Project description
A key concern for every experimentalist is whether there is sufficient statistical evidence to support or refute tested hypotheses. This problem is related to what is known as statistical power: the higher the power, the more likely it is to detect an effect if it exists. In the past, a way to increase power was to increase the number of participants/samples in an experiment. However, this is not an option when sample sizes are limited, as when studying marginalised and underrepresented participant groups or rare conditions.
In this project, we will use statistical modelling and analyses to improve power through study design. For example, in an experiment where participants are asked to respond to three different conditions A, B and C, what is the optimal order of presentation to maximise power? Crucially, the answer to this question depends on the participants’ behaviour, e.g. their waxing and waning of attention. This links the question of optimal design to the study of non-stationary stochastic processes, which have posed a significant challenge to modelling and analysing experimental data. By combining concepts from matrix optimality with the theory of non-stationarity, this project will not only advance our understanding of statistical time series, but will also help researchers to answer critical questions in a world that embraces heterogeneity and non-uniformity amongst participants.
The project will be undertaken in close collaboration with Prof Kathy Conklin in the School of English (Nottingham) and , Centre for Research in Applied Linguistics and Dr Dale Barr in the School of Psychology and Neuroscience (Glasgow).