I tied to make make a "linearHypothesis" to test joint significance with the "car" package in R. However I got the error massage "there are aliased coefficients in the model."
My regression runs over 6 periods and includes time-dummies, interaction terms of time-dummies and the treatment group and individual-dummies (to cover fix-effects).
I read that most comments suspect this error message occurs because of perfect collinearity. What are 'aliased coefficients'?
However I do not see why my variables should be perfectly correlated. (Moreover I thought that lm() automatically omit variables that are perfectly correlated). As the last answer to the link above indicates, the dummy variables (probably especially the individual-dummies)could be the problem. Does anybody know if "linearHypothesis" fails to omit one dummy as basis? The mentioned answer recommends to "remove one dummy manually". How do I do that when I have more than 100 dummies? Before I just made sure that the variable was of class factor and R considered it like dummy variables automatically. Now I would need to generate dummy variables for all entries and then impute them (except one) in the regression, as far as I understand. However this would become a very long regression code when mentioning so many dummy variables.
