Over Interpreted Data 1

Sometimes people make statements that are based on fact but which draw inferences that are not supported by the data. My first example of this is taken from an equality and diversity training package available in the UK. The statement is

‘Women in full time employment are typically paid 17% less than male counterparts doing the same job.’

This is presumably based on the Office of National Statistics finding that, in April 2008, the mean hourly pay for women in full time employment was 17.4% less than that for men. (In April 2009 the gap in mean hourly rates for men and women in full time employment was 16.4%). However, the Office of National Statistics also reports the gap in median hourly pay for those in full time employment. It was 12.6% in 2008 and 12.2% in 2009. It is very difficult to see how women can be paid 17% less than male counterparts doing the same job when the median gap is around 13%. Had the statement stopped at ‘Women in full time employment are typically paid 17% less than men’ it would have been difficult to take exception to it, apart from quibbles about the extent to which ‘average’ is ‘typical’, see, for example, How to Lie with Statistics by Darrell Huff . Adding ‘doing the same job’ makes the statement problematic. There have been numerous reports on the gender pay gap including the Kingsmill Review (2003) and the reports of the Women and Work Commission: Shaping a Fairer Future (2006) and Shaping a Fairer Future – a review of the recommendations of the Women and Work Commission three years on (2009) (available from www.equalities.gov.uk). The Government Equalities Office also has a Factsheet Tackling the gender pay gap available from their website. The causes of the gender pay gap are complex and include vertical and horizontal segregation, gender stereotyping and difficulties in combining career and family responsibilities. (Horizontal segregation refers to the tendency for women to be in lower paying occupations and vertical segregation to the tendency for women to be in lower paying positions within a specific occupation.) So, at least some of the gender pay gap arises because women tend to be doing different jobs from men. Even the Fawcett Society claims only ‘up to 40%’ of the gap is due to direct discrimination, i.e. employers paying women less than men to do the same job.

The gender pay gap is a real and important issue. Women are disadvantaged and their skills are not fully utilised. Pay gaps persist even in populations with similar educational backgrounds and within the same employment sector (Purcell and Elias, Higher education and gendered career development (2004), see this website, and Connolly and Anderson, UKRC Research Briefing Equal Measures: Investigating University Science Pay and Opportunities for Success (2006), available from the UKRC website. It is precisely because it is an important issue that we should try to make accurate statements. In this case it is not difficult to find the correct figures. A Google search on ‘UK gender pay gap’ turns up the Office of National Statistics data as the top link. At the very least an inaccurate statement reduces credibility. At worst it leads to people dismissing the problem. And it seems to me that if you really want to tackle a problem then an accurate statement of the problem is a good place to start.