Patrick Lazarevic

24 Followers
35 Following
36 Posts
Social scientist focusing on survey methodology and the measurement of health
Head Survey Methodologist at Austrian Socio-Economic Panelhttps://statistik.at/asep
My dissertationhttps://tinyurl.com/DissLazarevic
Twitterhttps://tinyurl.com/TwitterLazarevic
ResearchGatehttps://tinyurl.com/RGLazarevic

Der 17. Workshop der Panelsurveys im deutschsprachigen Raum wird am 13. und 14. März 2025 in Wien stattfinden!

Den Call mit allen Infos findet Ihr anbei, die Deadline ist der 07.01.2025.

Einsendungen und Rückfragen bitte an [email protected]. Wir freuen uns auf Euch!

Anything above the red line(s) are significant 'biases'. We see that education and age biases were not significant for men using MEHM(+) but for SRH. Nominally, this is also true for women. Also generally for income and optimism. Great! (9/11)

Great, but are these indicators any good?

At least they seem to reduce age biases. Using the same subjective question, younger people tend to underestimate their 'objective' health, older people overestimate. MEHM(+) reduces this, making the responses more comparable. (7/11)

It does! I extracted an indicator from MEHM and MEHM+. MEHM +also includes pain and multimorbidity, which are also important for rating one's health: https://twitter.com/PatLazarevic/status/1604802411561984000

GSEM is flexible for adding more variables if you have them! (6/11)

Patrick Lazarevic (@PatLazarevic) on X

Finally, I'm happy to announce that @amelieqv 's and my article on the determinants of self-rated health and its changes is open access! #SelfRatedHealth #epitwitter https://t.co/jViAI51ZaY

X (formerly Twitter)

Using DEAS-data of Germans aged 40-85 (by @DZA ) wanted to combine the MEHM into 1 indicator. Cronbach's α was great across age groups (esp. for a short scale!).

Next step: combining them using generalized structural equation modeling (GSEM) - does it work? (5/11)

To the extent that "health" is covered in this model, residuals represent non-health biases. This is because it removes indirect non-health effects on SRH via health indicators, leaving only direct effects of non-health on SRH: non-health biases. (3/8)
As for health changes, we saw a somewhat similar trend: changes in functioning, diseases, and pain were the most important health domains to explain changes in SRH - especially in older age groups. (5/5)
What about age? Strikingly, functioning, diseases/conditions, and pain were more and more important in older age groups. Apparently, in older age these health domains grow more relevant for how we view our health. (4/5)

Functioning was the most important domain for subjective health, followed by diseases and pain. This mostly replicates another study with European data (more here: https://twitter.com/PatLazarevic/status/124249134113060864).

Interestingly, diseases/pain/mental health were a bit more relevant for women. (3/5)

SRH can be viewed as a weighted combination of health information:

Ideally, people recall all health info about themselves and combine it into an overall rating (solid arrows). But maybe the weight of every health aspect differs between population groups (dashed arrows)? (2/5)