Quote of the day: Pinker on changes in social sciences

By Matt Nolan 09/10/2013


Via Noah Smith.  We have the following post on the Pinker vs Wieseltier debate on science and humanities (if you have a chance I would suggest reading the debates themselves as well).

The era in which an essayist can get away with ex cathedra pronouncements on factual questions in social science is coming to an end.

Very good, and Pinker’s co-operative version of science with the humanities seems appropriate to me (where instead we are merely asking about how to deal with certain propositions and using the best tools available).  I think Pinker won this debate, I am unsure why Wieseltier felt it necessary to take such an extreme position though – I think he initially believed Pinker was trying to force through a view based on the superiority of scientific authority (one that Pinker rules out in his initial article!), when he was really just suggesting the use of the scientific method (namely introducing a degree of the positivist view of theory creation) given the improvements in data availability and usability we have had.

As XKCD says:

But even within Pinker’s reasonable claims there is one area where I would be a touch careful – a direction I was hoping the debate would actually go in!   I would just caution being too confident about placing beliefs on the basis of ‘empirical fact’.  We should definitely use the information and update our beliefs, but the Duhem-Quine thesis is even more binding in the social sciences than it is in the physical sciences – due to the lack of natural experiments and that more complicated causal chains involved.  [Note:  Would have also enjoyed a free will vs determinism debate]

Language and rhetoric allows us to give this context and give alternate hypotheses and elements of heterogeneity in society a fair go – even looking straight at empirical data, the use of these allows us to know what we can’t measure, helps establish a limit to the use of data, and helps us ‘pick’ what we should be trying to measure!  We should definitely make use of empirical data, and use it to establish underlying premises – but when it comes to writing an essay or op-ed the premise established from data could conceivably be so far in the background of the argument (due to the conditional nature of its use) that essayists may appear to be making ex cathedra pronouncements on issues that ‘on the surface’ appear to be factually false, but are actually appropriate.

Just to take the example in the piece, the 30-60% of Americans saying they take the bible literary (our fact) just tells us that this is what they reported to someone – it is not a revealed preference.  For this we need to see actual behaviour when making choice.  We could argue that they were saying this “to sound good to the interviewer” – even unconsciously.  Given this, the true number is well lower, and our view that literal views in religion are not widespread survives as a premise for whatever claim we are making.  In this way, the writer really has to appeal to authority – they do not have the space to write this out, but they have the argument around it in their backpocket.  Directly old auxiliary hypotheses!

I’d also note I really don’t know terribly much about these things, and it would make a lot more sense to find someone who works in an interdisciplinary field which already does this.  Someone who has a lot of experience with empirical data and models, but also write about language and works in a history focused type field.  Someone with an Economics History background.  The answer is pretty clear for the economists out there – Deirdre McCloskey.