MODIFIED STEPWISE REGRESSION MODELS ON SIMPLIFYING THE ARWU’S INDICATORS

Author's Name: Farrah Pei-Chen Chang & Liang-Yuh Ouyang
Subject Area: Social Science and Humanities
Subject Management
Section Research Paper

Keyword:

Academic Ranking of World Universities, stepwise regression model, correlation coefficient, partial correlation coefficient, coefficient of determination.


Abstract

The Academic Ranking of World Universities (ARWU) uses six indicators and their relative percentage weights to measure the academic performance of universities. The sixth indicator, the universities’ per capita performanceis a weighted average of the scores obtained in the previous five categories, divided by the number of current full-time equivalent academic staff members. However, the data sources of the number of current full-time equivalent academic staffmembers are not consistent for all participating universities, which might lead to an inconsistent comparison of the global competition.In attempt to simplify the indicators, this paper usesa stepwise regression analysis for each ranking year, and constructs stepwise regression models from 2004 to 2016. Of the constructed five models throughout the ranking years, we find three models that have the better model fitting. Furthermore, the new scoring formulas generated from the three modified stepwise regression models are all adequate to replace the original scoring formula. As it is shown in our empirical study, the three modified scoring formulas all produce very similar results when compared with the original outcomes.

Download Full Paper