Document Actions

You are here: FRIAS Events Other Events Lecture - Wilfried Hinsch

Lecture - Wilfried Hinsch

Differences that Make a Difference – Statistical Profiling, and Fairness to Individuals
When Dec 07, 2020
from 01:00 PM to 02:30 PM
Where Zoom-Meeting
Contact Name
Attendees Universitätsoffen/ open to university members
Add event to calendar vCal
iCal

Differences that Make a Difference –
Statistical Profiling, and Fairness to Individuals
 

Cordial invitation! Please send an  to take part.

Vortrag - Wilfried Hinsch.jpg

It is often considered to be unfair to impose a disadvantage on somebody for the only reason that he or she belongs to a group of persons who share a common statistical feature by virtue of being a member of the respective group. Relevant features may be ‘crime-proneness’ or ‘probability for credit default’; relevant group memberships may be defined in terms of ethnicity, gender, age, social class or place of residence.
It has been suggested that it is unfair to discriminate people based on information that statistically correlates with group membership since not all members of the group will actually have the feature—like proneness to commit a crime or to default on a loan—that would justify the imposition of a disadvantage. It, therefore, seems that, from a moral point of view, differences between statistical features of persons alone should never make a difference: The principle of fairness seems to require that every individual case is considered on its own merits.
This argument, however, rests on the spurious assumption that there is a categorical and morally relevant difference between statistical knowledge about groups of persons and (supposedly) non-statistical knowledge about individuals. Indeed, what we know about individuals is statistical, and hence probabilistic, in nature as well.
Nevertheless, discriminating between persons on the basis of statistical information often will be—rationally and morally—objectionable: Firstly, because of defective statistical and probabilistic reasoning. People make mistakes in choosing an appropriate and sufficiently narrow comparison group when ascribing statistical features to individuals. Secondly, because morally relevant information is ignored. People do not always appreciate that even statistical features that justify discriminatory treatment do so only in a pro tanto fashion. Often, recourse to these features will be defeated by other morally relevant considerations. These considerations cannot be deduced from an abstract or procedural principle of fairness. Rather they are contextual in nature and involve substantive conceptions of individual harm and social justice.
The main goal of my contribution is to elucidate this last point and to sketch a line of reasoning which allows to appropriately distinguish between individual differences that should make a difference and those that should not.