Research on the use of incentives generally finds that unconditional incentives are more effective than conditional incentives. However, on a longitudinal study, after the first couple of waves, it may be that sending unconditional incentives to previous non-responding individuals within responding households is a waste of resources, which could be more effectively re-directed elsewhere. This experiment randomly allocated households to two groups. In one the incentive was sent unconditionally to every adult in the advance letter as usual. In the other, only those adults who participated in the previous wave were sent an unconditional incentive in advance. That is, if ff_condincw7 =2, conditional incentives were given to all respondents with ff_ivlolw = 2, 3 and any rising 16 year olds with ff_ivlolw = 5. Unconditional incentives were sent in advance to all respondents with ff_ivlolw = 1 and all 16 year olds with ff_ivlolw = 4.
Since the range of incentives is larger in the Innovation Panel than the main-stage, incentive cards were used instead of paper vouchers. Previous wave non-responding adults in the treatment group (ff_condincw7=2 & ff_ivlolw=2/3/5) who did respond at IP7 were given or sent their incentive post-interview. Where the adult was interviewed in person by an interviewer, this required the interviewer to write the incentive amount on the front of a blank incentive card, hand it to the respondent and make a note of the serial number and amount required in CAPI. When the interviewer dialled in, this triggered the Operations Department to activate that card with the appropriate amount. Where the individual was interviewed online or by telephone, this triggered the sending of a pre-activated incentive card with the correct amount.
Households within PSUs were randomly allocated to treatments.
Controlling variable(s): ff_condincw7 on record hhsamp
Group 1 – Unconditional Incentive control group
Group 2 – Unconditional/Conditional Incentive Treatment
NOTE that the condition/unconditional incentive experiment interacts with the incentive amount experiment and as the actual amount of incentive was determined by the value of ff_incentw7.
