With my colleagues at the IDRC Employment and Growth program we had a very interesting discussion about papers analysing factors determining labor market entry. Now the discussion happened over lunch, where discussions tend to be rather boisterous, and let me declare that my comments on ‘risk of marriage’ may have sounded somewhat chauvinistic. Interestingly, in this group combining Latinos, Asian, etc, married, unmarried, and divorced – we all seemed to agree about the notion of risk in that case – but we disagreed on what this means, and this split the group in economists and non-economists (but let me reassure our managers and funders we all went back to our desks and work after ….) My claim that marriage doesn’t do anything was met with complete dismissal, by some of my best friends who are economists.
Our substantive difference, I think, revolved around the idea of causation. Now let me quickly say that I do not (just) mean correlation is not causation. Development economist, since the great sociologist Durkheim have found increasingly sophisticated ways of understanding not just that pattern of events coincide, but also which of the two leads to the other. I ask my economist friends economists to give me a good list of improved methods, but I imagine this would include lagged variables (education impacts economic growth decades later), instrumental variables (do not rely on Wikipedia but this seems to makes sense), and of course Randomised Control Trials.
Let’s say that by the gold standard we can establish that marriage causes a change in women s access to labour markets – hopefully not a contentious claim. As a sociologist I still argue that the idea of causations is limiting. This causation is still within the analysis, the model, the trial – which is still, of course, an abstraction of real life. Marriage itself does not cause anything (I might be exaggerating slightly but bear with me). It is the actions that people take (after marriage) that cause things: husband and in-laws holding women back from going to work, women deciding to quit their jobs, employers trying to get rid of women or trying to create an inclusive work environment that allows for different needs, etc. And it is perceptions that inform those actions: the norms around whether women should work, what this would mean for companies profits, women’s aspirations and reactions in the environment, how power is divided etc.
I think the above is not insignificant. I think I am starting to see the importance of Naila Kabeer s warning not to turn gender into a dummy variable. The distinction seems equally relevant for recent debates around inequality and voting behavior and the populist backlash as well. It is not the variable in the analysis that causes anything, it is actions and behaviors that do that, and this in turn is informed by perceptions and the constraints in which they live, as another great sociologist Anthony Giddens has taught. The above also is not to dismiss the importance of the causal analysis mentioned here, but to say that it is not enough, may be insufficient to understand human behaviour. Perhaps an obvious point, but better inter-disciplinary seems a priority, still.
And finally, I also think this is relevant for policy, and to consider how evidence shapes policy and action. Even gold standard causal analysis does not tell us enough about what policies works, unless it understands how the causal factors shape perceptions and actions – in that sense it is still correlation, and hence as crucial. The evidence also needs to be used directly by practitioners, as I think Pasi Sahlberg argues with respect to education policy and evidence. It seems on gender equality, understanding what causes what and how is particularly important.
PS: I would love to hear if you are struggling with the same questions, and whether you think they matter for policy. Also, if you are interested to work in this research field, and are not discouraged by my image of our team lunches, please check outhttp://bit.ly/2kxvugq