This experiment aimed (a) to investigate differences in data quality arising from switching Innovation Panel members from face-to-face to web mode of survey administration, (b) to test methods for the mitigation of a hypothesized decline in quality associated with moving to web data collection and (c) to contribute to a general understanding of how web data collection methods can affect data quality. The experiment asks respondents to recall facts gathered contemporaneously at earlier waves and uses earlier wave data as a validation check. Half of respondents completing the survey on the web received a commitment pledge as an experimental treatment to encourage more accurate reporting of historical information. Randomisation was of households across the complete sample, i.e., regardless of PSU, into one of two experimental treatments.
The controlling variable is F_FF_ITEMW6 on record F_HHSAMP_IP. It is coded as:
1 No commitment pledge
2 Commitment pledge
Affected variables in the questionnaire are f_empv1 f_empv2 f_sf1recall f_sf5recall f_sf6crecall f_healthrecall f_hlwtrecall f_hlwtsrecall f_hlwtprecall f_hlwtkrecall f_recallease f_recalleffort f_webrecall f_webinterrupt on record f_indresp_ip.
