This Wave 4 experiment examines the utility of observation data provided by interviewers. It is limited to the refreshment sample cases. The focus was on questions of evidence for children in the household or for access to personal transport such as a car or van. Version A of these questions used the original interviewer observations, which read as follows, “Based on your observation, is it likely that this address has a car or van?” with response options “Definitely has a car / van”, “Likely”, “Unlikely”, “Definitely does not have a car/van”, and “Cannot tell from observation”. The items continue, “Based on your observation, is it likely that this address contains one or more children aged under 10 (including babies)?” with response options “Definitely has a child/children aged under 10”, “Likely”, “Unlikely”, “Definitely does not have a child/children aged under 10”, and “Cannot tell from observation”.
Version B of these items was written more objectively. First, “Standing outside, can you observe any signs of a car or a van belonging to this address?” with response options “Yes, probably belonging to this address”, “Yes, unsure whether belonging to this address”, and “No”. Next, “Standing outside, can you observe any signs of children under 10 (including babies) at this address?” with response options “Yes” and “No”.
Households were randomly allocated within PSUs. The controlling variable is d_ff_arfexpw4 on the record d_hhsamp_ip. The experimental allocation is as follows:
1 “Version A” original wording
2 “Version B” alternative wording.
The substantive information is contained in d_children_b, d_children_a, d_carvan, and d_carvan2.
