The 5 AP Stats FRQ Mistakes Costing You Points
AP Stats FRQs are scored on a four-level rubric (4/3/2/1) that rewards complete statistical communication — not just the right answer. Students with strong formulas often land at 2 or 3 because they skip conditions, forget context, or phrase conclusions wrong. Each mistake below is a specific rubric element graders mark against you.
days until your AP Stats exam
Thu, May 7 · Afternoon session
The 5-second diagnostic
Which of these sounds like your last AP Stats FRQ?
Pick the one that feels most true. We'll show you what it looks like in your response, which rubric level you get capped at, and the fix.
You list inference conditions without actually checking them against the data
What it costs you: Every inference FRQ requires naming the conditions (Random, 10% Rule, Large Counts/Normality) AND verifying them with the specific data. Students who list conditions without checking get capped at a 3 on a 4-level rubric. Students who skip them entirely cap at a 2.
What it sounds like
"I will use a two-sample t-test. The conditions are: random, independence, and normality. These are all met."
Named, not checked.
Scoring-ready rewrite
"Conditions for a two-sample t-test: (1) Random — both samples were randomly selected per the problem statement ✓. (2) 10% Rule — n₁ = 30 and n₂ = 25 are each less than 10% of the population sizes ✓. (3) Normality — both sample sizes are <30, but the boxplots provided show no strong skew or outliers, so we can proceed ✓."
Named + verified with specific data.
How to spot this in your own writing
For each condition you list, ask: did I point to the specific data that verifies it? If your conditions paragraph could be copy-pasted into any FRQ, you didn't actually check them.
You choose the wrong inference procedure (e.g., z-test when you need t, or one-sample when you need two)
What it costs you: Picking the wrong procedure is the fastest way to collapse an entire FRQ from 4 to 1. The rubric caps the score at the lowest level possible when the procedure doesn't match the question — even if the math is perfect.
What it sounds like
"To compare the average scores of two independent classes where the population standard deviations are unknown: I will use a two-sample z-test."
Wrong procedure (σ unknown → t, not z).
Scoring-ready rewrite
"To compare the means of two independent samples where population standard deviations are unknown and we use the sample standard deviations, I will use a two-sample t-test for the difference in means."
Correctly identifies t (not z) for unknown σ.
How to spot this in your own writing
Before computing anything, write the procedure name and check: (1) one sample or two? (2) mean or proportion? (3) is σ known or unknown? (4) paired or independent? Answering those four questions correctly eliminates the vast majority of procedure errors.
Your conclusion states 'reject H₀' without explaining what that means for the original research question
What it costs you: Every inference FRQ requires a conclusion (1) that addresses the original research context and (2) that links the statistical decision to a real-world claim. 'Reject the null' alone caps you at a 3 on a 4-level rubric.
What it sounds like
"Since p-value = 0.023 < 0.05, we reject the null hypothesis. There is significant evidence to reject H₀."
Statistical decision. No context.
Scoring-ready rewrite
"Since p-value = 0.023 < α = 0.05, we reject H₀. We have convincing statistical evidence that the mean recovery time for patients on the new drug is less than the mean recovery time for patients on the current treatment."
Decision + context-specific claim.
How to spot this in your own writing
After your statistical conclusion, does the next sentence mention the actual subject of the study (drug, class, candidate, method)? If not, you're not in context and you lose the top rubric level.
You describe the p-value as 'the probability that H₀ is true' or a similar misinterpretation
What it costs you: Misinterpreting the p-value — saying it's the probability the null is true, or the probability of the alternative — is one of the most heavily penalized statistical errors. It also signals to graders that the student doesn't understand inference logic, which can cascade into lower scores on related parts.
What it sounds like
"The p-value is 0.03, which means there is a 3% chance the null hypothesis is true."
Wrong. Describes the null, not the data.
Scoring-ready rewrite
"The p-value of 0.03 means that if the null hypothesis were true, we would expect to see a sample mean at least as extreme as our observed sample mean in about 3% of repeated samples."
Assumes H₀ true, describes data extremity.
How to spot this in your own writing
When you explain p-value, does your sentence start with 'assuming H₀ is true' or 'if the null were true'? That conditional framing is mandatory. 'The probability of H₀' is always wrong.
You leave FRQ 6 (Investigative Task) blank or answer only the first parts
What it costs you: The Investigative Task is designed to apply statistical reasoning to a novel situation — no single procedure fits. Students who recognize it as just 'harder inference' and give up on parts (c), (d), or (e) lose a full 4-5 points from what is often the difference between a 4 and a 5.
What it sounds like
"Part (c): I don't know how to do this. Part (d): [blank]. Part (e): [blank]."
Partial credit impossible if blank.
Scoring-ready rewrite
"Part (c): Even though this isn't a standard procedure, we can estimate the standard error by simulating samples under the null, or by using the formula for the given sampling distribution. SE ≈ √(p(1-p)/n) applied to the observed proportion gives SE ≈ 0.089. Part (d): [follow the reasoning from (c) to construct a reasonable interval or test] …"
Partial credit reasoning even when unsure.
How to spot this in your own writing
If you see blanks in parts (c), (d), or (e) of FRQ 6 in your practice work — especially when you got earlier parts right — you're leaving points on the table. Write SOMETHING statistical for each part, even if it's incomplete. Partial credit on the Investigative Task is huge.
Behind the scenes
What an AP reader actually does with your Stats FRQ
AP Stats readers score FRQs on a 4-level holistic rubric (4=Complete, 3=Substantial, 2=Developing, 1=Minimal) in roughly 90 seconds per FRQ. Missing context, missing conditions, or missing conclusions each knock you down a full level.
Student's inference FRQ conclusion
The p-value is 0.018, which is less than 0.05, so we reject the null hypothesis. Conditions: random ✓, independence ✓, normality ✓. There is evidence to reject H₀.
What the reader notices first
Conditions listed without being checked against the specific data in the prompt. Capped at level 3.
Conclusion is 'reject H₀' with no context about the actual research question. Capped at level 3.
GradGPT scores Stats FRQs at the rubric level. Trained on thousands of rubric-scored AP Stats responses. See which level you're earning on every part — before the real reader does.
Frequently Asked Questions
Yes. Most level-3 FRQs are capped by one of two things: conditions not checked, or conclusion not in context. Fixing those two habits alone typically moves 3s to 4s. Combined with solid MCQs, that reliably crosses the 5-threshold.
Yes. Listing conditions without checking them caps the score at a 3. Checking them against specific data in the prompt — and showing the verification — is required for level 4.
'Reject H₀' is the statistical decision. Conclusion in context adds a sentence linking that decision back to the original research question, using the actual subject matter (drug, candidate, method, etc.). Both are required for full credit.
Use z when the population standard deviation (σ) is known. Use t when σ is unknown and you're using the sample standard deviation (s). For proportions, always use z (the sampling distribution uses p̂).
A lot. It's worth 1/8 of your FRQ score and most students leave parts blank. Even partial credit on parts (c), (d), (e) can add 2-3 points that swing the final AP score from a 4 to a 5.
GradGPT uses the official College Board AP Stats 4-level rubrics. Paste your FRQ and get level-by-level scores — flagging unchecked conditions, missing context, and p-value misinterpretations. Under a minute.
Will you get a 5?
Upload one FRQ. See which rubric level you'd be capped at — and which fixes would take you to a 4.
