Why a Fixed SAT Weakness Comes Back Two Tests Later
Why a Fixed SAT Weakness Comes Back Two Tests Later: Blocked Drills vs. Mixed Practice
This guide is part of the complete Digital SAT Prep Guide.
The pattern goes like this. A practice test shows you missing three Standard English Conventions questions. You spend a week on punctuation drills, get to 19 out of 20, and move on. Two practice tests later, the same three misses are back, and you conclude that you "forgot" the rules.
You did not forget them. If someone showed you the rule right now, you would recognize it instantly. What went missing is something the drill never trained: the ability to look at a question that does not announce its type and decide, on your own, which rule it is asking about. That decision is a separate skill, and blocked drilling is designed in a way that skips it entirely.
What a blocked drill actually teaches
A blocked drill is twenty questions of the same type in a row — twenty semicolon questions, twenty Transitions, twenty systems of equations. It is how nearly every prep resource organizes practice, and it is the fastest way to learn a rule you do not know.
It is also, structurally, missing a step. On the real test a question arrives with no label. Answering it correctly requires three things in order: recognize what is being tested, retrieve the rule, apply it. A blocked drill does the first step for you before you start, so you practice steps two and three twenty times and step one zero times.
| Practice type | What you practice | Feels like | Holds after two weeks? |
|---|---|---|---|
| Blocked (20 of one type) | Retrieve the rule, apply it | Fast, satisfying, high accuracy by question 10 | Rarely — the recognition step was never trained |
| Mixed (types shuffled) | Recognize the type, then retrieve and apply | Slower, more misses, frustrating | Usually — every question rehearsed the full sequence |
| Full practice test | All three steps, plus pacing | Realistic | Yes, but too expensive to use as a drill |
This is not a quirk of the SAT. The research on interleaved practice — Rohrer and Taylor's 2007 study of shuffled math problems is the standard reference — shows the same shape every time: blocked practice produces better performance during the session and worse performance on a delayed test, and learners consistently rate the blocked version as more effective because it feels better. The feeling is the trap.
Why Standard English Conventions is the worst case
Every domain suffers from this, but Standard English Conventions is where it bites hardest, for a reason that sits in College Board's own test specification.
Within each Reading and Writing module, questions are grouped by domain — Craft and Structure, then Information and Ideas, then Standard English Conventions, then Expression of Ideas — and within Craft and Structure, Information and Ideas, and Expression of Ideas, questions testing similar skills are grouped together and ordered easiest to hardest. Standard English Conventions is the exception. Its questions are ordered easiest to hardest regardless of which specific convention is being tested.
In other words, the test itself interleaves this domain. A Boundaries question about a semicolon can sit directly after a Form, Structure, and Sense question about subject-verb agreement, which sits after a question about a colon. The only way to answer them is to recognize each one cold. A student who learned punctuation in blocked drills has practiced exactly the wrong thing for exactly the domain where it matters most.
The practical consequence: "my grammar is bad" is almost never one problem. It is usually three or four adjacent rules — comma versus semicolon between independent clauses, the colon rule, punctuation around an interrupting phrase, and one agreement rule — that look like a single weakness because they all produce misses in the same group. Each one is learnable in an afternoon. Telling them apart under time pressure is the skill, and the Digital SAT grammar rules guide lists the full set to tell apart.
The inverse problem: strong on full tests, weak on single questions
A smaller group of students has the mirror-image complaint. Their full-length practice scores are fine, but when they work through single questions from a bank, their accuracy drops.
Same mechanism, opposite direction. A full test hands you context that a single question does not. Position in the module tells you the domain — if you are on question 18 of a Reading and Writing module, you know you are past Craft and Structure and probably into Standard English Conventions. Position within the group tells you the difficulty. The two questions before a hard Inferences question were also Inferences questions, and they primed the approach. A question pulled from a bank arrives with none of that.
If this is you, your recognition step is leaning on module position rather than on the question itself. It will fail on any question that does not land where you expect, and on the real test the hardest question in a group is exactly the one that does not look like its neighbors. The fix is the same as for the regression problem: mixed practice with no positional cues.
The two-stage protocol that makes a fix hold
Neither practice type is wrong. They do different jobs, and the regression happens when students stop after the first one.
Stage 1 — Blocked, to learn the rule. When a diagnostic shows a specific gap, do a blocked set of 15–20 questions on that exact skill. Read the rule before you start. Review every miss until you can state, in one sentence, why the right answer is right. Stop when you hit 90 percent. This usually takes one session.
Stage 2 — Mixed, to keep it. From the next day, that skill only ever appears inside shuffled sets. A ten-question mixed set drawn from all four Reading and Writing domains, or all four Math domains, with the newly fixed skill making up two or three of the ten. No labels. The only way to get the semicolon question right is to notice that it is a semicolon question.
The ten-day retest. Ten days after Stage 1, do a twenty-question mixed set and check accuracy on the fixed skill specifically. If it held, it is real. If it dropped back, you have a recognition problem, not a rule problem, and the fix is more Stage 2, not a return to Stage 1.
The same structure applies in Math. A student who fixes exponential functions in a blocked Advanced Math set will meet the next one in the middle of a Module 2 with a Geometry and Trigonometry question before it and a Problem Solving and Data Analysis question after. Whether they notice it is an exponential problem — rather than, say, a percent-change problem in disguise — is the whole question. Our error maps guide covers how to log misses by cause so you can tell the two failures apart.
What to change in your practice this week
- Stop measuring progress by blocked-drill accuracy. Nineteen out of twenty on a semicolon set is the beginning of the fix, not the end.
- Add a daily mixed set of ten.* All domains, shuffled, no labels. This is the single highest-return change for a student whose fixes keep unfixing.
- Log the reason for each miss in a mixed set as either "did not recognize" or "recognized, applied wrong."* After two weeks, the ratio tells you which stage you are missing.
- Retest on a different Bluebook form, not the same one.* A retaken practice test hides regression, because you remember the questions. Bluebook has eight official forms, Tests 4 through 11; use them as measurements, not drills.
Take the diagnostic
The hard part of this problem is that a full practice test cannot tell you whether a miss was a rule failure or a recognition failure. Both look like a wrong answer on a Boundaries question. You need to see the same skill under both conditions — inside a blocked set and inside a mixed one — to know which fix you need.
The MySatCoach diagnostic measures accuracy at the skill level in mixed conditions, which is the condition the real test uses, so a "fixed" skill that only works when it is labeled shows up as a gap rather than a strength. For a student whose weaknesses keep returning, that is the number that has been missing.
Continue Your Digital SAT Prep
- The Complete Digital SAT Prep Guide
- Digital SAT Boundaries Questions
- Three Weeks Out at 1430: A Domain-Targeted Plan
Related Guides
Frequently Asked Questions
Why does my SAT grammar weakness keep coming back after I fix it?
Because the drill taught you the rule and the test asks you to recognize which rule applies. When you do twenty semicolon questions in a row, you never have to decide that a question is a semicolon question — the drill decided for you. On the real test, Standard English Conventions questions are ordered by difficulty regardless of which convention they test, so the recognition step is the whole problem, and it is the step you never practiced.
What is the difference between blocked and mixed SAT practice?
Blocked practice groups questions by type — twenty Transitions questions, then twenty Boundaries questions. Mixed or interleaved practice shuffles them so each question requires you to identify its type before solving it. Blocked practice produces faster gains during the session and worse retention two weeks later; mixed practice feels harder and holds. Use blocked practice to learn a rule and mixed practice to keep it.
Why am I better on full practice tests than on single questions?
Full-length tests hand you the context that single questions do not. Position in the module tells you the domain and roughly the difficulty, and the questions around a hard one prime the approach. A single question from a bank arrives with none of that. If you are strong on full tests and weak on isolated drills, your recognition step is leaning on module position, and it will fail on any real question that does not sit where you expect it.
When should I retest a skill I fixed?
Ten days after you first fix it, with a mixed set rather than a blocked one. A fix that survives a ten-day gap and a shuffled set is real. A fix that only shows up in blocked drills the same week is the rule sitting in short-term memory, and it will not be there on test day.