Generations and Gender Programme Preparatory Phase Project (GGP-5D)

Résumé 0

This technical guide presents the State of the art regarding multi-mode data collection, related to the Generations and Gender Surveys (GGS – Round II). Mode effects may come from selection of the respondents (mode selection effects) or from the way respondents understand the questions and answer (mode measurement effects). Mode effects are estimated using GGS Round II data, using the French GGS test and the Uruguayan GGS, which both have collected data on two modes, from different representative subsamples; internet and telephone in France, Internet and face-to-face in Uruguay. Three main data quality indices are computed and used. In both countries, non-response are more frequent in the internet questionnaires than in the questionnaires filled in with an interviewer; straight-lining, the tendency for some respondents to give always the same answer when facing a battery of similar questions, is also more frequent on internet. Finally, questionnaires filled in on Internet are more diverse, with less frequent responses. This is likely due to lower social desirability bias, as the respondents are not interacting with an interviewer. Based on these comparisons, the positive global assessment of the push-to-web strategy conducted by the GGP is confirmed. Nevertheless, new methods are arising and new methodologies are tested. Mode effects are one among many methodological challenges, and the ongoing methodological efforts of the GGS central hub and national teams must be pursued.