Hi guys,
I have an experiemnt where three groups of fish were used, two of them were treated with two doses of virus, while the thrid is a control group. I want to compare the means of gene expression levels between the three groups at different time points after challenge and I am going to use ANOVA..
I am dealing with 10 time points, where 9 fish were sampled/ each time point (3 fish /group).
My question is: can I test the normality and equality of variance on the whole data set (values from all time points)at the sam time?or shall I do it on each time point independently from the other time points?
Does the small number (3fish /group)obligate me to use non-parametric tests?
Thanks in advance
Why don't you model the time-course of gene expression using a regression model?? Should be logistic, peaking, or a combination thereof. This would allow you to test meaningful hypotheses. IMO, comparing expression levels between groups separately for each time point is nonsense. I strongly suggest you involve a local statistician, because the non-linear regression models you might need are not always simple to fit, and testing interesting hypotheses can be challenging.
[it can be hard, possibly even impossible to get answers to the right questions -- but that should never be an excuse to ask the wrong questions, just because these could be answered more easily!]
You can use factorial analysis with tow factors doses of virus factor , and time is factor by this you can compare between doses and time via interaction , then use regression
I think you can calculate mean for different time points and apply kruskal wallis test. Sample size is too small for ANOVA.
Ms ceren, I think the kruskal is one way anova and he have to factor , if he use kruskal he can't compare sample with time . Two why anova is the ideal solution .
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA