partial eta squared

[Rain]

Hello everyone,

I am new in statistics. I have done a statistical analysis with Spss
and I have got significant results in one two way anova with repeated measures.
but partial eta squared is small, could you please guide me about it. The results are:
factor 1) F(2,236)=5.750, p=0.004, partial eta squared=0.046
factor 2) F(1,118)=185.7,p=0.000,partial eta squared=0.61
interactions) F(2,236)=5.42, p=0.005, partial eta squared=0.044

Thanks in advance for your insight.

[Bruce Weaver]

Bear in mind that p-values are inversely related to sample size, whereas measures of effect size are not systematically related to sample size. So there is nothing "wrong" or unusual in what you are seeing, given your fairly large sample size.

HTH.

[Rain]

Thanks, so if the p shows significant results, and we got just 4% of effect size is not something unusual.

[Rich Ulrich]

On Tue, 6 Oct 2015 04:41:15 -0700 (PDT), Rain
<sanaz....@gmail.com> wrote:

>Thanks, so if the p shows significant results, and we got just 4% of effect size is not something unusual.

- Not for an N of 200.

Please do not make the mistake of saying that one
"significant" is not distinguishable from another
"significant" .... Your result of 4% is notably smaller
than your other result of 60%.

Eta-squared is a version/re-naming of R-squared.
If it helps, consider the square root, so that 4%
translates to (some sort of partial-) r= 0.20.

--
Rich Ulrich

[Bruce Weaver]

But note that SPSS reports (and the OP is talking about) partial eta-squared, not eta-squared. Rich, I know you know the difference. But in case other readers do not, there is a nice explanation here:

http://jalt.org/test/bro_28.htm

HTH.