How to Interpret Independent-Samples t-Test Output in SPSS (and Report It in APA 7)

SPSS prints an independent-samples t-test as two tables (three in SPSS 27 and later), and the main one packs two tests into a single grid: Levene's test on the left and two versions of the t-test on the right. Most mistakes come from reading the wrong part of that grid. This guide goes through every table cell by cell with a worked example, lists the misreadings that cost marks, and ends with the APA 7 write-up. If you only need the reporting format, the independent t-test APA guide has it.

What SPSS gives you

Analyze > Compare Means > Independent-Samples T Test produces, in this order:

1. Group Statistics: the N, mean, standard deviation and standard error of each group.
2. Independent Samples Test: Levene's test, then the t-test on two rows.
3. Independent Samples Effect Sizes (SPSS 27 and later): Cohen's d and two variants of it.

The worked example

A researcher compared anxiety scores (0 to 80, higher means more anxious) between 24 people who finished an eight-week mindfulness course and 26 people on the waitlist. In Define Groups, group 1 is Mindfulness and group 2 is Waitlist. Every number in the tables below was computed from one dataset, so they agree with each other the way real SPSS output does.

Table 1: Group Statistics

• N: the people with a valid score in each group. Check these against your design; a smaller N than you expected means SPSS dropped missing values.
• Mean: the mindfulness group averaged 37.58, the waitlist group 45.81.
• Std. Deviation: the spread of scores within each group, 10.85 and 10.16. These are the SDs you report.
• Std. Error Mean: the standard deviation divided by √N. It describes how precisely the mean is estimated, not how spread out the scores are, and you rarely report it.

Table 2: Independent Samples Test

In the SPSS Viewer this table has stacked headings: "Levene's Test for Equality of Variances" sits over the first two columns and "t-test for Equality of Means" over the rest. Here the headings are flattened into one row:

The first two columns are Levene's test, not your result. Levene's test asks whether the two groups' variances differ. Here F = .009 and Sig. = .925: no evidence that they differ, so the assumption of equal variances holds.

The two rows are two different t-tests.

• Equal variances assumed is Student's t-test. It pools the two variances, and its df is n1 + n2 − 2 = 48.
• Equal variances not assumed is Welch's t-test. It does not pool, and its df is adjusted (here 46.986), usually to a decimal.

Read one row only. Because Levene's Sig. (.925) is above .05, this example uses the top row.

The columns of the row you read:

• t = −2.768. The sign comes from the group order: SPSS subtracts group 2 from group 1. It is negative because the mindfulness group (group 1) scored lower. Swap the groups and it turns positive; nothing else changes.
• df = 48.
• Sig. (2-tailed) = .008. This is the two-sided p value, the one you report. SPSS 27 and later split this column into One-Sided p (.004 here) and Two-Sided p (.008) under a "Significance" heading; report the two-sided value unless you preregistered a one-sided test.
• Mean Difference = −8.224, the group 1 mean minus the group 2 mean (37.583 − 45.808).
• Std. Error Difference = 2.972. t is the mean difference divided by this: −8.224 / 2.972 = −2.77.
• 95% Confidence Interval of the Difference: −14.199 to −2.249. The waitlist group's mean is plausibly between about 2 and 14 points higher. The interval excludes 0, which matches p < .05.

Table 3: Independent Samples Effect Sizes

• Standardizer is the standard deviation the mean difference is divided by. For Cohen's d it is the pooled SD, 10.498.
• Point Estimate is d itself: −8.224 / 10.498 = −0.78. The sign follows the mean difference; the size, 0.78, is close to what Cohen called a large effect (0.80).
• SPSS prints two more rows, Hedges' correction (d adjusted for small-sample bias, slightly smaller) and Glass's delta (which divides by one group's SD instead of the pooled SD), plus a 95% CI for each. Only the Cohen's d point estimate is shown above. Cohen's d is the usual choice; Hedges' correction is preferred by some journals for small samples.

