4.2.5.0 SUMMARY The unit explained the concept of autocorrelation at first order, its possible causes and detection (with particular interest on the first-order autoregressive autocorrelation, denoted by AR (1)) using Durbin-Watson test. 4.2.4.0 CONCLUSION In this unit, autocorrelation is statistically explained as a random process that measures the linear correlation between values of the process at different times, as a function of time or of the time lag. The significances of autocorrelation for OLS are shown to be comparable to those of heteroscedasticity and have two forms of occurrences, which could either be positive or negative. Students are advised to use the further reading materials to look at more autocorrelation techniques and study more on the correlation between the error terms arising in time series data as indicated in the introduction section of this unit.