*******PRQ R&R MODELS*****

Models with mental health as DV (coded: +1 better compared to before 2020, 0 same, -1 worse) 
*vote choice is coded so that 1=voted trump, 0=biden, rest are dropped 
*polarization is coded so that people who think polarization is much greater than in the past get high scores (1-5 scale) 

ologit mentalhealth trump2020 [pweight=weight]
est store m1

ologit mentalhealth trump2020  polarizationperception partisanship7 libcon  male black hispanic asian nativeam age income educationlevel polinterest unemployed i.inputstate [pweight=weight]
est store m2
margins , at(trump2020=(0(1)1)) predict(outcome(1))

ologit mentalhealth trump2020  partisanship7 libcon  male black hispanic asian nativeam age income educationlevel polinterest unemployed i.inputstate [pweight=weight]
est store m3

ologit generalhealth trump2020 [pweight=weight]
est store m4

ologit generalhealth trump2020  polarizationperception partisanship7 libcon  male black hispanic asian nativeam age income educationlevel polinterest unemployed i.inputstate [pweight=weight]
est store m5
margins , at(trump2020=(0(1)1)) predict(outcome(1))

ologit generalhealth trump2020  partisanship7 libcon  male black hispanic asian nativeam age income educationlevel polinterest unemployed i.inputstate [pweight=weight]
est store m6

estout m1 m2 m3 m4 m5 m6 using healthpost2020_sixmodelsPRQ.xls, cells(b(star fmt(3)) se) stats(pr2 ll chi2 N)



*Models where biden won agreement measure is used in place of vote choice (it is coded so that +1=agree biden won, 0=neither, -1=disagree) 
*This measure is highly correlated with Trump vote choice (-.79), such that those who voted for Trump are much less likely to say that Biden won 
corr bidenwon_w1_3point trump2020

ologit mentalhealth bidenwon_w1_3point [pweight=weight]
est store m1b

ologit mentalhealth bidenwon_w1_3point  polarizationperception partisanship7 libcon male black hispanic asian nativeam age income educationlevel polinterest unemployed i.inputstate [pweight=weight]
est store m2b
margins , at(bidenwon_w2_3pt=(-1(1)1)) predict(outcome(1))

ologit mentalhealth bidenwon_w1_3point  partisanship7 libcon male black hispanic asian nativeam age income educationlevel polinterest unemployed i.inputstate [pweight=weight]
est store m3b

ologit generalhealth bidenwon_w1_3point [pweight=weight]
est store m4b

ologit generalhealth bidenwon_w1_3point  polarizationperception partisanship7 libcon male black hispanic asian nativeam age income educationlevel polinterest unemployed i.inputstate [pweight=weight]
est store m5b

ologit generalhealth bidenwon_w1_3point  partisanship7 libcon male black hispanic asian nativeam age income educationlevel polinterest unemployed i.inputstate [pweight=weight]
est store m6b

estout m1b m2b m3b m4b m5b m6b using healthpost2020_sixmodelsbidenwinv2PRQ.xls, cells(b(star fmt(3)) se) stats(pr2 ll chi2 N)


*Robustness Online Appendix Material 
ologit mentalhealth i.trump2020##trumpwonstate polarizationperception partisanship7 libcon  male black hispanic asian nativeam age income educationlevel polinterest unemployed [pweight=weight]
margins , at(trump2020=(0(1)1)trumpwonstate=(0(1)1)) predict(outcome(1))
marginsplot

ologit mentalhealth c.trumpmarginstate##i.trump2020 polarizationperception  partisanship7 libcon male black hispanic asian nativeam age income educationlevel polinterest unemployed [pweight=weight]
margins , at(trumpmarginstate=(-50(10)40)trump2020=(0(1)1)) predict(outcome(1))
marginsplot