Research

What I work on, and how.

My work sits at the intersection of work and organisational psychology, quantitative methodology, and the applied concerns of organisations that have hired me (or my students) to look at things carefully. It is mostly quantitative, mostly diary-based or multilevel, and increasingly interested in what the right answer to a question actually is, methodologically, before it tries to be useful.

Themes

Four threads that keep coming back.

Work–nonwork boundaries and hybrid work

How people draw, defend, and dissolve the line between their working lives and everything else, and what happens to that line when work moves into the home. I am interested in what predicts successful boundary management, when integration helps and when it hurts, and what the everyday rituals of segmentation look like in practice. The pandemic made remote and hybrid work an unavoidable research setting; the questions it raised are not going away. What does the loss of a commute buy people, and what does it cost? When does homeworking help those with caring responsibilities, and when does it quietly redistribute the work to them?

Current threads in this area

dual-process model of segmentation enactment
boundary management in portfolio work
workplace embitterment and rumination
homeworker wellbeing and uncertainty (COVID longitudinal)
two-wave study of job satisfaction and homeworking
diary study of homeworker recovery

Assessment centres

Long-running methodological work on what assessment centres actually measure. The thread through this programme is that many of the inferences we draw from ratings depend on unmodelled confounds, and that being honest about that changes which designs are worth running.

Current threads in this area

rater variance and dimension effects
the unconfounding of multisource performance ratings
task-based versus dimension-based scoring

Wellbeing at work

The daily texture of working life: engagement, burnout, rumination, embitterment, and the small organisational conditions that shift them. Much of this work is diary-based, capturing state and trait at the level of a single day or week.

Current threads in this area

workplace embitterment and illegitimate tasks
engagement crossover between colleagues
wellbeing during the COVID-19 transition

Psychometric evaluation of AI-generated ratings

As AI tools start producing ratings of people (of personality, performance, job-fit), the psychometric questions we have always asked of human raters apply, and then some. This stream treats algorithmic outputs the way the assessment-centre literature treats assessor judgments: as ratings with their own reliability, dimensionality, and bias structure, to be uncovered rather than assumed. The early questions are basic in the best way: are AI ratings stable across prompts and models, what do they actually measure, and what is the right criterion to triangulate them against when no gold standard exists?

Current threads in this area

reliability of LLM-generated personality ratings across prompts and models
generalizability-theoretic decomposition of AI rater variance
bias and drift in AI-scored performance assessments

Methods

The toolkit.

A short list of the methods I reach for most often. I am wary of method-led research: the question should pick the method, not the other way around. But here is what the questions have picked lately.

Daily diary studies (within-person, multilevel)
Multilevel / hierarchical modelling
Bayesian structural equation modelling
Vector autoregressive models for intensive longitudinal data
Survey design and measurement (including generalizability theory)
Assessment-centre design and psychometrics
Cross-sectional and longitudinal survey research
Meta-analytic thinking (less as a method, more as a discipline)