How to Report Statistical Results in APA Format

Getting the analysis right is only half the job. Reporting statistical results in APA format follows a set of conventions that reviewers know by heart, and deviations get noticed immediately. A missing degrees of freedom value, a leading zero where none belongs, or an unitalicized symbol signals carelessness before anyone reaches your conclusions.

This guide covers the formatting rules that apply to every statistic, then the specific notation for each test in a typical quantitative paper. It's about how the numbers are written. For what belongs in the section and how to organize it, see the guide to writing a results section.

APA Notation by Test: Quick Reference

This table covers the analyses in most quantitative papers. Each row shows the core reporting string and what should accompany it.

The Rules That Apply to Every Statistic

Four conventions govern almost everything you'll write. Get these right and most of the rest follows.

Italicize the symbol, not the number

Statistical symbols written in Latin letters are italicized. That covers M, SD, t, F, r, p, d, N, n, and R². The numbers that follow stay upright. Greek letters are the exception and are never italicized, so alpha, beta, chi-square, and eta-squared all stay in roman type.

This trips people up because word processors don't enforce it and statistical software doesn't output it. The italics have to be applied by hand, which is exactly why a manuscript missing them looks rushed.

The leading zero rule

Drop the zero before the decimal point when a value can never exceed 1. That covers p-values, correlation coefficients, and proportions of variance like eta-squared and R squared. So you write p = .03 and r = .48, not 0.03 and 0.48.

Keep the leading zero when a value can exceed 1. Means, standard deviations, test statistics, and Cohen's d all qualify, since d has no upper bound. So d = 0.63 keeps its zero while r = .63 does not, even though the digits look identical.

Decimal places

Two decimal places is the default for most statistics, including means, standard deviations, test statistics, and effect sizes. P-values take two or three, with three reserved for values close to the threshold where the extra digit carries information.

Round consistently across the whole paper. Reporting one mean to two decimals and another to three, with no reason for the difference, reads as inattention.

Degrees of freedom go in parentheses

Degrees of freedom follow the symbol in parentheses, before the equals sign: t(58), F(2, 87), r(118). Tests with two df values, like F, take both separated by a comma, with the numerator first.

Chi-square is the exception that also carries the sample size: χ²(2, N = 240) = 9.41. That's a rule reviewers in some fields check specifically, so it's worth getting right.

Reporting P-Values

Report the exact p-value rather than a threshold statement. Writing p = .034 tells the reader more than p < .05, and current APA guidance prefers it. The one time you use an inequality is when the value falls below .001, where you write p < .001 instead of chasing decimals.

Never write p = .000, which statistical software will happily display. A p-value is never exactly zero, only smaller than the software chose to print. That output should always be reported as p < .001.

Two related habits are worth avoiding. Don't describe a result as approaching significance, because a p-value of .07 either clears your threshold or doesn't. And reserve the word significant for its statistical meaning, using words like substantial or meaningful when you mean practical importance. For the underlying logic, see the guide to hypothesis testing.

Reporting Descriptive Statistics

Descriptive statistics usually appear in parentheses after the group they describe. The standard form pairs the mean with the standard deviation.

Use capital N for the total sample and lowercase n for a subgroup. Percentages take the percent sign with no space, and a number is given alongside it when the base isn't obvious. For skewed or ordinal data, report the median and interquartile range instead, since the mean misleads there. The reasoning behind that choice is covered in the guide to standard deviation and variance.

Reporting T-Tests

Give the group descriptives, then the test, then the effect size. A complete independent-samples report looks like this.

For a paired test, the descriptives describe time points rather than groups, and it helps to name the direction of change. State which test you ran when it isn't obvious from the design, and say whether the test was one-tailed, since two-tailed is assumed by default. The tests themselves are covered in the guide to t-tests.

One convention on the name itself. APA writes it as a t test without a hyphen when it's a noun. It hyphenates only when the term modifies another noun, as in t-test results. Journals vary on how strictly they enforce this.

Reporting ANOVA

ANOVA reports carry two degrees of freedom values: the numerator first, then the denominator. Effect size for ANOVA is usually partial eta-squared, and naming it explicitly matters because plain eta-squared is a different quantity.

A significant omnibus F tells the reader that at least one group differs, so report the post-hoc comparisons that identify which. Name the correction you applied, such as Bonferroni or Tukey, since an uncorrected set of pairwise tests inflates the error rate. Factorial designs report each main effect and each interaction on its own line, in that order. See the guide to ANOVA for the tests.

Reporting Correlation

Correlations take degrees of freedom rather than the raw sample size, and df is N minus 2 for a Pearson correlation. The coefficient drops its leading zero.

Because r is already a standardized effect size, no separate effect size is needed. Adding r squared is useful when you want to state the shared variance directly. When you report many correlations at once, a correlation matrix table is clearer than a paragraph of coefficients. For coefficient selection and reporting in depth, see using correlation coefficients in research papers, and for the mechanics, what the correlation coefficient measures.

