Document Type



The reality television show American Idol has some biases which display themselves in data charts and curves of contestants’ telephone voting patterns. Some of the biases examined include issues of geography and performance order. This paper shows that the format of the show leads to biases in favor of contestants who are from, and have large fan bases in, certain parts of the country. Another major flaw is a bias in favor of contestants who sing later in the show. I show that the order the contestants sing in is not random chance but instead carefully chosen by the producers. Using freely available data, these biases are exposed and ways to correct for them are explored.

Author Bio(s)

Jason Gershman, Ph.D., associate professor and coordinator of mathematics at the college, earned his Ph.D. in Statistics from Rice University in 2008. His research has recently been published in the proceedings of the Joint Statistics Meetings of the American Statistical Association and the International Conference on Bioinformatics and Computational Biology. His work on analyzing data from reality-television shows including American Idol has been featured in the Washington Post, Sun-Sentinel, Vancouver Sun, and Montreal Gazette. Gershman also has been featured on South Florida's Local 10 (ABC affiliate) and WSVN 7News (FOX affiliate) to talk about his research work in this area and provide expert commentary and predictions.