Research Program Department Methodology and Statistics

This broad research theme is devoted to further developing and advancing statistical methodology for social research. These methods aim to improve social research and to help answering challenging research questions that are relevant in today’s society. Specifically we contribute to the following topics:

• Latent class analysis with the goal to identify clusters/subgroups of individuals who differ in their attitudes, preferences, behaviors, emotions, psychological states, etc., or in how these develop over time. Latent class analysis is somewhat similar to factor analysis, but with the important difference that the underlying latent variable explaining the associations between the observed responses is categorical instead of continuous. After identifying how individuals belonging to the various clusters differ, we may wish to predict the cluster memberships, to study whether cluster memberships predict future behaviors or risks, or to study how cluster membership changes over time. For an introduction, see this video
• Social network modeling with the goal to better understand temporal social interaction behavior among actors in social networks (such as information-sharing networks of employees in large organizations, networks of children and teachers in classrooms, or communication networks). We use so-called relational event data which contains information about who is interacting with whom in a social network and when. By analyzing such relational event sequences, we can learn how the past affects the future or how social interaction behavior evolves in continuous time. Our work has contributed to a better understanding of integration processes of new workers in organizations and complex communication structures during emergencies (www.nettrek.net).
• Qualitative research methods with the goal to better understand lived experiences and complex social interactions (such as the experience of losing a job, perceptions on mental health and interpersonal dynamics). We use so called narrative data (such as interviews and observations), which contains rich descriptions of how people experience and interpret events. By analyzing such meaning-making processes in social contexts, this allows to generate knowledge on how and why (inter)actions and situations take place. 
• Survey research with the goal to better understand processes that impact on the quality of survey research data. We use advanced methods to research to what extent respondents assign similar meaning to questions used in survey research (e.g. measurement non-invariance). We develop and use tools to diagnose response style behaviors and study the impact of such behavior on the quality of survey data. We use mixed-methods designs to gain insight in processes made by respondents who have a tendency to respond in social desirable ways.

Staff members

In alphabetical order: 

• Dr. Francisco Bahamonde Birke
• Dr. Guy Moors
• Prof. dr. Jeroen Vermunt
• Dr. John Gelissen
• Dr. ir. Joris Mulder
• Dr. Katharina Loter
• Dr. Lizbeth Burgos Ochoa
• Dr. Melissa De Smet
• Dr. Tamarinde Haven

“Meta-research” effectively means “doing research on research”. We study the scientific system in psychology and in general to find its flaws and empirically test potential solutions. The scientific system has many different aspects, which means that we study a wide range of topics.

Some examples of our research interests are:

• Statistical power and interpreting statistical results
• Questionable research practices and publication bias
• Pre-registration
• Analytical reproducibility and replication
• Meta-analysis and systematic reviews

We aim to create impact with our work through “traditional” journal articles, but also through more “direct” forms of output. For instance, we have created a clear checklist to avoid p-hacking, founded the Journal for Open Psychology Data, and built free software tools (e.g., statcheck, a spellchecker for statistics, and p-uniform, a tool to estimate and correct for publication bias).

For more information about our research group, visit our website.

Staff members

In alphabetical order: 

• Dr. Caspar van Lissa
• Prof. dr. Jelte Wicherts
• Prof. dr. Marcel van Assen
• Dr. Marjan Bakker
• Dr. Melissa De Smet
• Dr. Michèle Nuijten
• Dr. Paul Lodder
• Dr. Robbie van Aert
• Dr. Tamarinde Haven

Psychometrics is the art and science of measuring psychological attributes. Many attributes in psychology, but also in the social sciences, cannot be directly observed. Think of intelligence, reading proficiency, personality characteristics, and attitudes towards social issues. We cannot see from the outside how intelligent a person is. Tests and questionnaires are often used to make these attributes tangible. However, to be useful, test scores must be reliable and meaningful. But not only that! Researchers also want to understand which psychological processes are involved and, for example, rule out that measurement instruments are biased towards certain groups. And there is even more than that. Recent developments in computer technology and AI opens new opportunities to collect richer data (e.g., response times) using more authentic tasks. The psychometricians in our department develop and study innovative (statistical) methods to further improve measurement in psychology and the social sciences.  

