Correlation Analysis

This chapter summarizes different correlation statistics that can be evaluated to determine the type of association between continuous random variables. It describes the different measures of correlation between variables related to multiple linear regression model (MLRM) and the concepts of partial and semipartial correlations. The partial correlation coefficient is a measure of the linear relationship between two variables after simultaneously controlling for the effects of one or more independent variables. If the correlation coefficient is calculated controlling for a single variable, it will be a first-order partial correlation. To determine if the correlation coefficient is equal to zero, one can use the strategy of additional sum of squares through the Fisher probability F-distribution. According to the control variables used to assess the correlation between cholesterol and glucose levels, the calculated value of these correlations changes from positive correlation to small negative correlation when fasting triglyceride levels are part of the control variables.