A tutorial on Bayesian nonparametric models

A key problem in statistical modeling is model selection, that is, how to choose a model at an appropriate level of complexity. This problem appears in many settings, most prominently in choosing the number of clusters in mixture models or the number of factors in factor analysis. In this tutorial, we describe Bayesian nonparametric methods, a class of methods that side-steps this issue by allowing the data to determine the complexity of the model. This tutorial is a high-level introduction to Bayesian nonparametric methods and contains several examples of their application.     

Natural ambiguities: Racial categorization of multiracial individuals

Understanding the perception of multiracial persons is increasingly important in today's diverse society. The present research investigated the process of categorizing multiracial persons as “Multiracial.” We hypothesized that perceivers would make fewer Multiracial categorizations of multiracials and that these categorizations would take longer than monoracial categorizations. We found support for these hypotheses across six experiments. Experiment 1 demonstrated that perceivers did not categorize morphed Black–White faces as Multiracial with the same frequency with which they categorized Black and White faces as Black and White (respectively), and that categorizations of multiracials as Multiracial took longer than monoracial categorizations. Experiment 2 replicated and extended these effects to real Black–White faces. Experiment 3 showed that these findings generalized to Asian–White faces. We used pixel variance analysis to show that these effects were not due to increased variance among Multiracial faces. The image analysis showed that the Black–White morphs and real biracials were actually less varied than either the Black or White sets of faces. Experiments 4 and 5 demonstrated that cognitive load and time constraints detrimentally affected multiracial, but not monoracial, categorizations. Experiment 6 showed that imbuing monoracial categories with importance decreases the use of the Multiracial category. Implications of these findings for understanding perceptions of multiracial persons are discussed.     

Treating stimuli as a random factor in social psychology: a new and comprehensive solution to a pervasive but largely ignored problem

Throughout social and cognitive psychology, participants are routinely asked to respond in some way to experimental stimuli that are thought to represent categories of theoretical interest. For instance, in measures of implicit attitudes, participants are primed with pictures of specific African American and White stimulus persons sampled in some way from possible stimuli that might have been used. Yet seldom is the sampling of stimuli taken into account in the analysis of the resulting data, in spite of numerous warnings about the perils of ignoring stimulus variation (Clark, 1973; Kenny, 1985; Wells & Windschitl, 1999). Part of this failure to attend to stimulus variation is due to the demands imposed by traditional analysis of variance procedures for the analysis of data when both participants and stimuli are treated as random factors. In this article, we present a comprehensive solution using mixed models for the analysis of data with crossed random factors (e.g., participants and stimuli). We show the substantial biases inherent in analyses that ignore one or the other of the random factors, and we illustrate the substantial advantages of the mixed models approach with both hypothetical and actual, well-known data sets in social psychology (Bem, 2011; Blair, Chapleau, & Judd, 2005; Correll, Park, Judd, & Wittenbrink, 2002).     

Why and when peer prediction is superior to self-prediction: the weight given to future aspiration versus past achievement

Peer predictions of future behavior and achievement are often more accurate than those furnished by the self. Although both self- and peer predictions correlate equally with future outcomes, peers tend to avoid the degree of overoptimism so often seen in self-predictions. In 3 studies, the authors tested whether this differential accuracy arises because people give more weight to past behavior when predicting others, but emphasize agentic information, in particular data about their aspiration level, when predicting the self. Studies 1 and 3 showed that the exact same participants rated past behavior more diagnostic of future performance when predicting another person but viewed aspiration-level data as more valuable when someone else was trying to predict them. In Studies 2 and 3 (predicting an upcoming exam score and performance in a lab task, respectively), participants gave greater weight in self-predictions to aspiration-level data than did a yoked peer, who instead gave greater weight to evidence of past achievement. This differential weighting explained why peer predictions tended to be less optimistic and, thus, more accurate. Discussion centers on strategies for predicting future behavior and why people may remain ignorant of their own incompetence despite feedback.     

Religious Involvement Measurement Model in a National Sample of African Americans

This study examined the factor structure of a brief measure of religious involvement developed previously in research with African American women. Telephone interview methods were used with a national sample of both African American women and men (N = 2,370). Confirmatory factor analyses supported the distinction between religious beliefs and religious behaviors factors and indicated that the factor loadings were equivalent for women and men. Women reported higher levels of religious involvement than men. These results support the validity of this relatively brief instrument for assessing these two dimensions of religious involvement for both African American women and men.     

IRTs of the ABCs: children's letter name acquisition

We examined the developmental sequence of letter name knowledge acquisition by children from 2 to five years of age. Data from 2 samples representing diverse regions, ethnicity, and socioeconomic backgrounds (ns = 1074 and 500) were analyzed using item response theory (IRT) and differential item functioning techniques. Results from factor analyses indicated that letter name knowledge represented a unidimensional skill; IRT results yielded significant differences between letters in both difficulty and discrimination. Results also indicated an approximate developmental sequence in letter name learning for the simplest and most challenging to learn letters--but with no clear sequence between these extremes. Findings also suggested that children were most likely to first learn their first initial. We discuss implications for assessment and instruction.     

Borrowing from Sleep to Pay Work and Family: Expanding Time‐Based Conflict to the Broader Nonwork Domain

We extend cross‐domain research by examining sleep, a domain within the larger nonwork domain that competes for time with work and family domains. We draw from scarcity theory and research on slack resources to contend that, because people cannot increase the amount of time they have, they borrow time from sleep in order to spend more time working and with family. Utilizing a Bureau of Labor Statistics survey of 10,741 participants, we find nonlinear and interactive effects of time spent working and time spent with family on sleep time, suggesting that the negative effects of work and family on sleep time are especially strong when demands for work and family are high. In an experience sampling field study of 122 working adults, we similarly find a nonlinear effect of work time on sleep time as well as an interaction between work time and family time in predicting time spent sleeping. Both studies indicate that as slack time resources become increasingly scarce, time spent working and time spent with family have increasingly powerful negative effects on time spent sleeping. Contrary to our expectations, we found no support for gender as a moderator of these effects.     

On the measurement of criterion noise in signal detection theory: the case of recognition memory

Traditional approaches within the framework of signal detection theory (SDT; Green & Swets, 1966), especially in the field of recognition memory, assume that the positioning of response criteria is not a noisy process. Recent work (Benjamin, Diaz, & Wee, 2009; Mueller & Weidemann, 2008) has challenged this assumption, arguing not only for the existence of criterion noise but also for its large magnitude and substantive contribution to individuals' performance. A review of these recent approaches for the measurement of criterion noise in SDT identifies several shortcomings and confoundings. A reanalysis of Benjamin et al.'s (2009) data sets as well as the results from a new experimental method indicate that the different forms of criterion noise proposed in the recognition memory literature are of very low magnitudes, and they do not provide a significant improvement over the account already given by traditional SDT without criterion noise.     

“Socializing” Achievement Goal Theory: The Need for Social Goals

Cultures shape the types of goals that students pursue in the classroom. However, research in achievement goal theory seems to have neglected this cultural aspect with its exclusive focus on individualistically-based goals such as mastery and performance goals. This emphasis on mastery and performance goals may reflect Western individualist psychological thinking. Thus, social goals, which may be more salient in collectivist cultures, are relatively neglected. There is a dearth of studies investigating the role of social goals in academic motivation, and the few studies that did explore them are somewhat problematic. This paper reviews research done within the achievement goal theory, considers the need for more studies on social goals, and concludes with the argument that social goals are important in understanding student motivation especially in collectivist cultures.