How do I correct for autocorrelation I will start by saying

How do I correct for autocorrelation. I will start by saying I would like to do it in either Excel, or SPSS.

I do not understand how to correct for autocorrelation. One website easily describes how to do it in excel however I could not find any sources to verify this:

https://analysights.wordpress.com/ta...crosoft-excel/

This is the site that explains how to correct, however my question is after you lag the dependent variables, why does it now become the new independent variable?
More importantly, it suggest that when I re-run the regression (simultaneously rerunning the Durbin-Watson test) that I will keep the dependent variable the same (Wage), and run it against the new independent variable \"Lag1Wage\".

I do not understand why I run the new lagged independent variable against the dependent variable and not against the \"time\" as I did in the initial regression.

Thanks

Solution

Autocorrelation is corrected by including lagged versions of the response variable as explanatory variables.

For example, suppose the following model has autocorrelation of order 1.

yt=b0+b1t

To correct it, you would refit the model with yt-1 as an explanatory variable:

yt=b0+b1t+b2yt-1