A system of classifying and measuring personality, emotion, and behaviour

This thesis addresses two long-standing dilemmas in personality psychology. The first dilemma is that of identifying the latent dimensions of personality. The second is to identify how best to measure psychological constructs. Novel solutions to both of these problems are presented in two articles. The first article (published), 'An atlas of personality, emotion, and behaviour' proposes a two-dimensional taxonomy, with strictly orthogonal dimensions affiliation and dominance. The second article (submitted for publication), 'Metrology applied to personality, emotion, and behaviour', proposes two quantitative measures. The measures are consistent with both the lexical hypothesis and metrology, the science of measurement. In study 1, methods included cataloguing adjectival descriptors of personality, abstract noun descriptors of feelings and emotion, and verb descriptors of behaviour. Sociobiological and neurobiological evidence was further used to identify two orthogonal dimensions, each of which was divided into five ordinal categories. Using the Delphi Method, 20% of the catalogued words were scored by clinical psychologists, whilst the remaining 80% of words were scored using a tailored network approach. A technique was then developed to visualise a wide range of existing psychological and social constructs in two dimensions. Finally, a simulation technique was then developed to identify an alternative approach to psychological testing. Results: The identified dimensions of affiliation and dominance were derived from the cataloguing of over 20,000 English language words, including 7,000 adjectival descriptors of personality, 3,000 abstract noun descriptors of emotion, and 8,000 verb descriptors of behaviour. All 20,000 catalogued words were able to be classified according to the ordinal scale. A wide range of psychological and social constructs was visualised and delineated, including the Dark Triad, Five-Factor Model, leadership, criminality, and many DSM-5 personality disorders. The simulation approach facilitated the formation of a psychological testing methodology that minimises the number of questions that must be asked to encompass a broad spectrum of personality, whilst minimising confounding and maximising statistical power. In study 2, two quantitative psychological measures were proposed that strictly conform to metrological standards and the lexical hypothesis. The first measures semantic distance, inspired by the small world problem more popularly known as 'six degrees of separation'. The second measures the geometric distance between constructs according to the atlas. Both measures are theory realistic and address known issues with existing measures of psychological constructs, such as definitional circularity and reification. The method involved a crowdsourcing study of all 1,506 IPIP items. Respondents (N=1,814) were asked to identify the single best adjectival descriptor relevant to each item. The responses were then measured according to both newly proposed quantitative measures. It was found that participant responses were significantly heterogeneous across many IPIP items, calling into question these items' suitability for psychological testing purposes. The crowdsource responses were further used to test the hypothesis that five-factor models are hierarchical. Results did not support the notion that the five-factor model is hierarchical, contrary to popular opinion. Considered together, the conclusion of both studies is that a two factor model of personality may have advantages over the prevailing five-factor model.