The problem of recognizing precursory seismicity rate changes consists of two steps. First, a background or normal rate must be determined and, second, anomalous rates must be recognized. In this paper, we address the first of these steps, determining a background seismicity rate. Two problems must be dealt with if a meaningful background rate is to be determined. First dependent events must be recognized and removed from consideration and, second, detection or reporting changes must be accounted for. A number of techniques are available for recognizing dependent events. We used the one developed by McNally (1976) in this work. The technique we used for recgonizing detection and reporting changes was developed by Habermann (1983). We demonstrate these techniques on seismicity from the Imperial Valley region of California. After a reasonable background rate is determined, we examine seismicity rates in this region prior to the Imperial Valley earthquake of 15 October 1979. The source area of the earthquake shows a possible precursory quiescence which started during July 1979. The time of this rate decrease is close to the time of a change in the seismic instrumentation in the Imperial Valley which makes it difficult to determine whether the decrease is real or related to detection. Several other seismicity rate changes observed in the Imperial Valley appear to be related to changes in the operation of the network there. These include quiescent periods which occur only if magnitude cutoffs are applied to the data and periods of low reporting of events with M ≧ 2.5. These observations suggest that magnitudes were not reported for some events during several time periods. All of these observations indicate that determination of detection histories is crucial for meaningful seismicity studies.