Analyzing and adjusting item ordering to minimize respondent attrition in web-based surveys: Lessons learned from the VOCES-19 study in Mexico

Silvana Larrea Schiavon, Population Council
Lina Lopez-Lalinde, Population Council
Juan Pablo Gutiérrez, National Autonomous University of Mexico
Isabel Vieitez, Population Council
René Nevárez, Population Council
Pablo López, Population Council
Nicole A. Haberland, Population Council
Cristina Mac Gregor, Population Council
Thoai Ngo, Population Council

Web-based surveys have become an essential tool for social science research during the COVID-19 pandemic. However, despite their convenience for quickly collecting information remotely from a large number of participants, online surveys also present an increased risk of participant attrition and consequently, the loss of important information. Researchers working on the VOCES-19 study in Mexico conducted a hazard and survival analysis during the first round of remote data collection to identify points of high dropout that were preventing respondents from reaching the study’s main section of interest measuring violence outcomes and finishing the survey, and then implemented a methodological experiment with subsequent respondents. 55,366 participants were randomized to either receive the survey with the original section order, or the same survey with the sections reordered based on the findings of the initial analyses. The new order was effective in preventing dropout: participants that were presented with this order were 10 percentage points more likely to complete the violence section and 7.5 percentage points more likely to reach the end of the survey than the group who received the original order. These methods can be useful during survey development and piloting in order to minimize attrition in online surveys.

Keywords: COVID-19, Methodology, Children and youth

See paper.

  Presented in Session 83. Data and Methods: A Medley of Perspectives