After a linear forecasting model is found for a time series, if ...

[wrbasek0]

After a linear forecasting model is found for a time series, if the Durbin-Watson statistic is less than dL this means that:
 
  A) no positive autocorrelation exists, the linear model is adequate.
  B) positive autocorrelation exists, a nonlinear model should be tried.
  C) no positive autocorrelation exists, a nonlinear model should be tried.
  D) positive autocorrelation exists, the linear model is adequate.

Question 2

Explain why it is important to construct scatter plots prior to conducting regression analysis.
 
  What will be an ideal response?

[cat123]

Answer to Question 1

B

Answer to Question 2

Simple linear regression, which includes the word linear in its name, assumes there is linear relationship between x and y. If there is a curvilinear relation of some type between x and y this will not be detected by the correlation of the regression results. It may be that x and y have a very strong relationship, but if it's curved, the strength of relation will be underestimated. It is even possible for the correlation to equal 0 when there is a strong curvilinear relation. So a scatter plot should be constructed prior to doing regression analysis to check for curvature in the relationship. If this is not done, the curvature should actually show up in the residual plot, but it would have been better to check on the front end rather than waste time on fitting a linear model to curved data.

[wrbasek0]

Wow, this really help

[kusterl]

Thanks for the timely response, appreciate it