regression analysis -- problems

regression analysis -- problems

	problem	Detect	Correct
Heteroskedasticity	Non-constant error variance (*2)	Breusche-Pagan	White-corrected standard errors
Autocorrelation	correlation among error terms	Durbin-Watson test (*1) for trend models Significance of the autocorrelations should be tested using the t-statistics for autoregressive models	Hansen method; adjusting standard errors
Multicollinearity	high correlation among Xs	if F-test significant, t-tests insignificant	dropping X variables

(*1) positive autocorrelation if DW < d₁

(*2) Conditional Heteroskedasticity
Variance of the error term is correlated with the values of the independent variables.