METHOD GUIDE
Paired Samples t-Test in SPSS
Checking the Differences, Reporting the Effect
The assumptions of this test concern the differences between the two measurements, not the measurements themselves. This guide shows what to check, how to read the three SPSS tables and how to report the mean change in APA 7 with its interval.
IN SHORT
What the Test Estimates and How to Report It
A paired samples t-test analyzes the differences between two linked measurements, such as the same participants at two time points, the same people under two conditions, or matched pairs. It estimates the mean difference with a confidence interval and asks whether that difference is compatible with zero. The data are one numeric outcome recorded twice per pair. To run a paired samples t-test in SPSS, keep both measurements as two columns on one row per pair, open Analyze, Compare Means (Compare Means and Proportions in recent versions), Paired-Samples T Test, and plot the difference scores before reading the result. OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, carries out and reviews this analysis through an in-house PhD-level statistician.
ANALYSIS SUPPORT
Analysis Support for This Method
The tables show what changed. The pages below cover the paired analysis carried out on your own data, and the signed-rank route for differences that a mean does not summarize well.
Paired Samples t-Test
The comparison of two linked measurements carried out within your study design, with the check on the differences, the effect size and its interval, and the reporting agreed for your data.
Wilcoxon Signed-Rank Test
A rank-based analysis of the within-pair differences, for an ordinal outcome or when the differences are too skewed for a comparison of means.
BEFORE YOU RUN IT
What to Settle About the Pairs Before Testing
Because the test works on one difference score per pair, the decisions below concern how the pairs were formed and what the differences look like, and they come before any output is read.
One Pair per Case Each row holds both measurements of one participant or one matched pair, in two columns. Long-format data are restructured first, and a pair missing either value drops out of the test.
The Differences Are the Data The distributional question applies to the difference scores. Compute them and look at a histogram or Q-Q plot, since two skewed columns can still give well-behaved differences.
What the Change Can Mean Two time points without a comparison group cannot separate an intervention from practice, regression to the mean or season, so change is described and not attributed to a cause.
One Scale, Two Occasions Both measurements must use the same instrument and units, or a difference is meaningless. Ceiling and floor effects at either occasion also limit how much change the scale can show.
IN SPSS
How to Run the Paired Samples t-Test in SPSS
This outline is for IBM SPSS and is deliberately short. Menu labels vary by version, and the order of the two variables decides the sign of every result.
- Confirm that each row is one pair and that both columns use the same scale.
- Compute the difference variable with Transform, Compute Variable, and plot it.
- Open Analyze, Compare Means (Compare Means and Proportions in recent versions), Paired-Samples T Test, and move the two variables into a pair.
- Keep the 95% confidence interval, and in recent versions request the effect size estimates.
- Run it and check the number of pairs in the first table before reading anything else.
- Read the correlations table for context, then the test table for the result you report.
READING THE OUTPUT
What the Three SPSS Tables and Their Rows Tell You
The paired output spreads over several tables, one of which is easily mistaken for the result, so each element is listed with what it shows and what to verify before writing.
| Output element | What it tells you | What to check |
|---|---|---|
| Paired Samples Statistics | The mean, N, standard deviation and standard error of each measurement | That N is the number of complete pairs, and that any lost pairs are accounted for |
| Paired Samples Correlations | How strongly the two measurements are related across pairs | That it is read as context for precision and not reported as the test result |
| Paired Differences, Mean | The average of the first variable minus the second | That the sign agrees with the order of the variables and with your wording |
| Std. Deviation and Std. Error Mean of the differences | The spread of the difference scores and the precision of their mean | That the spread of the differences is what you examined in the plot |
| 95% Confidence Interval of the Difference | The values for the mean change that the data are compatible with | That it is given in outcome units and agrees with the sign of the mean |
| t, df and Two-Sided p (Sig. (2-tailed) in older versions) | The test statistic, with degrees of freedom equal to the pairs minus 1 | That the degrees of freedom match the number of pairs you report |
| Paired Samples Effect Sizes (recent versions) | A standardized effect with its interval, for a standardizer you can choose | That the standardizer is named, since the choice changes the value |
REPORTING
The APA 7 Template for a Paired Result
The wording below is a frame with blanks, to be completed from your own output. It contains no result.
