AP Statistics Calculator Guide (2027): Every Procedure on Desmos and TI-84
The May 2027 AP Statistics exam runs in Bluebook with the Desmos graphing calculator built in, and you can still bring a TI-84. This guide puts both side by side for every calculation in the revised course: summary statistics, the normal, binomial and t distributions, every test and interval, chi-square, and regression. Then it covers what you actually have to write on a free-response question.
Which calculators you can use in May 2027
- Desmos in Bluebook. The exam is fully digital, free response included, and Bluebook has the Desmos graphing calculator built in. Only the Bluebook version is allowed on exam day, not desmos.com or the Desmos app.
- Up to two handheld graphing calculators from College Board's approved list, such as the TI-84 Plus CE, TI-Nspire CX II, or Casio fx-9750GIII.
- New for 2027: no non-graphing calculators. A scientific calculator with stats functions was fine in past years. It is not allowed for AP Statistics from 2027.
Not sure whether to buy a handheld at all? See the best calculator for AP Statistics, and whether Desmos is enough.
Practice in the right version of Desmos
College Board's version of Desmos is configured differently from the regular website. Practice in the College Board practice calculator so nothing surprises you. Everything below is typed straight into an expression line: type normaldist or ttest and Desmos formats it for you. Procedures like ttest open two collapsible panels under the expression: Confidence Interval (with a confidence-level box) and Hypothesis Test (with a null-hypothesis box and Left / Both / Right tail buttons). Open a panel to see the statistic, degrees of freedom and p-value.
The procedure reference
Want it on one page? Print the TI-84 and Desmos cheat sheet. In the examples, L is a data list you type as L=[12,15,9,20,14,18,11,16]. Menu paths for the TI-84 Plus CE are written by name. DISTR is 2nd then VARS.
1. One-variable summary statistics
| You need | Desmos | TI-84 |
|---|---|---|
| Everything at once | stats(L) | STAT ▸ CALC ▸ 1-Var Stats |
| Mean, median | mean(L), median(L) | x̄ and Med in 1-Var Stats |
| Sample SD (the one AP uses) | stdev(L) | Sx |
| Population SD | stdevp(L) | σx |
| Quartiles | quartile(L,1), quartile(L,3) | Q1, Q3 |
quartile, not quantile. For the list above, quartile(L,1) gives 11.5, which matches the TI-84 and the AP method. quantile(L,0.25) interpolates and gives 11.75. Use the wrong one and your IQR and outlier fences will be off.Check your work: mean, median & mode calculator · standard deviation calculator
2. Normal distribution (normalcdf and invNorm)
| You need | Desmos | TI-84 |
|---|---|---|
| P(a < X < b) | normaldist(μ,σ).cdf(a,b) | DISTR ▸ normalcdf(a, b, μ, σ) |
| P(X < x) | normaldist(μ,σ).cdf(x) | normalcdf(−1E99, x, μ, σ) |
| Value with area p to its left | normaldist(μ,σ).inversecdf(p) | DISTR ▸ invNorm(p, μ, σ) |
Example: normaldist(100,15).cdf(85,115) = 0.6827. inversecdf only takes the area to the left. For the cutoff for the top 10%, enter normaldist(100,15).inversecdf(0.9) = 119.22.
Check your work: z-score calculator
3. Binomial distribution (binompdf and binomcdf)
| You need | Desmos | TI-84 |
|---|---|---|
| P(X = k) | binomialdist(n,p).pdf(k) | DISTR ▸ binompdf(n, p, k) |
| P(X ≤ k) | binomialdist(n,p).cdf(k) | DISTR ▸ binomcdf(n, p, k) |
| P(a ≤ X ≤ b) | binomialdist(n,p).cdf(a,b) | binomcdf(n, p, b) − binomcdf(n, p, a−1) |
| P(X ≥ k) | 1-binomialdist(n,p).cdf(k-1) | 1 − binomcdf(n, p, k−1) |
Example with n = 10, p = 0.3: P(X = 3) = 0.2668, P(X ≤ 3) = 0.6496, P(X ≥ 5) = 0.1503. The “at least” case trips people up: P(X ≥ 5) is 1 minus P(X ≤ 4), not 1 minus P(X ≤ 5).
