Generalised-linear models and data input

Question

I am working with choice test data from an Ecological experiment and I need to decide if I am doing a GLM or GEE. I am running SPSS but also not sure about data input with these tests. Individuals were given a choice for each experiment (control or treatment/ 0 or 1), and there were multiple experiments.

Answer 1

Whether you should use a Generalized Linear Model (GLM) or a Generalized Estimating Equations (GEE) in your ecological experiment with choice test data depends on the nature of your data and the goals of your study. In multiple experiments, people choose between a control option and a treatment option. This means that both methods are useful. GLM is good for modeling binary responses to guess choices based on the conditions of an experiment, and you can add more predictors if you want to. But if the same subjects or groups are measured more than once in different experiments, there may be a correlation. In this case, GEE should be used because it takes correlations into account and gives more accurate parameter values. To choose between these two methods, you need to carefully look at the correlation structure in your data and make sure it fits with the goals of your study. When you put data into SPSS, you can make it easier for either GLM or GEE analysis by setting up your data so that each row represents a person's choice and putting important predictor variables in the same column. Consulting with a statistician who is familiar with the details of your study can help you choose the best way for your analysis.