Do female and male teachers differ in burnout? Do students taught with the new method score higher than students taught with the old one? Questions of this form appear in almost every quantitative thesis, and the test that answers them is the independent samples t-test. It is simple in principle, but its SPSS output table has two rows and about ten columns, and students routinely read the wrong row, ignore Levene’s test or write p = 0.000.
This article is part of our guide How to Write a Master’s Thesis by the Mastermind PhD team and relates to our Statistical Analysis service. Sources last checked: November 2026.
This guide covers when the test applies and when it does not, its six assumptions and how to check them, the SPSS steps, how to read the output column by column, how to compute and interpret the effect size, and how to report the result in APA style, finishing with the mistakes that undermine a results chapter.
What is the independent samples t-test? It is a parametric test that compares the means of two independent groups (different people in each group) on a continuous dependent variable, to decide whether the observed difference is larger than would be expected by chance in the population. It is also called the two-sample or Student’s t-test. The version that does not assume equal variances in the two groups is Welch’s t-test, which SPSS prints as the second row of the output.
When should you use the independent samples t-test, and when not?
Use it when the independent variable is nominal with exactly two independent categories (gender, public versus private sector, experimental versus control group at post-test only) and the dependent variable is continuous (a scale score, a test score, a time). Do not use it when:
- There are three or more groups. Running three t-tests inflates the Type I error rate; use one-way ANOVA.
- The two sets of scores come from the same people measured twice (before and after). Use the paired samples t-test.
- The dependent variable is nominal (use chi-square) or clearly ordinal, heavily skewed in a small group, or affected by outliers you cannot justify removing (use Mann-Whitney U).
What are the six assumptions and how do you check them?
| Assumption | How to check it | If it fails |
|---|---|---|
| Continuous dependent variable (interval or ratio) | Nature of the variable; averaged scale scores are treated as interval | Mann-Whitney U |
| Independent variable with two independent categories | Study design | ANOVA for more than two groups |
| Independence of observations | Each participant in one group only, no repeated measurement | A different design or test |
| No influential outliers | Boxplot for each group (Explore) | Correct entry errors; document any exclusion |
| Approximate normality within each group | Shapiro-Wilk, skewness and kurtosis per group | Rely on robustness if each group exceeds about 30; otherwise Mann-Whitney U |
| Homogeneity of variance | Levene’s test in the same output table | Read the “Equal variances not assumed” row (Welch) |
The normality check belongs in the results chapter before the test, in a short paragraph or table; examiners look for it.
📱 Ask a statistician about your data — Send a short description of your instrument and sample size on WhatsApp and get a suitable analysis plan. Every analysis comes with the raw SPSS file and a results report.
How do you run the test in SPSS?
- Analyze, then Compare Means, then Independent-Samples T Test.
- Move the dependent variable (the scale score) to Test Variable(s). Several scores can be entered at once.
- Move the grouping variable to Grouping Variable, click Define Groups and enter the two codes exactly as they appear in Variable View (for example 1 and 2).
- In version 27 and later, tick Estimate effect sizes to obtain Cohen’s d, Hedges’ g and Glass’s delta in the output.
- Click OK. SPSS produces two tables: Group Statistics and Independent Samples Test.
How do you read the output?
Group Statistics gives n, mean, standard deviation and standard error of the mean for each group. These values go into your thesis table.
Independent Samples Test is read in two steps:
- Levene’s test for equality of variances. If its Sig. value is greater than .05, the variances are treated as equal and you read the first row (Equal variances assumed). If Sig. is .05 or less, read the second row (Equal variances not assumed), which is Welch’s t-test with fractional degrees of freedom.
- The t result. From the chosen row take t, df, the two-sided p value (labeled Two-Sided p in recent versions and Sig. (2-tailed) in older ones), the mean difference and its 95% confidence interval. A confidence interval that includes zero corresponds to a non-significant difference.
A methodological note: Delacre, Lakens and Leys (2017) recommend using Welch’s test by default, regardless of Levene’s result, because it keeps the error rate at the nominal level when group sizes and variances differ and loses almost no power when they do not. Either strategy is acceptable in a thesis if it is stated before the results and applied consistently.
Why does APA ask for an effect size, and how do you compute Cohen’s d?
The p value tells you whether a difference is detectable; the effect size tells you how large it is. Cohen’s d is the difference between the two means divided by the pooled standard deviation. If your SPSS version does not print it, compute d = (M1 − M2) ÷ √[(SD1² + SD2²) ÷ 2], which is accurate when the groups are of similar size.
| Value of d | Interpretation (Cohen, 1988) |
|---|---|
| Around 0.20 | Small effect |
| Around 0.50 | Medium effect |
| 0.80 and above | Large effect |
With 600 participants, a difference of 0.15 on a 5-point scale can be significant while d = 0.12 signals an effect with little practical meaning. The discussion chapter is where that distinction has to be made explicit. The Publication Manual (APA, 2020) and the Journal Article Reporting Standards require an effect size, with a confidence interval where available, for every reported test.
