[Annabell Zentarra]

Yes, I know that and read your trobleshooting site. I found a handout by you (https://www3.nd.edu/~rwilliam/gologi...06_Handout.pdf) where you presented estimation results of a gologit2 model and the marginal effects. As I see, the marginal effects are in the opposite direction compared to the logits. I conclude, that the marginal effect is the effect to get in a lower category?

[Richard Williams]

As I see, the marginal effects are in the opposite direction compared to the logits.

I sort of forgot about that handout! But no, I don't see what you are basing that claim on.

Are you maybe comparing signs of coefficients with signs of marginal effects? The very first coefficient is 0.956 for yr89. That tells you that, on an all other things equal basic, somebody from 1989 would be more likely to be in the three higher coded categories and less likely to be in the lowest category than somebody from 1977. But if you look at all 4 values of the marginal effects for yr89 across the 4 outcomes, you see they sum to zero. Those from 1989 are less likely to strongly disagree and disagree and more likely to agree or strongly agree than are their 1977 counterparts.

[Annabell Zentarra]

Ok, that makes sense! Thank you for your help. I´ll try to fix the mtable problem tomorrow! Thanks again!