PLS Path Modelling with Latent Variables: Analysing School Survey Data Using Partial Least Squares‐Part II1,2

Abstract

Abstract: Noonan, R. & Wold, H. 1980. PLS Path Modelling with Latent Variables: Analyzing School Survey Data Using Partial Least Squares ‐‐ Part II. Scandinavian Journal of Educational Research 24,1‐24. PLS (Partial Least Squares) path analysis with latent variables provides a powerful tool for data analysis in low‐information situations. In an attempt to assess the magnitude of the effects of a variety of regional, home, student, teacher, and school factors on variation in student cognitive and affective outcomes in science, 59 variables in 16 blocks were entered into a PLS analysis. The IEA school survey data for Sweden were used. The specification and estimation of the PLS model are discussed. It is found that the existing variation in school and teacher variables has no tendency to cause significant variation in student outcomes, which is due in part to the high level of equality achieved in the Swedish school system. Variation in teacher training, however, appears to influence variation in achievement indirectly through its influence on teaching methods.