Saturday, May 6, 2017

Neil Wilson — Empirical Data Considered Harmful

It is like collecting data about the movements of a man in a straitjacket. That data can tell you very little, if anything, about what will happen when you take the straitjacket off.
When you hear somebody demanding that you match the ‘empirical evidence’ or implying your ideas don’t fit the data, then what they are really saying is that they want you to keep the straitjacket on and not change anything substantial.
Empirical data first has to be passed through a policy filter to discard that data incompatible with the new policy settings under examination. And that may very well leave you with an empty set.
Modern Money Matters
Empirical Data Considered Harmful
Neil Wilson

6 comments:

Ignacio said...

If you believe in magic and "invisible hands", if you beleive money grows on trees or under the earth and then you go and collect it, and if you believe that the whole system has a live of its own and is not made up by humans, then you believe data to be "neutral" and valid, just like any sort of physical phenomena measurement. They don't see the straitjacket neither their hands adjusting how tight it is, they have been stripped of eyes and are unable to observe. The whole thing is profoundly unscientific.

Is just like money, remember most of the neoclassical foundations settle on Say's law, a "law" which is patently false (Say himelf recognized so much), yet economists choose to ignore this uncomfortable truth and pretend we all live in some sort of bartering economy, but instead we use money, but it has no effect whatsoever (yeah, riiiiiight).

Dan Lynch said...

I'm not sure what Neil's point is? Climate change data is not perfect, therefore we should ignore it?

Empirical data is nearly always less than perfect and less than complete (and the same can be said for armchair theory). Before firm conclusions can be drawn from empirical data, the claim being made should be replicated by different people under different circumstances. The more data and the more eyes analyzing the data, the better.

All that said, empirical data carries far more weight better than unsupported armchair theory or unsupported old wives tales.

I do a lot of technical experiments & data analysis. I am continually frustrated that my data is compromised by Murphy's Law and by variables that I either failed to anticipate or was unable to control. The solution is not to condemn empirical data but to get more and better data, because there is no credible alternative.

NeilW said...

"The solution is not to condemn empirical data but to get more and better data, because there is no credible alternative."

Before aircraft were invented how would you propose collecting empirical data on how they fly in reality?

The point is that the data is tainted. Particularly in human situations where it is not a universal truth - unlike climate change which is bound by unchanging physical processes even if the chaos that causes the outcome is hard to pick apart.

If you rely on data that is fundamentally tainted by the policy settings in which it was collected, then it is inapplicable to the new proposed situation. So trying to ascribe 'weight' to it isn't science. It is religion. You are curve fitting to a belief using unsubstantiated claims.

Peter Pan said...

When you hear somebody demanding that you match the ‘empirical evidence’ or implying your ideas don’t fit the data, then what they are really saying is that they want you to keep the straitjacket on and not change anything substantial.

What they are saying is vague. It might have something to do with the interpretation of data. It might be ideologically driven. Without further elaboration, this sort of 'demand' can and should be dismissed.

It is like collecting data about the movements of a man in a straitjacket. That data can tell you very little, if anything, about what will happen when you take the straitjacket off.

Motion capture analysis would be useful in both circumstances. That would be an example where the initial methodology is maintained.

Issues pertaining to the collection and interpretation of data can be addressed in a climate of open inquiry. It can be addressed through specific objections in place of vague ones.

Six said...

Dan, how do we collect empirical data on climate change solutions without first enacting the solutions? You can only run sims and collect data from the sims which will omit billions of variables.

Noah Way said...

Empirical data first has to be passed through a policy filter

Just like propaganda has to be passed through an information filter.