E

e is a term used in the calculation of survey response rates; it represents the proportion of sampled cases with unknown eligibility that are estimated to be eligible cases.

To determine response and other outcome rates for surveys, all cases in the sample first need to be classified into one of four categories: (1) completed cases; (2) eligible cases, no interview (nonrespondents); (3) cases of unknown eligibility, no interview; and (4) not eligible cases (out of sample). Then the eligibility status of the unknown cases needs to be estimated. The proportion of unknown cases that is estimated to be nonrespondents (i.e. eligible cases with no interviews) is known as the e-rate and is represented as e in equations. For example, in the formula for Response Rate 3, according to the standards of the American Association for Public Opinion Research (AAPOR), the response rate is the number of complete interviews (I) divided by the number of complete interviews (I) plus the number of partial interviews (P), plus the number of nonrespondents due to refusals (R), noncontact (NC), and other reasons (O), plus the number of unknown known cases (unknown if household (UH) and other unknowns (UO)) times their estimated eligibility rate (e):

So if 55% of the cases in the unknown category were estimated to be eligible cases, e would be .55, and 55% of the unknown cases would appear in the base (denominator). The 45% estimated not to be eligible would be excluded from the calculation of this response rate.

In estimating e, AAPOR requires that “one must be guided by the best available scientific information” and “one must not select a proportion in order to boost the response rate.” AAPOR has documented eight general methods for estimating the eligibility rate:

• 1.
  Minimum and maximum allocations
• 2.
  The proportional allocation or CASRO method
• 3.
  Allocation based on disposition codes
• 4.
  Survival methods using either (a) the number of attempts only or (b) the number of attempts and other attributes of cases
• 5.
  Calculations of the number/proportion of eligible population compared to same in the realized sample
• 6.
  Contacting databases or information sources, such as telephone companies
• 7.
  Linking to other records, and
• 8.
  Continued interviewing, especially after the close of the data collection period

Cases of unknown eligibility are rare in some types of surveys, such as in-person, area probability samples, but are common in other surveys, such as random-digit dialing (RDD) samples, and mail and Internet samples. In RDD samples in general and 217 [Page 218]especially in RDD samples with minimal callbacks, the number of cases of unknown eligibility (due to “ring-no answer” and “busy” outcomes) will be appreciable, and as a result calculating the response rate will be notably influenced by estimates of e. The same will be the case in mail and Internet surveys because of those sampled cases for which nothing is ever heard back by the researcher.