A Comparison of Partial Least Squares (PLS) and Ordinary Least Squares (OLS) regressions in predicting of couples mental health based on their communicational patterns

By H. Farahani, A. Rahiminezhad, Laleh Same, Kobra immannezhad. 2010

93 citations recorded by the source corpus.

Abstract Purpose: The Goal of this study is to compare two regression models (PLS and OLS) in order to determine which one is more robust in a study of predicting couples mental health Method: The data used in this study are those driven from the study conducted by Imaninezhed te.al (2009). Total sample was 100 couples, the missing data were 10% and VIF ≥ 10 and low tolerance) Result: In the result of the OLS regression, R 2 is 6.4% (p = 0.001) and R 2 of regression model with centralized data is 10. 4% and adjusted R 2 regression model with centralized data is 10.4% and adjusted R 2 9.3% (p = 0.001.) In the PLS regression two components yields, R 2 and predicted R 2 were 70% and 49.4% respectively Conclusion: these findings indicated that the PLS model provides much more stable results than the OLS model when sample size is small and there are data missing values and multicollinearity.

Subfields: Computer Networks and Communications

View Paper: https://www.semanticscholar.org/paper/35ad64c1b2bd9d5667934018023e882fa803ea5c