In this paper I compare the use of crude occupational coding relative to detailed
occupational coding in stratification models, in particular status attainment models.
In an earlier project on data from the International Social Survey Programma 1987,
it has been shown that more detailed occupational coding sometimes produces equal
or even weaker effects than crude measure. To replicate this finding, I use data
from five surveys held in the Netherlands between 1996 and 2006, including the R3
ESS data. I estimate MTMM (multi-trait multi-method) models to determine the size
of random and systematic measurement error in relative and crude occupation measures,
for occupations: father’s, mother’s, and respondent’s first and current/last. Results
show that crude measurement performs equal or better than detailed measurent. However,
the big gain is in using the two measures at the same time in a latent variable model.
Presented by:
Harry Ganzeboom (Dept. of Social Research Methodology, Vrije Universiteit Amsterdam)
Date & time:
January 21, 2008 4:00 pm - January 21, 2008 12:00 am
External seminars home