carried out an ordered logistic regression but sample sizes were not equal, one is much larger than the other (490) compared to 224 and 219, the result for this group was non-significant could this be the result of a larger sample size? If not, are there other negatives to having such big differences in cohort sizes; in terms of validity or reliability etc
2
-
$\begingroup$ Could you give more details about your sample and model? $\endgroup$– NatWHCommented May 30, 2018 at 21:03
-
$\begingroup$ it is taken from the paper by Geraghty, Hann & Kurtev (2017) on chronic fatigue syndrome $\endgroup$– hippeaCommented May 31, 2018 at 6:30
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
With a big enough sample size, even tiny differences would be significant. Your sample sizes seem fine, as long as the model is not very complex.
As to why your results are not significant, it could be that your model is simply not very strong - look at the effect sizes.
answered May 30, 2018 at 21:15