Differential benefits of questionnaire redesign: implications for data quality and statistical burden across different respondent profiles

Abstract

The aim of this paper is to show how different respondents benefit differently from methodological improvements in the questionnaire design.

When dealing with questionnaires, respondents differ in both the type and number of difficulties they face, depending on structural characteristics, core variables, and response strategies. This leads to differences in completion performance, with some respondents more likely to introduce inaccuracies, especially in quantitative information.

If the questionnaire is one of the main sources of non-sampling error, a good questionnaire design is a primary tool for maximizing accuracy. Data from the survey on Research and Development in business enterprises (RS1), collected before and after major changes to the questionnaire design, have been compared to explore differences in completion behaviour. Both descriptive and logistic model analysis, carried out to assess the impact of design changes intended to facilitate and improve the use of the unit of measurement in expenditure questions, along with the analysis of burden indicators, have shown the existence of different sub-populations, both in terms of accuracy and perceived burden. Profiling respondents has provided valuable insights into further opportunities for improvement.