Skip to ContentSkip to Navigation

Detecting nonlinearity in the associations between depression and cortisol

PhD ceremony:Mr R.B. (Bart) Toonen
When:April 08, 2024
Supervisor:prof. dr. P. (Peter) de Jonge
Co-supervisor:dr. K.J. (Klaas) Wardenaar
Where:Academy building RUG
Faculty:Behavioural and Social Sciences
Detecting nonlinearity in the associations between depression and

This thesis contains a study after the applicability of a specific class of mathematical methods from physics and ecology within psychological research after the causes of depression – in particular, the associations between the hormone cortisol and the presence of negative affect. Prevailing research after these associations shows inconclusive results, which may be caused by the research methods that are used. Typically, participants are assigned to different groups and conclusions are based on group averages. However, when there are large differences between participants within a group, such averages contain only a limited amount of information. One way to circumvent this problem is to study participants individually during a longer period of time, instead of assigning them to groups. This results in series of measurements on a large amount of subsequent moments in time. Analysis of these measurements is usually done with methods that assume linear relationships between the observations. The question is whether these relationships are really linear. This thesis therefore explores the applicability of methods that assume nonlinear relationships. One of the analyses showed that cortisol values in the morning are mainly associated with negative-affect values that occurred during the previous day. The main conclusions in this thesis are that nonlinear methods are capable of detecting relationships that are impossible or hard to find with linear methods, but these methods need long series of measurements and it is yet unclear to what extent the outcomes are affected by measurement uncertainty.