Methodological Note #005: How Does LAPOP Calculate Response Rates?

The organization uses handheld electronic devices to conduct interviews, with enumerators recording each attempted interview into one of the following categories: 1. Respondent willing to proceed; 2. Respondent refused; 3. Respondent unable to be interviewed (e.g., blind, deaf, or ill); 4. Unsuccessful, other reason; 5. Household unoccupied, abandoned, destroyed, or doesn't exist. The response rates are then processed for quality control reasons, with some 35,000 attempts (17% of the 210,000 conducted this round) being excluded due to quality control issues. The remaining attempts are then classified into categories using AAPOR's standard definitions, which includes four groups: interviews, non-interviews with eligible respondents, non-interviews with respondents of unknown eligibility, and non-interviews with ineligible respondents. The attempts are then mapped to disposition codes, with codes 1.0 indicating a respondent willing to proceed, 2.1 indicating a respondent refused after eligibility screening, 3.2 indicating a respondent unable to be interviewed, and 4.0 indicating a household unoccupied, abandoned, destroyed, or doesn't exist. The response rates are then broken down by completion percentage, with interviews being coded as complete if the respondent answered more than 70% of questions, and partial if the respondent answered 50-70% of questions. Cases in which a respondent began the interview but answered less than 50% of questions are coded as break-offs. The codes are then formalized into well-known categories, with attempts coded as 1.1 being completed interviews, 1.2 being partial interviews, 2.1 being refusals, 2.3 being noncontacts and known eligibility, 3.2 being noncontacts and unknown eligibility, 3.9 being unknown, other, and 4.0 being ineligibles. The response rates are then estimated, with the organization adopting a conservative strategy of assuming that all attempts with respondents of unknown eligibility were with eligible respondents. However, this approach is not used, as it is considered too aggressive. Instead, the organization attempts to estimate the proportion of such attempts likely to be eligible, but this is not done due to the lack of fine-grained demographic information. The response rates for the 2016/17 AmericasBarometer round are then calculated, with a global response rate of 20%, varying across countries from 10% to 55%. The response rates for each country are also provided. The organization uses a probability-based sampling method, with the sampling quota being used to determine eligibility. The sampling quota is used to determine whether a respondent meets the sample requirements, with the respondent being coded as eligible if they meet the quota. The response rates are then calculated using the final disposition codes, with the codes being used to determine the eligibility of the respondent. The response rates are then broken down by completion percentage, with the completion percentage being used to determine whether the interview was completed or partial. The response rates are then formalized into well-known categories, with the categories being used to determine the eligibility of the respondent. The response rates are then estimated, with the organization adopting a conservative strategy of assuming that all attempts with respondents of unknown eligibility were with eligible respondents.