Kwanho Kim posted on Thursday, November 09, 2017 - 5:43 pm
For explanation, please imagine that y is attitude towards President Trump, x1 is the amount of Trump-related news, and x2 is the amount of Trump-related gossip news (in certain period; aggregated into a variable).
My goal is estimating the impact of each of x1 and x2 on y; I want to show that exposure to Trump-related gossip news has distinguishable effects on the attitude towrds Trump, when exposure to Trump-related news can affect the attitude towards Trump also.
To do this, I am trying to fit a model like:
y ON x1 x2 x2 ON x1
All variables can be observed or latent (I did not decide yet).
However, I expect there will be a collinearity issue between x2 and x1 due to nature of variables.
Could you please tell me whether potential collinearity can cause problem in this case? If so, what is your recommendation for dealing with the problem?