Regression – Identifying the Appropriate Test for Two Independent Variables and One Dependent Variable

Question

I have a dataset of bird foraging rates. I am interested in how two independent variables, SEASON and HABITAT, (both of which have two options: breeding/non-breeding, and urban/rural) respectively affect the rate of foraging (a continuous variable- attempts/hour). I originally considered two-way ANOVA for the test, in which case I would use a type-II ANOVA because I found no interaction of SEASON and HABITAT. However, I am working with an unbalanced design and my data is not normal so I would need a two-way non-parametric alternative. What test would be best? I also thought of a Kruskal Wallis, but from my understanding, that's not a two-way test.
Here's a contingency table to get an idea of the sample sizes I'm working with:

Answer 1

Imbalance and non-normality are not necessarily a problem to carry out an ANOVA:

• Balanced designs are more powerful, but that does not mean that data imbalance invalidates your analysis.
• The normality assumption is regarding the distribution of the residuals, not the data.