statsmodels.stats.oneway.anova_generic

statsmodels.stats.oneway.anova_generic(means, variances, nobs, use_var='unequal', welch_correction=True, info=None)[source]

Oneway Anova based on summary statistics

Parameters:

  • means (array_like) – Mean of samples to be compared

  • variances (float or array_like) – Residual (within) variance of each sample or pooled. If variances is scalar, then it is interpreted as pooled variance that is the same for all samples, use_var will be ignored. Otherwise, the variances are used depending on the use_var keyword.

  • nobs (int or array_like) – Number of observations for the samples. If nobs is scalar, then it is assumed that all samples have the same number nobs of observation, i.e. a balanced sample case. Otherwise, statistics will be weighted corresponding to nobs. Only relative sizes are relevant, any proportional change to nobs does not change the effect size.

  • use_var ({"unequal", "equal", "bf"}) – If use_var is “unequal”, then the variances can differ across samples and the effect size for Welch anova will be computed.

  • welch_correction (bool) – If this is false, then the Welch correction to the test statistic is not included. This allows the computation of an effect size measure that corresponds more closely to Cohen’s f.

  • info (not used yet)

Returns:

res – This includes statistic and pvalue.

Return type:

results instance