Survey of experts on the predictive validity of ordinary least squares multiple regression models
Ordinary least squares multiple (OLS) regression analysis is a method commonly used in the social sciences for estimating causal models from non-experimental, often survey derived, empirical data. While the authors of such models typically have the objective of deriving quantitative descriptions of relationships, this questionnaire is concerned with the scientific question of whether the models have predictive validity. That is, whether OLS models provide out-of-sample forecasts that are more accurate—have smaller errors—than forecasts from models developed using simpler alternative methods using the same data or a part thereof.
We welcome your participation in this questionnaire. Click here to complete.
Professor Kesten Green, University of South Australia