Multicollinearity in Regression Analysis: Comment

Process and outcome: gender differences in the assessment of justice

“…Our reading of the literature shows that there are empirical methods for examining multicollinearity. In particular, there is an hypothesis test for multicollinearity in a correlation matrix (Haitovsky, 1969;Rockwell, 1975). These papers present and develop a test of the null hypothesis that the determinant of the correlation matrix is zeroÐthat the matrix is singular/multicollinear.…”

Inequality in the Legal Order: Some Further Analysis and Commentary

“…Strictly speaking, multiple regression requires that the independent variables entered into a regression equation be orthogonal (independent). Seldom, if ever, however, is this assumption strictly met in actual research applications, with multiple regression proving to be quite robust even when independent variables are highly correlated (Farrar and Glauber,1967;Haitovsky, 1969). To determine if income inequality and percent non-white population are too collinear (dependent) to be entered into the same regression equation, a series of auxiliary regressions were performed.…”

Some characteristics of the Canadian audit industry*

“…This revenue is denoted by VMPL (for Value Marginal Product of Labor) and is estimated at the mean values of the relevant vadables, as follows: It may be of interest to individual firms to compare these industry averages to their own averages. Further, as the structure of the industry may change over time, repetition of this study may be a valuable avenue for future research.^Â ppendix We present in this Appendix tests proposed by Klein (1961, pp.64 and 101), Beaton and Glauber (1962), and Haitovsky (1969).…”

“…We also conducted a second test for multicoUinearity proposed by Haitovsky (1969), which is based on the Klein test but replaces the TJ, -by the partial correlation coefficients, ryj^^^icr between all pairs of explanatory variables. For the CES function: …”