Page 57 - IJPS-3-1

Yin N and Heiland F

limitation severity in these vignettes. European respondents, compared to Americans,
tend to classify a vignette as more work-limiting. American respondents are the least
generous overall in rating work limitations, followed by the Italians, the Belgians,
the Dutch, and the French. The Swedes and the Spaniards are at the other end of
the spectrum and inclined to rate a given work limitation as more severe. German
respondents are in the middle.
These patterns are notable in light of our discussion of disability program generosity.
More inclusive rating scales are associated with more generous disability regimes,
as indicated by the fact that the country ranking according to the inclusiveness of
disability rating is in line with the ranking in terms of the generosity of the country’s
disability system.
Figure 2-1 shows the correlation between the generosity of disability system and
disability vignette rating. The horizontal axis represents a country’s disability policy
generosity index. The vertical axis refers to the percentage of respondents in a country
who classify a vignette as not at all limited. Each diamond in the graph represents
the rating for a specific vignette. There are apparent variations between countries
with different disability policy generosity in classifying the severity of the same
disability vignettes. The difference between the U.S. and other European countries in
rating styles is striking. The graph indicates a weak and negative correlation between
disability policy and disability ratings. That is, more generous disability policy seems
to be generally associated with more inclusive disability classifying styles. Together
with Figure 2-2, the direction of the correlations shows that respondents under more
generous disability regimes are more likely to report the same vignettes as more
work limiting. The weak strength of the correlations may have to do with how the
demographic and socioeconomic characteristics of a population are related to the
perception of popular views. Next we will estimate the effects of disability policy
generosity on reporting heterogeneity while controlling for a series of respondents’
characteristics. We test whether disability policy generosity predicts reporting styles
and whether the predictive power is robust to including the respondents’ individual
level factors.

2.3 Covariates

We test a model with a detailed set of individual-level and country-level factors.
Specifically, the model includes standard demographic covariates: age dummies,
education (in years), dummies for being female, and a series of health indicators
(dummies for high blood pressure, diabetes, cancer, lung problems, heart problems,
arthritis, obesity, the number of ADL limitations, and the number of IADL limitations).
We also include dummies for (last) occupation (technical/sales/administrative
support, service, farming/fishing/forestry, precision production/craft/repair, operators/
fabricators/laborers, elementary occupation, and managerial/professional specialty as
omitted/reference occupation). Past occupation is expected to be an important predictor
of a person’s risk of having a work limitation. The severity of work limitations is a
function not only of the health problems but also of the type of work engaged.
With respect to variation at the country-level, we include a set of variables describing
how a country scores in terms of disability policy generosity (coverage and maximum
benefit level, disability level for full benefits, permanence of benefits, medical
assessment, vocational assessment, minimum disability level and sickness benefit, and
unemployment benefit level and duration).
2.4 Statistical Approach

Standard ordered regression models (e.g. ordered probit) are often used to analyze self-
reported work limitation on a five-point severity scale:
Pr (h = ) k = Pr (µ k−1 ≤ x β + ε < µ k ) (1)
s
s
i
i
i
International Journal of Population Studies 2017, Volume 3, Issue 1 51