Troubleshooting
Problem
What is the difference between "Pearson Correlation" and "Coefficient Correlations" when doing a linear regression analysis in SPSS? The procedure is as follows: Analyze > Regression > Linear. In the Statistics option, check " Covariance matrix" in Regression Coefficients and "Descriptives" -> Continue, then OK. The output contains two correlation matrices. The first appears to be just correlations among the variables. What is the second?
Resolving The Problem
The correlation matrix in the descriptive statistics is the same thing we'd get if we were to request Pearson correlations in a procedure such as CORRELATIONS (assuming the same missing data treatment, if there are any data missing). It's just the correlations among the variables involved in the regression, including the dependent variable.
The Coefficient Correlations table contains the covariance and correlation matrices of the estimated regression coefficients (excluding the constant or intercept term). These matrices are helpful in assessing the extent to which collinearity among predictors may cause problems in estimation of the regression coefficients.
The covariance matrix is also useful if we want to hand-compute tests of linear combinations of regression coefficients (though this can be done more easily using the LMATRIX subcommand in the GLM/UNIANOVA procedures)
Related Information
Historical Number
65733
Was this topic helpful?
Document Information
Modified date:
16 April 2020
UID
swg21477816