Versions before SPSS 27 have no effect-size table. Compute d by hand as the mean difference divided by the pooled SD, where pooled SD = √[((n1 − 1)SD1² + (n2 − 1)SD2²) / (n1 + n2 − 2)].

Which row do I read when Levene's test is significant?

If Levene's Sig. is below .05, read the Equal variances not assumed row and report its decimal df. In this example that row gives t(46.99) = −2.76, p = .008, 95% CI [−14.22, −2.23]: the same conclusion, as is usual when the groups are similar in size. Many methodologists recommend Welch's test every time, since it loses almost nothing when variances are equal; if you follow that advice, read the bottom row whatever Levene's says and say so in your method. The Welch's t-test guide has the write-up, and the Levene's test guide covers reporting the check itself.

Misreadings that cost marks

• Reporting Levene's Sig. as the t-test p. The .925 here is about variances. The t-test's p is in the Sig. (2-tailed) column, .008.
• Writing p = .000. SPSS rounds to three decimals, so .000 means p < .001. Write p < .001. A p value is never exactly zero.
• Reading both rows, or the wrong one. Pick the row from Levene's test (or from a Welch-always rule), then use only that row's t, df, Sig. and CI.
• Treating a negative t as an error. It only says group 1 scored lower than group 2. Keep the sign consistent across t, the mean difference, its CI and d, and say the direction in words.
• Reporting the Std. Error Mean as the SD. Report Std. Deviation.
• Rounding Welch's df to a whole number. Report 46.99, not 47.
• Quoting the one-sided p because it is smaller. A one-sided test must be planned before you see the data.

The APA 7 format template

An independent-samples t-test showed that [dependent variable] was significantly higher/lower in [group 1] (M = X.XX, SD = X.XX) than in [group 2] (M = X.XX, SD = X.XX), t(df) = X.XX, p = .XXX, d = X.XX, mean difference = X.XX, 95% CI [X.XX, X.XX].

The write-up for this example

An independent-samples t-test showed that anxiety scores were significantly lower in the mindfulness group (M = 37.58, SD = 10.85) than in the waitlist group (M = 45.81, SD = 10.16), t(48) = −2.77, p = .008, d = −0.78, mean difference = −8.22, 95% CI [−14.20, −2.25]. Levene's test indicated equal variances, F(1, 48) = 0.01, p = .925.

SPSS does not print Levene's degrees of freedom in this table; for two groups they are 1 and N − 2. Writing the absolute values (t(48) = 2.77, d = 0.78) is also accepted, as long as the sentence says which group scored higher and the signs within the sentence agree.

Let KyroStat do the write-up for you

Already have your output? Copy the Independent Samples Test table from the SPSS Viewer and paste it into the box on the independent t-test reporting guide: it returns the APA sentence and checks that t, df and p agree with each other. Or upload the spreadsheet to KyroStat, which runs the test, checks the assumptions, and writes the APA 7 sentence and table with the code that produced them.

Also useful for the write-up (affiliate links): Proofreading for your thesis or paper (20% off with code PAP20): Paperpal

Frequently asked questions

Which row of the Independent Samples Test table do I report? The "Equal variances assumed" row when Levene's Sig. is .05 or above, and the "Equal variances not assumed" row when it is below .05. If you use Welch's test regardless, always the bottom row.

Is Levene's Sig. the p value of my t-test? No. Levene's test checks whether the group variances are equal. The t-test's p is in the Sig. (2-tailed) column, or Two-Sided p in SPSS 27 and later.

Why is my t value negative? SPSS computes group 1 minus group 2, in the order you set in Define Groups. A negative t means group 1 had the lower mean.

What does Sig. (2-tailed) = .000 mean? That p is below .0005 and rounded to .000. Report it as p < .001.

Where is Cohen's d in SPSS? In the Independent Samples Effect Sizes table (SPSS 27 and later), in the Point Estimate column of the Cohen's d row. In older versions compute it by hand from the Group Statistics table.

Which related guides might I need? Often these: paired-samples t-test output, Mann-Whitney U output (the usual non-parametric alternative) and Cohen's d.