This sample of 1,500 households is used by researchers to test innovative ways of collecting data and developing new areas of research. The IP is used by researchers based at Understanding Society and external researchers, who can suggest new experiments in the annual Innovation Panel Competition.
After each wave of the IP the results of the experiments for that wave are published. Here are the results from the IP Wave 15…
Survey mode matters when it comes to data linkage consent
In this experiment, 663 survey respondents who indicated having a Twitter account were asked if they would consent to link their survey responses to their Twitter data. The participants who complete their survey via the web had a significantly lower consent rate than those who complete their interview face-to-face. Read more
Direct survey questions trigger consumers to spend more
This study looked at whether people would adjust their thoughts on spending and saving by asking how they respond to hypothetical windfalls or losses. A key difference between the questions was whether they were asked a direct question or a two-part question.
- The direct question was “Imagine you unexpectedly received a one-time payment of £[AMOUNT] today. How much of it would you spend over the next [DURATION] months?”
- The two-part format was “Now consider a hypothetical situation where you unexpectedly receive a one-time payment of £[AMOUNT] today. We would like to know whether this extra income would cause you to change your spending, borrowing and saving behaviour in any way over the next [DURATION] months.
Researchers found the direct question primes the respondent to report higher spending than the two-part format. Read more
Informal carers are better identified when more clarification over care work is added to survey questions
Participants should be given more explanation on caring duties in survey questions to ensure informal carers are identified.
These results support both the expectation that adding more clarification about what respondents should include as caring increases the number reporting they provide it; and that using a specific task-based questioning approach further identifies informal carers.
When asking in a non-specific way with minimal clarification, 15.5% of respondents report providing informal care (Control Group) however when clarification is added, or when a task-based approach is used, 19.2% and 24.1% of respondents respectively report providing informal care. Read more
‘Drinks’ or ‘units’ – terminology matters when measuring alcohol consumption
When survey participants were asked about their alcohol intake, participants who measure their consumption in ‘units’ as opposed to ‘drinks’ were more likely to report higher values. Given that the results suggest that respondents do differentiate between ‘drinks’ and ‘units’, Understanding Society will look to use units, as the NHS does in its standard assessment for drinking behaviour. Read more
Participants more likely to consent to data linkage when asked by an interviewer
This experiment aimed to look at how survey mode affects participants’ consent to the linking of their data to other sources of data. The results showed that the mode difference mirrored what previous research has found: at 77.3%, the consent rate is much higher in a personal interview (CAPI) than the web consent rate (41.6%), a difference of 35.6 percentage points. Read more
The success of repeated consent questions in surveys
This research aimed to find out whether repeating consent questions makes any difference to participants agreeing to receive text messages from the Understanding Society survey. Of those respondents who had not previously been asked the SMS consent question, 73% provided consent. Of the respondents who had been asked and declined consent, 53% provided consent.
Of the respondents who said they would not consent to be asked questions by SMS, the most common reasons were that they did not want to answer additional questions (39%), that they don’t use text messaging (33%), and that they were unclear what the purpose was (15%). Read more
Face-to-face interviewing influence participants’ decision to install body measurement app
Respondents who completed the Understanding Society interview with a face-to-face interviewer were most likely to install a body volume app during the interview (48.8%), followed by those who completed the survey online (32.6%). Telephone respondents were least likely to install the app (7.8%). However, web respondents were most likely to actually use the app (19.8%), followed by face-to-face respondents (15.7%). Read more
Interviewer-led health measurements are more accurate than self-reported measures
Can we rely on self-reported biomarker data? Researchers found a notably higher compliance rate (93.4%) for self-measurement in the presence of an interviewer, 37 percentage points higher than for respondents asked to self-measure. The measurements were also more accurate with the interviewer. The most striking difference was in waist circumference: the average waist measurement in the self-reported sample was approximately 30cm wider than the average for respondents who conducted measurements with the interviewer. Read more
Question order can affect how people report their national identity
When looking at how respondents report their national identity, this research found that if British is listed before English in a survey, there is a reported increase in British-only identity and decrease in English-only identity.
You can read more about all these experiments in this Working Paper: Understanding Society Innovation Panel Wave 15: Results from Methodological Experiments
Survey methodology
