Analyses

One-way ANOVA

Compare the means of three or more independent groups.

Edit on GitHub

When to use it

Use one-way ANOVA when a numeric outcome might differ across three or more separate groups. For example: do exam scores differ between lecture, online, and workshop teaching? The test asks whether the group means are all the same. It does not, by itself, say which pair differs. That is what a post-hoc test is for.

Two groups only: use the independent-samples t-test.

Assumptions

Scores are numeric, and each person is in one group. Within each group, scores should be roughly normal, and the groups should have similar spread.

Tensr stores Levene’s test on the result. The default is the median-centre version. You can switch it to Brown–Forsythe or turn it off. Levene’s p below .05 means the spreads differ. This procedure does not print a Shapiro–Wilk table. If the spreads differ, use Welch (use_welch, off by default) or the Kruskal–Wallis test.

Running it in Tensr

Analyze → Compare Means → One-Way ANOVA. In chat: “One-way ANOVA of score by method.” Tukey is not a separate menu item. Ask for Tukey and Tensr runs this ANOVA with post hoc set to tukey.
Group column is the grouping variable. Value column is the numeric outcome. Groups with fewer than two rows are dropped, and the report marks them excluded.
Post hoc starts at none. Effect size starts at eta squared. Homogeneity starts at Levene. Group descriptives start on. The chart starts as a box plot. Missing values start as listwise.

Options

Prop

Type

Reading the output

The same 96 students, 32 in each teaching method. Tukey compares the pairs. The method gap was built to be a few exam points against a within-group spread of about 8 points.

GroupnMeanStatus
Lecture3268.209included
Online3272.234included
Workshop3273.741included
SourceSSdfMSFpη²
Between Groups523.3512261.6754.033.0210.08
Within Groups6,034.4379364.886———
Total6,557.78795

One-way ANOVA: score by method (3 groups, n = 96). F = 4.033, p = .021 — significant at α = 0.05. Post-hoc: tukey_hsd. The primary result is significant (p = .021). This is large enough, in this sample, to treat the comparison this page is about as a real association rather than noise. Metrics: F statistic = 4.033; p-value = .021; eta squared = 0.08.

Reporting (APA 7)

One-way ANOVA: score by method (3 groups, n = 96). F = 4.033, p = .021 — significant at α = 0.05. Post-hoc: tukey_hsd. This result is significant (p = .021). The effect is moderate, so report its size with the p-value.

Coming from SPSS

Analyze → Compare Means → One-Way ANOVA.

SPSS puts Welch and the post-hoc tests in the same dialog. Tensr’s Welch switch is use_welch, and post hoc is post_hoc. SPSS Options also offers a means plot. Tensr’s chart control is output_chart (boxplot, means_plot, or none).

The rank alternative is the Kruskal–Wallis H test. Two groups: independent-samples t-test. A second factor: two-way ANOVA. A covariate: ANCOVA.