Which outcome evaluation methods can capture the scope of the change that is likely to occur?
Many outcome evaluation measures are relatively crude, which means that a large percentage of the intended audience (sometimes an unrealistically large percentage) must make a change before it can be measured. If this is the case, the evaluation is said to “lack statistical power.” For example, a public survey of 1,000 people has a margin of error of about 3 percent. In other words, if 50 percent of the survey respondents said they engage in a particular behavior, in all likelihood somewhere between 47 percent and 53 percent of the population represented by the respondents actually engages in the behavior. Therefore, you can conclude that a statistically significant change has occurred only if there is a change of 5 or more percentage points. It may be unreasonable to expect such a large change, and budgetary constraints may force you to measure outcomes by surveying the general population when your intended audience is only a small proportion of the population. • Which aspects of the
