How to Report Paired Samples T-Test in APA Format


How to Report Paired Samples T-Test in APA Format

You ran the test, SPSS printed its output tables, and now your supervisor wants one clean sentence in Chapter 4. Learning how to report paired samples t-test in APA format is easier than most students fear: two sentences, one set of numbers, five formatting rules — plus the exact template, the SPSS clicks, and a hand-worked hypothetical example below.

When to Use a Paired Samples T-Test

A paired samples t-test (also called a dependent samples t-test) compares the means of two measurements from the same people or matched units: pre-test vs post-test scores for one group, ratings before and after a workshop, or scores from matched pairs. If the two sets of scores come from different, unrelated people, use the independent-samples t-test instead — that is a different test and a different write-up.

The Assumptions Box (Check Before Reporting)

The test has four assumptions — check them before you write up anything:

  • Continuous (scale) dependent variable — means only make sense for interval or ratio data.
  • Independent pairs — each participant or unit contributes exactly one pair, and the pairs are independent of each other. (The same person appears in both conditions — that is the design. What must not happen is one person contributing two different pairs, or pairs influencing each other.)
  • Approximately normal differences — the assumption concerns the difference scores (post minus pre), not the raw scores in each condition.
  • No extreme outliers — a single wild difference score can distort the mean difference and the standard error.

To check normality, compute the differences first: Transform → Compute Variable, name the target variable something like Diff_PostMinusPre, enter Post minus Pre, then run Analyze → Descriptive Statistics → Explore on it. Our guide on how to test normality in SPSS walks through the Shapiro-Wilk result and Q-Q plot. If the differences are badly non-normal or your data are ordinal, the Wilcoxon signed-rank test is the standard non-parametric alternative — the FAQ below explains what it tests differently, and one condition it needs itself.

How to Run a Paired Samples T-Test in SPSS Step by Step

  1. Click Analyze → Compare Means → Paired-Samples T Test. (In SPSS 29+, the path is Analyze → Compare Means and Proportions → Paired-Samples T Test.)
  2. Move Post (the later, "after" score) into the Variable1 slot and Pre (the earlier, "before" score) into Variable2 under Paired Variables. The order sets the sign of every result, because SPSS always subtracts Variable2 from Variable1. With Post first, SPSS computes Post − Pre, and that is the order used throughout this article — including the worked example below. Swap the order and every sign flips: the mean difference, the t-value, and both confidence-interval limits.
  3. Click OK. SPSS produces three tables: Paired Samples Statistics, Paired Samples Correlations, and Paired Samples Test. In current versions (27 and later) you will also see a fourth: Paired Samples Effect Sizes.

Prefer syntax? Paste this from the dialog: T-TEST PAIRS=Post WITH Pre (PAIRED) /CRITERIA=CI(.9500) /MISSING=ANALYSIS. (Replace the variable names — and keep the order: the variable before WITH is Variable1, so this syntax also computes Post − Pre.)

How to Interpret the Paired Samples T-Test SPSS Output

Only the third table carries the test result itself, but each table earns its place:

  • Paired Samples Statistics — M, SD, N, and Std. Error Mean for each condition. You need the means and SDs for the APA sentence.
  • Paired Samples Correlations — the correlation between the two measurements (with its N and significance). This is not a test result to report in your sentence; it tells you how strongly the pairing is working. The higher this correlation, the smaller the SD of the differences — and the more power the paired design buys you over treating the scores as independent.
  • Paired Samples Test — the main results table: Mean (the mean difference, here Post − Pre), Std. Deviation and Std. Error Mean of the differences, the 95% Confidence Interval of the Difference, t, df, and the p-value. In current SPSS versions the p-value appears as two columns, One-Sided p and Two-Sided p — read Two-Sided p for a standard report. (In older versions the same value is labelled Sig. (2-tailed).)
  • Paired Samples Effect Sizes (current versions) — Cohen's d and Hedges' correction for the pair, each with a 95% confidence interval. For a paired test, SPSS's Cohen's d uses the SD of the differences as its standardizer — that specific version of d is called d_z, which the worked example explains.

Worked Example: Paired T-Test from Output to APA Sentence

This is a completely hypothetical, illustrative example (not real study results). Suppose a researcher tests 24 students before and after a six-week tutoring programme. SPSS reports:

  • Pre: M = 68.5, SD = 9.6
  • Post: M = 74.9, SD = 10.1
  • Mean difference (Post − Pre) = 6.4, SD of the differences = 7.2
  • Standard error of the difference = 1.47, t(23) = 4.35, p < .001
  • 95% CI of the mean difference = [3.36, 9.44]

Every number here is hand-checked: SE = 7.2 / √24 = 1.47; t = 6.4 / 1.47 = 4.35; df = 24 − 1 = 23; with t = 4.35 and df = 23 the two-tailed p is 0.0002, reported as p < .001; the 95% CI is 6.4 ± (2.069 × 1.47) = [3.36, 9.44]. One more consistency check: these SDs imply a pre–post correlation of about r = .73 — that is the value the Paired Samples Correlations table would show, and it is why the SD of the differences (7.2) is so much smaller than the spread of either set of scores alone.

