So I am looking at whether the effect of rainfall on crop yield for 8 different States is similar across these States i.e. I am testing whether the slopes between rainfall and yield is significantly different from each other among the eight States. For each State, I have yield and rain data for 48 years. I ran the following code in r one with interaction between rainfall and State and other without interaction and then compared the model.
mod1 <- aov(yield ~ rainfall*state)
mod2 <- aov(yield ~ rainfall + state)
anova(mod1,mod2)
If model shows that removing the effect of interaction significantly affect the fit of the model, then I can conclude that slopes of yield vs rainfall are different for different States. So far this is ok and please let me know if I am wrong in interpreting this.
My problem is this: my rainfall data (my independent variable) collected over time for each State will not be independent of each other. In this case, will be results will be wrong? What other set of analysis I need to do if my results are wrong?
Thanks for the advice.
