Comparative studies of occupational sex segregation have employed a variety of
measures to estimate the extent of segregation across labor markets. In this article, the
authors focus on two intrinsic limitations of the ratio index, which is derived from the
log-linear framework: singularity for totally segregated occupations and sensitivity near
the extremes. To capture the real essence of gender occupational segregation, it is necessary
to examine rather detailed occupational categories. Such detailed occupational
classification poses a problem for the ratio index since small occupations are more
likely to be mono-gender occupations. The authors propose an alternative modified
index that resolves both the singularity and the sensitivity problems by employing the
“first-order approximation” of the logarithmic function. The modified index makes it
possible to compute measures of microsegregation for detailed occupational categories.
The advantages of the proposed index for comparative microsegregation analyses are
illustrated and discussed.