5.5 Longitudinal/panel models

After our GUI is deployed (see Figure 5.1), the user should select Hierarchical Longitudinal Models in the top panel. Then, Figure 5.11 will be displayed, and the user can see the radio button on the left-hand side that shows the specific models inside this generic class.

The hierarchical longitudinal models tab allows for estimating models that account for within-subject correlation when the dependent variable is continuous (Normal), binary (Logit), or a count (Poisson).

The input files for hierarchical longitudinal models should first include the dependent variable, followed by the regressors and a cross-sectional identifier ($i=1,2,\dots,N$). It is not a requirement to have a balanced dataset: $T_i$ can be different for each $i$ (see Chapter 9 for technical details). Users can see templates of datasets in the folders DataSim and DataApp (see the Appendix for details) in our GitHub repository. When the dataset is uploaded, users will have a preview of it.

Users should also specify the fixed part equation and the random part equation, both in R format. If only random intercepts are required, do not enter anything in the latter part (see Figure 5.11). Users should also type the name of the cross-sectional identifier variable. The results displayed and the posterior graphs are associated with the fixed effects and covariance matrix. However, users can download the posterior chains of all posterior estimates: fixed and random effects, and the covariance matrix.

Figure 5.11: Hierarchical longitudinal models: Specification.