Thursday, April 06, 2017

BSing with phony precision

The problem with the claim that every H1B visa creates 1.83 new American jobs is the absurd precision with which this result is put forward. We are dealing with a topic of vast complexity, where a myriad of causal factors are interacting to produce the observed outcome, and for which we have no control group and no ability to do repeated experiments using a controlled environment.

"But we used the most sophisticated statistical techniques to produce our result!" the researchers might respond. Very amusing: I guarantee that, with the same data set, an anti-H1B visa group could produce a study that "proves" that for each H1B visa issued, .794321 American jobs are lost.

"1.83" is an attempt to snow the reader with precision: "Wow, if they can cite the number to 2 decimal places, they must really know what they are saying!"

Here is a claim I would be willing to believe: "Based on our research, we are pretty sure that each H1B visa issued creates between one and three new American jobs."

3 comments:

  1. According to the table in this article, the average human male contains 32376.50 Calories.

    https://arstechnica.com/science/2017/04/counting-calories-of-going-cannibal-on-a-paleo-diet-human-flesh-is-just-meh/?comments=1&post=33116215

    ReplyDelete
  2. Basically, similar to how a study showing 65% correlation sounds plausible, but when you claim 95% correlation, you are probably forcefully fitting the data between the sets.

    ReplyDelete
    Replies
    1. I read, and this should terrify you, an account of a government employee who was ordered by his boss to separate and sort the dependent and independent variables before doing a regression. You get stronger results that way it seems.

      Delete

Distraction Deterrents in Small Contexts

"distracted from distraction by distraction" - T.S. Eliot I've been reading a little on how Facebook and other social netwo...