The Pratt index is a useful and practical strategy for day-to-day researchers when ordering predictors in a multiple regression analysis. The purposes of this study are to introduce and demonstrate the use of the Pratt index to assess the relative importance of predictors for a random intercept multilevel model.
Liu, Yan; Zumbo, Bruno D.; and Wu, Amery D.
"Relative Importance of Predictors in Multilevel Modeling,"
Journal of Modern Applied Statistical Methods: Vol. 13
, Article 2.
Available at: http://digitalcommons.wayne.edu/jmasm/vol13/iss1/2