the FBI). The third column is the number of offenses known to police from the UCR (as reported in Jamieson and Flanagan, 1989:510). The last column computes the implied rate of reporting by comparing columns 2 and 3. The two estimates are fairly close for rape. For robbery, we would expect the reporting rates implied by comparing NCS to UCR data to be higher than that in the NCS itself, because business crimes are included in UCR. However, the opposite appears to be true. Since we need to use UCR data on the percentage cleared by arrest, we are using the last column of Table 18 to estimate the percentage of robberies reported to police. The UCR data do not include simple assaults, which apparently have a lower reporting rate. Thus, we have used the lower NCS estimate of 39.5 percent for simple assaults.

Based on Table 18 and estimates of the "clearance" rate by arrest and the final disposition of the case, we can estimate the percentage of victimizations that eventually lead to each stage of the criminal justice process. This is done in Table 19.

Combining the criminal justice processing costs per offense known to police from Table 17 with the estimate of offenses reported to police from Table 18 yields an estimated criminal justice processing cost per offense, shown in Table 20. Processing costs are estimated to be $5,750 for murder, $1,150 per rape, and $500 each for robbery and assault. These estimates are averaged over all offenses (including attempts), even those not reported to police.

In 1986, an estimated $991 million was spent by all levels of government on indigent defense, whereas the average cost per