Study to Advance Statistical Approaches to Understanding Behavior

A computer program being developed by an IHRP researcher will provide a new tool to measure variations in influences on human behavior, beginning with teen smoking.

After creating and testing this program over the next two years, Donald Hedeker, professor of biostatistics in the UIC School of Public Health, will share the program with other researchers.

Funded by the National Cancer Institute, Hedeker’s program will be used to analyze large volumes of data collected during a major IHRP study examining smoking patterns among adolescents. In part of that study, 461 teens used handheld computers to complete about 30 surveys over a week, which they did four times over two years. Each survey provided 30 to 40 in-the-moment observations about the teen’s moods, activities, behavior and social interactions.

Hedeker’s program will trace mood fluctuations among the study’s teens and identify connections with their smoking behavior.

Robin Mermelstein, IHRP director and lead investigator of the study examining smoking patterns among teens, is collaborating with Hedeker in this research.

Tobacco research suggests that smokers use cigarettes to stabilize or take the edge off their mood.

“Taking the edge off isn’t a matter of making your mood better or worse, it’s about making it constant across the day,” Hedeker said.

“With 30 or 40 measurements per subject, we can characterize not just whether they are high or low on average, but how erratic they are, or how constant they are across these 30 or so measurements,” he said of his new program.

Current statistical approaches and software do not allow for this kind of analysis.

“By and large, statistical models are meant for looking at averages,” he said.

Earlier this year, Hedeker, Mermelstein and colleagues reported in the journal Addiction that teens who smoked more frequently exhibited less smoking-related mood variation, when times they smoked were compared to random times they didn't, than less frequent smokers. The article described some of the statistical methods that will be at the heart of his computer program.

Hedeker will develop the program so that it can interface with statistical software packages commonly used by researchers, such as SAS or SPSS, and make it available for free to other researchers. He did similar work on software development in the 1990s with federal funding.

There is an increasing demand for such statistical programs that analyze complex data about human behavior. Technological advances, especially the Internet and handheld computers, have increased the efficiency and accuracy of data collection. But analytical methods don’t yet match the sophistication of these tools, said Hedeker, who sees great potential in software and methods like his for research into psychiatric disorders such as schizophrenia, the study of pain and symptoms, diet and exercise.

Hedeker is a member of the IHRP Methodology Research Core, which provides methodological expertise to IHRP researchers.