Gender Inequality - it's a jungle out there

Radio 4’s flagship, the Today programme, has fallen for a common misrepresentation of the gap in pay between men and women.

 Or is it a misrepresentation? It depends on who you are talking to, as various bodies interpret this key statistic in various ways.
Introducing an item on Today on July 29, Sarah Montague accepted at face value the assertion by the Women and Work Commission that women are paid, on average, 23 per cent less than men. But the Office of National Statistics quotes a figure of 12.8 per cent, just over half as much.
And if that isn’t confusing enough, the Equality and Human Rights Commission believes the gap is 17.1 per cent.
These differences matter, so it would be nice to achieve a common basis for estimating them. Until we do, journalists should beware of accepting whatever figure is thrown at them. Government bodies with different objectives can easily come up with different figures – and may honestly believe they are right.
Earnings comparisons are best made on the basis of  median, rather than mean, earnings. This is because the salaries at the top of the scale are often high enough to distort the mean. On measures of “average” salaries, most people earn less than average.
Both the ONS and the Women and Work Commission, which is part of the Government Equalities Office, do use the median. (The EHRC, just to be awkward, uses the mean – of which more later.)
So why the difference between the ONS and the GEO? The ONS only counts full-time work, whereas the GEO includes part-time work, three quarters of which is done by women. This increases the apparent gap in pay between the sexes.
Which is right? The Statistics Authority cogitated over this in a report published in June. Neither measure is satisfactory, it admits, but it does come down in favour of not combining the two, as the GEO does. Its recommendation is to present the two estimates, for full-time and part-time employees, separately.  
It publishes rather a striking table (Table 2 in the UKSA’s note) that actually shows that women working part-time earn 3.4 per cent more than men in median hourly earnings. This isn’t a figure you’ll find the GEO or the Women and Work Commission quoting very often.
Where does all that leave us? Of those in full-time work, women earn 12.8 per cent less than men. Women in part-time work earn fractionally more than men, but less than full-time men or full-time women. When full- and part-time work are combined, the preponderance of women in part-time work produces the 22.6 per cent gap headlined by the GEO and accepted without question by Today.

So what about the rogue figure of 17.1 per cent quoted by the EHRC? It uses mean earnings rather than median, justifying it by saying that women are over-represented at one extreme of the distribution and men at the other, which results (it says) in gaps calculated from the median understating the size of the problem.
It then goes on to claim, remarkably, that for women working part time, the gap is 35.6 per cent. This dizzying figure is achieved by comparing part-time women with full-time men – a comparison hard to justify on any rational basis. As the Statistics Authority remarks, this estimate “needs particularly careful explanation and justification if it is not to mislead”. (Translation: it’s misleading.)
The gender pay gap is a jungle, where journalists should not venture without careful preparation and a trusty guide. Discussions are going on between ONS and GEO to determine how the results should be presented in future, and we may see some results when the 2009 Annual Survey of Hours and Earnings is published in November.
But somehow I think it’s going to take a lot of persuasion to get the equality-wallahs to abandon the high estimates to which they are so attached. 35.6 per cent? Whew!