multi level mixed effect ologit
Hi
I have estimated a multi-level mixed effect OLOGIT model with stata. I know that software might have different assumptions about the coefficient signs, thus I am not sure if my interpretations are consistent with stata.
1. I assume negative coefficient of a variable means higher values of that variable are more likely to be in higher category of response variable.
2. Should we interpret the random intercept variation per cluster the same? For example, negative random intercept for a cluster means that the cluster is less likely to be in lower categories of response variable. Actually, in this case my data shows the opposite.
Please let me know what you think.
I have estimated a multi-level mixed effect OLOGIT model with stata. I know that software might have different assumptions about the coefficient signs, thus I am not sure if my interpretations are consistent with stata.
1. I assume negative coefficient of a variable means higher values of that variable are more likely to be in higher category of response variable.
2. Should we interpret the random intercept variation per cluster the same? For example, negative random intercept for a cluster means that the cluster is less likely to be in lower categories of response variable. Actually, in this case my data shows the opposite.
Please let me know what you think.
