A distinctive feature of Agile PAIR is its quantitatively oriented, multidisciplinary, and (special-)education–based intervention research. We link prediction models with the adaptation and use of intervention components, creating truly data-driven interventions. Agile PAIR researchers cover a wide breadth of intervention strategies, especially for digital interventions. In addition, there is a high concentration of researchers in quantitative methods in the educational and social sciences, as well as a concentration of analytical expertise—unique in the German-speaking world—in dealing with complex cross-sectional and longitudinal data on human experience and behavior. Based on this Agile PAIR at TU Dortmund Univeristy represents the emerging fields of Educational Data Science and Psychological Data Science.

From early 2025 onwards, Agile PAIR contributes to two current research areas: Agile intervention design and evaluation, and targeted methodological research.

Agile Intervention Design and Evaluation

Agile Interventions rely on high-quality predictions of developmental and intervention trajectories, as well as the identification of risk and resilience groups. We focus mostly on educational success, as well as related moderators and mediators like motivation. Mostly digital interventions support the identified risk groups through the use of targeted, individually adaptable prevention and intervention measures. A distinctive feature of intervention research in Agile PAIR is that prediction and intervention are designed in an agile manner, meaning that prediction and intervention models are adapted flexibly, iteratively, and incrementally.

Agile PAIR therefore seeks to answer the question: What works for whom under which conditions, and which causal mechanisms explain these effects? Adapting interventions based on individual data corresponds to the principles of “personalized learning” or “personalized education”. Similar to the concept of “personalized medicine,” such personalization of education enables improvements in effectiveness (in terms of educational success), which should be mediated through higher engagement in the respective learning processes. Based on initial assessments and predictive modeling, a person model can thus be created that serves as a basis for deriving intervention goals and designs as well as individualized support measures.

Targeted Methodological Research

Agile PAIR researchers develop tailor-made methods for education, psychology and adjacent fields. Backgrounds in biostatistics, psychometrics and data science enable interesting transfers of methodologies in several directions. An example are novel statistical methods with good inference properties for heterogeneous experimental settings or in situations with small sample sizes (e.g., rare subgroups). Further methods target complex situations such as missing values, high dimensions or unbalancedness, especially in longitudinal data collections. For unstructured data like text or process data, Agile PAIR researchers study current natural language processing approaches like large language models. To facilitate the analysis of large and high dimensional datasets, Agile PAIR studies sketching and coreset methods, data reduction techniques of significantly smaller size retaining the statistical power of the full data. 

Agile PAIR systematically adapts and extends quantitative methods for the data-driven intervention context. While the applied background leads to practically useful methodology, Agile PAIR honors a strong theoretical basis and works towards mathematical guarantees.