What method is recommended, use contrast or t.test

I have a set with 60 observations, in four groups t1,t2,m1,m2 each with 15 observations. If I want to compare t1,t2 to m1,m2 or t1,m1 to t2,m2 what I did was to create a new variable so I can group the observations in 2 groups t and m and on the second iteration groups 1 and 2 then I just used a t.test basically to compare the group means... I was told that what I should've done is use CONTRASTS instead because the t.test doesn't take variance into account.

Why is this? can someone please expand on this or if it is better to use t.test as I did?


TS Contributor
this might be a bit of a misundetstanding: the vanilla t.test assumes equal variances, but so does anova. Moreover, you can run a t.test without the equal variance assumption. You might want to adjust for false positives though -e.g. by applying the Bonferroni correction for your t-tests.