Simon’s Paradox

Nov 22 / Network Capital
Trends and analyses that appear in small data-sets may disappear when amalgamated with a larger data-set. Do not assume data as a fact in itself. Always look at the variables and the data-set and beware of even the strongest correlations.

Subscribe to read more

In the year 1973, University of California, Berkeley did a study on gender bias in graduate admissions for the university. According to the admissions data, it was clearly evident that men were favoured by the admissions committees. Men had a 45 per cent acceptance rate, while women had a 35 per cent acceptance rate. After looking at university level performance, the researcher next looked at individual departmental records. The results here were shocking. They showed that six out of eighty-five departments were significantly biased against men. On the other hand, only four departments were actually biased against women.

The research then concluded that women tended to apply for more competitive departments which caused a steeper acceptance rate for them. Whereas men applied for easier-to-get-into and less-competitive departments that relaxed their acceptance rates.

Named after Statistician Edward Simon, the Simon’s Paradox is a popular social sciences and medical sciences phenomena in which there is a paradoxical relationship between the smaller data groups and the larger data-set.In addition to highlighting the deviation in the data-set trends, Simon’s Paradox is an extremely relevant method to test the validity of trends, causal relations and parameters of a data-set. It is a mechanism to prevent exploitation or incorrect analysis of a limited data-set.