Construct a data set for which the paired t-test statistic is very large, indicating that when this analysis is used, the two population means are different, but t₀ for the two-sample t-test is very small so that the incorrect analysis would indicate that there is no significant difference for the means.