0
$\begingroup$

I'm currently reading this paper about the adherence of health check-ups in the korean population and try to make sense out of the given numbers. In particular I am confused about the adherence to health check ups (in percent) by sex in table 1 and supplementary figure one.

A quick example:

The supplementary figure shows 5288 male participants, where 394 and 3327 took part in an opportunistic and national health check up, respectively. In percentage this makes up 7.45% and 62.92% of the 5288 male participants or at most 70.37% in total (double-counting is possible).

Table one shows 68.1% (total), 7.2% (opportunistic) and 60.3% (national).

This is the first time I am really working with a paper. What are possible sources for this disparity in the numbers? The section "2. Definitions of Socioeconomic and Health Check-up Variables" describes the partitioning into these groups.

Improve this question
$\endgroup$

1 Answer 1

Reset to default
3
$\begingroup$

I believe the explanation is this (haven't read the whole paper): The "Statistical Analysis" section starts with "To represent the general Korean population with minimal bias, sampling weights were applied to account for the complex sampling." The Table represents estimators computed from the sample; you can see this from the fact that confidence limits are given. Given that sampling weights were applied, these estimators are not plain relative frequencies; some people from underrepresented segments of the sample had a higher weight in these computations. Therefore (I believe) the estimators deviate from the plain frequencies that you computed.

Improve this answer
$\endgroup$
3
  • $\begingroup$ Thanks a lot for the answer! This totally makes sense. I read the part with the sampling weights but did not draw a connection. It's all coming together now :) $\endgroup$
    – user7802048
    Feb 4, 2021 at 22:17
  • $\begingroup$ One further question: Figure 1 shows results (p values) from a trend analysis ( I guess something like pearsons correlation between paired samples) with very small p-values. I was not able to reproduce these values (in R) using the (four) numbers in Figure 1. May sampling weights be the cause of this as well? I'd be happy to ask this as a new question if you have no answer (or time to look into it). $\endgroup$
    – user7802048
    Feb 4, 2021 at 22:21
  • $\begingroup$ I'd think so. I can't tell exactly because I didn't put much time into figuring out what exactly they did there, and also of course I don't know what you did in order to reproduce it, but I can well imagine that sampling weights are involved there. $\endgroup$
    – Christian Hennig
    Feb 5, 2021 at 10:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.