Abstract
Most surveys are affected by nonresponse and, in the case of household surveys, this happens when individuals or households do not provide the requested information. This phenomenon needs to be considered when assessing the survey quality. R-indicators are a valuable tool to evaluate the impact of nonresponse on survey results. In this paper, the R-indicator is used to assess and analyze the representativeness of the sample for the ICT Households Survey, conducted by NIC.br. The ICT Households survey sample was originally designed to provide results for Brazil’s regions, and it is not possible to ensure the quality of the sample for estimating statistics for smaller areas such as federative units (states). R-indicator methodology is usually employed as a measure of discrepancy between the selected/planned and realised samples. Here it is used to determine whether, in the planning stage, the sample would produce estimates with good precision for unplanned domains, such as federative units. The results of the estimated indicators revealed that the respondent and planned samples of ICT Households survey can be considered representative for Brazil and major region levels. At the federative unit level, however, there is evidence of a gap between the planned and respondent samples and the target population.
Get full access to this article
View all access options for this article.