Abstract
Outliers have been of constant concern for statistician. These outliers must be handled carefully while analyzing data, particularly, when the data points are measurements based on experiments conducted in scientific labs. Various studies have been conducted on the comparison of outliers detection tests when the data points were assumed to follow normal distribution or skewed exponential or gamma distribution. However, the data may follow some heavy tailed symmetric distribution. One important heavy tailed distribution is Cauchy distribution which has applications in different fields of science including Physics, engineering etc. In the present study, we introduce two tests, in univariate case, for detecting one or more outliers while sampling is done from Cauchy distribution. The proposed tests make use of the robust statistics, namely, median, quartile deviation and the variance based on "sample free of suspected outlier(s)". We simulate critical values and powers of the tests to compare them with the tests available in the literature. We compare their efficiency using power criteria.