The effect size — read this before you quote it. The standard effect size for a paired test is Cohen's d_z: the mean difference divided by the SD of the differences: d_z = 6.4 / 7.2 = 0.89, large by Cohen's (1988) guidelines (.20 small, .50 medium, .80 large). SPSS's Effect Sizes table reports this same value, with a 95% CI of [0.41, 1.36] here, plus Hedges' correction (g_z = 0.86), which adjusts d slightly for small samples.

But d_z comes with a warning. Because its denominator shrinks as the pre–post correlation grows, d_z is not directly comparable to the Cohen's d of a between-subjects study. Standardised by the average SD of the two conditions instead, the same data give d_av = 6.4 / 9.85 = 0.65. Same students, same improvement — two different effect sizes, both correct, answering slightly different questions. Use d_z when the comparison is within paired designs (and for power analysis of a paired test); use d_av when you need comparability with between-subjects effects or a meta-analysis — and always write which one you reported.

Treat the benchmarks as rough guidance, not cutoffs: some authors extend the scale (Sawilowsky, 2009, adds 0.01 = very small, 1.2 = very large and 2.0 = huge), so pick the convention your supervisor or journal uses and stay consistent.

How to Report a Paired Samples T-Test in APA Format

Here is the template (two sentences — the first names the test and variables, the second carries the result):

A paired-samples t-test was conducted to compare [variable] in the [condition 1] and [condition 2] conditions. There was a significant difference in the scores for [condition 1] (M = [mean1], SD = [SD1]) and [condition 2] (M = [mean2], SD = [SD2]) conditions; t(df) = [t-value], p = [p-value].

A paired-samples t-test was conducted to compare students' test scores before and after a six-week tutoring programme. There was a significant difference in the scores for the pre-programme (M = 68.5, SD = 9.6) and post-programme (M = 74.9, SD = 10.1) conditions; t(23) = 4.35, p < .001.

Add one sentence with the effect size and confidence interval:

Scores increased by a mean of 6.4 points, 95% CI [3.36, 9.44], a large effect (Cohen's d_z = 0.89).

For a non-significant result, keep the same sentence and swap in: "There was no statistically significant difference ... t(df) = [t-value], p = [p-value]."

The APA 7th formatting rules behind these sentences:

  1. Italicise the statistical symbols: t, p, M, SD, d, N, n. (Two standing exceptions in APA style: Greek letters and subscripts that act as labels are set in regular type.) The name of the test written out in ordinary prose — "a paired-samples t-test was conducted" — is not set in italics as a phrase.
  2. Degrees of freedom in parentheses right after t: df = N − 1 (here 24 − 1 = 23).
  3. Match decimals to your measurement. Report means and SDs to one or two decimal places — the example uses one (68.5, 9.6) because the test scores carry one-decimal precision; whichever you choose, use it consistently. Report the t-value to two decimals, and the p-value to three decimals — except when that rounds to zero.
  4. Drop the leading zero for p-values (write .009, not 0.009), because a p-value cannot exceed 1. Means, standard deviations, and d keep their leading zero where the value could exceed 1 (write 68.5 and 0.89, not .89 — a d above 1 is possible, as Sawilowsky's scale shows).
  5. Never write p = .000. SPSS prints .000 for very small values; always report p < .001 instead.

The same reporting logic extends to other tests — for correlations, the APA sentence follows the identical pattern (italicised statistic, df, p), as shown in our guide to reporting Spearman's correlation in APA format.

Common Mistakes When Reporting a Paired T-Test

  • Using it for independent groups. Comparing two different groups (boys vs girls) is not paired data — that needs the independent-samples t-test.
  • Checking normality on the raw scores. The assumption is about the difference scores, not each condition separately.
  • Reporting only the t and p. Supervisors also want the means, SDs, the direction of the difference, and increasingly the effect size.
  • Writing "insignificant" — a non-significant result is "not statistically significant", never "insignificant".
  • Forgetting the sign convention. SPSS reports Variable1 − Variable2. If you enter Pre first and Post second, SPSS prints the mean difference as −6.4 and t(23) = −4.35 for the example data — the negative sign simply means "the second score was higher". Either enter the variables in the order that matches your write-up (this article uses Post − Pre), or report the negative values and let your wording carry the direction ("scores increased from pre to post"). Never silently drop the minus sign.

Frequently Asked Questions

What is the difference between a paired and an independent samples t-test?

A paired (dependent) samples t-test compares two measurements from the same people or matched pairs — before vs after. An independent samples t-test compares means from two different, unrelated groups. The design, not the numbers, decides which test you run.