Research interests:

• Mixture Item response theory
• Response-time models
• Individual change assessment
• Continuous test norming
• Comparative judgements
• Dynamic testing
• Adaptive testing

Staff members

In alphabetical order: 

• Dr. Inga Schwabe
• Dr. Jesper Tijmstra
• Prof. dr. Klaas Sijtsma
• Dr. Leonie Vogelsmeier
• Dr. Lieke Voncken
• Dr. Maria Bolsinova
• Dr. Wilco Emons

Data science is a very broad term that covers all aspects of dealing with data: It not only concerns analysis of data but, for example, also creating data (e.g., web scraping, turning non-numerical data such as texts, pictures, and music into numbers suitable for data analysis), legal and ethical aspects of collecting and analyzing data, storage and communication of data. With respect to data analysis, data science is often associated with large datasets and intensive computations. 

Some active research topics in our department are:

• Natural language processing and text mining
• Machine learning, e.g., for theory formation in the social and behavioral sciences
• High-dimensional (multi-domain) data and regularization

Staff members

In alphabetical order: 

• Dr. Bennett Kleinberg
• Dr. Caspar van Lissa
• Dr. Inga Schwabe
• Dr. Katrijn Van Deun
• Dr. Li Zeng
• Dr. Lizbeth Burgos Ochoa

Bayesian statistics is a powerful and flexible framework for parameter estimation and hypothesis testing which has gained popularity in the social and behavioral science in the last decades. Prior distributions are used to reflect our uncertainty before observing the data. After observing the data, our prior is updated using Bayes’ theorem to obtain the posterior distribution which reflects our uncertainty after observing the data. Thereby, Bayesian statistics can be seen as a principled probabilistic framework for updating our uncertainty about a statistical model with unknown parameters in light of the observed data and for combining different sources of information in case of uncertainty.

We contribute to the following topics in the field of Bayesian statistics:

• Bayesian hypothesis testing using Bayes factors.
• Bayesian psychometrics.
• Bayesian time-series modeling (e.g., for experience sampling modeling).
• Bayesian regularization methods for identifying true nonzero effects in case of many (many) predictor variables.
• Bayesian Gaussian processes for nonlinear statistical science.
• Bayesian Gaussian graphical modeling for studying psychological networks. 

Staff members

In alphabetical order: 

• Dr. Jesper Tijmstra
• Dr. Joran Jongerling
• Dr. ir. Joris Mulder
• Dr. Maria Bolsinova

There is a growing interest in social and behavioral research in studying processes as they unfold over time, as opposed to just studying the static outcomes of these processes. This requires measuring several times a day, on multiple days, all while participants go about living their everyday lives (using for example Experience Sampling methods (ESM) or Ecological Momentary Assessment methods). These types of data provide many exciting opportunities, like truly personalized insights and intervention, but also have their own set of challenges. Frequent measurements require new types of measurement procedures and instruments, and the large amounts of data bring with them new analytical difficulties to tackle. Intensive Longitudinal Methods are a relative young and exciting field in which much remains to be discovered!

We contribute to the following topics in the field of Intensive Longitudinal Methods:

• Psychometrics for Intensive Longitudinal Data.
• Dynamic Structural Equation Modeling of Intensive Longitudinal Data
• (Vector) Autoregressive Modeling of Intensive Longitudinal Data
• Network modeling of Intensive Longitudinal Data
• Non-linear modeling of Intensive Longitudinal Data.

In addition, the MTO department is actively involved in the Tilburg Experience Sampling Center (TESC), a multidisciplinary network of experts on Intensive Longitudinal Data and Methods that focuses on team-science and live-long learning in the context of this new and exciting type of research.

Staff members

In alphabetical order: 

• Dr. Joran Jongerling
• Dr. Katrijn Van Deun
• Dr. Leonie Vogelsmeier
• Dr. Melissa De Smet