Support #2194
open
Analysing sub populations
50%
Description
I am conducting an analysis of wave 1 and 2 Understanding society data, and I intend to analyse a subpopulation of the study sample which is made up of individual respondents who 1)identify as an ethnic minority 2)completed the extra five minutes questions and 3)has ever consumed alcohol.
Currently I have dropped all respondents who do not meet the above criteria from my dataset. However, when doing some reading around accounting for complex survey design in stata I found some information which states that dropping cases from survey data can lead to incorrect calculation of standard errors. (https://stats.oarc.ucla.edu/stata/seminars/survey-data-analysis-in-stata-17/)
The article states that no cases should be dropped from the dataset and instead the subpop command stata should be used to restrict analyses to the subpopulation of interest.
Is this guidance for analysing subpopulations relevant to the Understanding society data? And do you recommend this method for analysing subpopulations?
Updated by Understanding Society User Support Team about 10 hours ago
- Category changed from Data analysis to Data documentation
- Status changed from New to Feedback
- % Done changed from 0 to 50
- Private changed from Yes to No
Hello Evie
I recommend taking a look at the Weighting guidance section in our user guide: https://www.understandingsociety.ac.uk/documentation/mainstage/user-guides/main-survey-user-guide/weighting-guidance/. In particular, the subsection “representing the population” may be helpful: https://www.understandingsociety.ac.uk/documentation/mainstage/user-guides/main-survey-user-guide/representing-the-population/. You also find information there about the extra 5 minutes questionnaire.
Additionally, please review the subsection “How to use weights – Analysis guidance for weights, PSU, Strata” for examples on setting up your dataset to achieve unbiased standard errors: https://www.understandingsociety.ac.uk/documentation/mainstage/user-guides/main-survey-user-guide/how-to-use-weights-analysis-guidance-for-weights-psu-strata/
I hope this information is helpful.
Best wishes,
Roberto Cavazos
Understanding Society User Support Team
Updated by Evie Gates about 8 hours ago
Understanding Society User Support Team wrote in #note-1:
Hello Evie
I recommend taking a look at the Weighting guidance section in our user guide: https://www.understandingsociety.ac.uk/documentation/mainstage/user-guides/main-survey-user-guide/weighting-guidance/. In particular, the subsection “representing the population” may be helpful: https://www.understandingsociety.ac.uk/documentation/mainstage/user-guides/main-survey-user-guide/representing-the-population/. You also find information there about the extra 5 minutes questionnaire.
Additionally, please review the subsection “How to use weights – Analysis guidance for weights, PSU, Strata” for examples on setting up your dataset to achieve unbiased standard errors: https://www.understandingsociety.ac.uk/documentation/mainstage/user-guides/main-survey-user-guide/how-to-use-weights-analysis-guidance-for-weights-psu-strata/
I hope this information is helpful.
Best wishes,
Roberto Cavazos
Understanding Society User Support Team
Hi Roberto,
Thank you for your response.
I have reviewed the documents you attached and from what I can understand providing I appropriately account for PSU and strata in my analyses, then my standard errors will be calculated correctly.
These documents however do not provide any guidance on what methods to use to create a subpopulation dataset, thus I am still unsure if the method I have used is correct.
Best Wishes,
Evie Gates