Worked example: from the SPSS output to the APA sentence
A sample of 300 teachers (142 men, 158 women); the dependent variable is a burnout scale score from 1 to 5.
Group Statistics: men M = 3.21, SD = 0.78; women M = 3.49, SD = 0.64.
Independent Samples Test: Levene’s test Sig. = .041, below .05, so the variances are unequal and the Welch row is read: t = 3.35, df = 274.8, two-sided p = .001, mean difference = 0.28, 95% CI [0.12, 0.44]. The interval excludes zero.
Effect size: d = 0.28 ÷ √[(0.78² + 0.64²) ÷ 2] = 0.28 ÷ 0.713 = 0.39, a small-to-medium effect.
The APA sentence: “An independent samples t-test with Welch’s correction for unequal variances showed that women (M = 3.49, SD = 0.64) reported significantly higher burnout than men (M = 3.21, SD = 0.78), t(274.8) = 3.35, p = .001, d = 0.39, 95% CI of the difference [0.12, 0.44]; the effect was small to medium.” The sentence contains seven elements: both means and standard deviations, t, df, p, the effect size and the direction of the difference. These are what an examiner checks.
How do you present the result in APA style?
In the text, use the pattern above. For a non-significant result, report it with the same completeness: “Burnout did not differ significantly between the two sectors, t(298) = 1.13, p = .260, d = 0.13.” Non-significant findings are findings and are written with the same precision.
In a table, give one row per group with n, M and SD, followed by shared columns for t, df, p and d. The table number in bold sits above the italic title: “Table 8, Independent Samples t-Test for Burnout by Gender.” When Welch’s row was used, print the fractional degrees of freedom as SPSS shows them (274.8) and explain in the note: “Note. Welch’s correction was applied because Levene’s test was significant.”
APA number rules that apply here: two decimals for means, standard deviations, t and d; two or three decimals for p; exact p values unless p < .001; no leading zero for p; statistical symbols in italics; and n rather than N for a group.
Which mistakes weaken the result?
- Writing p = 0.000; the correct form is p < .001, because a probability is never exactly zero.
- Always reading the first row without looking at Levene’s test.
- Comparing three qualification levels with three successive t-tests.
- Omitting the effect size, or describing a significant difference in a large sample as “a large difference”.
- Switching to a one-tailed test to rescue a non-significant result without a directional hypothesis stated in advance.
- Testing single Likert items instead of the scale score.
- Not reporting the two group sizes; 12 principals against 280 teachers calls for Welch’s test and a comment on balance.
📱 Ask on WhatsApp now — 5 years of experience across all disciplines and 14 academic services. Two-stage payment (80/20); revisions are free until supervisor approval, within 30% of the chapter.
Frequently asked questions
When should I use Mann-Whitney U instead of the t-test?
When one group is clearly non-normal and smaller than about 30, when the dependent variable is explicitly ordinal (a rank or a single Likert item), or when outliers cannot be justified for removal. Mann-Whitney compares distributions of ranks rather than means, so you report medians with it.
What if the two groups are very different in size, such as 40 and 260?
The test does not require equal sizes, but unequal sizes make Levene’s result more consequential. Read the Welch row when variances differ, state both sizes explicitly, and interpret cautiously when the smaller group is below 30.
Do I need a t-test if my research questions are purely descriptive?
No. If every question asks “what is the level of …”, descriptive statistics are sufficient. The t-test is for a question or hypothesis that compares two independent groups.
What is the difference between the independent samples and the one-sample t-test?
The one-sample t-test compares a single group mean with a fixed value, such as the scale midpoint of 3 or a known norm. The independent samples t-test compares two groups with each other. Some theses use the one-sample version to judge whether a scale mean differs from the theoretical midpoint.
How do I report t-tests for several scales in one table?
One row per scale with the two means and standard deviations, then *t*, *df*, *p* and *d*, and a note stating which rows used Welch’s correction. If several comparisons test the same hypothesis, mention whether you adjusted the alpha level (for example with a Bonferroni correction) or justify not doing so.
Sources
- Field, A. (2018). Discovering Statistics Using IBM SPSS Statistics (5th ed.). SAGE.
- Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences (2nd ed.). Lawrence Erlbaum.
- Delacre, M., Lakens, D., & Leys, C. (2017). Why psychologists should by default use Welch’s t-test instead of Student’s t-test. International Review of Social Psychology, 30(1), 92–101. https://doi.org/10.5334/irsp.82
- American Psychological Association. (2020). Publication Manual of the American Psychological Association (7th ed.); Journal Article Reporting Standards: Quantitative. https://apastyle.apa.org/jars/quantitative
- Laerd Statistics. Independent t-test using SPSS Statistics. https://statistics.laerd.com/spss-tutorials/independent-t-test-using-spss-statistics.php
- IBM. IBM SPSS Statistics documentation: Independent-Samples T Test. https://www.ibm.com/docs/en/spss-statistics
Sources last checked: November 2026. This article is general educational guidance and does not replace your university regulations or your supervisor’s instructions. How we prepare our content.