To report a t-test in APA 7 format, write it as t(df) = t-value, p = p-value, followed by the effect size (Cohen's d) and a 95% confidence interval — for example, t(48) = 2.34, p = .023, d = 0.66, 95% CI [0.42, 5.18]. If you've just run your analysis and you're staring at your output wondering which numbers go where (and in what order), you're in the right place — this is the single most common formatting mistake supervisors flag in draft theses.
Key Takeaways
- The core APA 7 format for a t-test is t(df) = value, p = value, with the test statistic and p italicised and a leading zero dropped from p (write
.023, not0.023). - APA 7 requires you to report an effect size (usually Cohen's d) alongside significance — a p value alone is no longer sufficient.
- Report p values to two or three decimal places; for very small values write p < .001 rather than p = .000.
- Always pair your statistics with the group means and standard deviations (M and SD) so readers can interpret the direction of the effect.
- A full APA 7 t-test result includes the test statistic, degrees of freedom, p value, effect size, and a confidence interval — in that logical order.
What is the correct APA 7 format for a t-test?
The correct APA 7 format for a t-test is: t(degrees of freedom) = test statistic, p = probability value, effect size, confidence interval. Written in a sentence, it looks like this:
Participants in the treatment group scored significantly higher (M = 24.6, SD = 4.1) than those in the control group (M = 19.8, SD = 3.9), t(48) = 4.21, p < .001, d = 1.20, 95% CI [2.51, 7.09].
Three formatting rules trip people up most often. First, italicise t, p, M, SD, and d — but not the numbers or parentheses. Second, drop the leading zero before any value that can't exceed 1 (so p = .03, not 0.03; d can exceed 1, so it keeps its digit). Third, put degrees of freedom in parentheses immediately after the italic t, with no space.
What do the numbers in a t-test actually mean?
Each element of the report answers a different question for your reader:
- t value — how large the difference between groups is relative to the variability in your data. Bigger (in absolute terms) means a stronger signal.
- Degrees of freedom (df) — roughly your total sample size minus the number of groups. For an independent-samples t-test with 50 people, df = 48.
- p value — the probability of seeing a difference this large if there were truly no difference. Below .05 is the conventional threshold for "statistically significant."
- Cohen's d — the effect size. It tells you how big the difference is in practical terms: 0.2 is small, 0.5 medium, 0.8 large.
- 95% CI — the range in which the true difference between groups plausibly lies.
A p value tells you whether there's an effect; the effect size and CI tell you how much it matters. APA 7 insists on both, because statistical significance without magnitude is misleading in small or very large samples alike.
A worked example, step by step
Imagine you're comparing exam scores between students who used flashcards and those who didn't. You have 30 students per group (N = 60).
- Flashcard group: M = 78.4, SD = 8.2
- Control group: M = 71.9, SD = 9.1
- Independent-samples t-test output: t = 2.91, df = 58, p = .005
- Cohen's d = 0.75, 95% CI of the difference [2.03, 10.97]
Walking through it: the df is 58 (60 participants − 2 groups). The t of 2.91 means the 6.5-point difference is nearly three standard errors away from zero. The p of .005 is well below .05, so the difference is statistically significant. Cohen's d of 0.75 is a medium-to-large effect — flashcards made a real, practically meaningful difference. The confidence interval [2.03, 10.97] doesn't cross zero, which reinforces significance.
Written up in APA 7:
Students who used flashcards scored significantly higher on the exam (M = 78.4, SD = 8.2) than those who did not (M = 71.9, SD = 9.1), t(58) = 2.91, p = .005, d = 0.75, 95% CI [2.03, 10.97].
That one sentence contains everything a reviewer needs. If you ran this in StatRyx, the APA-formatted sentence is generated for you — effect size, CI, and all — so you can paste it straight into your results section.
How do I report different types of t-tests?
The notation is nearly identical across t-test variants; only the wording and degrees of freedom change.
| T-test type | APA 7 example | Note on df |
|---|---|---|
| Independent-samples | t(48) = 2.34, p = .023, d = 0.66 | N − 2 |
| Paired-samples | t(29) = 3.10, p = .004, d = 0.57 | pairs − 1 |
| One-sample | t(24) = 2.05, p = .051, d = 0.41 | N − 1 |
| Welch's (unequal variances) | t(45.3) = 2.18, p = .034, d = 0.62 | often non-integer |
For a paired-samples t-test, name the comparison explicitly (e.g. "pre-test versus post-test"). For Welch's t-test, don't be alarmed by non-integer degrees of freedom like 45.3 — that's expected and correct; report it as given.
How should I report a non-significant t-test?
Report a non-significant t-test with exactly the same structure — do not hide it or write only "there was no significant difference." APA 7 expects the full statistics regardless of the outcome:
There was no significant difference in anxiety scores between the two conditions (M = 14.2, SD = 3.6 vs. M = 13.8, SD = 4.0), t(38) = 0.34, p = .736, d = 0.11.
Reporting the non-significant effect size (here, a tiny d = 0.11) is good practice — it signals the effect was genuinely small, not just undetected due to low power.
Common APA 7 t-test mistakes to avoid
- Writing p = .000.* Statistical software rounds, but a probability is never exactly zero. Write p < .001* instead.
- Keeping the leading zero on p. APA 7 drops it: p = .04, never p = 0.04.
- Omitting the effect size. This is the most common reason results sections get sent back for revision under APA 7.
- Forgetting the means and SDs. Without them, your reader can't tell which group scored higher.
- Italicising the parentheses or numbers. Only the letters (t, p, M, SD, d) are italic.
If you're still deciding whether a t-test is even the right test for your data, see our guide on Mann-Whitney U vs the t-test — the choice hinges on whether your data meet the normality assumption.
Stop calculating this by hand — run it free in StatRyx → Try StatRyx