Reporting an Independent Samples t-Test in SPSS
Row, df, CI and Effect Size
Reporting an independent samples t-test in SPSS starts with which row of the output is quoted. The lower row carries fractional df, and the sentence then holds both group means, the mean difference with its interval and a named standardized effect size.
IN SHORT
What a Complete t-Test Report Holds
A t-test report gives the mean and spread of each group, the difference between the means with its confidence interval, the t statistic with its degrees of freedom and p, and a standardized effect size with an interval of its own. For reporting an independent samples t-test in SPSS, the values come from the Group Statistics and Independent Samples Test tables of Analyze, Compare Means, Independent-Samples T Test, and in recent releases from an Independent Samples Effect Sizes table. Menu names shift slightly between versions. A report of this kind is checked against the saved output by in-house PhD-level statisticians at OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research.
In This Guide
ANALYSIS SUPPORT
Analysis Support for This Method
An outcome scale or a variance pattern that makes the t-test doubtful changes what the report can claim. Unequal variances alone are handled within the t-test, while an ordinal outcome points to a comparison on ranks.
Independent Samples t-Test
The comparison of two group means for your design, with the variance question settled on the data and a sentence that carries the interval, not only the p-value.
Mann-Whitney U Test
A rank-based comparison for an ordinal outcome, or for a continuous one when the distributional assumptions of the t-test are not tenable.
BEFORE YOU REPORT IT
Choices That Show Up in the Sentence
These four choices are visible to a reviewer in the finished sentence, so they are settled before any figure is copied.
Which Row Is Quoted Fix the row in advance, on the design, and not on the p-value of Levene's test. The row quoted is named in the text, with the reason in a short clause.
Direction of the Difference The mean difference is the group entered as Group 1 in Define Groups minus the group entered as Group 2. Say which group was subtracted from which, so that the sign can be read.
Which Standardizer Cohen's d and Hedges' g divide by a pooled standard deviation, while Glass's delta divides by one group's. Choose the measure the question calls for and name it.
Group n, Not Total N The n beside each mean is the count with a valid outcome in that group. Cases with no recorded outcome drop out of that group's count, so n can sit below the number recruited.
IN SPSS
From Three Printed Tables to One Sentence
The row decision itself is argued in Independent Samples t-Test in SPSS Explained, and this section starts from the tables that IBM SPSS printed once that decision was made.
Take Each Group's Mean and SD
From the Group Statistics table, take N, Mean and Std. Deviation for each group.
Tip: Take the SD and not the Std. Error Mean printed beside it. The two columns sit side by side, and the second is much smaller, so a slip is easy to make.
Pick the Row and Say Why
In the Independent Samples Test table, take the equal variances not assumed row, and state in the text that the Welch adjustment to the df was used.
Watch: Choosing a row after reading Levene’s p-value makes the final test a two-stage procedure with error rates of neither t-test. The row is fixed in advance and named.
Take t, df, p and the Difference
From that row, take t, df, the two-sided p (Sig. (2-tailed) in older releases, Two-Sided p in recent ones), Mean Difference and the 95% Confidence Interval of the Difference, lower and upper.
Check: The df in the lower row carries decimals. Report it as printed, to one or two places, and do not round it to a whole number to match the upper row.
Take the Effect Size and Its CI
In releases that print it, take Cohen’s d or Hedges’ correction with its 95% confidence interval from the Independent Samples Effect Sizes table.
Watch: The table standardizes by the pooled SD, even beside the lower row. When the group variances differ clearly, Glass’s delta may suit better, and the text names the measure reported.
Settle the Direction Wording
Write which group had the higher mean and which group was subtracted from which in the mean difference.
Check: A negative mean difference beside a sentence that reads as positive is a common slip. Read the sign against the order of the groups in Group Statistics.
Reconcile Text, Table and Output
Enter the figures in the sentence and the group means table below, then check the paragraph against the saved tables.
Check: The mean difference equals the first group’s mean minus the second’s, within rounding. A larger gap between the two points to figures taken from different runs.
READING THE OUTPUT
Where Each Value in the SPSS t-Test Tables Goes
The tables below come from three printed sources, and this one pairs each value with its destination so that the row decision stays visible.
| Output value | Where SPSS prints it | Where it goes and what to check |
|---|---|---|
| N, Mean, Std. Deviation | Group Statistics | The M, SD and n cells of the group table, with the SD taken and not the standard error |
| Levene’s Test, F and Sig. | Independent Samples Test, left columns | Not quoted to justify a row; kept in the methods only if the analysis plan included it, and then described, not used |
| t | Chosen row of Independent Samples Test | The statistic, written with its df and the row it came from |
| df | Same row; decimals in the lower row | The df exactly as printed for that row, with its decimals when the lower row is used |
| Sig. (2-tailed), or Two-Sided p in recent releases | Same row | The p-value, with a printed .000 given as p < .001 and larger values kept to three decimals |
| Mean Difference and Std. Error Difference | Same row | The difference and its sign, with the order of subtraction stated; the standard error is not the SD |
| 95% CI of the Difference, Lower and Upper | Same row | The interval, written beside the difference as the range of values the data are compatible with |
| Cohen’s d or Hedges’ correction, Point Estimate and CI | Independent Samples Effect Sizes | The named effect size with its interval, and the standardizer stated when it is not the pooled SD |
REPORTING
The Sentence and the Group Means Table Together
The sentence names the row quoted, and the table repeats the group summaries so the decision can be traced through both.
