when comparing the means of two populations, it is often preferable to use a two-sample t test, rather than a one-way anova f test because the non-pooled t statistic does not require the assumption that the populations have equal .