Binary correlation coefficient and measures of risk

5^th International Conference on Biometrics & Biostatistics

October 20-21, 2016 Houston, USA

Jake Olivier

University of New South Wales, Australia

Posters & Accepted Abstracts: J Biom Biostat

Abstract :

The binary correlation coefficient �? is a natural extension of Pearson��?s correlation coefficient for the 2�?�?2 table. However, �? is not often used as a measure of association due to constraints on the marginal probabilities that also constrain the possible values of �?. The motivation for constructing �? is scaling the covariance for two binary random variables to the interval [-1,1] using the Cauchy-Schwarz inequality. However, it can be shown that �? is usually bounded by a narrower interval and there exists a better inequality for the covariance of two binary random variables. It will be demonstrated that an improved correlation coefficient can be constructed using this inequality. Additionally, it will be demonstrated that the binary correlation can be transformed to epidemiological measures such as the relative risk, odds ratio and hazard ratio when knowledge of the marginal distributions are known.

Biography :

Email: j.olivier@unsw.edu.au