one way anova vs. multiple comparisons

I'm using matlab to study significance different in mean for deviation in the thre directions 8vrt, lng and lat). with one way anova i get no significance (0.0532) but with multiple comparisons i get significance between lng and lat directions. when i apply anova on just lng and lat i got significance (0.0006) how can it be possible?


Less is more. Stay pure. Stay poor.
Do you control for multiple comparisons in the subsequent pairwise comparisons (adjust your level of significance to account for changes in chance)? Are the subsequent comparisons t-tests?