Reporting a Paired Samples t-Test in SPSS
Change, Interval and Effect Size
Reporting a paired samples t-test in SPSS puts the mean change and its interval first, then t with its df and a named effect size. Each is read from the printed tables while keeping track of which occasion was subtracted from which.
IN SHORT
What Changed, by How Much and With What Precision
Two measurements on the same cases leave one quantity to report, the mean change, and the rest of the report is built around it: a 95% confidence interval for that change, t with its degrees of freedom, an exact p-value and a named effect size on a standardized scale. For reporting a paired samples t-test in SPSS, take these from the Paired Samples Test and Paired Samples Effect Sizes tables (Analyze, Compare Means, Paired-Samples T Test; the effect sizes are an option in recent releases, and menu labels vary). At OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, an in-house PhD-level statistician prepares or checks this reporting, and what the change means is for the researcher to decide.
In This Guide
ANALYSIS SUPPORT
Analysis Support for This Method
A mean change summarizes the data well when the individual changes are not badly skewed or driven by outliers. Ordinal change, or three or more occasions, calls for a different analysis, settled before any reporting.
Paired Samples t-Test
The change in a mean between two measurements on the same cases, analyzed with its interval and a standardizer that suits the design.
Wilcoxon Signed-Rank Test
A rank-based comparison of paired measurements, for ordinal outcomes or when the differences do not support a summary by their mean.
BEFORE YOU REPORT IT
What the Change Score Depends On
Four decisions made before any value is copied determine the sign, the interval and the effect size that the report can defend.
Direction of Subtraction SPSS subtracts the second variable from the first, so the order of the pair decides whether an increase prints as positive or negative, and the sentence follows that order.
Complete Pairs Only A case missing either measurement drops out of the test, so the N in the report is the number of complete pairs, with the number lost stated beside it.
Which Standardizer By default, recent SPSS releases standardize by the SD of the differences, and some let the dialog choose another standardizer, so the Standardizer column is read before d is quoted. The average-variance version answers a different question.
Planned or Exploratory A comparison fixed in advance is reported as the planned test, and any further pair of occasions is labeled exploratory, with the number of comparisons made stated.
IN SPSS
Which Table Supplies the Mean Change and Its Effect Size
The pairs and the distribution of their differences are checked first, in Paired Samples t-Test in SPSS Explained, and the values below then come from one run of IBM SPSS.
Take the Means and N of the Pairs
From the Paired Samples Statistics table, take each measurement’s mean and standard deviation and the N, which counts complete pairs.
Check: Compare N with the number of participants measured on the first occasion. The gap is the attrition or missingness that the text should state.
Take the Mean Difference and CI
From the Paired Samples Test table, take the Mean under Paired Differences and the 95% Confidence Interval of the Difference, Lower and Upper.
Watch: The interval belongs to the mean difference, not to either occasion’s mean. Placing it beside the means invites a reader to compare the wrong intervals.
Take t, df and the p-Value
In the same row, take t, df and the p-value (Sig. (2-tailed) in older releases, Two-Sided p in recent ones).
Tip: SPSS also prints a one-sided p. Report the two-sided value unless a direction was stated in the analysis plan before the data were seen.
Take the Effect Size and Standardizer
From the Paired Samples Effect Sizes table in recent releases, take Cohen’s d or Hedges’ correction with its interval, and read the Standardizer column.
Check: Note the Standardizer value. If it equals the SD of the differences, the effect size lives on the scale of the change scores, and the text should say so.
Confirm the Direction of Change
Read the order of the two variables in the table and check that the sign of the Mean agrees with the sentence you are about to use.
Watch: Entering the pair in the opposite order reverses the sign of the mean, t, d and the interval limits, while p and the size of d stay the same, so a second run is a simple check of the direction.
Reconcile the Occasion Means With the Change
Enter the values in the paragraph and in the table of occasions below, and compare both with the saved output.
Check: Take the occasion means from the Paired Samples Statistics table, which uses complete pairs only. Means taken from all cases will not subtract to the reported difference.
READING THE OUTPUT
The Paired Samples Tables and Where Their Values Land
One paired run prints four tables in recent releases, but only three hold values for the result, while the correlation table gives context only.
| Output value | Where SPSS prints it | Where it goes and what to check |
|---|---|---|
| Mean, Std. Deviation and N | Paired Samples Statistics | The occasions table; N is the number of complete pairs, not of participants recruited |
| Correlation | Paired Samples Correlations | Context for precision only; it is not reported as the test result |
| Mean under Paired Differences | Paired Samples Test | The mean change, signed by the order of the variables, with the direction put in words |
| 95% Confidence Interval of the Difference | Paired Samples Test | Beside the mean change, in the outcome’s own units and not beside either occasion mean |
| t and df | Paired Samples Test | t(df) in the sentence; df is the number of pairs minus 1 |
| Two-Sided p (Sig. (2-tailed) in older releases) | Paired Samples Test | The exact p; a printed .000 becomes p < .001, and a one-sided p is not substituted |
| Cohen’s d or Hedges’ correction, Point Estimate | Paired Samples Effect Sizes, recent releases | The effect size, named by type, with its interval and the standardizer printed beside it |
REPORTING
The Change Leads, and the Occasions Stand Beside It
The paragraph reports the mean difference and its interval as the result, and a small table lets the reader judge that change against the level at each occasion.
