Fitting Curves with Periodic and Nonperiodic Trends and Their Interactions with Intensive Longitudinal Data

Abstract

In longitudinal series with only a few waves of measurement, growth or change over time can often be well estimated by non-complex time-graded polynomials. However, with more intensive schedules of measurement, fluctuation patterns tend to be more complex, and modeling them in a parsimonious and flexible way becomes tough. Looking upon ideas from functional data analysis, the authors prove how to build multilevel regression models for intensive longitudinal data that manifest a mix of periodic and nonperiodic trends. In this chapter, they apply the multilevel regression models characterized by Walls et al., but in a different approach. They account for complex patterns of change in intensive longitudinal data by a mix of period effects, nonperiodic effects, and autocorrelated noise.