In Brief

A paper in this month's issue of Psychological Methods (Vol. 10, No. 2) provides a new technique for researchers seeking to use structural equation modeling to study how pairs of people interact.

Many questions in diverse areas of psychology, from developmental to clinical to social, involve understanding how people affect and influence one another, says co-author Erik Woody, PhD, a clinical psychologist at the University of Waterloo in Waterloo, Ontario.

For example, explains co-author Pamela Sadler, PhD, of Wilfrid Laurier University, also in Waterloo, a psychologist might bring two people who've never met into a lab and observe how they work on a problem together. The psychologist might ask questions like: Do they tend to take on complementary roles? How does their impact on each other change over time?

Many behavioral researchers are aware that the statistical technique of structural equation modeling can help them answer questions like those. In the past, however, it only worked when the people in the pairs were distinguishable by some characteristic, such as gender, that was relevant to the research--like male-female pairs in a study about couple interactions.

The problem, says Woody, is that for an interchangeable pair--say a pair of two women--either participant could be assigned to be "participant one" or "participant two" in the data analysis. But flipping the designation around, as should be logically possible, gives two different answers.

The researchers' new technique, which builds on the work of psychologist David Kenny, PhD, an expert on research methods at the University of Connecticut, provides answers that aren't affected by those assignments.

"In the past, sometimes, researchers have tried to avoid the problem by ignoring the fact that people in interchangeable dyads influence each other," says Sadler. "Our article offers a unified and conceptually motivated SEM approach to these kinds of models."