Statistical Aspects of Group Dynamics : Multilevel Methods for Emergent Processes in Teams
Author
Summary, in English
Teams are dynamic systems that develop, adapt, and change as members interact and respond to their environments. Theory in organizational research emphasizes that team phenomena are multilevel, temporal, and often nonlinear. Yet, the statistical methods commonly used to study teams have lagged behind these theoretical advances, limiting empirical progress. This dissertation addresses this methodological shortcoming through three papers.
In Paper I, we critically examine existing approaches to the empirical estimation of consensus emergence, the process through which initially diverse individual perceptions converge into a shared team perspective. We introduce a formal statistical definition of consensus emergence and demonstrate common pitfalls, such as conflated variance components and model misspecification.
In Paper II, we extend heterogeneous variance models by integrating Gaussian processes. This framework provides a flexible way to capture nonlinear changes in variability over time, thereby allowing richer insights into how convergence and divergence unfold within teams.
In Paper III, we turn to the evolution and consequences of emergent states. Using the development of new venture teams as an empirical context, we propose a joint modeling framework to study how trust trajectories are shaped by significant events and, in turn, how trust predicts member departure. The model further accounts for non-ignorable missing data through a shared-parameter specification.
Together, these contributions advance the methodological toolkit for studying emergent team phenomena. By aligning statistical models more closely with theoretical advances, the dissertation provides researchers with tools to rigorously examine how collective states form, evolve, and influence outcomes in dynamic organizational settings.
In Paper I, we critically examine existing approaches to the empirical estimation of consensus emergence, the process through which initially diverse individual perceptions converge into a shared team perspective. We introduce a formal statistical definition of consensus emergence and demonstrate common pitfalls, such as conflated variance components and model misspecification.
In Paper II, we extend heterogeneous variance models by integrating Gaussian processes. This framework provides a flexible way to capture nonlinear changes in variability over time, thereby allowing richer insights into how convergence and divergence unfold within teams.
In Paper III, we turn to the evolution and consequences of emergent states. Using the development of new venture teams as an empirical context, we propose a joint modeling framework to study how trust trajectories are shaped by significant events and, in turn, how trust predicts member departure. The model further accounts for non-ignorable missing data through a shared-parameter specification.
Together, these contributions advance the methodological toolkit for studying emergent team phenomena. By aligning statistical models more closely with theoretical advances, the dissertation provides researchers with tools to rigorously examine how collective states form, evolve, and influence outcomes in dynamic organizational settings.
Department/s
Publishing year
2025
Language
English
Full text
- Available as PDF - 647 kB
- Download statistics
Document type
Dissertation
Publisher
Lunds universitet
Topic
- Probability Theory and Statistics
Keywords
- emergent states
- heterogeneous variance models
- multilevel modeling
- Gaussian processes
- joint modeling
Status
Published
Project
- The Statistics of Entrepreneurship
Supervisor
- Jonas Wallin
- Frédéric Delmar
ISBN/ISSN/Other
- ISBN: 978-91-8104-639-7
- ISBN: 978-91-8104-638-0
Defence date
21 October 2025
Defence time
10:15
Defence place
EC3:207
Opponent
- Ellen Hamaker (Professor)