Reporting Regression

Regression reporting has two layers: the model as a whole, then each predictor. Report the model first.

Lead with the unstandardized coefficient b, since it's the one interpretable in your variables' units. Standardized betas belong alongside it when you want to compare predictors measured on different scales. For models with several predictors, report adjusted R² as well. Put the coefficients in a table rather than prose once you pass three or four.

Logistic regression reports odds ratios rather than raw coefficients, always with their confidence intervals. An interval spanning 1 means the predictor isn't significant, no matter how the point estimate looks. The models are covered in the guides to simple linear regression, multiple regression, and logistic regression.

Reporting Chi-Square and Non-Parametric Tests

Chi-square carries both degrees of freedom and the sample size inside the parentheses, and the Greek chi stays upright.

Rank-based tests follow the same pattern with their own statistics: U for Mann-Whitney, H with degrees of freedom for Kruskal-Wallis. The important difference is in the descriptives. Because these tests work on ranks, describe the data with medians and interquartile ranges rather than means and standard deviations. Reporting a mean alongside a rank-based test reintroduces the sensitivity to skew you switched tests to avoid. See the guide to non-parametric tests.

Effect Sizes and Confidence Intervals

Journals across psychology, education, and health now expect an effect size with every significance test, and many expect a confidence interval too. Both go at the end of the reporting string, after the p-value.

Name the specific measure rather than writing effect size generically, since Cohen's d, Hedges' g, and partial eta-squared aren't interchangeable. Confidence intervals use square brackets with the limits separated by a comma, and the confidence level stated once: 95% CI [0.02, 1.80]. If every interval in a paragraph is at 95 percent, say so at the start rather than repeating it.

One thing effect sizes fix that p-values can't: significance depends heavily on sample size, so a trivial effect can reach significance in a large study. Reporting the magnitude is what keeps that visible. See the guide to effect sizes for which measure pairs with which test.

When to Use a Table Instead of Text

Prose works for a handful of statistics. Past that, a table is clearer and journals prefer it. The usual triggers are a correlation matrix, a regression with several predictors, and descriptives across multiple groups or time points. Any set of results the reader will want to compare across rows belongs in one too.

• Don't duplicate. If a value is in a table, don't restate it in the text. Highlight the finding and point the reader to the table.
• Number and title every table. Tables are numbered in order of first mention, with a brief italicized title beneath the number.
• Reference every table in the text. A table that no sentence points to will be queried in review.
• Note what the symbols mean. Use a general note beneath the table to define abbreviations and state the significance convention.

Common Mistakes

• Writing p = .000. Software output, not a real value. Report p < .001.
• Leaving symbols unitalicized. Statistical software doesn't italicize, so it has to be done by hand.
• Leading zeros on p-values and correlations. Both are bounded by 1, so both drop the zero.
• Omitting degrees of freedom. A test statistic without its df can't be evaluated or replicated.
• Reporting significance without an effect size. Most journals now treat this as incomplete reporting.
• Saying a result approached significance. A p-value either clears the threshold or it doesn't.
• Mixing eta-squared and partial eta-squared. They're different quantities and partial values run larger. Label which one you mean.
• Interpreting in the results section. Report the numbers there and save what they mean for the discussion.

Where to Go Next

This guide covers notation. For what belongs in the section and how to organize it, see writing a results section. For APA citation and reference formatting, which follows a separate set of rules, see the APA citation guide. For the sample size decisions that shape what you can report, see calculating sample size. The guide to Type I and Type II errors covers the errors behind those decisions.

When the analysis is written up, the results section is where a subject-matter editor earns their place. A reviewer who spots a p-value reported as .000, or an effect size that doesn't match the test, will question everything around it. Editor World's academic editing, dissertation editing, and journal article editing services include review of statistical reporting. You choose your own editor by field, so the person reading your results knows your discipline's conventions.

Frequently Asked Questions

Page last reviewed: September 2026. Content reviewed by Editor World editorial staff. Editor World, founded in 2010 by Patti Fisher, PhD, graduate of The Ohio State University, provides professional editing and proofreading services for academic researchers, doctoral candidates, faculty, business professionals, and authors worldwide. 100% human editing, no AI at any stage. BBB A+ accredited since 2010 with 5.0/5 Google Reviews and 5.0/5 Facebook Reviews. 16 years in business with 140 million+ words edited for over 8,000 clients in 65+ countries. Multiple Gold and Bronze Stevie Award winner. Native English editors from the United States, the United Kingdom, and Canada. Less than 5% of applicants are accepted to the editor panel. Recommended by the Boston University Economics Department, University of San Diego, University of Michigan, UCLA, University of Missouri, and more.