Skip to main content

How to report p-values in APA style

APA has specific and slightly unusual rules for statistics: no leading zero on a p-value, three decimals, and never p = .000. Here is the full set, with a template for every common test.

7 min read · Last reviewed 7 August 2026

The rules, in one place

  • No leading zero on any number that cannot exceed 1. That covers p, r and partial η². Write p = .032, not p = 0.032. Numbers that can exceed 1 — t, F, M, SD, Cohen's d — keep their zero.
  • Two or three decimal places for an exact p-value. p = .03 and p = .032 are both acceptable; three is more common and more informative near the threshold.
  • p < .001 below that, rather than a smaller figure.
  • Never p = .000. This comes straight out of SPSS and is always wrong — a p-value is never zero. Report p < .001.
  • Italicise the statistic, not the parentheses or the numbers: t, F, p, r, M, SD, N, d. Greek letters are the exception — χ² and η² stay upright.
  • Spaces around operators. p = .03, not p=.03.
  • Report an effect size and a confidence interval alongside the test. APA 7 treats this as expected, not optional.

A template for each test

TestFormat
t-testt(28) = 2.14, p = .041, d = 0.78
ANOVAF(2, 27) = 4.10, p = .028, η² = .23
Chi-squareχ²(1, N = 98) = 1.02, p = .312, V = .10
Correlationr(18) = .44, p = .052
Z-testz = 2.31, p = .021
Regression coefficientb = 1.24, SE = 0.41, t(96) = 3.02, p = .003

Two details in there are easy to miss. Chi-square carries the sample size inside the parentheses alongside its df — χ²(1, N = 98) — which no other test does. Correlation reports df in the parentheses, which is n − 2, not n: twenty pairs give r(18).

Putting it in a sentence

The statistics go inside the sentence that states the finding, not instead of it. A result reported as bare numbers makes the reader do the interpretive work.

Participants in the training condition scored higher (M = 52.3, SD = 8.1) than controls (M = 47.9, SD = 9.4), t(53.5) = 1.90, p = .062, d = 0.50, 95% CI [−0.24, 9.04].

Note the fractional degrees of freedom. That is a Welch correction, it is correct, and it should be reported as-is rather than rounded — APA asks for the df the test actually used.

For a non-significant result, resist "there was no difference." The honest phrasing names what you failed to find and lets the interval speak:

The difference was not statistically significant, t(53.5) = 1.90, p = .062, 95% CI [−0.24, 9.04]. The interval is consistent with anything from a negligible difference to one of about nine points, so this study does not resolve the question.

Confidence intervals

Square brackets, a comma between the limits, and the confidence level stated once: 95% CI [0.04, 1.57]. Do not repeat the percent sign inside the brackets, and do not write it as a range with a dash — a dash is ambiguous when a limit is negative.

If you report several intervals at the same level in one paragraph, APA lets you state the level once at the start and use [lower, upper] thereafter.

Tables and multiple results

When you have more than about three results, they belong in a table rather than the prose. In an APA table, p-values get their own column with the same rules — no leading zero, three decimals, < .001 at the floor.

The asterisk convention (* p < .05, ** p < .01, *** p < .001) is still permitted but is discouraged in APA 7 when exact values will fit. Asterisks throw away the difference between p = .049 and p = .001, which is the information a reader most needs.

The errors reviewers catch

  • p = .000 — the single most common. Use p < .001.
  • A leading zero on p or r — p = 0.03 should be p = .03.
  • Missing degrees of freedom — t = 2.14 is not reportable on its own.
  • n instead of df for a correlation — twenty pairs give r(18).
  • "Approached significance" for p = .06. Either it cleared your threshold or it did not; report the exact value and the interval and let the reader judge.
  • No effect size — expected in APA 7 for every test where one is defined.

Straight from the calculator

Every calculator on this site emits the APA string for its result, formatted to these rules, with a copy button. The p-value calculator produces it from a statistic; the t-test, A/B test and chi-square calculators produce it from raw data along with the effect size and interval you need to report next to it.

Other style guides differ. AMA and Vancouver keep the leading zero (P = 0.03) and AMA capitalises the P. If you are writing for a medical journal, check its instructions to authors rather than assuming APA.

Keep reading

Ready to run the numbers?

Our calculator shows the shaded distribution, the exact p-value, and a plain-English reading of what it supports.

Open the P-Value Calculator

Frequently asked questions

How do you write a p-value in APA style?

With no leading zero and two or three decimal places: p = .032. If the value is smaller than .001, write p < .001 rather than the exact figure. The symbol p is italicised, there are spaces around the equals sign, and p = .000 is always wrong — that output comes from software rounding and should be reported as p < .001.

Why is there no zero before the decimal point?

APA drops the leading zero on any quantity that cannot be greater than 1, which covers p-values, correlations and proportions of variance such as partial eta squared. Statistics that can exceed 1 — t, F, means, standard deviations, Cohen's d — keep it. The rule is about the range of the quantity, not about it being a statistic.

How do I report a non-significant result?

Report the exact p-value and the confidence interval, and describe what you failed to find rather than claiming there was no effect. "The difference was not statistically significant, t(53.5) = 1.90, p = .062, 95% CI [−0.24, 9.04]" is correct; "there was no difference" is not, because a non-significant result cannot establish absence. Avoid "approached significance" — APA reviewers routinely flag it.

How do you report a chi-square in APA?

χ²(df, N = sample size) = statistic, p = value — for example χ²(1, N = 98) = 1.02, p = .312. Chi-square is the only common test that carries the sample size inside the parentheses. The Greek letter is not italicised, unlike t, F and p. Report Cramér's V or phi alongside it, because χ² grows with sample size and says nothing about strength on its own.

Should I still use asterisks for significance?

APA 7 permits the convention but prefers exact p-values wherever they fit, including in tables. Asterisks collapse p = .049 and p = .0001 into categories and discard the difference. If a table is too dense for exact values, footnote the asterisk levels clearly — but exact values in a column are usually the better choice.

Do other style guides use the same rules?

No. AMA style writes a capital P and keeps the leading zero, so P = 0.03. Vancouver style also keeps the leading zero. Some journals cap reported precision at two decimals, and some still ask for asterisks. Check the target journal's instructions to authors before assuming APA conventions apply.