I'm working on some socio demographical (Age, tenure, salary, etc...) data to predict absenteism. I have all the employee data, so i have the entire population.
When i run a descriptive analysis for my age variable per absence category, here is what i have :
abs age mean (0) 35,41259805 (1) 35,14586814 (2) 35,77493261
So i was thinking about running an anova to be sure that this difference is actually significative or not (witch seem to be the case here)
But i was told that this can be done only when the data follow a normal law. Does this also apply for census data ?