Understanding Society User Support

Dear Piotr,

Thank you very much for your feedback, this is now clear.

All the best,
Laura

Good afternoon,

Sorry for the follow-up question, but new doubts have emerged while reconstructing families.

I am using the COVID-19 collection, wave 6 (November 2020). I found that linking family members using the information on cohabitants provided by respondents within the COVID-19 study can lead to imperfect matching, as it is not based on identifiers but rather on respondents' reported information about the age and sex of their cohabitants. Given this limitation, I was planning to use Calendar Data 2020, as it ensures that the data are collected as close as possible to the November 2020 COVID-19 wave.

To this end, since Calendar Data contains repeated values for some respondents, I selected the most recent record (in terms of wave) for each respondent and used it to link family members in the COVID-19 data via the egoalt file of Calendar Data 2020.

Do you think this procedure is viable? Or do you believe that filtering out repeated respondents in Calendar Data 2020 could introduce issues, even if it is only used for linking purposes?

Thank you in advance for your support and clarification.

Best wishes,
Laura

Hi Laura,

Have you tried identifying which mainstage wave is closest in time by comparing the interview dates, the interview COVID date, and the mainstage interview dates?

Best wishes,
Piotr Marzec
UKHLS User Support

Dear Piotr,

Thank you for your reply. This could be also an option. I was just wondering in principle and more generally if it is a fair procedure to pick just one respondent-wave out of the two in Calendar Data, but apparently yes. Thanks!

I have an additional concern: when using the household level identifiers from wave 10 (j_hidp) provided in the COVID-19 data, I see that there are respondents in households with more than one person , who completed the wave 10 questionnaire (j_ioutcome == 11) but are not present in either j_egoalt, j_indall or j_indresp. Why is it the case?

Thank you very much again!

Best,
Laura

Hi Laura,

In principle, the Calendar Year datasets are derived from the mainstage data, so you shouldn’t expect any discrepancies between the two, though you would need to verify this to be certain.

The 2019 Mainstage data, which you download alongside the Covid-19 Survey data, was released to facilitate analysis and provide users with access to the interviews from the mainstage survey, which were not yet available at the time of the first Covid-19 dataset release. However, it has not been updated since 2021, making discrepancies with the current mainstage releases more likely.

Therefore, I recommend using the most recent release of the main UKHLS survey.

Best wishes,
Piotr Marzec
UKHLS User Support

Dear Piotr,

Thank you for your reply. I see some discrepancies between the j_hidp provided within the COVID-19 survey and the j_hidp in the mainstage wave 10, but I will trust more j_hidp in wave 10 for my analysis following your advice.

Thank you again.

Best wishes,
Laura