Understanding Society User Support

Hi there,

Thanks a lot for suggested reading. It's really helpful. I was hoping to get further clarification on one or two points if possible? From FAQ 14, it states that: "DO NOT use ONLY the year 1 sample, or ONLY the year 2 sample." This gives me the sense that I should not, as I asked in my initial query, restrict my analysis to YEAR 2 respondents ('year 2' - months 13-24), and so cannot look at 2023 using wave 14. Instead, I'll need to wait for wave 15 to do any analysis of 2023. I just wanted to confirm that's correct, or does the above FAQ 14 recommendation only apply to pooling data from different waves? In other words, if I just wanted to look at the GPS-GB sample (year 2 sample) in all the UKHLS waves, could I still look at 2023? I only ask because FAQ 13 mentions that "Please also note that if you use months 13-24 you are excluding Northern Ireland from your analysis." Which feels like it is suggesting one can restrict one's analysis to the year 2.

I may have just misunderstood the FAQs, and apologies if so.

Best wishes,

James

James,

Thank you for your question. Are you aware of our calendar year data releases? It comes earlier than main release and should be suitable for you to study 2023. We are currently working on its release.

Year 2 only sample is not representative of UK. It exclude NI completely, so only has GB information, and many proportions are off due to different selection probabilities in year 1 and year 2. Your best bet would be to use calendar year when it is released, but there is a possibility of very careful correction for the different selection probabilities. You would need to correct these for Bangladeshi, for all 5 ethnic groups that we boost, for all immigrants, and new immigrants (since wave 2 and since wave 6 separately). You could do this by matching yer 2 distribution to previous full wave information, or to official statical distributions.

On multilevel use, this can be helpful:
https://www.understandingsociety.ac.uk/documentation/mainstage/user-guides/main-survey-user-guide/analysis-guidance-for-weights-when-fitting-multilevel-models/

Hope this helps,
Olena

Hi Olena,

Thanks a lot for the message. It sounds like the best plan would be to wait for the 2023 calendar release. Thank-you for flagging this option up. One final question: from the release dates of previous calendar year datasets, it looks like the 2021 calendar year dataset was released at the start of January 2024, and the 2022 calendar year dataset was released January 2025. Would we be expecting a similar pattern for the 2023 calendar year release, expected around January 2026? And would this be after the WAVE 15 release? Any insight would be hugely helpful for planning our research on this.

Thanks so much again,

James

Hi Roberto,

That's really helpful. Thanks so much.

Best wishes,

James