Media Blogpost: Statistics in Law

Statistics in Law (Media Post) 

BAZEMORE ET AL. v. FRIDAY ET AL.(1986) 

Many times in class we have discussed the enormous influence that statistics have in the media, yet, we haven’t discussed how influential statistics can be on the legal system.

Bazemore v. Friday was a legal case involving pay discrimination between black agents and white agents in the North Carolina Extension Service.  On the case, the black agents in question presented multiple regressions that showed that the black agent’s salary was lower than the white agents’.

The media, which covered this case extensively, was amazed by the Court of Appeals decision that the regressions presented by the plaintiff were “unacceptable evidence of discrimination” given that on the regression they did not account for many variables that affected income such as years of experience and education. 

A regression analysis is a statistical technique to estimate the relationships among variables, which can be used for statistical inference. In every regression model, a dependent variable is established and  the regression is estimated by the correlation of many other parameters .

In the case of Bazemore v. Friday, the regressions presented used the salary of the agents as the dependent variable, with other parameters used as explanatory variables. 

The court was right on its decision of not allowing the regressions to be used as evidence due to the parameters used. Given that the regressions omitted various important variables that affected the salaries, the omitted variable bias might overcompensate or under compensate the results of the regressions; thus, leading to a possible wrong conclusion. 

I find this case incredibly relevant on the misuse of statistic due to the nature of it. So far in class, we have discussed how the collection of data might be bias, yet, this case is an example on how statistical analysis can be modified to misguide decisions.

References:

http://supreme.justia.com/cases/federal/us/478/385/