Noncooperation Rate

Noncooperation occurs when a research unit is able to cooperate but clearly demonstrates that it will not take required steps to complete the research process. The [Page 518]noncooperation rate compares the number of research units that refuse to cooperate to the number of all potentially eligible units. Noncooperation, along with noncontacts, compromises the majority of survey non-response. In survey research, noncooperation often takes three basic forms:

• 1.
  Household refusals refer to the refusals that occur shortly after the household has been requested to participate in the research and before a designated respondent has been selected from within the household. The reasons of household refusals often involve “Not interested,” “Don't have time,” and “Don't do surveys.” No comment hang-ups during a telephone interviewer's introduction of the survey often occur, especially when the interviewer is speaking English and the respondent lives in a non-English-speaking household.
• 2.
  Respondent refusals refer to a refusal to participate by the designated respondent selected within the household. Respondent refusals might be harder to convert, since the refusal was given by the person who should be interviewed and not by a gatekeeper within the household.
• 3.
  Breakoffs, which are a form of partial completion, refer to instances in which the respondent does not continue through the major part of the questionnaire. Breakoffs are different from other partial interviews because the proportion of questions completed in other partials may be considered as sufficient response, depending on the definition of response rate used. The standard of being a sufficient partial response is pre-determined by the researchers; the researcher should always provide a clear definition. For example, legitimate partial completions might be defined as cases with 50% to 94% of the questions answered. Breakoffs are considered as noncooperation, similar to refusals, as some of these are respondents merely hang up on the interviewer after questioning has started without saying anything more.

The most common way to calculate noncooperation rate (NCR) is to use the proportion of all cases in which a sample unit refuses to complete an interview out of all potentially eligible cases; this is comparable to a refusal rate. The numerator includes cases in which the household refuses to start the questionnaire, or the identified respondent refuses to start, or the identified respondent refuses to complete the interview. The denominator is the number of all eligible cases, as indicated in the following formula:

NCRl is the most conservative approach to calculating the noncooperation rate.

Other versions of the noncooperation rate differ in the composition of the denominator. The following formula (NCR2) includes estimates of the proportion of cases of unknown eligibility that actually are eligible. By estimating such a proportion (i.e. e), researchers aim to make a more precise computation of all potentially eligible units. However, the estimation of e must be guided by the best available scientific information on what share the eligible cases make among the unknown cases, and one must not select a proportion simply in order to decrease the noncooperation rate. The basis for the estimate must be explicitly stated and detailed.

A third type of noncooperation rate calculation (NCR3) discards all cases of unknown eligibility. It means either a special case of NCR2, in which e is assumed to be zero (i.e. that there are no eligible cases among the cases of unknown eligibility) or the rare case in which there are no cases of unknown eligibility. This formula generates the maximum nonresponse rate, since the denominator is the smallest among the three

...