Moral Foundations Theory is a functionalist theory on morality, postulating that morality is intuitive, hereditary, and pluralistic, resting on a set of moral foundations. When it was introduced, there were five foundations: Care and Fairness, sorted under the Individualizing Foundations, and Loyalty, Authority, and Purity, sorted under the Binding Foundations. Together, all five foundations create a moral system in which the needs of the individual and the larger social group can be balanced. Accompanying the theory was the Moral Foundations Questionnaire (MFQ), holding two subscales of equal length, the Relevance and the Judgement subscale, which measured the foundations as five statistical factors. However, psychometric investigations of the MFQ have shown that two latent dimensions corresponding to the Individualizing and Binding Foundations seem more appropriate than the theoretical five. To clarify this issue, this thesis aims to investigate the MFQ more deeply. However, to understand the questionnaire, we must study psychometric methodology and the hypothetical construct. To achieve this, both aspects are discussed, and three empirical papers and one software application are presented.

In brief, while Paper I corroborated a well-known connection between rigid black-and-white cognition and conservative values, operationalized by the Binding Foundations, the paper also hinted that the two latent dimensions are sufficient when studying moral differences. Paper II, reporting a psychometric investigation of the MFQ, confirmed that the MFQ holds the two latent dimensions, and it was also noted that items measuring Loyalty were not exclusive to any of the two dimensions. As a response, the D3mirt software package was developed and used in Paper III to offer a visual three-dimensional model that can account for latent overlaps in the items. Adding two items on loyalty to the Relevance subscale, Paper III showed that the subscale is dominated by the two latent dimensions and that loyalty, operationalized as Collectivism, can be said to be a facet.

Given that the two dimensions in question are well documented in the literature, the conclusion is that both MFQ and MFT do not offer anything new but should be seen as subsections of a much broader field. In summary, this thesis shows the importance of considering both method and theory so that the psychometric models, D3mirt or others, can be understood from a larger context outside the minimal picture the models themselves present.