I am running a multilevel binary logistic in SPSS.
The predictor variable in question is a categorical variable with four levels. In the first fixed effect table, it seems like the predictor variable is insignificant (p=.128). The table has F, df1, df2, and sig. Doesn't it mean there is no overall effect of the predictor? However, in the coefficient table, it says that two levels are significant (p=.035 and .045) among the four levels. For example, in the first table with the title "Fixed Effects", the variable "Text" is insignificant. But in the second table with the title "Fixed coeffeicients", the levels of the variable "Text=3"and "Text=4" are significant. By the way, there was a two-way interaction effect with that variable.
How can I interpret this result? Should I focus on the first table and say the predictor is insignificant? Or, since there is an interaction anyway, should I describe the significant levels of the predictor variable? Is there any book or thesis that I refer in this case? Thank you in advance.
