Study 3 Sample Clauses

Study 3. As the data collected in this study was a subsample of Study 2, the contributions were the same.

Study 3. The emotion response system in anorexia and bulimia nervosa: facial expression, subjective experience and physiology OVERVIEW This chapter describes a study which extends the work in Chapter 3. As well as measuring facial expression and subjective response to emotional stimuli this study will measure physiology, namely skin conductance. Therefore, it provides the opportunity to investigate more objectively the emotion response system in people with an ED and to note any discrepancies across this system.

Study 3. Factorial simulations In Study 3, we further evaluated K-means and standard K-modes by conducting simulations with a factorial study design. The data set design factors were number of individuals (i.e. patients, objects) in the dataset, underlying subtype structure (subtype structures varied in overlap), relative distribution of individuals across subtypes, the probability of being present for “high probability” attributes, and conditional (within-subtype) dependence of the “high probability” attributes. These design factors were fully crossed, creating 108 cells, and eight datasets were created for each cell (for a total of 864 datasets). Each clustering algorithm (K-means, K-modes) was performed for K = 2, 3, . . . 9 on each artificial data set, yielding a set of eight partitions of the data set by each algorithm. Recovery of true subtype structure for each of the eight partitions in a set was measured by ARI. The best ARI of these eight was used as an “observation.”