Monitoring process and non-sampling errors control in IT-LFS sample survey

Abstract

This study presents strategies to minimise non-sampling errors in the IT Labour Force Survey. Emphasis is placed on the rigorous monitoring of interviews and processing data to ensure the reliability and accuracy of the collected information. Ongoing data analysis is conducted to promptly identify and address inconsistencies or biases, thereby enhancing data quality. In the context of the LFS, this takes on a very high value due to the complexity of the defining framework and the breadth of the concepts involved.

Several strategies are developed to ensure these objectives: i) instructional notes and ad hoc training sessions are systematically implemented to enhance the proficiency of interviewers and ensure standardised data collection procedures; ii) real-time remote monitoring of CATI interviews: listening to the conversation between the interviewer and respondent, as well as viewing the interviewer’s screen in real-time, through remote tools; iii) on-site monitoring: ISTAT staff periodically visit the interview sites to observe interviews, ensuring adherence to quality standards directly; iv) post-interview monitoring and data analysis to evaluate the logical flow of the questionnaire as well as the consistency of the collected information with the data time series.

To assess the impact of interviewers on the quality of survey results, a synthetic indicator was developed to highlight areas significantly influenced by interviewer effects (as measured by the intraclass correlation coefficient), support corrective measures through evaluations at the individual-interviewer level, and plan targeted training activities.

These strategies help identify and correct potential sources of non-sampling error in relevant process parameters and ensure a thorough evaluation of both methodological and content-related aspects of the interviews, enabling timely interventions when necessary.