Dear colleagues! I have two groups of animals which I compare using 8 day long behavioral paradigm. For each day I have two data points (day and night). Thus I have one independent factor (animal group) and 16 observations. ANOVA revealed highly significant interaction of factor*observation. What post-hoc should I use to check at which time points groups differed significantly?
I attach a graph which shows how analysed factor changes with time (x axis) in both groups (red and blue).
@Kirill Below thread on the Research Gate might be helpful to your cause.
https://www.researchgate.net/post/How_do_I_conduct_post-hoc_tests_on_a_one_way_repeated_measures_ANOVA_with_no_between-subjects_factors
Attached are my favorites. I like Bonferroni it requires only one division by calculator to do.
Best wishes, David Booth
The most popular post hoc tests are the Tukey test and Bonferroni test, Tukey test is generally used when you have a large number of comparaisons, the Bonferroni test has a good statistiscal power when the number of comparaisons is low.
There is an obvious "daytime"-effect during the first 5 day. However, it seems to vanish after day 5, just the time when the lines of the groups cross. This all looks very strage and conclusions about treatment effects will be difficult to derive or to interpret.
First I wonder if you modelled this day-to-night variataion (or why you ignored it). Then I wonder why you assume that subsequent readings are independent (they are very obviousely not; but you wrote abot ANOVA and 16 observations, what implies that each time-point was modelled as an independent group in ana ANOVA design).
It seems that the read-out increases initially at some rate and levels off into a plateau. So these would be the interesting things to test: initial rate and plateau value. I'd use a non-linear mixed model. Not sure how to model the da/night differences since they seem to be pronounces during teh first 5 day only and then seem to vanish.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA