Hello.
I'm performing a length-weight relationship in two populations of caridean shrimps (they are from two distinct seasonal periods), the pooled data was divided into males, females, non-ovigerous females and ovigerous females. I performed the equation and obtained the b slopes of each sub-sample analyzed, but now I want to compare those slopes to know if there is any different between them. I read that I could do an ANCOVA to compare, but I'm not sure if there is another methodology. Thank you.
I think you are right and that ANCOVA is the best method for the task.
You could indeed use ANCOVA/linear regression (after checking your model assumptions, i.e. linear relationship between length and weight (in model 1), normality and homoscedasticity of residuals and independence of observations).
As observations from the same population are probably not independent, you should add population as a variable in your model.
There are two possibilities:
Model 1: model the relationship between length and weight, including an interaction with sex.
Since you are interested in the difference of length-weight slopes between sexes, you should add an interaction effect in your model.
The model looks something like this:
length ~ weight + sex + population + weight*sex
If the interaction effect is significant, the length/weight slope differs with sex. To identify the which of the four groups differ significantly from each other, you can carry out a post-hoc test on the interaction term (e.g. Tukey test). If the interaction is not significant, you cannot assume that the l/w ratio is significantly different between any of the four groups.
Model 2: use length/weight ratio as a response variable
Another way would be to use the length/weight relationship directly in a model:
length/weight ~ Sex + Population
Here, we do not need an interaction term, as the relation between the l/w ratio (which is the beta in Model 1) and the four ‘Sex’-groups is implicitly part of the model. If ‘sex’ is significant, you can again perform a post-hoc test on ‘sex’.
If you are using R, this link might be helpful: https://rcompanion.org/handbook/G_09.html
- Badges
- Science topic
- Similar topics
- Entomology
- Resistance