Step 3:. When possible, assure confidentiality to each participant, explaining the system by which this is accomplished.

Individuals are more likely to respond to a questionnaire - and to be candid - if they know that their name will not be connected to the results. In some cases, it is adequate simply to have people fill out a questionnaire anonymously. However, the quality of many evaluations depends on making links among several sets of data. For example, you might want to track how teachers' opinions change over time or connect these opinions to those of their students. Also, you need to be able to remind participants who do not respond in a timely fashion.

Tracking questionnaires requires setting up an ID code for each participant for use during the data-gathering and analysis phases (and, in the case of computer administration, a secure log-in system). All data coming from an individual participant should have that person's ID code on it. Eventually, the individual's name can be dispensed with, and all the data can be dealt with in terms of identity codes.

There are several ways to assure participants that their responses will be confidential. You can briefly explain that all participants will be known only by their IDs and that their names will never be used in the presentation of data. You also can emphasize (if appropriate) that the analysis results will aggregate the responses of groups of individuals.

If necessary, solicit the advice of a human subjects committee. Most organizations that carry out research projects with human subjects have such committees.