Eduwin Pakpahan is Associate Professor in Mathematics in the School of Engineering, Physics, and Mathematics, Northumbria University, where he leads the Applied Statistics and Data Science (ASDS) research group and is Theme Lead for Mathematics.

His research examines how disadvantage accumulates across a life course and surfaces decades later as cognitive decline and multimorbidity, and how far associations observed in cohort data can bear a causal reading. He works with harmonised cohort studies of ageing — ELSA, SHARE, HRS and CLHLS — and with data from low- and middle-income settings, including the Indonesia Family Life Survey. He is developing dementia risk prediction models for low- and middle-income countries, where instruments built in high-income cohorts typically lose calibration and formal diagnosis is sparse or absent.

The methodological work sits in longitudinal data analysis: mixed-effects and latent variable models for repeated measures, with attention to heterogeneity in individual trajectories, informative dropout, and confounders that are themselves time-varying. His doctorate, completed at the University of Perugia (Italy), was on causal inference in Gaussian graphical Markov models. That training still shapes his approach: he treats a model's assumptions as claims to be scrutinised rather than caveats to be listed.

A longstanding interest concerns the relationship between quantitative and qualitative evidence. Causal claims in population health seldom rest on statistical estimates alone; they depend on mechanism, context and judgement that survey data document only in part. He is interested in how the two kinds of evidence can be made to constrain each other rather than proceed in parallel.

Before joining Northumbria in 2020 he was at Newcastle University and the University of East Anglia, and earlier a research associate at the European University Institute in Florence. His first degree in Statistics is from Bogor Agricultural University, Indonesia. He is a Fellow of the Royal Statistical Society. He welcomes enquiries from prospective PhD students and from colleagues with longitudinal or cohort data that raise causal questions.

Mixed-effects models and longitudinal data analysis — heterogeneity in individual trajectories, attrition and informative dropout, measurement invariance over time

Causal inference with graphical models — instrumental variable methods, Gaussian graphical Markov models, time-varying confounding in observational data

Latent variable and multivariate models for constructs that are measured only indirectly

Life-course models of ageing, cognition and health inequality

Dementia risk prediction in low- and middle-income settings

Cross-country harmonised cohort studies (ELSA, SHARE, HRS)