NIPALS Path Modelling with Latent Variables

Noonan, R. & Wold, H. 1977. NIPALS Path Modelling with Latent Variables Analysing School Survey Data Using Nonlinear Iterative PArtial Least Squares. Scand. J. educ. Res. 21, 33‐61. NIPALS (Non‐linear Iterative PArtial Least Squares) path modelling with latent variables provides ways of treating the 'soft' data typical of the social sciences. In low‐information, non‐experimental situations, the NIPALS family of methods can be used to estimate parameters of variables measured using multiple indicators in accordance with a causal path model. To illustrate the use of NIPALS procedures, some simple NIPALS models are analysed and compared with traditional models. IEA school survey data are used. NIPALS path modelling with latent variables is seen to provide a useful alternative to the traditional variance‐added approaches to the analysis of school survey data. 1The present article constitutes a partial report of the project Causal Models in Low Information Situations under the leadership of Professor Herman Wold. The project was supported by the Tercentenary Fund of the National Bank of Sweden. Notes 1The present article constitutes a partial report of the project Causal Models in Low Information Situations under the leadership of Professor Herman Wold. The project was supported by the Tercentenary Fund of the National Bank of Sweden.