Statistical modelling

Statistische modellen

It is very rare that raw data – data as they are produced – can be used straight away to answer a question. Apart from cleaning the data, it is most often necessary to summarize data, and study relations between different variables that are collected. Moreover, there often is uncertainty about whether the data that are collected really describe reality. Statistical modeling techniques are used to study relations, quantify evidence and helps to answer questions in a better way.

At the department of Methods and Statistics, one key research topic is the use of latent variables for combining multiple variables. This includes, for example, the following techniques: item response theory (IRT), multilevel analysis, and structural equation models (SEM). Informative hypotheses enable researchers to answer research questions in an informative matter, and allow research questions to be answered that cannot be answered with traditional statistical methods. Evidence synthesis methods are used to combine multiple studies and/or different types of information (e.g. findings from multiple sources). In the last few years, one important focus area of the department is on the study of causal relations, often in combination with intensive longitudinal data. In these models, observations are collected over long periods of time to understand change and development over time. In these areas the department closely collaborates with the Dynamics of Youth strategic theme.

More information about the specific research themes within this area: