This experiment tested different ways of asking respondents for consent to link data from their LinkedIn profiles to their Understanding Society survey data. LinkedIn is a platform that serves mainly as a professional networking site, including allowing individuals to search for jobs.
This experiment included two aspects. The first experimental condition tested the impact of placing this consent question early or late in the questionnaire. The second condition tested the impact of including a motivational statement prior to the consent request. Half of respondents received a motivational statement similar to those used to reduce item missingness on other survey questions (Al Baghal and Lynn 2015) and now regularly used in Understanding Society. The statement was placed before the consent question: “The data this would provide is key to this study. This data will enhance the understanding of your survey responses.” The control group received the consent question without this statement.
For the ongoing Wave 14 sample (identified by hhorig<19), the allocations to treatments were generated at the individual level by the questionnaire script. The experiments are controlled by the following variables on the record n_indresp_ip:
n_lkincon (1/2 each):
1 LinkedIn consent is early
2 LinkedIn consent is late
n_linkstat (1/2 each):
1 LinkedIn consent statement included
2 LinkedIn consent statement not included
For the wave 14 refreshment sample the allocations were however made at the household level before the start of fieldwork. This was done to test the differences in experiment variances based on whether the allocations occur at the household-level (so everyone in the same household receives the same treatment) or at the individual-level (so treatment is allocated in during the survey to each individual independent of household) impact outcomes, including precision, differently. These allocations are recorded in the following variables on record n_hhsamp_ip:
n_ff_lkincon (1/2 each, allocation stratified by sampleorig, ff_incentw14, ff_gridmodew11, l_hhmodes):
1 LinkedIn consent is early
2 LinkedIn consent is late
n_ff_lkinstat (1/2 each, allocation stratified by ff_lkincon, sampleorig, ff_incentw14, ff_gridmodew11, l_hhmodes):
1 Additional consent statement included
2 Additional consent statement excluded
The variables affected by these experiments are on record n_indresp_ip:
lkinhavear, lkinlinkear, lkinurlear, lkinurleardk, lkinforenear, lkinsurnear, lkinempear, lkintitear, lkinlocear, lkinschear, lkinhavlat, lkinlinklat, lkinurllat, lkinurllatdk, lkinforenlat, lkinsurnlat, lkinemplat, lkintitlat, lkinloclat, lkinschlat