Reporting paired t-test results in APA style puts the mean difference with its 95% interval first, followed by t with its degrees of freedom, the exact p-value and the effect size. SPSS prints the first variable of the pair minus the second, so when [Time 1] was entered first, the printed mean, t and d change sign and the interval limits change sign and swap places before they fit the frame below.
[Outcome] was compared between [Time 1] and [Time 2] for the ___ participants measured on both occasions. [Outcome] [increased / decreased] from M = ___ (SD = ___) to M = ___ (SD = ___), a mean difference of ___, 95% CI [___, ___], t(___) = ___, p = ___, d = ___, 95% CI [___, ___], standardized by [the SD of the differences]. The mean difference is [Time 2] minus [Time 1], so a positive value means the outcome was higher at [Time 2]. [One sentence on what a change of this size means for the study.]
| Table 1. [Outcome] at [Time 1] and [Time 2] (n = ___ complete pairs) | |||
|---|---|---|---|
| Occasion | M | SD | n |
| [Time 1] | ___ | ___ | ___ |
| [Time 2] | ___ | ___ | ___ |
| Difference ([Time 2] minus [Time 1]) | ___ [___, ___] | ___ | ___ |
| Note. The difference is [Time 2] minus [Time 1], with signs reversed from the output where [Time 1] was entered first. t(___) = ___, p = ___, d = ___ [___, ___], standardized by [the SD of the differences]. The interval beside the difference is a 95% confidence interval. Means and SDs are from complete pairs only, and ___ cases with a missing occasion were excluded. | |||
A paired effect size in APA style names its type and its standardizer together. By default, recent SPSS releases print Cohen’s d standardized by the SD of the differences and, beside it, Hedges’ correction, the small-sample adjustment usually reported as Hedges’ g. A value standardized by the SD of change scores is not on the same scale as a d from two independent groups, and effect sizes from different designs are not pooled without saying so. The mean difference with its interval remains the primary result, because it keeps the outcome in its own units.
COMMON MISTAKES
A Change Score Misread on the Page
Each is found by checking the paragraph line by line with the printed output, and each alters what the change appears to be.
Sign Read as Direction A negative mean difference prints whenever the first variable is lower, so an increase over time is negative in a before-and-after order. The text states the order and the direction in words.
Correlation Reported as the Result The Paired Samples Correlations table describes how the two measurements relate across cases and is not the test. Quoting it as the finding replaces the change with a different statistic.
Standardizer Not Named A d with no standardizer cannot be compared with one from another study, since the denominator may be the SD of the differences or an average of the two SDs. The report names it.
Pairs Lost Without Mention Cases missing one occasion leave the test silently, so the N behind the statistic is smaller than the sample. The text gives the number of complete pairs and the number lost.
p Without an Interval A p-value alone says nothing about how large the change is. The mean difference with its interval is the estimate, and the p-value is secondary to it.
Independent-Groups Wording Describing the result as a difference between groups misstates the design. The same cases were measured twice, and the sentence says that the change was within the same participants.
Which way does the sign of your change run? Independent Statistical Review confirms the sign, the complete-pair N and the standardizer behind each paired result.
FREQUENTLY ASKED QUESTIONS
Effect Sizes and Intervals in a Paired Report
Which effect size to quote, what to do when a release prints no effect size table and how to word a change whose interval includes zero.
Should I report d or Hedges' g from the SPSS table?
Either, provided the label matches the value; Hedges’ correction is the better choice in small samples because it reduces the upward bias of d. Name which one appears in the text and keep the same measure across every paired comparison in the paper, so that readers are not left to guess.
What if my SPSS release prints no effect size table?
The mean difference and its interval are still the main result, and the standardized effect is a short step away. Releases without the table leave that step to a calculation from the printed values, or to another program, and the report names the standardizer used so that the value can be interpreted.
How is a change reported when its interval includes zero?
With the same elements as any other result: the mean difference, its interval, t, df, the exact p-value and the effect size. An interval that includes zero shows that no change, and changes of either sign up to its limits, remain compatible with the data, so the text says the change was not distinguishable from zero in this sample and never that nothing changed.
Where does a statistician come in once the paragraph exists?
An in-house PhD-level statistician at OnlineSPSS can check the paragraph and table against the Paired Samples output, confirm the sign and the standardizer, or prepare the reporting text from the saved output within Results Reporting & Interpretation. Whether the change matters in practice is the researcher’s judgment.
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
Which Way Did Your Outcome
Move, and Does the Text Say So?
A change score earns trust when the interval, the effect size and its standardizer all come from one printed table. A paragraph that carries all three can be set beside the output and followed line by line.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014