A comparison of participant, Smartphone app and interviewer measurement of body size

2022-06-15

2022-10-15

1. Uptake of an App to measure body volume
2. Percent of body fat, waist and hip measurement

We randomly split the Innovation Panel of Understanding Society into 6 groups receiving different incentive schemes and feedback to encourage the use of a body volume measurement app. Additionally, we use the random allocation to mode of interview (face-to-face v web) to compare self-measurement and interviewer measured outcomes.

The 6 treatment arms for encouraging app usage are determined by 2 incentive schemes (conditional vs unconditional monetary incentives) and 3 feedback offers (total body fat, visceral body fat, or no feedback).

We will then compare takeup and body volume measurements across different treatments.

There are two connected experiments focused on collecting body size measurement.
1.On the full sample of the Understanding Society Innovation Panel, we are experimenting with ways of encouraging participants to download, use and return data from a body volume app. The app requires participants to take two pictures of themselves, which will create silhouettes of their body shape (viewed from the front and side) and enable the calculation of a range of body volume measures – such as total and visceral body fat, waist and hip measurement, and the lengths of body parts. Instructions are provided depending on whether they want to take the pictures themselves or have someone else take the pictures of them. There are two experiments to encourage uptake:
a) Incentives: participants will be randomly allocated to receive either £5 added to the unconditional incentive sent with the survey invitation or promised a conditional £5 bonus if they complete the app measurement.
b) Feedback: participants will be randomised into three groups: those who are offered feedback of their total body fat, those where feedback of their visceral body fat, or those for whom no feedback is mentioned.
We will investigate which type of incentive and feedback offer leads to the best uptake (highest response, lowest response bias) for the App.

This experiment will be nested in a general mixed mode design, so we will also be able to assess if mode (web v face-to-face) results in the higher uptake and if there is an interaction of mode and treatments outlined above.

2.As noted above the innovation Panel has a mixed mode design based on random allocation to mode. Hence a random subset will be interviewed in person so that we can compare self-measurement and interviewer measured outcomes on a group of participants randomly allocated to face-to-face interviews. In addition to the Body volume app, all participants will be sent a tape measure and asked to measure their waist and hips according to set protocols. For the random subset allocated to face-to-face interviews, interviewers will measure their height, weight and body fat, waist and hips according to standard anthropometric protocols. Hence for this random subset of participants, a comparison will be possible between the app measured body fat and body fat measured by bioimpedance scales. For the waist and hip measurements, a three-way comparison will be possible between self-measurement, the app calculated measures and measures taken during the face-to-face interview.

The randomisation to treatment will be a random allocation of households within primary sampling units prior to fieldwork commencing, such that all members of a household are assigned to the same treatment. The allocation is done by a computer programme.

For all treatments, the randomization will be done at the household level.

For the two experiments on the body volume app, prior to fieldwork beginning, households will be randomly allocated to one of the two incentive groups and one of the three feedback groups. These allocations will be crossed.

One third of households are allocated to face-to-face interviews and two thirds to web first interviews.

Yes

1500 households

2000 adults

a) For the incentive experiment, approximately 750 households (1000 adults) will be in each treatment group randomly allocated to conditional or unconditional incentives.
b) For the feedback group 500 households, 667 adults, will be randomly allocated to each of the three treatment groups: no feedback (control), total body fat (treatment 1), visceral fat (treatment 2)
c) Approximately 500 households 667 adults will be randomly allocated to receive both the self measurement requests and the face-to-face interview, while 1000 random households will only be asked to complete the self measurement.