Training model and interviewer performance in the permanent population census

Abstract

Training interviewers is crucial in survey design to minimize interviewer effects, which impact non-sampling errors (partial non-response, refusals, and measurement errors) and data quality. Istat has developed expertise in training interviewer networks and continuously works on strategies to improve and harmonize its training model, adapting to societal changes and respondents' lifestyles.

This study examines the effect of interviewer training on survey performance to improve the training program. The data analyzed relate to the performance of 5,894 interviewers trained for the 2023 Italian Population Census "List" survey.

To explore the relationship between training and fieldwork performance and identify interviewer profiles, we used a statistical classification method (decision tree), considering socio-demographic and geographical factors as potential covariates.

The results represent a crucial step toward the ongoing use of statistical methods to enhance the data collection process.