Introducing an intrateam longitudinal approach to the study of team process dynamics

Abstract

In this article, we introduce an intrateam longitudinal approach to study the temporal dynamics of team processes and its relations to antecedent and consequence variables. We compare this approach with the conventional interteam longitudinal approach (e.g., repeated-measures [M]ANOVA, random coefficient modelling, latent growth modelling) and discuss the conceptual and methodological differences between the two approaches. Whereas the interteam approach follows a sample-to-cases order of inference and assumes random deviances of individual teams' change patterns from the sample-level pattern, the intrateam approach follows a cases-to-sample order of inference and allows for qualitative differences in individual teams' change patterns. In the intrateam approach, each team's change trajectory is directly measured and then used in the next-step multivariate analyses. We argue that the intrateam approach is more compatible with the current conceptualization of team processes as team members' interactions over time (Marks, Mathieu, & Zaccaro, 2001) and with the reasoning underlying the Input–Process–Output (IPO) framework. Next, we illustrate the intrateam approach and apply both approaches in an empirical longitudinal study of team conflict and team satisfaction (N = 42). The results show the contrast between the two approaches and added value of the intrateam approach in the study of team process dynamics.

Keywords:

• Interteam longitudinal approach
• Intrateam longitudinal approach
• Team conflict
• Team satisfaction

Notes

¹As Marks and colleagues (2001, p. 357) conclude after a literature review of team processes, “the essence of the construct [of team processes] lies in team interaction” and “different forms of team processes describe the types of interactions that take place among team members during the course of goal accomplishment”.

²In the article, the terms “interaction patterns over time”, “change patterns”, “development patterns”, “growth trajectories”, and “temporal dynamics patterns” have the same meaning and are interchangeably used.

³Here, we focus on studies in which teams are the unit of analysis. However, all the longitudinal methods and approaches discussed can be applied to studies in which persons and organizations are the unit of analysis.

⁴The “rule of thumb” suggests a minimum number of 20 measurement moments in order to generate reliable estimation.

⁵The letter t represents the total number of measurement moments in a study.

⁶This is supported by bibliographic analyses. For example, among the 20 articles with a longitudinal research design published in the European Journal of Work and Organizational Psychology since 1996, six use two measurement moments, whereas eight articles uses three moments, and only one used four moments. The other articles describe case studies or review longitudinal research. A search in the PsycLit database reveals 37 additional journal articles on longitudinal team research (longitudinal and team as title words); 14 of these studies have two measurement moments, eight have three moments, six have four moments, and four have more measurement moments (mostly for a dependent variable). Not all of these studies deal with work teams, and only some of them deal with team processes. These counts were collected on 24 October 2011. Although it seems that the two-moment design is the most popular to date, we consider that the two-moment design gives rather limited information on temporal dynamics and the three-moment design is the simplest longitudinal design to study the temporal dynamics of team processes.

⁷It is a rule of thumb that at least five observations (or data points) are needed in order to estimate one parameter. Therefore, the required sample size depends on how many parameters are to be estimated in specific models. When a single categorical variable is used to represent several dynamic patterns (e.g., in MANOVA), fewer teams are required than when each pattern (or each pattern category) is represented by a dichotomous dummy variable (e.g., in linear regressions with multiple dummy variables).

⁸We are aware of the debate among team researchers (Moreland, 2010; Williams, 2010) over whether dyads are groups or not. In this article, we follow Williams' (2010) argument and consider dyads as groups or teams for two major reasons. First, in our view, the phenomenon of our interest in the empirical study, that is, team conflict and team satisfaction does exist at dyad level and we have added team size as a control variable. Second, the generally accepted definition of groups or teams includes two-person groups or teams (e.g., Kozlowski & Bell, 2003).

⁹In this article, we take .10 as the cut-off point for the significance level. We use this rather lenient standard, because the purpose of the article is to introduce the intrateam longitudinal approach and to make the first endeavour (to our knowledge) to compare whether the intra- and interteam longitudinal approaches produce differential empirical results.

¹⁰The mean difference between the inverted-U-shape group and the U-shape group is at the .018 significance level; that between the inverted-U-shape group and the continuous increase group is at the .04 level. The mean difference between the continuous decrease group and the U-shape group is at the .06 level; that between the continuous decrease group and the continuous increase group at the .064 level.

¹¹In the group of high-satisfaction teams, between-subjects effect is significant for the three conflict types, F _TC(1,22) = 664.48, p < .001; F _RC(1,22) = 495.87, p < .001; F _PC(1,22) = 410.21, p < .001. So is in the group of low-satisfaction teams, F _TC(1,18) = 431.53, p < .001; F _RC(1,18) = 384.95, p < .001; F _PC(1,18) = 288.35, p < .001.

¹²We use the term “estimation” for abstracting information of within-team changes over time from the data of each team's temporal dynamics. It differs from estimating the sample-level temporal dynamics from the discrete data points of individual teams across measurement moments as in the interteam approach.