Support #2078

Choosing the correct group and sample for weight

Added by Albert Ward about 2 months ago. Updated 10 days ago.

Start date:
% Done:




I'm trying to choose a correct weight for some analysis I'm doing. I'm matching variables from the BHPS to wave 9 of the UKHLS. These include some variables at the household level. What weight would I choose for this, esp. re: the group and sample/timeline. When you have conflicting levels for the group, do you choose the lowest (I.e. the individual level). And I'm a little confused about the sample/timeline, particularly the difference between ub and ui.

So I was thinking i_indscui_lw ?



Updated by Understanding Society User Support Team about 2 months ago

  • Category set to Weights
  • Status changed from New to In Progress
  • Assignee changed from Understanding Society User Support Team to Olena Kaminska
  • % Done changed from 0 to 10
  • Private changed from Yes to No

Many thanks for your enquiry. The Understanding Society team is looking into it and we will get back to you as soon as we can. We aim to respond to simple queries within 48 hours and more complex issues within 7 working days.

Best wishes,
Understanding Society User Support Team


Updated by Understanding Society User Support Team about 1 month ago

  • Status changed from In Progress to Feedback
  • Assignee deleted (Olena Kaminska)
  • % Done changed from 10 to 80

Hi Albert,

We would need some more information to advise on the weight to use. For instance, you would need a more precise starting point for your analysis – what is the earliest wave you plan to use in your analysis?

From the information you posted, it is clear that you want to use some measures from the BHPS and wave 9 of Understanding Society. In addition, you will use variables from the household and the individual interviews. It seems that the weights i_indin91_lw or i_indin01_lw could suit your analysis. If the earliest wave of your analysis lies between waves 1 (1991) and 10 (2000) of the BHPS, you can use the i_indin91_lw longitudinal weight; otherwise, you can use the i_indin01_lw longitudinal weight.

Please note that these weights refer to different populations. The BHPS started in 1991, covering the Great Britain household population. In 2001, a sample of households from Northern Ireland was selected. Therefore, if your analysis starting point is before 2001 and you use the i_indin91_lw weight, the analysis will be restricted to Great Britain. If your analysis starting point is 2001 or later and you use the i_indin01_lw weight, the analysis will cover the UK population.

Regarding the question about using information from different instruments (e.g., household and individual questionnaires), you should use the weight designed for the lowest level.

I leave you a link to a document that can provide practical information about selecting your weight (pp. 4-5) and other useful tips.

Best wishes,
UKHLS User Support Team


Updated by Olena Kaminska about 1 month ago


Additionally, this can answer many of your questions:

Hope this helps,


Updated by Albert Ward about 1 month ago

Thanks for this! Yes, the earliest wave at least is BHPS wave 9. So the prefix is based on the final wave for longitudinal weights, and the sample the earliest sample?

That's helpful on NI, thank you.

And just to be clear: if I'm combining questions from the self-completion questionnaire and main interview, I should choose sc ?



Updated by Albert Ward about 1 month ago

Sorry, a final question: with regards to cross-sectional weights, if I am matching questions BHPS wave 9 to a ukhls wave, for instance, does the same principle apply, so I_indin91_xw ?



Updated by Understanding Society User Support Team 27 days ago

  • Assignee set to Olena Kaminska

Updated by Albert Ward 27 days ago

That was worded poorly: what I meant was that, I assume that if I am matching variables from previous waves and then doing some analysis at that wave, I should use longitudinal waves in that case. So, I_indin91_xw as a weight does not exist?


Updated by Olena Kaminska 26 days ago


So, all you mentioned in num4 message messages is correct, but if you use more than one wave you have to use lw weight. Only if all information comes from one wave then you can use xw weight.

Hope this helps,


Updated by Albert Ward 26 days ago

Thanks again. One final question: I thought I might need these weights: I_indsc91_lw and I_indsc01_lw, but it seems they don't exist. Why is that? Is there something I'm missing here?


Updated by Olena Kaminska 25 days ago

BHPS never provided self-completion weights, so we kept with that tradition. But sample size would be very small by now for such weights as it would require answer to self-completion in all waves (over 30+ years). You have two options: use indin weight as a suboptimal, or create your own tailored weight as demonstrated here:

Hope this helps,


Updated by Albert Ward 25 days ago

Ah, I see. That makes sense thanks.


Updated by Understanding Society User Support Team 10 days ago

  • Status changed from Feedback to Resolved
  • % Done changed from 80 to 100

Also available in: Atom PDF