Thursday, September 10, 2015

The Triangle Method of Research Design

How do you start thinking about a research project? I find that an adaptation of Baron and Kenny’s (1986) mediating-effects design is helpful. That design showed three variables, as shown here. A student of mine is interested in the effect of the quality of early intervention programs in a couple of European countries on family empowerment. So her independent variable is the quality of programs, and her outcome variable is empowerment. To make the question interesting but also to see determine what might explain (i.e., mediate) or alter (i.e., moderate) the effects of quality on empowerment, she needs to think about a third variable. Is it a demographic variable, like family socio-economic status or the severity of the child’s disability? Is it another psychological variable, like the respondent’s locus of control or parenting beliefs? Is it a specific characteristic of the independent variable (i.e., quality), like dosage—how much service time the family experiences—or the number of professionals?

By inserting the third variable, the study becomes interesting and relevant and not (yet) too complicated. I usually then ask how the mediator or moderator, pictured here, works for different conditions. These are often background variables, such as demographics. How do these relationships work for different SES levels or different child ages, for example? They might be more socio-demographic, such as different countries, different states, or different program types. 

When starting to think about a research project, remember the triangle!

Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of personality and social psychology, 51, 1173.