How to Report a T-Test in APA Format

Report your t-test results with precision and clarity using APA 7th edition guidelines. Learn the exact format, common mistakes to avoid, and how to write results for tables and prose.

Key takeaway: APA style requires you to report the test statistic with its degrees of freedom, the p-value, and an effect size. Write t(df) = X.XX, p = .XXX, d = X.XX, with specific rules for decimals, italics, and when to use "less than" notation.

Why This Matters

A t-test result by itself is incomplete. Reporting t = 2.31 tells your reader nothing without the degrees of freedom, and a p-value alone doesn't convey the size of the effect. APA format ensures your results are reproducible and comparable across studies. When you follow the convention, readers know exactly what test you ran, how many participants you studied, and whether the effect is meaningful.

This guide goes beyond the single-sentence answer on the hub page. Here you'll see three complete worked examples, formatting rules, table layout, and answers to the questions that stumble most writers.

The APA Reporting Formula

t(df) = X.XX, p = .XXX, d = X.XX
t
The t-statistic, always italicized, rounded to 2 decimal places.
df
Degrees of freedom, in parentheses, not italicized. May be a whole number (e.g., 24) or fractional in Welch's t-test (e.g., 37.42).
p
The p-value, italicized. Use format .XXX (leading zero omitted). If p < .001, write "p < .001" instead of ".000".
d
Cohen's d, the effect size, italicized. Rounded to 2 decimal places. Optional but strongly recommended.

This is the minimal format. You may add a 95% confidence interval around the mean difference, but it is not required by APA for a paper. In tables or condensed results, you can omit the effect size, but it must appear somewhere in your results section.

Example 1: One-Sample T-Test

Scenario: You measure the height of 25 university students and test whether their mean height differs from the national average of 170 cm.

The Data

  • Sample size: n = 25
  • Sample mean height: M = 172.4 cm
  • Sample standard deviation: SD = 4.2 cm
  • Population value (μ₀): 170 cm

Calculation

  1. SE = 4.2 / √25 = 4.2 / 5.0 = 0.84

    Standard error: divide the sample SD by the square root of n.

  2. t = (172.4 − 170) / 0.84 = 2.4 / 0.84 = 2.857

    t-statistic: divide the difference between sample mean and population mean by the SE.

  3. df = 25 − 1 = 24

    Degrees of freedom: n minus 1.

  4. p (two-tailed) ≈ 0.0093

    From a t-distribution with 24 df, the two-tailed p-value for t = 2.857 is approximately .0093.

  5. d = (172.4 − 170) / 4.2 = 2.4 / 4.2 = 0.571

    Cohen's d: divide the mean difference by the sample SD.

APA Result Sentence

Written form: "The sample mean height (M = 172.4, SD = 4.2) significantly differed from the population mean of 170 cm, t(24) = 2.86, p = .009, d = 0.57."

What To Notice

  • The t-statistic and p-value are italicized; df is not.
  • The p-value has no leading zero: .009, not 0.009.
  • The t-statistic is rounded to 2 decimal places: 2.86 (not 2.857).
  • The effect size (d = 0.57) is included, rounded to 2 decimal places.
  • The word "significantly" is paired with the α level if you need to state it: "significantly differed at α = .05".

Example 2: Independent-Samples T-Test

Scenario: You compare test scores between two groups of students: those who received tutoring and a control group.

Part A: Equal Variances Assumed (Standard Case)

The Data

  • Tutored group: n₁ = 30, M₁ = 85.2, SD₁ = 8.5
  • Control group: n₂ = 28, M₂ = 78.9, SD₂ = 7.8
  1. sp² = [(30−1)×8.5² + (28−1)×7.8²] / (30+28−2) = 3743.37 / 56 = 66.84

    Pooled variance: sum of weighted squared deviations divided by total df.

  2. sp = √66.84 = 8.176

    Pooled SD: take the square root of pooled variance.

  3. SE = 8.176 × √(1/30 + 1/28) = 8.176 × √0.0690 = 8.176 × 0.2627 = 2.148

    Standard error: multiply pooled SD by the square root of 1/n₁ + 1/n₂.

