Independent-samples t-test in SPSS

Imagine you want to know whether students from two different fields of study experience different levels of exam stress. Or whether men and women differ in their average self-esteem. If you want to compare two independent groups, you need the independent-samples t-test in SPSS.

What does the independent-samples t-test test?

Definition:
The independent-samples t-test tests whether the means of two independent groups differ significantly.
Null hypothesis: $H_0: \mu_1 = \mu_2$
Alternative hypothesis: $H_1: \mu_1 e \mu_2$

The groups must not overlap—each person must belong to only one group.

---

Assumptions

Before you get started, check whether the following conditions are met:

• The dependent variable is measured on a metric scale (e.g., stress, performance, self-esteem)
• The grouping variable is dichotomous (e.g., male/female, degree program A/B)
• The groups are independent
• The distributions of the dependent variable are approximately normal
• The variances should be roughly equal (this is tested automatically)

---

Example: Exam stress in two degree programs

Imagine you have an SPSS file with the following variables:

• stress (metric, 0 = no stress, 10 = maximum stress)
• studiengang (1 = Psychology, 2 = Business Administration)

You want to know: Is there a difference in average stress levels?

---

Step-by-step guide in SPSS

1. Check the data

Make sure that:

• stress is a metric variable
• studiengang is a grouping variable with exactly two groups

2. Run the test

Go to:

Analyze → Compare Means → Independent-Samples T Test...

3. Enter the variables

• Drag stress to the Test Variable(s) field
• Drag studiengang to the Grouping Variable field
• Click Define Groups…
  → Enter, for example, 1 and 2

Confirm with OK.

---

Interpreting SPSS output

The output consists of two tables:

Table 1: Group-specific statistics

Group	N	Mean	Std. Deviation
Psychology	30	7.0	1.0
Business Administration	30	6.5	1.2

Table 2: Test results

Levene’s test	Sig.	t	df	Sig. (two-tailed)	Mean difference	95% CI
F = 1.21	.276	1.86	58	.068	0.5	[-0.04; 1.04]

---

Interpretation

Statistic	Meaning
Levene’s test Sig. = .276	Variances are not significantly different → equality can be assumed
t = 1.86, df = 58	Test statistic with the associated degrees of freedom
Sig. (two-tailed) = .068	p-value – no significant difference at the 5% level
Mean difference = 0.5	Mean difference: Psychology students perceive more stress
Confidence interval contains 0	Difference is not significant

Conclusion: There is no significant difference at the usual 5% level. A difference is possible, but it is not sufficiently supported by the statistical evidence.

---

Optional: One-tailed hypothesis

By default, SPSS calculates a two-tailed test. If you want to test whether one group has more than the other, you can only do so with theoretical justification in advance and an adjusted interpretation.

Example: You expect psychology students to experience more stress. In that case, you may halve the p-value (if t > 0):

• p (two-tailed) = .068
• p (one-tailed) = .034
  → significant at the 5% level, but only permissible if there was a clear hypothesis beforehand.

---

Summary

Step	What you do
1	Prepare the SPSS file with an outcome variable and a grouping variable
2	Menu: Analyze → Compare Means → Independent-Samples t Test
3	Define the groups and run the test
4	Check Levene’s test (homogeneity of variance assumption), then interpret the t-test
5	Interpret the p-value and, if applicable, take a directional hypothesis into account

---

Q&A to think along

Question 1: When can you use this test?
Answer: When you want to compare two independent groups on a metric outcome variable.

Question 2: What is Levene’s test used for?
Answer: It tests whether the variances in the groups are equal. If the p-value is > 0.05, equality can be assumed.

Question 3: What does a p-value of .068 mean?
Answer: There is no significant difference—but a trend can be discussed.

Question 4: What do you need to consider when formulating directional hypotheses?
Answer: They must be justified in advance. The one-tailed p-value is obtained by halving the two-tailed value—but only if t points in the expected direction.