On gender, STI and statistics

There are many kinds of statistics that relate to gender in and around science, technology and innovation. But before going through some of the useful sources, some words of caution are necessary. Some scientists and policy-makers may give the impression that if there are numbers involved than that is better than if there are not – but numbers do not in themselves equal truth. Indeed just because there is a way of measuring something does not mean that that the measures are relevant, accurately measured or meaningful.

Similarly, just because there are no statistics does not mean there is no problem. In the launch of the EU Report on National policies on women and science in Europe: A report about women and science in 30 countries, by Teresa Rees in 2002, Hilary Rose famously said “No statistics, no problem, no policy. Statistics help identify problems and can monitor the effectiveness of remedies” … now, for the first time, we have the data. In developing more gender-aware statistics the work of the Helsinki Group on Gender in Research and Innovation, comprising experts on gender and statistics in science, is central.

Within research itself, I have more than once been involved in research projects in which statistics derived from questionnaires suggest few, if any, differences between women and men working at the same organisational level, but when you then interview them you find great difference in, for example, their work-family situation, and their experiences of professional recognition or sexual harassment.

At the same time, there are some areas of inquiry that definitely require good quality statistics. An obvious one is pay. The gender pay gap persists in science, as elsewhere, and is remarkably slow to change. To study this, you need statistics. Another area where statistics are essential is monitoring what actually happens in recruitment, appointments and promotions. This is partly about who is appointed at the end of the recruitment process, but it also about the question of who applies, how do people hear of jobs, who is shortlisted, who does that shortlisting, who is interviewed and by whom, who is offered jobs, who accepts, and then of course on what conditions, with what pay. To keep track of this, science organisations need to have monitoring systems, at different organisational levels.

So, it is important to be aware of, on one hand, the wealth of statistical resources available, and, on the other, the need for more gender-aware statistics. Useful sources here are: Birgitta Hedman, Francesca Perucci, & Pehr Sundström (1996). Engendering Statistics: A tool for change. Stockholm: Statistics Sweden; Westbrook, L., & Saperstein, A. (2015) New Categories Are Not Enough: Rethinking the Measurement of Sex and Gender in Social Surveys. Gender & Society, Vol. 29, No. 4, 534–560; and Bruno, I. (2009). The “Indefinite Discipline” of Competitiveness Benchmarking as a
Neoliberal Technology of Government. Minerva, Vol. 47, No. 3, 261–280. 

These kinds of debates on gender and statistics can be located in relation to basic questions on the nature of social categories, and how categories including gender categories are complicated by intersectionalities. Gender statistics often only deal with women and men, girls and boys, and often do not deal well with further gender positions, such as transgender, intersex, non-binary, and queer, with sexualities, or with intersections with other social divisions, such as age, class, and ethnicity.

There are many kinds of resources to be considered. Let’s start with those at a very general contextual level. These, amongst others, provide broad gendered societal statistics and more focused gendered statistics on education, science, technology and related fields, for example horizontal occupational and disciplinary distributions, and vertical, hierarchical distributions, by gender. Key institutions here are:

• the UN (both in general, and in relation to ‘women and science’
• UNDP 
• UNECE 
• FAO 
• ILO 
• World Bank

Then of course there are the various sources from the EU and the EC (for example, http://ec.europa.eu/eurostat/statistics-explained/index.php/Gender_statistics). One of many EU reports that addresses this in terms of research funding is The Gender Challenge in Research Funding . Two further major particular EU resources need to be mentioned here.

First, there is the work of EIGE, and its ‘Gender Statistics Database’ and ‘Gender Equality Index’. This includes very useful contextual information, and also advice on understanding gender statistics.

The second is the She Figures reports which summarise the state of gender distributions in science  The She Figures have been a spur to national comparisons. There is, perhaps unsurprisingly, very great variation in the extent national authorities have given full attention to gender statistics in science, technology and innovation. In addition, it is important to note the resources available at country level in Europe. A good example here is the Norwegian site: http://eng.kifinfo.no/c62415/seksjon.html?tid=62420.

Outside Europe, useful sources are: Gender differences in science, technology, engineering, mathematics and computer science (STEM) programs at university, by Darcy Hango, produced from Statistics Canada in 2013; and Gender Differences in Science, Technology, Engineering, and Mathematics (STEM) Interest, Credits Earned, and NAEP Performance in the 12th Grade, by Brittany C. Cunningham, Kathleen Mulvaney Hoyer and Dinah Sparks, produced from the US National Center for Education Statistics in 2015.

There are many further more specific areas where gender and statistics are important and are becoming more so. One is the use of bibliometrics in evaluations, of individuals, research groups, even whole universities. These kinds of statistics have a very uneven usefulness.  They may be more indicative of quality and impact in some disciplines, especially STEM, natural science and medicine, and far less in humanities and social sciences. Using one set of metrics in isolation, such as Web of Science, Scopus, Google Scholar, is very dangerous indeed. This is not least because some metrics focus only or mainly on journals to the neglect of chapters and books, and also have limited coverage of some disciplines, including those where women are well represented. The study of these metrics, their reliability across disciplines, and their gendering and frequent gender bias, has become a specialist area of scientific research in itself. Useful sources here include: Cameron, Elissa Z., White, Angela M., & Gray, Meeghan E. (2013). Equal Opportunity Metrics Should Benefit All Researchers. Trends in Ecology & Evolution, Vol. 28, No. 1, 7-8; Maliniak, Daniel, Powers, Ryan M., and Walter, Barbara F. (2013). The Gender Citation Gap in International Relations. International Organization, Vol. 67, No. 4, 889-922. (); and  Sabaratnam, Meera, and Kirby, Paul, and 200 signatories (2014). Why Metrics Cannot Measure Research Quality: A Response to the HEFCE Consultation.

In sum, it is increasingly necessary not just to demand gender statistics in STI, but also to be able to recognise in a critical way their uses, limitations and even possible abuses. For example, within science itself, the tendency persists to publish experimental results that are positive in showing a significant finding rather than those that are negative or inconclusive (Song, F., et al. (2010). Dissemination and Publication of Research Findings: An updated review of related biases. Health Technology Assessment, Vol. 14, No. 8), and as we all know, correlation does not mean causation! Happy statisticking!