Analyses

Hotelling's T²

Compare two groups on several numeric outcomes at once.

Edit on GitHub

When to use it

Use Hotelling’s T² when you have exactly two groups and several numeric outcomes. It is the multivariate version of the independent-samples t-test. For example: do lecture and workshop students differ on recall and confidence at the same time?

Three or more groups: use MANOVA. One outcome: use the independent-samples t-test.

Assumptions

The outcomes are numeric. The two groups are separate. The outcomes should be roughly multivariate normal, with a similar covariance pattern in each group.

Tensr does not test those assumptions. It does compute η² from the T² statistic and prints it when it is available. You cannot turn that off.

Running it in Tensr

Analyze → Correlate → Hotelling's T². In chat: “Hotelling’s T-squared of recall and confidence by method.”
Group column must contain two groups, or you pass exactly two labels in group values. Dependent columns are the numeric outcomes.
There is no post-hoc option and no equal-covariance switch.

Options

Prop

Type

Reading the output

Two tutorial groups drawn from the same marking scheme, on score and confidence. The groups were not built to differ.

VariableA meanB meanDifference
score68.57568.935-0.36
confidence48.40751.355-2.948

Hotelling's T² comparing A vs B on 2 variable(s) (score, confidence). The primary result is not significant (p = .276). This is not large enough to treat the comparison this page is about as a reliable association. The result is non-significant: the data are still compatible with no effect. Metrics: T² = 2.653; F = 1.31; p-value = .276; η² = 0.033.

Reporting (APA 7)

Hotelling's T² comparing A vs B on 2 variable(s) (score, confidence). This result is not significant (p = .276). Report the estimate with that p, and do not describe the pattern as a reliable effect.

Coming from SPSS

The path stored in Tensr is Analyze → Compare Means → Hotelling. The menu you actually open is Analyze → Correlate → Hotelling's T².

SPSS often reaches this test through Multivariate GLM with two groups. Tensr prints T², the converted F, p, η², and the two group means. It does not print a covariance matrix or a univariate t for each outcome.

Three or more groups, or a follow-up ANOVA per outcome: MANOVA. One outcome: independent-samples t-test.