  4. t = (85.2 − 78.9) / 2.148 = 6.3 / 2.148 = 2.931

    t-statistic: divide the difference between group means by the SE.

  5. df = 30 + 28 − 2 = 56

    Degrees of freedom: n₁ + n₂ − 2.

  6. p (two-tailed) ≈ 0.0049

    From t-distribution with 56 df, the two-tailed p-value for t = 2.931 is approximately .0049.

  7. d = (85.2 − 78.9) / 8.176 = 0.771

    Cohen's d: divide the mean difference by the pooled SD.

APA Result Sentence

Written form: "The tutored group (M = 85.2, SD = 8.5) scored significantly higher than the control group (M = 78.9, SD = 7.8), t(56) = 2.93, p = .005, d = 0.77."

Part B: Unequal Variances (Welch's T-Test with Fractional df)

Sometimes group variances are substantially different. Instead of pooling, Welch's t-test uses separate variances and produces fractional degrees of freedom. Here's an example:

The Data

  • Group A: n₁ = 25, M₁ = 92.1, SD₁ = 12.3
  • Group B: n₂ = 20, M₂ = 85.4, SD₂ = 6.9
  1. SE = √(12.3²/25 + 6.9²/20) = √(6.052 + 2.381) = √8.433 = 2.904

    Welch SE: take the square root of (variance₁/n₁ + variance₂/n₂).

  2. t = (92.1 − 85.4) / 2.904 = 6.7 / 2.904 = 2.307

    t-statistic: same formula as standard independent-samples t.

  3. df ≈ 39.05

    Welch–Satterthwaite df: calculated using a complex formula that accounts for unequal variances. Notice the fractional value—this is normal for Welch's test.

  4. p (two-tailed) ≈ 0.0254

    From t-distribution with 39.05 df, the two-tailed p-value for t = 2.307 is approximately .0254.

  5. d = (92.1 − 85.4) / √[(12.3² + 6.9²)/2] = 6.7 / 9.845 = 0.681

    Effect size: for Welch's test, use the average of the two group variances.

APA Result Sentence (Welch's Test)

Written form: "Group A (M = 92.1, SD = 12.3) scored significantly higher than Group B (M = 85.4, SD = 6.9), t(39.05) = 2.31, p = .025, d = 0.68. Due to unequal group variances, Welch's t-test was used."

Fractional df in APA Format

In Welch's t-test, df values like 39.05 or 37.42 are normal and expected. Always report them with at least two decimal places. Some writers round to the nearest whole number, but reporting the calculated value is more precise.

Example 3: Paired-Samples T-Test

Scenario: You measure anxiety levels in 32 participants before and after a mindfulness intervention. Each person is their own control (pre-post design).

The Data

  • Sample size: n = 32 pairs
  • Mean difference (post − pre): M_diff = 4.7 (anxiety decrease)
  • SD of differences: SD_diff = 6.8
  1. SE = 6.8 / √32 = 6.8 / 5.657 = 1.202

    Standard error of differences: SD_diff divided by √n.

  2. t = 4.7 / 1.202 = 3.910

    t-statistic: mean difference divided by SE.

  3. df = 32 − 1 = 31

    Degrees of freedom: number of pairs minus 1.

  4. p (two-tailed) ≈ 0.0004

    From t-distribution with 31 df, the two-tailed p-value for t = 3.910 is approximately .0004 (highly significant).

  5. d = 4.7 / 6.8 = 0.691

    Cohen's d for paired samples: mean difference divided by SD of differences.

APA Result Sentence

Written form: "Anxiety scores decreased significantly from pre-intervention (M = 28.3, SD = 7.4) to post-intervention (M = 23.6, SD = 5.9), t(31) = 3.91, p < .001, d = 0.69."

Pre-Post Reporting

For paired samples, report both pre and post means and SDs in the text (as shown above) or in a table. This gives readers the raw effect sizes and context for the statistical result.

