heteroscedasticity of residuals in regression suggests that there is a. lack of independence in successive errors. b. nonnormality in the errors. c. multicollinearity among the predictors. d. nonconstant variation in the errors.