Discrimination and Statistical Arguments

[Una versione Italiana di questo post è disponibile]

It is recent news that President Trump issued an executive order limiting access to USA for people that were born in seven specific countries have a “high jihadist presence”.

The decision caused a series of problems for many people, even for holders of valid visa or green card, that could not (re-)enter USA.

There were huge protests both in the USA and in other countries. Obviously also on the social networks there have been several criticisms. Most criticisms are based on an argument that I’d call “statistical” which, though formally correct, stands on a premise that is totally wrong.

Statistical argument

A sample of such argument can be found in several tweets based on the following image:


Or other  tweets, are based on the following image (or variations thereof):


I didn’t check the numbers but we can consider them correct.

The premises

This kind of arguments that I called “statistical” basically tell us that the executive order just had the wrong target.

The point is that such argument is based on the premise that there could be any “right” targets for a ban. Such a premise is absolutely wrong and unacceptable.

It is unacceptable from the point of view of any inclusive democracy that has the civil and human duty of welcoming the troubled foreigners. As it is unacceptable from a Christian point of view, that many local “thinkers” (often hypocritically) recall.

Also, the premise is logically wrong: such kind of premises would bring to ban e.g. Italian because Italy has widespread Mob. And extending a little bit (not too much), in the era of big data, it is the same kind of premises that would ban you because of what you wrote on your Facebook log in the previous years. We are returning to the pseudo-science of phrenology and Lombroso’s criminal anthropology.

The very idea of banning human beings on the basis of their origin, religion, gender or any other like feature must be rejected as a whole, while any statistical argument does implicitly support it.