Probability-Based Web Panels for Official Statistics: Basic Insights and Analysis of the Bias of Survey Estimates

Gregor Čehovin, Vasja Vehovar
Statistika, 106(2): 164–179
https://doi.org/10.54694/stat.2025.28

Abstract
The past three decades of survey research have revealed a shift from traditional probability-based surveys to various types of web surveys. In official statistics, the probability-based web panels (PWPs), for which respondents are recruited once and then incentivized for repeated participation in web surveys, play a particularly important role in this process. This article provides insights into the use of PWPs for official statistics. First, a survey of European Union national statistics offices revealed that one-quarter had already implemented or were planning to implement PWPs; the main barriers to their implementation were lack of knowledge and expertise. In addition, we evaluated the quality of the estimates in the Slovenian PWP (1KA Panel) that replicated questions from 12 traditional probability-based surveys. The findings showed that 205 of 651 PWP estimates (31%) exhibited relative bias exceeding 10%. Biases varied substantially across survey topics, indicating the selective suitability of PWPs for official statistics.

Keywords
Probability-based web panels, survey estimates, relative bias, nonresponse error, measurement error, coverage error, processing error