Support #1928

Health condition (hcond) for new entrants and missing values (including -8 inapplicable)

Added by Alexander Labeit 6 months ago. Updated 8 days ago.

Data documentation
Start date:
% Done:



I'm interested in constructing health status and especially certain health conditions over time.
Therefore I have created variables and a longitudinal panel dataset with
pidp wave hcond* (e.g. hcond1 for asthma, etc.)

Health conditions variables for new entrants:
In wave 2 new entrants were not asked the health conditions module and no health conditios variables are available
but should be available in all other waves 1, 3 following.

I have idenditified for every respondent when an individual entered US as a respondent ('new entrant').
I have also studied the relating documentation for the health conditions:
FAQ_HCOND_Changes_Waves_Table_20230214.xlsx and FAQ_HCONDS_Changes_Waves_Description_20230214_v5
As described in the documentation material health conditions variables are not asked and therefore missing in wave 2
but I have found in wave 1 and wave 3 (and the following waves) for respondents which enter the panel (entrants): some frequency values a (-8 inapplicable).
I would expect as values 1,0,-1,-2,-7,-9 but not -8 ("Valid skip" - This information is missing because the person was never asked this question as they were not eligible for it.)

What I do not understand is why the value -8(inapplicable) is given because every respondent should answer the health conditions as new entrant except in wave 2
If it helps I can send you which cases I have identified with the info about pidp/wave/hcond


Updated by Understanding Society User Support Team 5 months ago

  • Status changed from New to Feedback
  • % Done changed from 0 to 50
  • Private changed from Yes to No

Dear Alexander,

The definition of new entrants in the routing of the hcond questions is based on the following condition: ff_ivlolw <> 1 And Ff_everint <> 1. From what I can see, only cases that do not meet this condition were outside of the universe what means that the routing worked as intended. However, ff_everint equals 1 also for cases who joined UKHLS in wave 3, but had been previously interviewed only in BHPS. Does your variable capturing new entrants work on the assumption that new entrants are only respondents never previously interviewed in UKHLS, and not UKHLS OR BHPS? In other words, is it similar to fwintvd_dv from xwavedat?

Best wishes,
Piotr Marzec
UKHLS User Support


Updated by Alexander Labeit 5 months ago

An additional question for respondents who have been interviewed before in one of the BHPS samples and are now being interviewed in US using the variable ff_everint (fed forward ever interviewed) for deciding about an existing health condition.

There is the information from memorig available to which sample a respondent belongs: values 3 to 6 means belonging to the BHPS GB/Scotland/Wales/NI sample) and I can identify these respondents.
Also, additional information is available using ff_everint (fed forward ever interviewed) if a respondent was interviewed previously with value 1.
Respondents who have had their (first) interviews in one of the BHPS samples and are still later interviewed in the US sample could be identified using these both variables.

For answering the question to decide if a health condition like asthma exists for respondents in the US and who are now for the first time interviewed within the US sample and who have been interviewed before in the BHPS sample
-> information from earlier waves of the BHPS could be used as below:
For example, if the asthma health condition hcond1 is given as inapplicable (-8) in the first appearance of an US wave and if ff_everint (fed forward ever interviewed) has value 1:
I could potentially try to use this information from the BHPS sample to carry forward if the respondent has an existing health condition like asthma if asked for the first time within the US:
I would use info from the variabe hlprxe(suffer asthma attacks/short of breath) from waves BH11, BH16
This approach could be used to decide if hcond1 exists (has a doctor or other health professional ever told you that you have any of these conditions?)
I should have in mind that the wording of the asked question is not identical.


Updated by Understanding Society User Support Team 5 months ago

  • % Done changed from 50 to 80

Hi Alexander,

Yes, absolutely, you can take this information from the BHPS health questions, for some respondents this is the only source of information - the issue is that the universe for hcond in wave 3 (based on the condition of never being interviewed) for the continuing BHPS sample members was not optimal because even though they had been previously interviewed, they had been never asked the UKHLS hcond questions, as these questions newly introduced in the UKHLS, so not asked in the BHPS and should have been asked in the UKHLS. So, to reiterate, using the BHPS information is the only way for these respondents. However, as you noted, there are some important differences between BHPS and UKHLS in how health questions were asked, e.g. the wording is not the same, some health conditions are missing etc., so you need to decide whether the degree of comparability is satisfactory for you research.

Hope this helps.

Best wishes,
Piotr Marzec
UKHLS User Support


Updated by Understanding Society User Support Team 8 days ago

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

Also available in: Atom PDF