Statistics:
Understanding Data

Marie D. Thomas, California State University, San Marcos

The Tables Are Turned:
A Psychologist Becomes a Research Subject

PROLOGUE

FOR 12 MONTHS I WAS a participant in a research project that was designed to compare the effects of “traditional” and “alternative” diet, exercise, and stress-reduction programs (Riegel & others, 1996). Volunteers who were randomly assigned to the traditional program were taught to eat a high-fiber, low-fat diet; do regular aerobic exercise; and practice a progressive muscle relaxation technique. Participants who were randomly assigned to the alternative program received instruction in yoga and in a meditation technique, along with a diet based on body type and tastes. I was randomly assigned to the no-treatment control group, which was monitored throughout the year for weight and general health but received no diet, exercise, or stress-reduction intervention.

The participants in the study were drawn from a large medical group. Invitations to participate in the study were sent to 15,000 members of the medical group. Out of that initial pool, 124 volunteers were recruited, and about 40 were randomly selected for each group—the traditional, alternative, and no-treatment control groups. The participants included men and women between the ages of 20 and 56. A total of 88 subjects lasted the full year. The researchers were pleased that so many of us stayed with the project; it isn’t easy to get people to commit to a year-long study!

Data collection began even before participants found out the group to which they had been randomly assigned. We were mailed a thick packet of questionnaires covering a wide range of topics. One questionnaire asked about our current health status, use of prescription and over-the-counter medications, use of vitamins, and visits to both physicians and alternative health care practitioners. Another questionnaire focused on self-perceptions of health and well-being. Here we rated our mood, energy level, physical symptoms, and health in general. A lifestyle survey requested information about diet (how often did we eat red meat? how many servings of fruits and vegetables did we consume a day?), exercise (how many times per week did we do aerobic exercise?), and behavior (such as cigarette smoking and consumption of alcoholic beverages). The lifestyle survey also assessed psychological variables such as levels of stress and happiness and how well we felt we were coping.

At our first meeting with the researchers, we handed in the questionnaires and were told which of the three groups we had been assigned to. We returned early the next morning to have our blood pressure and weight measured and to have blood drawn for tests of our levels of cholesterol, triglycerides, and glucose. The two intervention groups also received a weekend of training in their respective programs. In addition to daily practice of the techniques they had been taught, people in the traditional and alternative groups were expected to maintain a “compliance diary”—a daily record of their exercise, diet, and relaxation/meditation activities. The purpose of this diary was to determine whether health outcomes were better for people who practiced the techniques regularly. At first I was disappointed when I was randomly assigned to the control group because I was especially interested in learning the alternative techniques. However, I was relieved later, when I found out how much detailed record keeping the intervention groups had to do!

The researchers accumulated even more data over the year-long period. Every 3 months, our blood pressure and weight were measured. At 6 and 12 months, the researchers performed blood tests and asked us to fill out questionnaires identical to those we’d completed at the beginning of the project.

The study included many variables. The most important independent variable (the variable that the researcher manipulates) was group assignment: traditional program, alternative program, or no-treatment control. The dependent variables (variables that are not directly manipulated by the researcher but that may change in response to manipulations of the independent variable) included weight, blood pressure, cholesterol level, self-perceptions regarding health, and mood. Since the dependent variables were measured several times, the researchers could study changes in them over the course of the year.

This study can help to answer important questions about the kinds of programs that tend to promote health. But the purpose of describing it here is not just to tell you whether the two intervention programs were effective and whether one worked better than the other. In the next couple of sections, I will use this study to help explain how researchers use statistics to (1) summarize the data they have collected and (2) draw conclusions about the data. The job of assessing what conclusions can be drawn from the research findings is the domain of inferential statistics, which I’ll discuss later in this appendix. We’ll begin by exploring how research findings can be summarized in ways that are brief yet meaningful and easy to understand. For this, researchers use descriptive statistics.