade 4 | 42% | Electric circuits; mixtures; friction; magnets; energy resources; the water cycle; weathering and erosion as slow change |
| Grade 5 | 31% | Circuits that do something; solutions and separating mixtures; day and night from Earth's spin; energy transformations; balanced and unbalanced forces at work |
| Grades 6–8 | 24% | Distance–time graphs; the periodic table; physical properties; mixtures and solutions; classification levels; stars and the solar system — the heart of the Texas Grade 8 paper |
So a student who aces every form could still sit the Texas Grade 8 paper never having been tested on a distance–time graph, a circuit, or the periodic table (beyond the one or two chemistry-extension questions). That is the gap behind your instinct that the Texas tests feel stronger.
What fixes it:
| Step | What it means |
|---|---|
| 1 · Add a state-strength block to each form | Roughly a quarter of each form's questions, drawn from the invisible clusters above, weighted to what Texas actually samples. The bank grows; the NGSS questions stay. |
| 2 · Anchor where real papers exist | Texas publishes released papers for Grades 5 and 8 only. Those two forms get calibrated item-for-item against the released papers — difficulty, format and topic mix — per the standing ruling for the Grade 5 exit test. Grades 3, 4, 6 and 7 inherit the style, using released Grade 5/8 questions filtered to what's taught by that grade. |
| 3 · Keep the best question wins | A test samples the atoms; it never covers them all. Blueprint by topic coverage, then fill each slot with the strongest available question — whichever framework it came from. |
What the existing bank gets right: fifteen parallel forms per grade (retakes never reuse a form); rich scenario clusters — the beaver dam, the hive, the toy-truck trials read as well-built, age-right stimuli; eight question formats; complete NGSS coverage; answers keyed; already running live as QTI. Writing that from scratch again buys nothing.
What it can't do as-is: see the Texas-only quarter-to-half of each course (section 5); match our Grade 3 order (section 1); serve as five subject tests at middle school (section 3). All three are blueprint problems, not question-quality problems — solved by adding questions and re-sorting forms, not by replacing them.
One caution before full trust: these questions were authored outside our question rules (the claim-key form, option discipline, distractor standards). Before the bank becomes the mastery instrument, run our independent question-quality pass on a sample per grade — if a strand's sample fails, re-author that strand's questions only.
About 195 questions carry broken labels instead of standards. Mostly on Grade 8 forms: fragments like "(protons", "balancing", "EXT-MSChem-U1L2" where a standard should be. The questions themselves are fine — the labels need a cleanup pass. This matters beyond tidiness: the proposed queue-based serving of mastery tests needs exactly one clean knowledge-component tag per question, and these labels won't support it.
The tests use eight question formats — multiple choice, typed answers, drag-to-match, highlight-the-text, dropdown-in-sentence, ordering, matching, and custom interactions. Our mastery tests are multiple choice plus typed answers. Students will meet the other formats cold unless the mixed practice sets before each test deliberately use them. Worth one line in the question-writing guidance.
Forces and motion (balanced and unbalanced pushes 3-PS2-1, predicting repeating motion 3-PS2-2, magnets and static without touching 3-PS2-3, solving a problem with magnets 3-PS2-4); life cycles 3-LS1-1; animal groups 3-LS2-1; inherited traits and their variation 3-LS3-1, environment shaping traits 3-LS3-2; fossils 3-LS4-1; variation helping survival 3-LS4-2; organisms suiting habitats 3-LS4-3; surviving environmental change 3-LS4-4; weather records 3-ESS2-1; world climates 3-ESS2-2; weather-hazard defences 3-ESS3-1; and the three engineering-design standards 3-5-ETS1-1/2/3. All taught in our Grade 3 except the engineering trio (our Grade 4–5) and variation-helps-survival (our Grade 5).
Energy of moving objects, energy transfer and collisions (4-PS3-1…4); waves, light and information (4-PS4-1…3); animal and plant structures and senses (4-LS1-1/2); weathering, erosion and rock layers (4-ESS1-1, 4-ESS2-1/2); energy resources and hazards (4-ESS3-1/2); engineering design. All taught by the end of our Grade 4.
Grade 5 material (particles too small to see 5-PS1-1…4, gravity pulling down 5-PS2-1, energy from the Sun to food 5-PS3-1, plant growth from air and water 5-LS1-1, matter cycling in ecosystems 5-LS2-1, Earth's spheres and water 5-ESS2-1/2, star brightness and sky patterns 5-ESS1-1/2, protecting Earth 5-ESS3-1) plus a cumulative sample of every Grade 3 and 4 standard. All taught by the end of our Grade 5.
Between them the three grades cover all 59 middle-school standards; Grade 8 re-tests everything and adds a small chemistry extension (periodic table organisation, atomic structure, counting and balancing atoms in equations). Our 6–8 courses teach all 59 — the engineering-design standards via the sub-domain added 29 September — with the three sliver-gaps listed above.