Biases in evaluations are an essential part of human experience, and unsurprisingly, they have been studied across multiple and often disparate literatures. The three papers I present here further our understanding of biases by integrating across different theoretical frameworks and research areas. In Chapter 1, I present advances in research methods and practices, which include discussions on statistical power, how to measure and control Type I error, and preregistrations. This paper sets the basis for the how and why of the methods used in the subsequent chapters. In Chapter 2, I dive into intergroup contexts to investigate a frequently overlooked confound in the literature of implicit bias: the exemplar-category confound. In common research practices in the field, implicit measures include presentations of specific exemplars (e.g., faces of Black people), whereas explicit measures often focus on responses to abstract social categories (e.g., feeling thermometer towards the category Black people). Results of four experiments suggest that previously obtained implicit-explicit dissociations using the Implicit Association Test may be at least partly driven by the exemplar-category confound. In Chapter 3, I turn into interpersonal contexts to investigate evaluative biases in the romantic relationships literature. First, I successfully developed a paradigm to reliably manipulate people’s evaluations of traits, and afterwards, I used this paradigm to conduct the first experimental tests of four key theoretical accounts. In conclusion, this work develops and disseminates new methods and paradigms, challenges longstanding assumptions in the field, and furthers our understanding of biases and their consequences.

People have ideas about the attributes (i.e., traits or characteristics that vary along a dimension) that they like in others (e.g., “I like intelligence in a romantic partner”), and these ideas about liking are called summarized attribute preferences (Ledgerwood et al., 2018). But where do summarized preferences come from, and what do they predict? Across four studies, we examined how people form summarized attribute preferences and whether they predict situation selection. We showed participants a series of photographs of faces and assessed both their experienced liking for an attribute (or functional attribute preference) as well as their inference about how much they liked the attribute in the abstract (their summarized attribute preference). Our results suggest that summarized attribute preferences—despite being (weakly) grounded in functional attribute preferences—were affected by incidental aspects of the context in which people learn about them (i.e., the overall likeability of the pool of faces). Furthermore, we observed a double dissociation in the predictive validity of summarized and functional attribute preferences: Whereas summarized attribute preferences predicted situation selection at a distance (e.g., whether to join a new dating website based on a description of it), functional attribute preferences predicted situation selection with experience (e.g., whether to join a new dating website after sampling it). We discuss theoretical and methodological implications for the interdisciplinary science of human evaluation.