In independent samples t-test APA format, the statistic is reported as t(df) = ___, p = ___ with the mean difference and its interval. A two-group results table APA style adds the n, mean and SD of each group and a note with the test. The frame below uses the equal variances not assumed row.
An independent samples t-test compared [outcome] between [Group A] and [Group B], using the equal variances not assumed row. [Group A] (M = ___, SD = ___, n = ___) scored [higher / lower] than [Group B] (M = ___, SD = ___, n = ___), t(___) = ___, p = ___, mean difference = ___, 95% CI [___, ___], Hedges’ g = ___, 95% CI [___, ___]. The degrees of freedom are fractional because that row does not assume equal variances, and the effect size uses the pooled SD.
| Table 1. [Outcome] by [Group], With the Mean Difference | |||
|---|---|---|---|
| Group | n | M | SD |
| [Group A] | ___ | ___ | ___ |
| [Group B] | ___ | ___ | ___ |
| Difference ([Group A] minus [Group B]) | ___ | 95% CI [___, ___] | |
| Note. t(___) = ___, p = ___; Hedges’ g = ___, 95% CI [___, ___]. The equal variances not assumed row is reported, and SD is the standard deviation, not the standard error. n is the number of cases with a valid outcome. | |||
Hedges’ g is Cohen’s d with a small-sample correction, and in the SPSS table it is the row labeled Hedges’ correction. Both divide the mean difference by the pooled SD, so their interpretation is in standard deviation units. Conventional benchmarks are rough guides, and the size is judged against the field’s measurement scale. Releases that print no effect size table need the value computed from the printed means, SDs and n. A wide interval signals that the size of the difference is poorly determined, even when p is small.
COMMON MISTAKES
Slips That Separate a Sentence From Its Output
A line-by-line comparison of the finished sentence with the three printed tables shows where each of these has crept in.
Row Picked After Levene Moving to the lower row only when Levene's p fell below .05 conditions the test on a test. The row is fixed in advance and named in the text.
df Rounded to a Whole Number Fractional df belong to the lower row. A whole number beside a t from the lower row suggests the upper row was copied, and the p-value then comes from the wrong place.
A Sign Without an Order A mean difference of negative value says nothing until the reader knows which group came first. The order of subtraction is written beside it.
g or d Left Unnamed An effect size after the interval that does not say whether it is Cohen's d, Hedges' g or Glass's delta cannot be compared with another study.
Interval Missing A t statistic and p alone leave out the quantity readers use, the mean difference and the range of values it is compatible with. Both belong in the sentence.
SE Placed Under SD Group Statistics prints Std. Deviation and Std. Error Mean in adjacent columns. The second copied under an SD heading shrinks the apparent spread of the group.
Is the right row in your draft? Independent Statistical Review checks the quoted row, the df and the interval and the effect size against the printed tables.
FREQUENTLY ASKED QUESTIONS
Levene, Three Estimates and the Sign of the Difference
Most of these start from the printed tables themselves: the Levene columns, the three effect size rows and the sign of the mean difference.
Do I report Levene's test if I use the lower row?
It is not required when the lower row was the plan from the start. If a journal or a thesis template asks for assumption checks, give the Levene result as descriptive information, and let it leave the row unchanged. The sentence names the row quoted and the reason for it in a short clause.
Why does the effect size table show three estimates?
They are Cohen’s d, Hedges’ correction and Glass’s delta, which differ in the standardizer or in a small-sample adjustment. Quote one, name it, and give its interval. Hedges’ g is the usual choice for small groups, and Glass’s delta is considered when one group serves as a reference with its own variance.
Which group is subtracted from which in the difference?
The group entered as Group 1 in Define Groups, minus the group entered as Group 2. Group Statistics lists the groups in that same order. State the order next to the mean difference so that a negative sign is not mistaken for a decrease in the wrong group.
What is a draft t-test sentence compared with during a check?
The saved Group Statistics, Independent Samples Test and effect size tables. An in-house PhD-level statistician at OnlineSPSS sets the quoted row, the df, the interval and the effect size beside them, and where needed prepares the sentence and group table from them. This is part of Results Reporting & Interpretation, and what the difference means is argued in the paper’s discussion.
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
Is the Reported Row
the Right One?
The row quoted, its fractional df and the named effect size each sit in a printed table, and a request can start from those tables whether or not a draft sentence exists yet.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014