More: binomial distribution on TI-84 and Desmos, with “at least” and “between” worked through · combinations calculator · Unit 2 study guide
4. t critical values (invT)
For a 95% interval with 7 degrees of freedom, Desmos: tdist(7).inversecdf(0.975) = 2.3646. TI-84: DISTR ▸ invT(0.975, 7). For a C% interval the area is (1 + C)/2, so 0.975 for 95% and 0.995 for 99%.
5. One-proportion z-test and interval
| You need | Desmos | TI-84 |
|---|---|---|
| Test of H₀: p = p₀ | zproptest(x,n).null(p₀), then read .score and .p | STAT ▸ TESTS ▸ 1-PropZTest |
| Confidence interval | zproptest(x,n).conf(0.95), then .lower and .upper | STAT ▸ TESTS ▸ 1-PropZInt |
Example: 58 successes in 100 trials, testing p = 0.5, gives z = 1.6 and a two-sided p-value of 0.1096. .p is two-sided. Use .pright or .pleft for a one-sided test, or click Left or Right in the panel’s Tails buttons. More on tails and p-values: how to find a p-value on Desmos and the TI-84.
6. Two-proportion z-test and interval
Desmos: zproptest(x₁,n₁,x₂,n₂). TI-84: STAT ▸ TESTS ▸ 2-PropZTest or 2-PropZInt. Example: 45/100 vs 30/90 gives z = 1.643 and p = 0.1004.
.stderr into a test. The z statistic uses the pooled proportion, as AP requires. The .stderr value Desmos shows (0.0703 here) is the unpooled standard error used for the interval. If you show the z calculation by hand, use the pooled standard error.Check your work: confidence interval & hypothesis test calculator · Unit 3 study guide
7. One-sample t (including paired data)
| You need | Desmos | TI-84 |
|---|---|---|
| t-test from a list | ttest(L).null(μ₀) | STAT ▸ TESTS ▸ T-Test (Data) |
| t-test from summary stats | ttest(n, x̄, s).null(μ₀) | T-Test (Stats) |
| t-interval | ttest(L).conf(0.95), then .lower and .upper | STAT ▸ TESTS ▸ TInterval |
| Paired data | ttest(L₁-L₂).null(0) | Store L₁−L₂ in L₃, then T-Test on L₃ |
Example with the list above, testing μ = 12: t = 1.834, df = 7, two-sided p = 0.1092, and the 95% interval is (11.31, 17.44). With summary statistics the order is n, mean, SD: ttest(25,52.1,4.8).null(50).pright = 0.0193.
8. Two-sample t-test and interval
Desmos: ttest(L₁,L₂) or ttest(n₁,x̄₁,s₁,n₂,x̄₂,s₂). TI-84: STAT ▸ TESTS ▸ 2-SampTTest or 2-SampTInt, with Pooled: No.
Both report a decimal degrees of freedom, such as df = 13.05. That is the Welch–Satterthwaite value. By hand, AP teaches the conservative df = min(n₁−1, n₂−1), which gives a slightly larger p-value. Both are accepted, and our inference calculator shows the conservative method so you can follow every step.
Related: Unit 4 study guide
9. Chi-square test for homogeneity or independence
Desmos: enter each column of the two-way table as a list, as in chisqtest([20,30],[25,25],[15,35]). The panel shows the observed and expected counts, each cell's contribution, χ² = 4.167, df = 2, and p = 0.1245.
TI-84: enter the table as matrix [A] (2nd ▸ x⁻¹ ▸ EDIT), then STAT ▸ TESTS ▸ χ²-Test with Observed [A]. The expected counts are written to [B], which you need for the “all expected counts at least 5” condition.
