[We’re pleased to welcome Paul Bliese of University of South Carolina. Paul recently published an article in Organizational Research Methods entitled “Understanding Relative and Absolute Change in Discontinuous Growth Models: Coding Alternatives and Implications for Hypothesis Testing” with co-author Jonas W.B. Lang.]
Jonas and I became interested in the topic because we kept encountering “transition events” that could lead to discontinuous change and wondered how to statistically model the events. For instance, a combat deployment represents a potential transition event in the career of a soldier. Likewise, unexpectedly changing a complex task on a participant in a lab represents a transition event that could be frustrating and impede performance. As a final example, letting sleep deprived participants get a full night’s sleep is a positive transition event that should improve cognitive performance (but may not do so equally for all participants). In all these examples, some pattern of responses is interrupted by the transition event; however, where the models are really useful is in trying to understand the patterns of change after the transition event because individuals rarely react in the same way.
When Jonas and I got into writing the manuscript we were really surprised by how some minor coding changes surrounding TIME could produce parameter estimates that had quite different meanings. In fact, I realized that if I had figured out all the details that went into the submission years ago, I probably would have specified and tested hypotheses differently in my own publications where I used the approach. My hope is that other researchers will use the manuscript as a resource to study other transition events and that the examples will help provide better specificity to the types of hypotheses researchers can propose.
The abstract for the paper:
Organizational researchers routinely have access to repeated measures from numerous time periods punctuated by one or more discontinuities. Discontinuities may be planned, such as when a researcher introduces an unexpected change in the context of a skill acquisition task. Alternatively, discontinuities may be unplanned, such as when a natural disaster or economic event occurs during an ongoing data collection. In this article, we build off the basic discontinuous growth model and illustrate how alternative specifications of time-related variables allow one to examine relative versus absolute change in transition and post-transition slopes. Our examples focus on interpreting time-varying covariates in a variety of situations (multiple discontinuities, linear and quadratic models, and models where discontinuities occur at different times). We show that the ability to test relative and absolute differences provides a high degree of precision in terms of specifying and testing hypotheses.
You can read “Understanding Relative and Absolute Change in Discontinuous Growth Models: Coding Alternatives and Implications for Hypothesis Testing” from Organizational Research Methods free for the next two weeks by clicking here. Want to stay current on all the latest research from Organizational Research Methods? Click here to sign up for e-alerts!