## INTRODUCTION

Investigators in the health and biological sciences use a tremendous variety of experimental techniques to gather data. Molecular biologists examine fragments of DNA in test tubes and chromatographs. Physiologists measure the function of organ systems with transducers in whole animals. Clinical investigators observe the response of people to medical and surgical treatments. Epidemiologists collect information about the presence of diseases in whole populations. The types of data collected by each of these investigators are as different as the methods used to collect them. These differences are so great that the observations and methods of one discipline can be incomprehensible to people working in another. Once the data are collected, however, they are analyzed using common statistical methods.

Because few biomedical investigators have received much formal training in statistical analysis, they tend to use very basic statistical techniques, primarily the t test. The t test and related methods such as analysis of variance are known as univariate statistical methods because they involve analyzing the effects of the treatment or experimental intervention on a single variable, one variable at a time. Although there is nothing intrinsically wrong with such an analysis, it provides only a limited view of the data.

To understand the limitations of a univariate approach to analyzing a set of data, consider the following example: Someone approaches you, says that he or she is feeling short of breath, and asks you what you think the problem is. You first note that the person is male, then recall everything you know about the relationship between gender and shortness of breath. Second, ask the person's age; the person is 55. Knowing that the person is 55, recall everything you know about shortness of breath in people who are 55 without taking into account the fact that you know the person is a man. Third, ask if the person just exercised; the person walked up two flights of stairs. Knowing that the person walked up two flights of stairs, recall everything you know about exercise and shortness of breath without taking into account the gender or age of the person. Fourth, find out if the person smokes; the person does. Knowing that the person smokes, recall all you know about smoking and shortness of breath without taking into account the gender or age of the person or the fact that the person just climbed two flights of stairs. Fifth, ask the person his favorite food; it is deep-fat fried egg yolks. Knowing that the person loves deep-fat fried egg yolks, recall all you know about diet and shortness of breath without taking into account the gender or age of the person, the fact that the person just climbed two flights of stairs, or the fact that the person smokes.

So what can you say? Gender alone tells very little. The fact that the person is 55 could indicate that he is in the early ...

### Pop-up div Successfully Displayed

This div only appears when the trigger link is hovered over. Otherwise it is hidden from view.