Common APA Formatting Mistakes

Wrong Right
t(24) = 2.86, p = 0.009, d = 0.57 t(24) = 2.86, p = .009, d = 0.57
The t and p are not italicized. Always italicize t, p, and d.
t = 2.86, p = .009 t(24) = 2.86, p = .009
Degrees of freedom are missing. Always include df in parentheses immediately after t.
p = 0.009 p = .009
P-values include a leading zero. Omit the leading zero when p < 1.0.
p = .000 p < .001
Reporting extremely small p-values as ".000" is imprecise. If p < .0005 (or your software's minimum), write p < .001.
"The results proved that treatment is effective." "The results provide evidence that treatment is effective."
Using the word "proved" or "proves" after a t-test. T-tests assess evidence, not proof. Use "provides evidence for" or "is consistent with".
t(24) = 2.86, no p-value or effect size reported. t(24) = 2.86, p = .009, d = 0.57
Only the t-statistic is reported. Report p and effect size (d) to complete the picture.

Reporting in a Table vs. in Text

Sometimes you'll present results in a table instead of prose. The same APA rules apply, but the layout changes.

Example: Results Table (Multiple T-Tests)

Suppose you tested reading scores across three experimental conditions. You'd present the descriptive stats and t-tests in a table like this:

Condition n M SD
Standard instruction 28 72.4 8.6
Scaffolded instruction 26 78.9 7.1

Note. Standard vs. Scaffolded: t(52) = 3.15, p = .003, d = 0.87.

The key differences from in-text reporting:

  • Means and SDs go in columns so readers can compare groups at a glance.
  • The statistical test results (t, p, d) go in a note below the table.
  • Numeric cells are right-aligned.
  • Use a simple 3-line table (APA's recommended format): top rule, header rule, bottom rule, with horizontal rules only where needed.

FAQ

Do I have to italicize t and p?

Yes. In APA style, t, p, d, F, and other statistical symbols are italicized when they represent variables or parameters. This is a formatting rule, not optional. If you're writing in Word, Google Docs, or an online editor, select the symbol and press Ctrl+I (or Cmd+I on Mac), or use the italic button in the toolbar. Some equation editors (like Word's built-in equation editor) handle italics automatically.

How many decimal places should I use for the p-value?

Report p-values to three or four decimal places: .023, .0048, .156. The exact number depends on your field's convention, but three decimals is standard in psychology. Never report p = .0000 or a string of zeros; instead, write p < .001. If your software reports p < .0001, you may write either p < .001 or p < .0001; be consistent within your paper.

What if my p-value is less than .001?

Write p < .001. Do not write p = .000, p = .0000, or p < .0001 unless your field or journal has a specific rule otherwise. p < .001 is the APA standard and communicates that the result is significant at a very strict threshold. If you want to be more precise, you may report the actual computed value (e.g., p < .0001), but .001 is sufficient and is the most common choice.

Do I need to report a confidence interval around the mean difference?

A 95% confidence interval is not required by APA, but it is increasingly expected and provides valuable information. If you report a CI, it typically appears after the p-value: t(24) = 2.86, p = .009, d = 0.57, 95% CI [0.08, 1.06]. Some authors include the CI for effect size (d) and some for the raw mean difference; check your journal's guidelines. For basic papers or theses, reporting d without a CI is acceptable.

How do I report a non-significant result?

Use the same format, and avoid language that implies you "proved the null hypothesis." Instead of "there was no effect," write: "The difference was not statistically significant, t(24) = 1.52, p = .14, d = 0.31." This reports the results clearly while using the appropriate statistical language (fail to reject the null, not accept it). Effect size is still important even when p > .05, because it tells readers the practical magnitude of the difference.

Should I report one-tailed or two-tailed p-values?

Report two-tailed p-values by default unless you pre-registered a one-tailed hypothesis before collecting data. Most t-tests in practice are two-tailed. If you conduct a one-tailed test, you must state it explicitly in your methods section and in your results: "one-tailed p-value" or "α = .05 (one-tailed)." Failing to state this is a common error and can mislead readers.

Related Tools & Resources

Need to compute your t-test or verify your calculations? These CampusBreeze calculators and guides are designed to work together:

Sources