Say I have a categorical variable concerning the voting intention for ten parties, and I want to run two separate (logit) models for the voting intention of two parties in particular, Party A and Party B. I've thus created two different DVs: a dummy for A (Vote for A=1, All others=0) and a similar dummy for B. These are my two very simple models (dataset is the same for both models):
Model 1= logit VotePartyA EconomyIndex Sociallndex Ideology age i.gender i.education i.job i.area
Model 2= logit VotePartyB EconomyIndex Sociallndex Ideology age i.gender i.education i.job i.area
My hypothesis concerns the first two IVs, both continuous (and standardized). Party A is economically-oriented, whereas Party B is socially-oriented. Their preferred policy Index is positive and significant, but I also expect that the effect of the EconomyIndex on the vote for Party A will be greater than it will be the effect of the SocialIndex on the vote for Party B. To test this hypothesis I've run a test of equality between the coefficient of EconomicIndex for Party A (in Model 1) and the coefficient of SocialIndex for Party B (in Model 2), i.e. on Stata, after using suest, I've run:
test [Model1_PartyA]EconomyIndex=[Model2_PartyB]SocialIndex
My question is very trivial: is it correct to validate my hypothesis using a test of equality of coefficients of two different variables across two models that have the same independent variables but - clearly - differ on their DVs?
I've found a lot of examples for comparing the coefficients of the same variable across different models, or comparing different variables in the same model, but I am struggling to find something similar... So I was wondering if I messed something up, which probably I did.
Many thanks in advance!
