Santa Claus, the Gender Wage Gap and Over-Aggregation

In the December 24, 2009 edition of The Tennessean, Steve Edmondson has an op-ed piece regarding the gender gap in compensation.  The article is in response to the December 18 piece by Don Corn appearing in Tennessee Voices.  Mr. Edmondson states:

Belief in Santa Claus by children is charming and fun, but at some point the reality that Santa is nothing more than an entertaining myth becomes apparent and believing in the myth would be a reason for concern in an adult.  Adults that believe in the myth that women do not get equal pay for equal work... should also be a reason for concern.
In his article, Mr. Corn states that on average, women earn 26% less than men.  Mr. Edmondson objects to this statistic, saying that "the use of this statistic is misleading and is not related to reality in the workplace."  While I do believe that there are some instances of gender discrimination in compensation, I agree with Mr. Edmondson that this statistic is misleading.

According to Highlights of Women's Earnings in 2008 published by the U.S. Department Labor, Bureau of Labor Statistics, July 2009, women's earnings were approximately 80% of men's among full time wage and salary workers.

Differentials exist when comparing median weekly earnings by gender and race:

However, the above statistics refer to earnings in the aggregate - that is, across all occupations and industries.  When the aggregate earnings statistics are broken out by major occupation group, we see that the majority of women's earnings (51.3%) come from two occupational groups: professional and related occupations and office and administrative support occupations.  Men's earnings have a more even distribution across all major occupation groups.

When one compares the median weekly earning of all workers by major occupation group, one sees that there are differences by major occupation group:

If one were to break out occupations within the major groups, one would see that there are different median weekly earnings within group.  For example, within the Professional and related occupations group, the median weekly earnings of workers in computer and mathematical occupations was $1,242, while the median weekly earnings of workers in community and social services occupations was $788.  Not only are there differences in earnings within major occupation group, there are differing proportions of men and women employed in those sub-groupings within major occupation group.

As an economic and statistical expert specializing in quantitative analyses of alleged employment discrimination, I recognize that discrimination is a very real phenomenon.  But I also recognize that in order to draw an inference of discrimination (or not), one needs to compare apples to apples.  The reliance on an aggregated statistic that lumps all workers into one giant category and neglects to account for occupational differences, as well as the numerous other valid predictors of compensation, contributes nothing to the discussion of the gender wage gap.