Syndicate content

Add new comment

Submitted by Igor Louboff on
Most authors (no matter their position in the debate) seem to rely on a common assumption, namely that there actually is a methodological approach that allows a rigorous assessment of the types of causal questions we are interested in. In other words, the starting point adopted by most scholars and practitioners consists in claiming that there is a way to isolate the influence of a particular causal factor and to measure its consequent effects. The striking point here is the apparent absence of any discussion over how warranted the latter assumption might be when dealing with policy questions requiring the analysis of complex systems. When detailed and correctly exposed, arguments related to causal complexity are generally well accepted (e.g. impossibility of isolating a particular factor acting as part of a much wider set of interacting influences -virtually most cases in social sciences). However, this supposedly collective wisdom seems to vanish as we step into individual argumentations. My point is quite simple: by focusing on which particular methodology should be favoured under a given set of conditions, we totally ignore and discard the possibility that the causal question being assessed might in fine be unaddressable. Shifting the focus from methodologies to policy questions appears to allow solving redundant concerns, such as the one pointed out above. Why do we observe opposite results when employing different methodologies? Is it that one particular methodology is less appropriate than the other in capturing reality? Or is it that the several layers of uncertainty underlying both methodological approaches result in the generation of large approximations? Moreover, many interventions entail only small changes (particularly in microfinance), making approximations even more hazardous. In brief, while the methodological debate is a healthy one, wider attention should be given to the types of causal questions we can actually address with a limited range of uncertainty. Anyone working with causal inference instruments in the study of social phenomena knows how much uncertainty lies behind any causal assessment, and knows how many questionable (and often unwarranted) assumptions are generally required to support the most basic inferential conclusion. Despite this awareness, the debate remains focused on the study of ‘methodologies’ rather than turning to the study of ‘policy questions’ and their underlying causal structure.