“…It is common in such settings to model the rows of as a user‐specified vector‐valued function of time , the th row of then being . Polynomial bases are prevalent, particularly in the foundational work of Potthoff and Roy (1964), Rao (1965), Grizzle and Allen (1969) and others, but splines (Nummi & Koskela, 2008) or other basis constructions (Izenman & Williams, 1989) could be used as well. In longitudinal studies, model () might be used when modeling profiles, while model () could be used when modeling just profile differences.…”