Introduction

Variability abounds in our environment, such as in daily weather patterns, the number of accidents occurring on a certain highway per week, the number of patients in a hospital, the response of a group of patients to a particular type of drug, or the birth/death rate in any town in a given year. The “observed” occurrences in nature or in our man-made environment can be quantitatively described using “statistics” such as mean value, standard deviation, range, percentiles, and median. In the fields of biology and medicine, it is very difficult, and often impossible, to obtain data from every individual that possesses the characteristic of interest that we wish to observe. Say, for instance, we want to assess the success rate of a coronary drug-eluting stent. To do this, we must define our population of interest – patients who underwent coronary stent implantation. It is an almost impossible task to track down and obtain the medical records of every individual in the world who has been treated with a drug-eluting stent, not to mention the challenges faced in obtaining consent from patients for collection and use of their personal data. Therefore we narrow our search method and choose only a part of the population of interest (called a sample), such as all Medicare patients of age >65 that received a coronary stent from May 2008 to April 2009.