At Wave 1, an experiment was conducted to help develop efficient data collection instruments concerning household consumption. There is methodological concern about whether respondents consider all possible categories of expenditure to arrive at reports of household consumption. The way respondents formulate answers could also be affected by categorical cues in the question text. This three-way split-ballot (different questions) experiment in the household questionnaire addressed these concerns.
There was random allocation of households within PSUs to one of three treatments:
Group 1 Question about overall expenditure without cues
Group 2 Question about overall expenditure with detailed categorical cues
Group 3 Separate questions about amounts of expenditure broken out into reporting on each category rather than an overall figure.
The variable in the data that controls allocation to treatments is a_group3 on the record a_hhsamp_ip.
The substantive information from each experimental treatment can be found in the following variables on the record a_hhresp_ip:
Group 1 a_xpall_g1
Group 2 a_xpall_g2
Group 3 a_xpfood1_g3, a_xpfood2_g3, a_xpfood3_g3, a_xpfood4_g3, a_xpfdout_g3, a_xpaltob_g3, a_xpclftw_g3, a_xphealth_g3, a_xptrans_g3, a_xptel_g3, a_xprec_g3
At Wave 6, a further experiment was incorporated to address concerns about the way respondents report information on expenditure and consumption. While the experiment was administered in the household questionnaire, the questions were about “benefit unit” consumption rather than household consumption. For the purposes of this research, a ‘benefit unit’ was defined as an adult, their spouse or partner, and any dependent children under the age of 18 living in the household. The experimental questions were only asked if the household respondent qualified as responsible for paying the bills, such as rent, mortgage, gas or electricity. The PNO of this person is listed in item f_consper on the record f_hhresp_ip.
Households were randomly allocated to two experimental treatments. In one treatment (Version A), respondents are asked to give total benefit unit expenditure by adding up a set of expenditure categories using a showcard to trigger recall of expenditure on each category. In the alternative treatment (Version B), respondents provided an amount of expenditure for each expenditure category, then reconciled the total amount spent for accuracy.
The variable in the data that controls allocation to treatments is f_consexpw6 on the record f_hhsamp_ip taking a value of 1 if Version A and 2 if Version B.
The substantive information from each experimental treatment can be found in the following variables on the record f_hhresp_ip:
f_consper and f_consintro for all treatments
Version A f_constotatshh; f_constotatsmm; f_constotatsss; f_constotal_a; f_constotatehh; f_constotatemm; f_constotatess; f_constrat1 to f_constrat5; f_consoth_code; f_opusltshh; f_opusltsmm; f_opusltsss; f_opusl1; f_opusl2; f_opusltehh; f_opusltemm; f_opusltess
Version B f_expbrktshh; f_expbrktsmm; f_expbrktsss; f_expmort; f_expbills; f_exptrans; f_expfood; f_expclothes; f_expchild; f_expdiy; f_exphealth; f_exphobby; f_exptreats; f_exphols; f_expgive; f_expoth; f_breaktot; f_breakad; f_expbrktehh; f_expbrktemm; f_expbrktess