Write the core as t(df) = ___, p = ___, with the mean difference and its 95% CI [___, ___] and a standardized effect with its interval, and name the standardizer. Give the mean and standard deviation at each occasion and the number of complete pairs, then say in words in which direction the measure moved and by how much in its own units. State the order of subtraction so that a reader can tell an increase from a decrease. Treat paired differences in SPSS as descriptive of change unless the design supports more, and report how any incomplete pairs were handled.
COMMON MISTAKES
Mistakes Reviewers Find in Reported Paired Tests
Six slips recur in reviewed manuscripts and theses on paired designs, and most of them start before the test table is read.
Normality Checked on the Columns Testing each of the two measurements for normality asks the wrong question. The assumption concerns the differences, so the difference scores are the ones to plot and inspect.
The Correlation as the Result The p-value in the correlations table tests whether the measurements are related. It does not test whether they changed, and quoting it as the main finding misreads the output.
Separate Tests per Group When two groups each have before and after measurements, separate paired tests in each group do not show that the groups changed differently. That question needs one model that compares the change between groups.
Pairs Lost Silently A participant missing one measurement leaves the analysis without any message in the main table. The number of pairs is reported, and the reason for any drop is stated.
Change as Treatment Effect A before and after difference in one group is open to practice effects and regression to the mean. The wording follows the design and does not claim that an intervention caused the change.
Standardizer Not Named Standardized effects computed with the standard deviation of the differences differ from those using the average standard deviation of the two measurements, so the text names which one is used.
Reported a before and after change already? Independent Statistical Review checks the pairing, the assumptions and the wording of the change before you submit.
FREQUENTLY ASKED QUESTIONS
Questions About Paired Samples Output
Short answers to what researchers ask about the differences and the output tables, from the sign of the difference and skewed differences to confidentiality and the assessment.
Why is the mean difference negative in my SPSS output?
SPSS subtracts the second variable of the pair from the first, so the sign follows the order in which the two variables were entered. If the later measurement was entered first, an increase appears as a negative mean difference with a negative interval. The t value changes sign, but the p-value stays the same. State the order of subtraction in the text, and check the sign against the two means before writing that scores rose or fell.
What if the differences are clearly skewed?
A few options exist, and the choice follows from the question. A transformation changes the scale on which the change is expressed. A signed-rank analysis addresses a different question from a comparison of means, and a model for the original scale may be better. For an ordinal outcome, a rank-based method is usually the better fit than the mean.
Why does SPSS print a correlation table for a paired test?
The SPSS paired samples correlations table shows how strongly the two measurements move together, which indicates how much the pairing removes person-to-person variation and why a paired design can be more precise than comparing two independent groups. The table is context for precision. The test result and its interval are in the paired test table.
Can OnlineSPSS analyze my before and after measurements?
Yes. An in-house PhD-level statistician can confirm that a paired approach fits how the measurements were taken, carry out the analysis with the check on the difference scores, and set out the change with its interval. An analysis you have already run can be reviewed instead. For the tables and the reporting alone, see Results Reporting & Interpretation. The research question and conclusions remain with you.
How do you protect my research data and confidentiality?
Your data and project details are handled confidentially, and project information is not published. Researcher names and institutions are not used publicly without permission. An NDA is available where required and can be arranged before sensitive materials are shared. Read more about confidentiality and data security.
What happens when I request an assessment, and how long does it take?
You share your research question, design and relevant materials, such as the dataset or the output you have, and a statistically trained Research Coordinator reviews the request and responds the same day. The analysis, deliverables, timeline and one fixed project fee are then confirmed before any work begins, so the timeline follows from the agreed scope.
Still have a question? Find more answers in our full FAQ.
NEXT STEP
Want the Paired Change
Checked Before You Submit?
Send the two measurements as they sit in your file, the number of complete pairs and any output you have. The analysis and its reporting are agreed with you before any work begins.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014