Appendix C: Calculating a Response's Net Effect

What is a response's net effect, taking into account changes in the problem that would have occurred anyway (as shown in the control area), and displacement or diffusion (as shown in a nearby, similar area)? For each of the three areas–response, control and displacement/diffusion–you take pre- and postresponse measures that might show an increase or a decrease in their problem levels.

Because the three areas can have different base levels of crime, you must standardize the changes in crime from before to after the response. To do so, for each area, you divide the difference in crime by the amount of crime in the before period. The result is a proportional change in crime. The formula is as follows: (crime after–crime before)/(crime before) = proportional change in crime.

The net effect is the sum of the three proportional changes. But because we are dealing with crime and other harmful activities, we are interested in declines. We treat a decline in the problem in the response area as a negative number (since the before number is greater than the after number). Similarly, we treat a decline in the displacement/diffusion area as a negative number. (A decline in that area indicates diffusion, whereas an increase indicates displacement.) We treat a decline in the control area as a positive number, and an increase as a negative number. This ensures that if the control and response areas' problem levels change in the same direction by the same amount, the net effect will be zero (assuming no displacement or diffusion).

Table C.1 shows the sign (positive or negative) to use for each area, depending on the direction of change from before to after. Totaling these changes, using the appropriate sign, provides the response's net effect. Note that if the response area has an increase, the control area has an even greater increase, and there is no displacement or diffusion, then the net effect is negative, suggesting that the response kept the problem level lower than it would have been otherwise.