Analytical model for percentages

Hi everyone. I'm new to the forum and have a question. I'm working with a data set about the police, where entire shifts of police work was observed and coded into data. Specifically, I'm examining teamwork and collaboration amongst patrol officers. The unit of analysis is the officer him/herself. One way to look at this is to create a variable that measures the percent of observed time (eight hour shifts) that an officer performs this behavior. I find that there is a range of zero to around 40 percent, with the average somewhere around 12 percent.

I originally proposed analyzing this data with the Poisson (or negative binomial) model. The problem is that it is designed for integer counts, such as 1, 2, and 3. It will not work for percentages, which are continuous. The Poisson model does refer to rates, but these are the count in a particular period of time. In my case all the time periods are identical, so the rate is just implicit. I need to find another analytical model. Any suggestions?


Last edited: