Statistical analysis of longitudinal data using multi-level model?

This is my first post, I'm just wondering if I can use this approach?

I want to research the opinion of 8 countries(data is a proportion), over the period of three occassions. The countries being sub-level 1, and the sub-level 2 being the time periods. I want to capture the estimated effect of the predictor time on the predicated variable Y(opnion of sample). With the intent of answering the question, Has opinion Y changed over the time period? Is there difference in how much each of the changed(slope and intercept).

My fear is in that the state of the data for the dependent variable opinion is in the data format proportions. Do I have to use logistic regression?

Thanks in advance if someone offers some input!:wave:
Hi there!

The dependent variable (the variable of interest ) is the aggregated opinion from a survey.
From the beginning were there 4 answers, categorical, agree, agree strongly, disagree strongly.
I reduced it to agree - disagree, binomial.
The variable has observable limits of 0 to 1. It's a ratio basically.

The independed variables are of the same nature, except the categorical variable of which country the record is concerning. There is also a time variable which has been centered around the value of 0( standard procedure)

I want to study how this opinion varies over time and between groups(countries). Hence, there is no purpose to view this ratio as a probability.

Hope you understood with this knowledge, if not. Let me know what was unclear and I'll try to explain further.

Should I scrap the multi-level model and go for a logistic model? Writing this as a thesis^^ Just second doubting myself alot now.
I guess I could log-transform the variables if I have problems with normality, but the initial plots give no such indication.
Yeah I was looking in to non-linear multi level models.

Talked to my professor today, im switching to individual person data and switching to a logistic regression...... Can close the thread...

Hey atleast I learned alot about normal multi-level models:D