Survey Response Rates
From ICE Primer: A Tobacco Control Research Methodology Primer
Since methods of recruitment differ, the types of outcomes will differ accordingly, and so will the construction of outcome rates which describe the success of recruitment. AAPOR and CASRO have developed standard response rate calculations in an attempt to make it easier for research teams and organizations to compare results.
(Links to AAPOR and CASRO sites.)
Some investigators find it useful to distinguish contact rates, eligibility rates, cooperation rates, and overall response rates. Separate sets of rates may be computed for recruitment and for follow-up, or for households and for individuals.
An important principle to follow when providing outcome rates and response rates is to make the formulae for producing them clear and complete.
For example, suppose we wish to enumerate and determine the smoking status of all members of a sample of households, by asking a household informant face-to-face, and then select one adult smoker in each sampled household (if there is such a member) to interview. Then, at the household level, we could define:
- household contact rate = the number of households contacted (where the interviewers spoke with someone), divided by the number of households approached (doors knocked upon)
- household enumeration rate = the number of households in which the members were enumerated, divided by the number of households contacted
In reporting outcome rates, it is important to show the numbers in the numerator and denominator of each rate.
The enumeration rate here is a kind of household cooperation rate. The overall household response rate might be taken to be the product of the household contact rate and the household enumeration rate.
- household eligibility rate = the number of households containing a member eligible for the survey, divided by the number of households in which the members were enumerated
For example, if only smokers are eligible for the survey, the household eligibility rate is the proportion of enumerated households which contain smokers.
At the individual level, we could define:
- individual response rate (conditional on selection) = number of individuals interviewed, divided by number of individuals selected for interview following household enumeration
Then an unconditional individual response rate might be considered to be the product of the conditional rate and the household response rate, if it is necessary to compare rates over different protocols.
An example of a different protocol would be that of a telephone survey for which in a contacted (and engaged-household an eligible respondent is selected if any exist. In that case the individual response rate is often taken to be the number of completed interviews divided by an estimate of the number of eligible individuals who could have been obtained under this protocol from the telephone numbers attempted, had the calls been answered and cooperation obtained. The individual response rate is the product of:
- estimated eligible household contact rate = number of eligible households contacted, divided by an estimate of the number of eligible households in the telephone sample
- eligible individual cooperation rate = number of completed interviews, divided by number of eligible households contacted
The estimate of the number of eligible households in the telephone sample would be the number of known eligible households plus an estimate of the number of eligible households among those of unknown eligibility.
