Page 46 - IJPS-11-6

International Journal of
Population Studies Gender gaps in reporting limitations

3.2. Self-reported work limitations in the U.S. and 2007 that are not captured in the present model are likely at
Europe accounting for reporting heterogeneity play in explaining the gender differences in reported work
Tables 4 and 5 present estimation results for the model disability.
of severity of one’s own work disability with and without In the left panel of Table 5, the estimation results based
adjustment for reporting scale heterogeneity. The model on the SHARE 2004 data show that the coefficient for
explains the self-reported work disability on a five-point women has changed little in response to adjusting for the
scale. The model “without adjustment” is a HOPIT model reporting heterogeneity. This is not surprising given the
that does not allow the cut-points to vary with respondents’ fact that men and women in Europe during that year did
characteristics. It is similar to a standard ordered probit not apply different reporting styles for work disability, as
model. The model “with adjustment” is a HOPIT model shown in the left panel of Table 3. Thus, reporting styles
allowing for reporting scale variation. The former model is did not explain the observed gender gap in self-reported
rejected by the data, as is evident from a comparison of the work disability among Europeans in 2004. However,
log pseudo-likelihoods at the bottom of each table. in 2006, things were different: The observed gender
In the left panel of Table 4, comparing the estimates differential in work disability has diminished after we
from the models with and without scaling adjustment corrected for the reporting heterogeneity (right panel of
based on the HRS 2004 data, one of the most salient Table 5). It means that the observed lower probability of
changes is the coefficient for women. Before adjusting for having (self-reported) work disability among European
reporting heterogeneity, women seem to be significantly women compared to men in 2006 is mostly due to the fact
less likely to have (self-reported) work disability than men, that women during that year apply much higher thresholds
everything else being equal. The gender differential in (self- in classifying and reporting disability.
reported) work disability loses its statistical significance Angelini et al. (2011), also using the 2004 and 2006
and becomes much smaller in magnitude when scaling waves of the SHARE data and anchoring vignette
heterogeneity is accounted for. It suggests that the observed approach, documented that surprisingly large fraction of
lower likelihood for women to have (self-reported) work individuals changed their self-reported work disability
disability is mainly due to the fact that women apply a status within 2 years, and found that these dynamics can
stricter criterion in reporting disability severity than men. be largely explained by the fact that respondents changed
That is, for a potentially identical work limitation, women the way that they assessed the severity of work disability
would likely report it as less serious than men would. After problems over time. This evidence clearly supports our
accounting for the reporting style difference between claim that individual response scales might vary over time,
genders, the gender difference in disability diminishes. with median thresholds shifting to the right. This finding
Another noticeable change after the scaling adjustment is implies that, given the same health status, one person might
that non-Hispanic blacks do not seem to have significantly rate herself as work-disabled in 1 year but not in the next.
more (self-reported) work disability than non-Hispanic
whites. The disappearing racial difference in work disability 4. Discussion
is primarily attributable to the lower threshold used by In this study, we analyzed a set of disability vignettes
non-Hispanic blacks in rating disability. related to pain, cardiovascular health, and depression from
The right panel of Table 4 shows the estimation results longitudinal survey data in the U.S. and eight European
for respondents’ own work limitations based on the HRS countries over two waves (2004 and 2006/2007). Our
2007 data. We found that the reporting heterogeneity findings show distinct gender differences in classification
explains only part of the residual gender gap in work across nearly the entire work disability severity spectrum.
disability. Before adjusting for the reporting heterogeneity, Men tend to rate identical work limitations as more severe
after controlling for a large array of demographic, than women do. This gender differential in disability
socioeconomic, and health variables, women appear reporting styles exists in both the U.S. and Europe
less disabled than men. The gender differential in work and is consistent across both waves. The differential
disability has become smaller once we adjusted for the remains robust after controlling for various factors such
reporting heterogeneity and accounted for the fact that as demographics, socioeconomic status, health, and
women apply tougher standards and are less likely to employment status. Moreover, we observed that reporting
report work disability than men. However, different than styles change over time. However, once we accounted
in 2004, the gender difference in reported work disability for this reporting heterogeneity, the gender gap in self-
is not fully explained in 2007 by reporting heterogeneity reported work limitations often diminishes or disappears.
and the factors accounted for in the model. More factors in This suggests that the lower rate of self-reported work

Volume 11 Issue 6 (2025) 40 https://doi.org/10.36922/ijps.1969