Comparing Groups - Are my stats tests biased?


I am testing for statistical significance between lipid (oil) yield of two separate algal species split into four groups total. Each species has been split into two treatment groups, one with fewer nutrients and one with more. Therefore I have four groups to test statistical differences between.

When I run ANOVA on all four I get no significance, but when I only compare treatments within a species (so 1 v. 2 or 3 v. 4) I get statistical significance. What I'd like to know is which makes more sense. Is it even correct to be testing for significance BETWEEN species? Should I accept the first ANOVA result for all groups or should I be testing within species?

Thanks in advance!


TS Contributor
You did not report your exact research question(s),
but it looks like you should do a two-factor analysis
of variance (2x2).

With kind regards