Discrimination or favouritism: a tricky and interesting question

I’ve often considered this distinction at the back of my mind, but never really given it much explicit thought. While actively hostile discrimination – for instance on the basis of race of gender – is still around, there’s not much of it about. On the other hand people not only like those they’re already familiar with, but are more comfortable with those who are like them. This is clearly discrimination (in one sense) and it’s clearly bad for you if you’re not in the favoured group. But it’s not driven by hostility except in so far that one might say that we retain some instinctive hostility towards or fear of strangers.

Anyway here’s a paper that explores the distinction between what it calls ‘discrimination’ on the one hand and ‘favouritism’ on the other, or if you feel like a little greek, endophilia versus exophobia. It looked at the marks examiners gave university papers when the nationality and/or gender of the student matched theirs and when it didn’t.

It is clear that there is substantial endophilia by nationality in the grading. A student who matches the grader’s nationality receives a score that is 0.17 standard deviations higher when her name is visible than when it is not.  . . . This effect is also economically important: 1t is equivalent to moving from the median score to the 57th percentile of the distribution of scores. Its magnitude is similar to that of the effect of large differences in teacher quality on students’ test scores that was found by Rivkin et al (2005).

While favoritism by nationality exists in grading, there is no apparent exophobia by nationality: The estimated impact of being visible when not matching by nationality is small and positive. The results of estimating the regression examining gender matching are shown in Column (2) of Table 3. Although the point estimate suggests the existence of endophilia, we cannot reject the hypothesis that it is zero. For non-matches there is exophilia, but here too the impact is statistically insignificant and also minute.

On average grading seems gender-neutral in all dimensions. . . . Neither male nor female graders exhibit significant endophilia or exophobia, and for both men and women the absolute impacts are tiny. Again, there is no sign of either statistically significant or important differences in behavior depending on the match or non-match of the grader’s and student’s gender.


  1. I[]
This entry was posted in Economics and public policy. Bookmark the permalink.
Notify of

Inline Feedbacks
View all comments