I am doing some research on the employment status of women in Pakistan and want to know the socioeconomic and demographic determinants. I have selected one of the variables (type of residence, which is a dummy i.e 1 for "rural"and 0 for "urban") but in the descriptive statistics employment ratio for rural there are more than urban which may be caused by the logic that Pakistan is an agri-based economy and the majority of the people work in fields related to agri-work. But in logistic regression estimates the estimated coefficient is negative which shows the negative effect on employment for rural. See results attached. Could anyone guide me on how to resolve this problem or if it is correct the is there any logic to make sense of it?