President Elect’s Column

George Tasca
Giorgio A. Tasca, Ph.D.

How I Learned to Stop Worrying and Love Statistics 

One of the things that I have noticed over the decades of providing, training, and supervising group therapy is that each group appears to have its own qualities and trajectory. That is, each group seems to have its unique characteristics and growth patterns that set it apart from other groups of its kind. (I don`t mean to imply that there aren`t similarities across groups, but only that in many ways each group is unique). We know from group research that sometimes the group’s path is determined by its pre-group history (personality characteristics of the individuals that pre-dated their membership in the group), sometimes it is affected how members get along with others of the group in which they happen to find themselves (group composition), and sometimes the trajectory is affected by qualities of the leaders. These constituent components (individual characteristics, group composition, and leadership) then interact in complex ways. Gary Burlingame and his colleagues referred to these complex levels of interaction as the structural aspects of groups (member to member, member to leader, and group as a whole). The results of interactions across these structural aspects over time result in what some group researchers refer to as emergent properties of groups. This is akin to what the gestalt psychologists refer to as the “whole being more than the sum of its parts”. It is only recently that group researchers have the tools to catch up to these complexities of groups.

For practice oriented group psychologists and for group psychotherapists the concept of group emergent properties from its structural elements was implicitly known. However, there wasn`t a whole lot of research to quantify, demonstrate, or test these fundamentally group concepts. For decades, group researchers did their best by borrowing methods from individual psychology and psychotherapy to study group phenomena. This severely limited what the researcher could do and could say about groups and the individuals that made them up. For example, in the past, when we studied if an individual`s personality affected their outcomes in group, we simply did what individual therapy researchers did – correlated a pre-treatment personality scale score with change in an outcome. This method essentially ignored the group – even though we knew that the group (composition, leadership, group as a whole) likely interacted with individual personality and outcome. Until very recently there has been little research on how the group affects an individual’s experience of cohesion or alliance and vice versa. New research shows that if the individual and group agree on their experience of the alliance, then the individual`s outcomes are better. This is an example of group researchers finally being able to test what clinicians implicitly knew to be true.

This small revolution in group research has come about because of advancements in statistical theory and methods, and because of powerful computing capacity that is now readily available on anyone`s laptop. For example, multilevel modeling (a statistical advancement in regression equations) has transformed how we conduct group therapy research. For the first time, we are able to: take into account the impact of the group on the individual, test hypotheses about member to member or member to group interactions, and model the unique trajectory that each group takes across time – just to name a few. For years group practitioners have been far ahead of researchers in terms of theorizing about how groups work and advancing the need for more groups. Finally, group research methods have caught up to these rich theories. I envision a day in the near future in which group researchers not only test group concepts, but by way of testing these ideas they will also lead the development of new theories and models of how groups work and how they can be more effective. For that we need young group psychologists who are just as comfortable running a computer model as they are running a therapy or work group.