I was recently invited by the University of Luxembourg as a keynote speaker at an International Conference, ‘Gender Variations in Educational Success: Searching for Causes’. It quickly became apparent that boys’ achievement with respect to girls’ is an international, hot topic. National and political concerns in Germany, Switzerland and Luxembourg have created a mass of research dominated by multivariate analysis and structural equation modelling. Scholars from these countries drew heavily on data available from the Programme for International Student Assessment (PISA) to search for the ‘causes’ of boys’ underachievement. Alongside PISA data they used national public examination and test scores, psycho metric measures of cognitive competencies in, for example, reading, mathematics and problem solving, scales of well-being and data gathered from questionnaires designed by researchers to capture, for example young people’s motivations to learn. The availability of international data sets that allow cross-national comparisons in boys’ and girls’ achievements, competences, well-being and motivations has produced a new generation of highly talented researchers trained in multivariate modelling. However, some of the comments and questions at this international conference indicated that many of the younger participants believed that multilevel modelling was the only legitimate or ‘scientific’ way to conduct empirical studies into the causes of educational underachievement. Given the way funding bodies assess research proposals, it is becoming increasingly difficult to undertake qualitative research studies in education. It appears that across Europe a hegemony has emerged around statistical modelling that reflects a long tradition of imaging that quantitative methods are ‘objective’ while qualitative methods are ‘subjective’. We could ask, has gender and education research finally come of age as a proper ‘scientific’ field or is something else happening?
It seems timely and appropriate to reconnect the new wave of gender and education research to the history of the field of gender and education studies in Europe. Why? Because there a danger that quantitative methods that use scales, questionnaires and mathematical modelling will further divorce research from the real life experiences of young people and teachers who participate in everyday life that is full of contingencies as well as more visibly, stable rituals of practice. As feminist scholarship has shown, we will understand ‘causes’ only partially through statistics despite advances in mathematical modelling. There is a danger that gender will be reduced to sex group. As far back as 1968 Robert J. Stoller made the distinction between sex and gender, which is worth retaining.
Within the UK and in other western European countries such as France as well as in Scandinavian countries, the field of gender and education research emerged within political movements campaigning for women’s emancipation. The field reached a critical mass in the 1980s when enough feminist sociologists such as Madeleine Arnot, Gaby Weiner, Miriam David in the UK, Tuula Gordon in Finland and psychologists such as Valerie Walkerdine in the UK and Sandra Bem in the US, to name but a few, established theories about how gender dynamics were recreated through institutional practices in schools and universities. Ethnographic studies of schooling burgeoned in the 1980s and were published in academic journals such as Gender and Education. Later, studies of masculinity and schooling grew rapidly following Connell’s book Masculinities in 1995. The aims of qualitative studies of gender were to understand how gender emerges in everyday flow of classrooms and school practices and further how gender can act as socio-cultural resources that can be taken up and used by students and teachers in interactions and practices. Research demonstrated how such practices often contribute to the reproduction of gender roles and inequalities in education and so in life.
As a social and cultural phenomena gender cannot be captured by sex-group categories such as ‘boy’ and ‘girl’. Furthermore, we need to be constantly vigilant not to slip from statistical inferences to ‘causes’. We know too much about statistical modelling to forget that models can, and often do, reflect the categories that we first use to create a data base and second that statistical models often reflect artefacts of the mathematical equations used in the specific design of, for example, a regression analysis (e.g. Bem, 1993). We can use multivariate modelling to create all kinds of associative links that bear no relation to what happens in classrooms, in teacher-student interactions or in life. Furthermore, there is a grave danger that badly designed self reporting questionnaires will serve simply to reinforce stereotypes such as boys are active, inquisitive and cheeky, risk takers while girls are passive, compliant and boringly good. If a questionnaire asks a boy or a girl if they prefer, to move or sit, mathematics or English, it is fairly predictable that the results will reflect the cultural gender norms in that society. We cannot reduce gender to a factor or assume that sex group categories (boy, girl) correspond to socio-cultural gender norms of masculinity and femininity. There seems to be an unintended amnesia about critiques of studying gender in education using statistical analysis. Could this be because the scholars who forwarded these critiques were women and often feminists?
This is a plea for dialogue across boundaries, be they national, methodological or generational. In one sense this challenge goes to the heart of the central problem within this field of research, namely that power relations and gender are always deeply entangled. Does multi level modelling ‘feel’ like proper science because it aligns with ’hard’, ‘objective’ masculinity while ‘soft’ qualitative methods align with femininity? There is a grave danger that new theoretical work on masculinities which use qualitative methods (e.g. Connell, Epstein, Frosch) and the new wave of quantitative work on statistical modelling around boys’ education in Europe will develop into independent silos. Will young women making their way into the academy using ‘big mathematics’ defend themselves by avoiding qualitative methods? This would be a great shame. If we can’t tackle, or have no motivation to tackle, the gender dynamics within the field of educational research, including the status afforded to qualitative and quantitative methods (cf. Flyvberg, 2008) what chance have we of influencing this dynamic in everyday life? The organisers of the international conference hosted by Luxembourg University had specifically set out to reach-out across these boundaries by actively seeking participation from a range of countries using a range of methods. This conference was an important step in the right direction.
If we wish to understand the causes of gender variation within education we need nuanced approaches that use a range of methods and instruments that can genuinely engage with the flows, contingencies and historical legacies of everyday classroom and school life. Of course, we also need to recognise that to study phenomena, researchers are obliged occasionally to reify and stabilise the flow of everyday life. Only by stabilizing phenomena using categories can we gather a snapshot of life that can then be subjected to complex, quantitative analysis. Yet, we need to be fully aware that the categories, including ‘boy’ and ‘girl’ that we create on an SPSS data bases are abstractions that have scientific use yet are blunt instruments for understanding gender dynamics in human life. We need to remember that whatever factors we choose to focus on, there will be a greater variation within sex-group than between sex-group (Duveen and Lloyd, 2000). The choice of categories requires considerable care. Indeed, we could say that the way we choose to ‘cut’ phenomena (Barad, 2007) is an ethical issue.
Gabrielle Ivinson, Cardiff University.