Refusal

In survey research, a refusal occurs when a request to participate in a survey is declined. In addition, some respondents who do participate can refuse to answer a particular question (sometimes called item nonre-sponse) or can refuse to finish the survey (sometimes called a partial or breakoff interview). The refusal rate is calculated as the proportion of contacted people who decline the survey request. Some researchers include partial interviews as refusals in the refusal rate calculation, others do not.

Refusals are important because they are a major type of nonresponse that can potentially introduce error in survey estimates. With refusal rates increasing both in the United States and worldwide, the reasons for refusals and how they are handled are important concerns for survey researchers.

In telephone surveys, the vast majority of refusals in surveys come shortly after the phone has been answered. As such, these nonresponders can be at the household level or at the respondent level. If the refusal occurs before the appropriate respondent within the household can be determined (either by random selection or other eligibility requirements), it is considered a household-level refusal. If the appropriate respondent has been determined and he or she is the person actually refusing the survey, it is considered a respondent-level refusal. A partial interview or breakoff is a respondent-level refusal.

For in-person and telephone surveys, interviewers may use a refusal report form to record any discernable details about the household or respondent, such as gender, age, and race. If the sample is from a client list or panel sampling frame, researchers may be able to estimate parameters related to nonresponse (e.g. demographics, past purchase behavior, or other list variables). These parameters from refusers can then be compared with the obtained sample to estimate the presence and impact of any potential nonresponse bias. Differences can be adjusted through survey weighting.

In self-administered surveys (e.g. mail survey, Internet survey), refusals make up some portion of those who do not return the questionnaire (nonrespon-ders). For most self-administered studies, researchers know very little about why they did not participate because there is no interaction with an interviewer; for example, did they never receive the questionnaire or did they receive it but refuse? Depending on how the sample was obtained or constructed, researchers may or may not know which respondents did not participate or any other details about them. If parameters from nonresponders of self-administered surveys can be determined, again they can be used to determine potential nonresponse bias and used in survey weighting adjustments.

The reasons for refusal are varied. It is often difficult for researchers to ascertain the reasons for refusals because many do little more than hang up the phone ("immediate hang ups") or simply never respond in any way in the case of mail and Internet sampling. If there is an interaction with the interviewer, refusers usually do not communicate much aside from declining. Some may cite objections due to invasion of privacy, being reached at a bad time, length of interview, topic saliency, poor past survey experience, or a belief that the request is a telemarketing effort. Others may have a language barrier that prevents them from participating effectively. If the interviewer can determine (or at least suspect) a language barrier exists, this household can be recontacted by a bilingual interviewer.

Regardless of the reasons given, often the main reason for a refusal is that the interviewer has reached the household at an inconvenient time. In most mediated surveys, interviewers are initiating the contact with the household by proactively contacting it rather than a respondent returning an interviewer's call to take the interview, usually on a toll-free number. As such, the interviewer's request is often "interrupting" something at the household level. Unless specifically asked not to, researchers will typically recontact households that refused initially and will make a second request of the household to participate at a later date in the field period. Perhaps the interviewer will reach the household at a more convenient time or even talk to another household member than originally refused, possibly resulting in a completed interview. If a household refuses a second time, it is usually considered a "final" refusal and is not contacted again.

Although the Federal Trade Commission exempted survey and opinion research from the National Do Not Call Registry guidelines, most telephone survey organizations maintain internal do-not-call lists. Therefore, if a refuser asks to be placed on the do-not-call list, this information is recorded and this household will not be contacted again by that organization. Technically, if this household number is sampled in a subsequent survey by the same organization, it should be coded as a refusal even though it was not contacted.

In self-administered surveys, nonresponders are often sent reminders or additional requests to complete the survey. This could be a second copy of the questionnaire sent in the mail (the respondent could have misplaced the first one) or, in the case of an email invitation, a link to the online survey. Again, if respondents refuse and request not to be contacted further, they do not receive reminders or future survey requests. In fact, in most online surveys the invitation to participate includes simple instructions for how to opt out of any future requests or reminders.

Because handling refusers and potential refusers is a key part of an interviewer's job, training is especially important. A successful survey completion depends, in part, on the rapport established between the interviewer and the respondent. Poorly trained interviewers with nonneuttal attitudes can hurt this relationship and lead to increased refusals. And given that the vast majority of refusals in telephone surveys occur in the first seconds of the interviewer-respondent interaction, the interviewer has very little time to develop rapport, anticipate potential barriers, and alleviate respondent objections. Refusal avoidance training focuses on how to avoid refusals by detecting respondent objections and proactively addressing them in an effort to persuade respondents to participate.

In most in-person surveys, interviewers have more time to develop rapport before the household member who opens the door refuses to cooperate. Furthermore, advance contact is more effective in in-person surveys as all sampled addresses can be mailed an advance letter. In telephone surveys of the U.S. general public only about half of the residences that are sampled can be matched to an accurate mailing address; but when this is possible sending an advance letter prior to placing the first call reduces the proportion of refusals appreciably.

In addition to learning how to disarm a potential refusal situation, interviewers can also be trained to [Page 702]"convert" initial refusals into completed interviews. This technique is called refusal conversion. Typically, sample households who have initially refused a survey request are called back by an experienced and trained interviewer. Some interviewers who are particularly adept at this type of respondent interaction may even be considered refusal conversion specialists. Converting refusals is important because it reduces nonre-sponse, saves costs, and may also reduce the potential bias it can introduce.