Explaining development in dependent variable with logistic regression


I want to explain an increase in volunteering (dependent variable) by using a set of determinants. I'm using odds ratios.

In the table below, you see 'year' in the first row. This shows that people are more likely to volunteer in 2020 than in 2010 when controlling for gender and age (OR = 1.96). I was told that when the OR of 'year' decreases when a variable is added, it means that this variable explains (a part of) the increase in volunteering.

The OR of 'year' changes slightly each time I add a determinant. However, every OR of 'year' falls within the confidence interval of the OR of 'year' in Model 2. Does this mean that I can't explain anything based on this analysis? Because the OR of 'year' does not decrease significantly?



Active Member
overall its good that the effect is consistent across these models. What do you mean by 'can't explain anything'. The model explains something or you wouldn't have p-values < 0.05.


Less is more. Stay pure. Stay poor.
Agree consistency is good. You could do a goodness of fit model. But truly you cannot make any comments on years changing unless you truly know the data generating process. what you were told is likely malarkey, since there are multiple reasons for a variable to change including mediation, collider bias, etc.