Hi all
I am doing one study to see the effect of Racim on oral health related quality of life (OHIP-outcome). I need some help. While building up the model i added interaction term Racism*gender and I am getting female exposed to racism had higher odds (AOR:1.72) of having Bad OHIP vs males experiencing no racism. By changing the reference category (AOR for women experiency racism :1.26 vs males expereincing racism)





However when i am doing stratified analysis- i am getting males with racism having bad OHIP vs males expereiencing no racism AOR: 1.71 (95% CI: 1.20-2.86).

For females the same scenario AOR: 1.49(95%CI: 1.04-2.14) for females experiencing racism vs female experiencing no racism

But when i am comparing stratified results with interaction; as per stratification males had higher odds ratio compared to females while interaction giving me higher odds ratio for females. Why I am getting different results?

At univariate females had 1.2 OR for OHIP but racism is more common among males (OR:1.57)

On the basis of above output which model i should opt: interaction or stratified?