All time references are in CEST
The debating chamber – big strategic survey research questions for community debate |
|
| Session Organisers | Mr Andrew Phelps (UK Office for National Statistics) Ms Laura Wilson (UK Office for National Statistics) |
| Time | Tuesday 18 July, 09:00 - 10:30 |
| Room | |
Session Details
Do you want to debate with a peer, or have your own say, on a survey related topic? This session will provide a forum for the community to debate some of the big strategic questions of the day. Note: this will complement rather than replace or duplicate the usual conference plenary panel discussion.
The session invites future, strategic thinking rather than specific project-based presentations. The session will facilitate more diverse participation using an online tool (such as Slido.com) to manage audience questions.
Applications can be submitted under two formats:
1. Solo 'have your say’ applications. You will present or pitch your thoughts on a topic for 5 minutes followed by 25-30 minutes audience debate and discussion.
2. Paired debating applications. You and another person will present your opposing thoughts on a topic for 5 minutes each followed by 20-30 minutes audience debate and discussion. Some topics may be more suitable for a two-speaker approach.
Things to note:
- Speakers will need to demonstrate in their application that they have suitable knowledge and experience to be able to speak to the subject strategically
- If submitting a paired debate, you will need to source your opposer and jointly submit an application (session coordinators will not pair applications themselves)
- In your application, please provide a justification for why you feel the topic will be important to discuss
Suggested topics of interest:
As a starting point we suggest the following strategic topics, but welcome applications on any future strategic theme:
- Field interviewing
- Survey design and survey quality (e.g., including representation, measurement, incentive strategy, analysis, evaluation, qualitative approaches, questionnaire development etc)
- Research priorities
- Alternative data sources (big data, admin data, survey paradata, etc.)
- Data equity/inclusive survey data
- Technology innovations: AI, etc
- Training, capability and capacity building
Keywords: Strategic, discussion, good practice, innovation, future focus
Miss Faye Gracey (Administrative Data Research Wales) - Presenting Author
Administrative data research (ADR) will replace the vast majority of surveys, if not all surveys in time. All surveys should be set up to allow the data collected to be deposited in a Trusted Research Environment (TRE) where it can be linked to administrative data. By administrative data we mean the information created as people engage with public services, like education attainment, GP records, and car tax.
Putting survey data in a TRE can enable researchers to learn more about their respondents and whether they achieved their target sample. For example, we compared the Annual Population Survey (APS) to Census 2021 for England and Wales in the cross government ONS led Integrated Data Service. This did raise questions about the reliability of the APS survey findings. Depositing survey data in TREs also opens up more research opportunities to answer broader research questions, by linking survey data to other datasets.
Using administrative data can be more cost effective, reliable, timely, and less burdensome on data subjects than a survey. For instance, we were able to link the school workforce census with vaccine records to help understand the level of disease protection among the school workforce. We were able to do this quickly in the SAIL databank, we did not need to burden school staff by asking them to complete a survey, our results were more reliable as they were not limited to those who were willing and able to complete a survey, and the approach was cheaper than any type of survey.
Administrative data is being made more available for research in TREs. ADR is set to play an increasing role in social science research in the future. ADR UK has a key role in making this happen.
Professor Peter Lynn (University of Essex) - Presenting Author
This is a solo "have your say" proposal.
In a world of ever-decreasing survey response rates, ever-increasing costs associated with seeking response from a probability sample, and ever-expanding sources of non-survey social data, the value of probability sampling as a vital basis for social surveys is being questioned. The cry goes up, "What is the point of having a probability sample if only 20%, or maybe 10%, of sample members are actually going to respond?"
My position is that probability sampling remains just as important in a low-response multi-source world as it was in a high-response limited-source world. I will argue this from a total error perspective. I will explain how probability sampling rules out some sources of error that would otherwise be present and allows other sources to be assessed and adjusted for. I will suggest that probability sampling provides certain assurances that are much-needed if social surveys are to continue to provide vital information for democratic societies. Viva probability sampling!
Justification: This is a hot topic, frequently raised by survey commissioners and funders and by academics. In the UK, in 2022 the SDCNet identified the role of non-probability sampling as a priority area for R&D.
Knowledge and experience: I have worked as a survey methodologist for 37 years, first in a survey agency and then in academia. I have published more than 100 articles in peer-reviewed journals and book chapters, several of which are on sampling topics. I have been a member of the European Social Survey Sampling Panel for 23 years and have Chaired the panel for the last 8 years, during which time sampling requirements have come under scrutiny. I have served on several advisory committees including the (UK) National Statistics Methodology Advisory Committee.