Dear Stata users,
I am working with survey data on attitudes towards specific population subgroups, for simplicity let's say high-skilled immigrants and low-skilled immigrants.
Each respondent answers questions regarding his/her attitudes towards both subgroups. I have an outcome variable (out) where respondents have to specify their attitudes towards each immigrant group separately on a scale from 1 to 10 - one outcome variable measuring attitudes towards high-skilled immigrants (out_high) and one outcome variable measuring attitudes towards low-skilled immigrants (out_low).
Similarly, I have different indicator questions (ind) that capture potential concerns regarding immigrants, for example whether the respondent thinks that immigrants take away jobs (ind_jobs), that they undermine the host country's norms (ind_norms), etc. - again asked separately for high- and low-skilled immigrants.
I want to analyze if different concerns affect attitudes towards the two immigrant groups in different ways. First, I standardized outcome and indicator variables. Next, I run two separate regressions. For model 1 I only use outcome and indicator questions that ask for high-skilled immigrants. For model 2, I only use outcome and indicator questions that ask for low-skilled immigrants:
reg out_high ind_jobs_high ind_norms_high
reg out_low ind_jobs_low ind_norms_low
I would like to compare coefficients between the two models, to see whether the association between the different concerns and overall attitudes is stronger/weaker when considering a specific immigrant group.
I thought about the -suest- command, but it does not seem to work as both dependent and independent variables are different.
Anyone knows how to do it?
Thanks!
I am working with survey data on attitudes towards specific population subgroups, for simplicity let's say high-skilled immigrants and low-skilled immigrants.
Each respondent answers questions regarding his/her attitudes towards both subgroups. I have an outcome variable (out) where respondents have to specify their attitudes towards each immigrant group separately on a scale from 1 to 10 - one outcome variable measuring attitudes towards high-skilled immigrants (out_high) and one outcome variable measuring attitudes towards low-skilled immigrants (out_low).
Similarly, I have different indicator questions (ind) that capture potential concerns regarding immigrants, for example whether the respondent thinks that immigrants take away jobs (ind_jobs), that they undermine the host country's norms (ind_norms), etc. - again asked separately for high- and low-skilled immigrants.
I want to analyze if different concerns affect attitudes towards the two immigrant groups in different ways. First, I standardized outcome and indicator variables. Next, I run two separate regressions. For model 1 I only use outcome and indicator questions that ask for high-skilled immigrants. For model 2, I only use outcome and indicator questions that ask for low-skilled immigrants:
reg out_high ind_jobs_high ind_norms_high
reg out_low ind_jobs_low ind_norms_low
I would like to compare coefficients between the two models, to see whether the association between the different concerns and overall attitudes is stronger/weaker when considering a specific immigrant group.
I thought about the -suest- command, but it does not seem to work as both dependent and independent variables are different.
Anyone knows how to do it?
Thanks!
Comment