Background: which chi-square tests are still on the exam
10. Linear regression, r, and residuals
| You need | Desmos | TI-84 |
|---|---|---|
| Least-squares line | Put the data in a table, then type y₁ ~ m x₁ + b | STAT ▸ CALC ▸ LinReg(a+bx) L₁, L₂ |
| r and r² | Shown under the regression; also corr(L₁,L₂) | Turn Stat Diagnostics on (MODE screen on a TI-84 Plus CE) |
| Residuals | Created automatically as e₁; click “Plot” for the residual plot | 2nd ▸ STAT ▸ NAMES ▸ RESID |
Check your work: linear regression calculator · Unit 5 study guide
The fallback: p-values from the distributions alone
If the test functions ever fail, the distribution functions still get you every p-value. Work out the test statistic from the formula sheet, then:
- z statistic, two-sided:
2·normaldist(0,1).cdf(-|z|). For z = 1.6 that is 0.1096, matching the built-in test. - t statistic, right tail:
1-tdist(df).cdf(t). Double it for a two-sided test. - Chi-square:
1-chisqdist(df).cdf(χ²).
The statistics formulas cheat sheet has every test statistic formula.
What to write on the free-response section
The calculator does the arithmetic. It does not earn the points. Graders want the same communication whichever calculator you used:
- Name the procedure, such as “two-sample t-test for a difference in means”, and check its conditions.
- Label probability calculations. The 2026 scoring guidelines accepted labeled calculator syntax, listing Desmos-style
normaldist(mean = 109, standard deviation = 16)andbinomialdist(n = 10, p = 0.2459)next to TI-stylenormalcdf. Write the distribution, its parameters and the bounds. A barenormalcdf(85,115,100,15)with no labels is risky. - Report the test statistic, df where relevant, and p-value, then give the conclusion in context.
Practice the full write-up with the AP Stats FRQ practice tool, which has a 10-point self-scoring rubric.
Removed for 2027: you can skip these
Desmos and the TI-84 still have tools for topics that are no longer tested. Don't spend study time on the chi-square goodness-of-fit test (chisqgof, χ²GOF-Test), the t-test for a regression slope (LinRegTTest), or geometric distribution functions. See every topic removed for 2027.
As an Amazon Associate, MathInSite earns from qualifying purchases.
Frequently Asked Questions
-
Can I use Desmos on the 2027 AP Statistics exam?
Yes. The May 2027 AP Statistics exam is fully digital in Bluebook, and the Desmos graphing calculator is built into the app. Only the Bluebook version counts: the desmos.com website and the Desmos phone app are not allowed. You may also bring up to two approved handheld graphing calculators.
-
Can Desmos do a chi-square test or a two-proportion z-test?
Yes. The College Board version of Desmos runs every inference procedure in the 2027 course, including chisqtest for homogeneity and independence and zproptest with two samples for a two-proportion z-test. Older guides that say Desmos cannot do these are out of date.
-
Does Desmos output count as showing work on an AP Statistics FRQ?
Not by itself. You still need to name the procedure, check conditions, and state the test statistic and p-value in context. For probability calculations, the 2026 scoring guidelines accepted labeled calculator syntax such as normaldist with the mean and standard deviation written out, alongside TI syntax like normalcdf. Bare keystrokes or an unlabeled answer are not enough.
-
Why does Desmos give a different p-value than my two-sample t-test by hand?
Desmos and the TI-84 use the Welch-Satterthwaite degrees of freedom for a two-sample t procedure, which is usually a non-integer larger than the conservative df of the smaller sample size minus one taught for hand calculation. The larger df gives a slightly smaller p-value and a slightly narrower interval. Both approaches are accepted; just be consistent.
Sources
- College Board AP calculator policy
- AP Statistics exam page (AP Central)
- College Board Desmos calculator configuration (SY2026–27)
- Desmos help: Inference and Probability Distributions
- 2026 AP Statistics scoring guidelines
Desmos syntax tested in the College Board practice calculator. Updated for the May 2027 exam. Last reviewed October 2026.