Odds ratio became different direction after adding other variables in the logistic regression model

I am doing a logistic regression to test out the relationship between problematic behaviour and a number of variables e.g. self-centeredness, delinquent peers, childhood adversity and parenting.
If I put "self-centeredness" alone into the model, it is

But then, if I added other variables altogether, it becomes

My question is, how should I interpret the change of direction of the odds ratio (from 1.045>1, to 0.879 <1). Can it be explained by any statistical phenomenon? Thank you!


Less is more. Stay pure. Stay poor.
If you are just fishing through a bunch of models, yeah things are going to change if the variables are not independent. The best approach is to map out a structural causal model or path diagram based on theory to understand the relationships and changes. There could be multiple causes, but we don't know the context of your data. Historically people would say there is collinearity between variable, but this is a piss poor description. In actuality it could be variables are partially or fully mediated or even share a common effect, etc. There isn't an easy solution unless you know how variable effect each other based on theory or physiology.