What if the difference scores are not normally distributed?

Do not panic or switch tests automatically. With moderate samples the paired t-test is reasonably robust to mild non-normality. If the departure is severe, the Wilcoxon signed-rank test is the standard alternative — but note two things: it tests something slightly different (the ranked differences, interpreted as a median difference only when the differences are roughly symmetrically distributed — an assumption of its own), so confirm it answers your research question before switching.

Can I report a one-tailed p-value?

Only if you predicted the direction of the difference in advance and stated so in your hypothesis. Current SPSS versions print both a One-Sided p and a Two-Sided p column, so use the One-Sided p value itself if your pre-stated directional hypothesis justifies it — never halve a two-sided p-value by hand, and confirm your supervisor's or journal's convention first.

Key Takeaways

  • The paired samples t-test is for two measurements on the same (or matched) participants; different groups need the independent t-test.
  • Check normality on the difference scores, not the raw conditions.
  • Run it via Analyze → Compare Means → Paired-Samples T Test (Compare Means and Proportions in SPSS 29+) — and enter the variables in the order you want subtracted (this article: Post − Pre).
  • Report: test name, both means and SDs, t(df), and the p-value — plus the mean difference with its CI and a named effect size (d_z or d_av) for a complete write-up.

References

American Psychological Association. (2020). Publication manual of the American Psychological Association (7th ed.).

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Lawrence Erlbaum Associates.

Lakens, D. (2013). Calculating and reporting effect sizes to facilitate cumulative science: A practical primer for t-tests and ANOVAs. Frontiers in Psychology, 4, 863. https://doi.org/10.3389/fpsyg.2013.00863

Sawilowsky, S. S. (2009). New effect size rules of thumb. Journal of Modern Applied Statistical Methods, 8(2), 597–599. https://doi.org/10.22237/jmasm/1257035100

Need help with your thesis or SPSS analysis? WhatsApp Farzana at +92 326 2047345.

Struggling with SPSS for your thesis? Get the step-by-step guide SPSS Made Simple — 30-page ebook + Excel templates pack. Buy it here: https://whop.com/researchmadesimple/spss-made-simple-a-step-by-step-guide-for-psychology-social-science-students/

Comments

Post a Comment

Popular posts from this blog

How to Test Normality in SPSS: A Step-by-Step Guide

Image
How to Test Normality in SPSS: A Step-by-Step Guide Before you run a t-test, ANOVA, or Pearson correlation, you need to know how to test normality in SPSS . Many parametric procedures rest on a normality assumption, but the assumption does not always apply to your raw scores: what exactly needs to be normal depends on the analysis you plan to run. This guide explains what the assumption really covers, shows you the two built-in normality tests in SPSS, where to click, how to read the output, and what to do if the assumption is not met. Why Check Normality Before Running Parametric Tests in SPSS Normality means scores cluster symmetrically around the mean and taper off evenly on both sides, forming the familiar bell curve. But "the data must be normal" is too simple, and it is worth being precise about what each procedure actually assumes: Independent-samples t-test: the assumption concerns the distribution of scores within each group . It matters most when group si...
Read more

Multiple Regression in SPSS: Run and Interpret It

Multiple Regression in SPSS: Run and Interpret It Multiple regression in SPSS is one of the most requested analyses by thesis students, and one of the most misreported. It lets you predict one continuous outcome variable from two or more predictors at the same time — for example, predicting exam scores from both study hours and sleep hours. In this guide, you will learn exactly when to use it, which assumptions to check before running it, the step-by-step SPSS clicks, and how to read every output table without guessing. A clear worked example walks you through the interpretation, and you will finish with a copy-ready APA-style write-up. When to Use Multiple Regression in SPSS Use multiple regression when you have: One continuous outcome variable (the dependent variable), such as exam scores, salary, or test performance. Two or more predictor variables (the independent variables), which can be continuous or categorical (categorical predictors must be coded as dummy variables). I...
Read more

One-Way ANOVA in SPSS: Run and Interpret It

Image
If you need to compare the average scores of three or more groups, an independent t-test is not enough. Running several t-tests one after another inflates the chance of a false positive. The correct tool is a one-way ANOVA in SPSS, which tests all groups at once in a single analysis. In this guide, you will learn when to use it, how to run it step by step, how to read every table in the output, and how to report your results in APA style. When to Use a One-Way ANOVA Use a one-way ANOVA when you have one categorical independent variable with three or more groups and one continuous dependent variable. The word "one-way" simply means there is one grouping factor. Common examples include comparing exam scores across three teaching methods, comparing customer satisfaction across four store locations, or comparing anxiety levels across three treatment groups. If you only have two groups, use an independent t-test instead. If you have two grouping factors, you need a two-way ANOV...
Read more