My regression looks like this:
Predictors <- c("lage", "llen", "lwid", paste0("me", 1:19), paste0("art", 1:302), paste0("hal", 2:25))
MLR_Formula <- as.formula(paste("lprice ~", paste(Predictors, collapse = " + ")))
Hedonic_Regression <- lm(MLR_Formula, data = Contemporary_Art_Dataset, subset = sold == 1)
Now I'm trying to run some f-tests on the groups of dummies (me, art and hal), but there is always an 'aliased coefficients' error:
library(car)
linearHypothesis(Hedonic_Regression, paste0("me", 1:19, " = 0"))
Fehler in linearHypothesis.lm(Hedonic_Regression, paste0("me", 1:19, " = 0")) :
there are aliased coefficients in the model.
I've tried multiple things to either identify the aliased coefficients or exclude them, but nothing has worked so far:
alias_info <- alias(Hedonic_Regression)
print(alias_info)
should show aliased coefficients, however, there seem to exist none
should show linear combinations between multiple predictors, however, there seem to exist none
library(car)
NA_arts <- c( "art1", "art11", "art26", "art50", "art54", "art59", "art61", "art92", "art101", "art127", "art142", "art154", "art164", "art168", "art192", "art210", "art218", "art257", "art272", "art294")
all_arts <- paste0("art", 1:302)
valid_arts <- setdiff(all_arts, NA_arts)
hypotheses <- paste0(valid_arts, " = 0")
linearHypothesis(Hedonic_Regression, hypotheses)
manually excluded all dummies that are NA (e.g. bc they aren't included in the subset), didn't change the error message
- manually excluded one reference dummy, didn't change the error message either
For your help I'm very grateful :))