t-test Calculator with APA 7 Results
Independent-samples, paired-samples and one-sample t-tests from raw data or summary statistics. Get t, df, p, the confidence interval, Cohen’s d, Levene’s test, and the results sentence written in APA 7 style, ready for your thesis.
Use one-tailed only if the direction was fixed before collecting data.
How to use this calculator
1. Choose the test. Independent samples compares two different groups. Paired samples compares the same people twice. One sample compares a group with a fixed value, such as a scale midpoint.
2. Paste your data. One value per line, or copy a column straight from Excel or SPSS. If you copy two columns at once, the calculator splits them. If you only have means and standard deviations from a paper, switch to summary statistics.
3. Name the groups. The names you type are used in the APA sentence, so write what your thesis calls them.
4. Read the reported line. For independent samples the calculator runs Levene’s test and marks whether Student’s or Welch’s t is the one to report, the way SPSS shows both rows.
5. Copy the APA sentence. It goes into your results chapter as it is. The full chapter, with every hypothesis and the tables, is what the Chapter 4 Interpreter writes from your SPSS output.
Which t-test do you need?
An independent-samples t-test compares the means of two separate groups: men and women, treatment and control, two schools. The people in one group are not the people in the other.
A paired-samples t-test compares two measurements of the same participants: before and after a training, a pre-test and a post-test. Each row of your data is one person measured twice.
A one-sample t-test compares the mean of one group with a fixed value. In questionnaire research that value is usually the midpoint of the scale, for example 3 on a 1–5 Likert scale, to ask whether agreement is above neutral.
How the statistics are calculated
Independent samples, equal variances (Student): t = (M₁ − M₂) / (sₚ √(1/n₁ + 1/n₂)), where sₚ is the pooled standard deviation and df = n₁ + n₂ − 2. Unequal variances (Welch): the standard error is √(s₁²/n₁ + s₂²/n₂) and the degrees of freedom are adjusted with the Welch–Satterthwaite formula, which is why they are not a whole number.
Paired samples: the differences dᵢ = xᵢ − yᵢ are computed first and the test is a one-sample t-test on those differences, t = Mᵈ / (sᵈ / √n), df = n − 1.
One sample: t = (M − μ₀) / (s / √n), df = n − 1.
The two-tailed p-value is the probability of a |t| at least this large if the population means were equal. The 95% confidence interval is the mean difference ± tₑᵣᵢₜ × SE. Cohen’s d divides the mean difference by the pooled SD (independent), by the SD of the differences (paired, the value SPSS labels Cohen’s d) or by the sample SD (one sample); Hedges’ g corrects d for small samples. All values are checked against scipy to six decimals.
Assumptions to check first
Independence. Each observation must come from a different participant (independent samples) or each pair from one participant (paired). Students who share a class or a family are not fully independent; that is a design issue, not something a test can fix.
Normality. The dependent variable (or the differences, for paired samples) should be roughly normal in each group. With 30 or more per group the t-test is robust to moderate departures. With small samples check it with a Shapiro-Wilk test; if it fails, use the Mann–Whitney U test (independent) or the Wilcoxon signed-rank test (paired) instead.
Homogeneity of variance (independent samples only). Levene’s test checks whether the two groups have similar spread. If it is significant (p < .05) the variances differ and you report the Welch row, which SPSS labels “equal variances not assumed”. Many methodologists now recommend Welch by default; this calculator shows both and marks the one to report.
How to report a t-test in APA 7
The pattern is: the test, what was compared, the descriptives of each group, then t(df) = value, p, the confidence interval and the effect size. Statistics are italicised, numbers use two decimals, and p is given exactly to three decimals unless it is below .001. The calculator writes this sentence for you; change the group names and the name of the measure so it reads like your study.
A results chapter repeats this for every hypothesis and adds an APA table per test, plus the descriptive statistics, the assumption checks and the reliability of each scale. See a real one in the Chapter 4 Interpreter demo.
Frequently asked questions
Should I report Student’s or Welch’s t?
Report Student’s t when Levene’s test is not significant (variances are similar) and Welch’s t when it is. If you only have summary statistics, Welch is the safer choice when one SD is more than twice the other or the group sizes differ a lot.
One-tailed or two-tailed?
Two-tailed, unless your hypothesis stated a direction before the data were collected and your supervisor agreed. A one-tailed test halves the p-value, which is why examiners look at it closely.
My data are not normal. What now?
With 30 or more per group, carry on; the t-test is robust. With small, clearly skewed samples use the non-parametric alternative: Mann–Whitney U for independent samples, Wilcoxon signed-rank for paired samples, and report the medians.
Can I use this with SPSS output?
Yes. Paste the raw column from Data View, or type n, mean and SD from the Group Statistics table into summary mode. The t, df and p match SPSS’s Independent Samples Test table, both rows.