I'm writing my thesis on the effects of tobacco control policies on tobacco cognitions. One of the cognitions is pro-smoking health beliefs and is a dependent variable in a binary generlized estmimation equation.
Descriptives
Over time, that is after policy implementation, the percentage of people reporting pro-tobacco cognitions decreases (31.4% pre-policy implementation vs 21.1% post-policy implementation).
GEE
The results of the binary GEE show a significant increase in odds for reporting pro smoking health beliefs. (OR=1.84 95%CI=1.19-2.83)
My question is, how come that the descriptives show a strong decrease oafter policy implementation while the GEE suggest the opposite.
Thank you for reading my question.
Kind regards,
Eli
