Simple Interactive Statistical Analysis
Response Rates
Input.
Program presumes that the input consist of positive (integer) numbers without decimals.
Explanation.
Often response rates in survey research are calculated simply by dividing the number of completed interviews by the number of individuals who were selected to participate in the research. However, this method is too simplistic and does not do justice to the complexity of research design, sampling process and the practical difficulties of contacting and assessing potential survey participants. Particularly in the case of telephone interviewing (CATI) and postal surveys these difficulties might be considerable and using a simple way of response rate calculation might lead to an unrealistic estimate of response rates.
Potential respondents consist of a number of groups:
eligible and completed. These are the positive responders who complete the interview and should be in the numerator and in the denominator of the sample size calculation. Fill in an integer number in the first box.
eligible and uncooperative. The refusers, should be included in the denominator of the sample size calculation. Fill in an integer number in the second box.
ineligible and cooperative. If our research is about females then males do not form part of the sampling frame, males should not be incorporated in the sample size calculation. Often it is necessary to contact the respondent and obtain their cooperation to assess if they meet the inclusion criteria and are eligible for the research. For example, when we are interested in a particular age group, sexual preference group or cultural group this might be required. For the program add group three and four (next category) together, they are both ineligible, then fill in an integer number in the third box.
ineligible and uncooperative. Are not included in sample size calculations. For the program add group three (previous category) and four together, they are both ineligible, then fill in an integer number in the third box.
unknown. This is a difficult
group which consists of two main categories. A) People who can be contacted
but who refuse cooperation to such an extent that eligibility cannot be assessed.
B) People who cannot be contacted at all. In postal surveys it might be that
people do not reply after a number of reminders while in telephone surveys
it might be that the phone rings on the other side but is never answered.
These unknowns further consist of two group. Eligibles, they should
be included in the denominator of the sample size calculation, and ineligibles,
they should not be included in the sample size calculation. However, we do
not know the proportions of ineligibles against eligibles in the unknown
category and have to use some estimation procedure. Fill in the number of
unknowns in the fourth box. Is an integer number.
This program uses the procedure proposed by the
Council of American Survey Research Organizations (CASRO). This program divides
the unknowns between eligibles and ineligibles using the same relative
proportions as observed in the sample overall. The formulae proposed by CASRO
is:
response rate=completed/(eligible+ eligible/(eligible+ineligible)*unknown)
It should be considered that to enable a good sample size calculation to be made using the CASRO procedure as much information as possible needs to be collected from all respondents, eligibles and ineligibles alike. Some postal survey researchers ask people not to send a reply if they are not eligible. This to save on postal charges and administration costs. However, this strategy will make it impossible to ever assess response rates.
