The main purpose of this exercise is to convince you that LSD should not be used in anova.sav you find simulated data for a normally distributed variable Z ~ N(µ,1). In groups 1 to 6 µ=0, while in group 7 µ=0.5.

Run an ANOVA using only groups 1 to 6 (You can hide group 7 from the analysis by data followed by select cases and then you filter out group 7 by if GROUP~=). Also run the posthoc tests LSD and Bonferroni.

LSD performs t-tests that differ from the ordinary t-test by using a standard deviation based on all groups (and not just the two involved in the comparison) and the degree of freedom is associated to that estimate.