Fieldwork compression experiment

Currently the main Understanding Society survey collects each wave of data over the course of two years. That is, in any given calendar year, half of the sample are completing the survey for wave t. The other half already completed wave t the year before and are now completing wave t+1. However, it is proposed that in a future wave, Understanding Society will compress fieldwork into a single year. This compression will mean that half the sample will skip the rotating content of one survey wave. As an alternative, these respondents could be asked to complete a longer survey that asks some or all of the rotating modules they would otherwise miss.

This experiment tests two ways in which participants are asked to complete these additional modules, and two sets of additional questions, a longer or shorter version. This leads to four experimental groups, as well a control group. This experiment also includes the use of additional incentives, either unconditionally or conditionally offered, depending on the design allocated to.

With the continuous, longer interview, the advance letter noted that the interview would be longer than usual, but during the interview no indication was given as to when the respondent was answering the additional modules. The letter noted that because the interview was longer, the incentive would include an additional unconditional £5.

With the potential ‘break-off’ point, there was a request towards the end of the standard IP13 interview to complete an extra set of questions. The participant was informed at this point that if they completed the rest of the interview, they would be given (in CAPI) or sent (in Web) an additional £5 voucher (conditional).

The control group received the standard IP13 questionnaire and no additional incentive.

The longer interview contained all of the modules planned for wave16, except one, and took about 15-25 minutes depending on respondent routing. The shorter interview contained five modules, all except one being rotating modules that would normally be asked of the whole sample, i.e. containing no routing, and took about 10 minutes. The one exception in both the longer and shorter version is that the Wealth and Assets module was dropped and replaced by a module on Panel Conditioning (see Section 10.7).

The allocation was randomised at the household level, with 1/5 of households allocated to each condition. The controlling variable is ff_compressionw13 on record m_hhsamp_ip:

1          Continuous longer interview, full set of rotating modules

2          Potential break-off request, full set of rotating modules

3          Continuous longer interview, reduced set of rotating modules

4          Potential break-off request, reduced set of rotating modules

5          Control

The variables from the additional modules are:

fwcintro, jbperfp, jbonus, jbrise, tujbpl, tuin1, jbpen, jbpenm, jbpeny4, penmcn, penmpy, penmtp, penspb, wktime, wkends, jbflex, jbfxuse, jbfxinf, wktech, wktecho, wktechinet, wkhome, wkaut1, wkaut2, wkaut3, wkaut4, wkaut5, depenth1, depenth2, depenth3, depenth4, depenth5, depenth6, jblkcha, jbxpcha, jblkchb, jbxpchb, jblkchc, jbxpchc, jblkchd, jbxpchd, jblkche, jbxpche, jbsec, envhabit1, envhabit2, envhabit3, envhabit4, envhabit5, envhabit6, envhabit7, envhabit8, envhabit9, envhabit10, envhabit11, carmiles, flyes, nflyin, nflyeu, nflyos, hrs_slph, hrs_slpm, tslp_30m, tslp_wak, tslp_cgh, med_slp, tsta_awk, slp_qual, trcarfq, trbusfq, trtrnfq, trbikefq, needcar, cyclepath, hubuys, hubuyshhm, hufrys, hufryshhm, humops, humopshhm, huiron, huironhhm, hupots, hupotshhm, hudiy, hudiyhhm, hupayhswrk, hucdress, hucdresshhm, hucbed, hucbedhhm, hucunwell, hucunwellhhm, hucplay, hucplayhhm, huchomework, huchomeworkhhm, hucferry, hucferryhhm, huboss, howlng, chargv, charfreq, charam, save, saved, savreg, savlt, workdis, journeysat, workpark, workparkpay, ppen, ppent, ppyrs, ppreg, ppram, pprampc, volun, volfreq, volhrs, scacfwc, scrffwc, scunfwc, scenv_ftst, scenv_crlf, scenv_grn, scenv_bccc, scenv_pmep, scenv_meds, scenv_crex, scenv_tlat, scenv_nowo, scenv_fitl, scenv_noot, scenv_canc, scopecl30, scopecl200, scwemwba, scwemwbb, scwemwbc, scwemwbd, scwemwbe, scwemwbf, scwemwbg, scopfama, scopfamb, scopfamd, scopfamf, scopfamh, condthink, conddo, conddo2, condatt, whypart, whypart2.