Graphical Method:

Although the true values of the error terms are not available to the researcher, the estimated values of the same terms are their proxies and can be used. Usually these residuals are generated automatically  after an OLS  estimation  by most statistical software packages. Very often a visual examination  of the residuals gives us an indication of the presence of autocorrelation. One can simply plot  the estimated residuals, against time, also called the time  sequenceplot. If autocorrelation exists this will usually show up  as several small (large) followed by several large (small) errors. That is, the consecutive values of the residuals will be highly correlated.

Alternatively we can also plot the values of,  against .  If this shows up as a linear pattern then there is autocorrelation. A positive relationship implies positive autocorrelation and vice versa. This  is arguably a crude method of examination. The following methods are statistically more powerful in detecting autocorrelation.