How to Perform One-Way ANOVA in SPSS: Step‑by‑Step Guide with Real Data

Introduction

SPSS (Statistical Package for the Social Sciences) is a software dedicated to statistical analysis. The steps for performing a one‑way ANOVA in SPSS are outlined below.

Open SPSS, load the data file, and select the dependent variable under the “Analyze” menu.

Choose “General Linear Model → Univariate” to access the one‑way ANOVA dialog, select the factor, and click OK.

SPSS generates a report containing means, variances, and other statistics for each factor and its effect on the dependent variable.

A sample study investigates how three learning methods (independent, group, and teacher‑guided) affect student scores. The recorded averages and variances are:

Independent learning: mean 85, variance 10

Group learning: mean 90, variance 15

Teacher‑guided: mean 95, variance 20

The ANOVA concludes that learning method has a significant impact on scores.

Case Details

To examine the effect of three treatments (A, B, C) on patients' ALT levels, 45 patients were randomly assigned to three groups (15 each). ALT values (U/L) were measured after treatment.

Data (excerpt):

Group A: 6.6, 6.8, 8.5, 9.8, 9.6, 11.6, 11.2, 16.5, 15.5, 14.0, 16.7, 16.0, 18.9, 18.0, 19.5

Group B: 25.5, 24.6, 28.9, 26.4, 23.2, 24.5, 25.6, 27.6, 32.2, 30.4, 30.0, 26.6, 33.4, 34.1, 33.6

Group C: 10.5, 11.3, 7.4, 6.9, 17.0, 15.3, 16.8, 9.1, 6.7, 20.3, 18.6, 13.9, 9.5, 12.7, 6.2

Analysis

The data represent three independent groups with quantitative ALT measurements. Assuming independence, normality, and homogeneity of variances, a one‑way ANOVA is appropriate. Significant overall differences warrant post‑hoc multiple comparisons to identify which groups differ.

SPSS Operations

Enter data into SPSS (1 = Group A, 2 = Group B, 3 = Group C).

Select Analyze → General Linear Model → Univariate (assuming normality).

In the main dialog, move ALT to the Dependent Variable box and Group to Fixed Factor(s).

Click Options , check “Descriptive statistics” and “Homogeneity tests”, then OK.

For post‑hoc tests, click Post Hoc , move Group to the Post Hoc Tests box, and select methods such as LSD, Bonferroni, Tukey, Šidák, Scheffé, Dunnett, etc. Choose Dunnett with Group A as control and a two‑sided test.

Result Interpretation

Descriptive statistics show means, standard deviations, and sample sizes for each group.

Levene’s test (F = 0.791, p = 0.460) indicates equal variances.

The ANOVA table shows a significant group effect (F = 68.810, p < 0.001), confirming differences among treatments.

Post‑hoc comparisons (Bonferroni, Dunnett, S‑NK) reveal:

Group B vs. A: Mean Difference ≈ 15.16 U/L, p < 0.001.

Group B vs. C: Mean Difference ≈ 16.29 U/L, p < 0.001.

Group A vs. C: No significant difference (p > 0.05).

Homogeneous subsets (S‑NK) place Groups A and C together (no significant difference) and Group B in a separate subset.

Conclusion

Mean ± SD ALT levels: A = 13.28 ± 4.39 U/L, B = 28.44 ± 3.65 U/L, C = 12.15 ± 4.64 U/L. The overall effect of treatment on ALT is statistically significant (F = 68.810, p < 0.001). Pairwise comparisons show B differs significantly from both A and C, while A and C do not differ.

References

Yang Chao, “One‑Way ANOVA – SPSS Tutorial”, https://www.mediecogroup.com/zhuanlan/lessons/164/