Non-linear structural equation modeling

• Research in the social sciences often includes hypotheses concerning interactive or nonlinear effects on a given outcome latent variable. When it comes to estimating such effects, however, there is lack of consensus on how to do so properly, particularly when performing structural equation modeling (SEM). A plethora of methods have been proposed and discussed, including those de-scribed in Algina and Moulder (2001), Jaccard and Wan (1995), Joreskog and yang (1996,1997), Yang-Jonsson (1997,1998), Klein and Moosbrugger (2000), Klein and Muthen (2002), Marsh,Wen, and Hau (2004), Ping (1996a, 1996b), Schumacker and Marcoulides (1998), and Wall and Amemiya (2001, 2003). Most approaches to latent variable interactions are based on a product indicator methodology originated by Kenny and Judd (1984) that requires a level of technical and computational sophistication that renders them quite inaccessible to the average practitioner. The focus of this chapter is on the discussion of a technically straightforward approach using latent variable scores to estimating interactive and nonlinear effects within SEM. The next section, we will first describe LVS approach in a theoretical framework and in succeeding section, the approach will demonstrate in a practical manner using an empirical data. To illustrate how latent variable scores can be used to estimate nonlinear relationship between latent variables we will use Reading comprehension model as an example.

Ämnesord

NATURVETENSKAP  -- Matematik -- Sannolikhetsteori och statistik (hsv//swe)

NATURAL SCIENCES  -- Mathematics -- Probability Theory and Statistics (hsv//eng)

Nyckelord

Non-linear Structrual Equation Modeling

Statistics

Statistik

Statistics

Statistik

Publikations- och innehållstyp

vet (ämneskategori)

kap (ämneskategori)