Longitudinal Design Choices: Relating Data to Analysis

This chapter discusses challenges that emerge when repeated measures are introduced to the sampling plan. The chapter distinguishes between four types of data: single measurement data (one measure per person), panel data (many people measured a few times), N = 1 time series data (one person measured many times), and N > 1 time series data (several people measured many times). The chapter also distinguishes between four different types of analysis: cross-sectional analysis, between-persons analysis, withinpersons analysis, and fixed-effects analysis. Likely, results from one type of analysis will not generalize to results from another type of analysis, and the interpretation of results strongly relies on the type of data analyzed as well as the content of the variables included. The chapter discusses the interpretation of relationships estimated through these four types of analyses and ties these to the data types that can be used. Finally, networks estimated from longitudinal data are introduced through graphical vector auto-regressive models (GVAR): in N = 1 time series data a contemporaneous network and a temporal network can be obtained, and in N > 1 longitudinal data a third between-persons network can be obtained.