Extracting Moral Foundations From Text Narratives: Relevance, Challenges, and Solutions

• In Event: Advances in Text Analysis Methods

Abstract

Moral Foundations Theory (MFT; Graham et al., 2012) and the Model of Intuitive Morality and Exemplars (MIME; Tamborini, 2012) contend that moral judgments and decision-making are built on a universal set of basic moral intuitions. A large body of research across diverse disciplines including media psychology has supported many of the MFT’s and MIME’s central hypotheses. Yet, an important prerequisite of this research - the ability to extract latent intuitive moral content represented in media stimuli with a reliable and standardized procedure - has not been systematically studied. In this paper we subject different extraction procedures to rigorous tests, highlight challenges by identifying a range of reliabilities, and provide solutions that maximize the reliability and validity of moral intuition extraction. In five content analytical studies we demonstrate that traditional content analytical approaches lead to moral intuition extraction with rather low reliabilities; variation in coding reliabilities can be predicted by both text features and characteristics of the human coders, and reliability is largely unaffected by the intensity and detail of coder training. We show that a coding task with simplified training and an annotation technique that treats moral foundations as fast, spontaneous intuitions leads to plausible and acceptable inter-rater agreement, and potentially to more valid moral intuition extractions.

Authors

• Rene Weber, U of California - Santa Barbara

• James Michael Mangus, U of California, Santa Barbara

• Richard Huskey, The Ohio State University

• Ori Amir, U of California, Santa Barbara

• Reid Swanson, U of Southern California

• Andrew Gordon, U of Southern California

• Peter Khooshabeh, US Army Research Lab

• Lindsay S Hahn, Michigan State U

• Ron Tamborini, Michigan State U