How do you test for autocorrelation?
A common method of testing for autocorrelation is the Durbin-Watson test. Statistical software such as SPSS may include the option of running the Durbin-Watson test when conducting a regression analysis. The Durbin-Watson tests produces a test statistic that ranges from 0 to 4.
What does autocorrelation tell you?
Autocorrelation refers to the degree of correlation of the same variables between two successive time intervals. It measures how the lagged version of the value of a variable is related to the original version of it in a time series. Autocorrelation, as a statistical concept, is also known as serial correlation.