Two Approaches to Analyzing Data of Project Teams

Abstract

In this chapter, we compare and discuss several statistical approaches to analyzing data obtained from team members. As an illustrative example, we analyze data obtained from 636 team members across 87 teams from 8 companies located in Poland using ordinary least squares multiple regression analysis, multiple regression with clustered standard errors, and mixed modeling. In the example, we analyze the relationship between team size and team gender composition concerning team climate. While a model building approach to hypothesis testing yields most similar results for multiple regression an a clustered standard error approach, a multilevel model yields results more similar to OLS regression when testing the significance of individual predictors, suggesting a clustered standard error correction is more prone to a type II error rate when testing model coefficients than an equivalent multilevel model. Finally, the implications of these observations to team data analyses are discussed.