A measure of the degree of correlation among explanatory variables in a regression analysis. This commonly occurs for nonexperimental data. Parameter estimates will lack reliability if there is a high degree of covariation between explanatory variables, and in an extreme case, it will be impossible to obtain estimates for the parameters. Multicollinearity is especially troublesome when there are few observations and small variations in the variables.