Refusal Rate

The refusal rate is the proportion of all potentially eligible sample cases that declined the request to be interviewed. Before calculating the refusal rate, researchers must make some decisions about how to handle the types of nonresponse in the calculation.

There are two types of refusals that are generally included in the refusal rate calculation:

• 1.
  Household-level refusals: These refusals occur before the appropriate respondent within the household can be determined (either by random selection or other eligibility requirements). That is, the person refusing may or may not be the selected respondent within that household.
• 2.
  Respondent-level refusals: These are refusals that occur after the appropriate respondent has been determined. That is, the selected respondent has been identified and that individual is the one actually refusing.

Partial interviews (“partials“), or breakoffs, can be classified as another type of respondent-level refusal. Partials are when respondents begin an interview but do not finish it. They may be refusing to complete the questionnaire, or they may have other circumstances that interfere with finishing it (e.g. they initially ran out of time and were not recontacted to finish the interview during the field period). Depending on how many questions were answered, researchers may treat some breakoffs as completed interviews and others as respondent refusals or noncontacts as a final disposition.

There are three generally accepted refusal rate calculations that are included in the American Association for Public Opinion Research's 2006 Standard Definitions. In all three refusal rate calculations, the numerator is simply the total number of refusals (household- and respondent-level). But each of the calculations differs according to which dispositions or call outcomes are included in the denominator.

Refusal Rate 1. The denominator includes all possibly eligible cases still in the sample frame, regardless of whether or not the eligibility of the case could be determined. Thus, the denominator includes completed interviews, refusals, noncontacts (e.g. callbacks), and cases of unknown eligibility. This rate is the most conservative of the three refusal rate calculations.

Refusal Rate 2. The denominator is similar to Refusal Rate 1 except it uses only a proportion of the unknown cases, not all of them. This proportion is an estimate (e) of how many unknown cases would likely be eligible (e.g. is a household and meets other survey criteria). In estimating this proportion, researchers must be guided by the best available [Page 708]information on what share of eligible cases is likely among the unknown eligibility cases.

Refusal Rate 3. The denominator excludes all cases of unknown eligibility. In other words, this rate is the proportion of refusals among those sample cases known to be eligible sample units for the survey. Noncontact and other rates can be calculated in a manner similar to this rate so that, when summed, all will equal the total nonresponse rate.

The survey refusal rate has been increasing in the past 20 years in the United States and elsewhere in the world. This has happened despite the efforts of many survey researchers around the world. There is no easy or apparent solution to this decline other than to continue to investigate its causes and possible solutions.

...