Ecological Momentary Assessment of Weight Stigma and Eating Behavior in Everyday Life.

BACKGROUND: Weight stigma is widespread, but the existing literature on its harmful consequences remains largely limited to lab-based experiments and large-scale longitudinal designs. PURPOSE: The purpose of this study was to understand how weight stigma unfolds in everyday life, and whether it predicts increased eating behavior. METHODS: In this event-contingent ecological momentary assessment study, 91 participants reported every time they experienced weight stigma and documented whether they ate, how much they ate, and what they ate. These reports were compared against a timepoint when they did not experience stigma. RESULTS: Participants reported a wide variety of stigmatizing events from a variety of sources, with the most common ones being the self, strangers, the media, and family. Multilevel models showed that participants were no more likely to eat post-stigma (vs. the comparison point), but if they did eat, they ate more servings of food (on average consuming 1.45 more servings, or 45% more). Moderation analyses indicated that this effect was amplified for men versus women. CONCLUSION: Experiencing weight stigma appears to beget behavioral changes, potentially driving future weight gain, placing individuals at ever more risk for further stigmatization.

This study looked at how weight stigma in everyday life impacts eating. People reported on episodes of weight stigma and their eating in the next 30 min. Weight stigma came from many different places, including family, strangers, media, and even themselves. Even though people did not necessarily eat more after weight stigma episodes, if they did eat, they ate significantly more food­about 45% more. This relationship was stronger in men than in women. The study also explored whether different kinds of people react differently to weight stigma. Weight stigma experiences led to even more food eaten among people who tended to have high buy-in about negative stereotypes of heavier people, as well as people who thought weight was an important part of their identity. The opposite was seen among people who worried the most about experiencing weight stigma in the future. These findings suggest that experiencing weight stigma may not always prompt people to eat immediately, but when they do eat, they tend to eat more, challenging the idea that weight stigma motivates people to eat less.

Power to Detect What? Considerations for Planning and Evaluating Sample Size.

ACADEMIC ABSTRACT: In the wake of the replication crisis, social and personality psychologists have increased attention to power analysis and the adequacy of sample sizes. In this article, we analyze current controversies in this area, including choosing effect sizes, why and whether power analyses should be conducted on already-collected data, how to mitigate the negative effects of sample size criteria on specific kinds of research, and which power criterion to use. For novel research questions, we advocate that researchers base sample sizes on effects that are likely to be cost-effective for other people to implement (in applied settings) or to study (in basic research settings), given the limitations of interest-based minimums or field-wide effect sizes. We discuss two alternatives to power analysis, precision analysis and sequential analysis, and end with recommendations for improving the practices of researchers, reviewers, and journal editors in social-personality psychology. PUBLIC ABSTRACT: Recently, social-personality psychology has been criticized for basing some of its conclusions on studies with low numbers of participants. As a result, power analysis, a mathematical way to ensure that a study has enough participants to reliably "detect" a given size of psychological effect, has become popular. This article describes power analysis and discusses some controversies about it, including how researchers should derive assumptions about effect size, and how the requirements of power analysis can be applied without harming research on hard-to-reach and marginalized communities. For novel research questions, we advocate that researchers base sample sizes on effects that are likely to be cost-effective for other people to implement (in applied settings) or to study (in basic research settings). We discuss two alternatives to power analysis, precision analysis and sequential analysis, and end with recommendations for improving the practices of researchers, reviewers, and journal editors in social-personality psychology.

Does the follow-your-passions ideology cause greater academic and occupational gender disparities than other cultural ideologies?

Five preregistered studies (N = 1934) demonstrate that the prevalent U.S. ideology to "follow your passions" perpetuates academic and occupational gender disparities compared to some other cultural ideologies. Study 1 shows that the follow-your-passions ideology is commonly used by U.S. students in making academic choices. Studies 2-5 find that making the follow-your-passions ideology salient causes greater academic and occupational gender disparities compared to the resources ideology (i.e., the idea that one should pursue a field that leads to high income and job security). In Study 4, the follow-your-passions ideology causes greater gender disparities even when compared to a cultural ideology that aligns more with the female gender role (i.e., communal ideology). In Study 5, a moderated mediation analysis supports the hypothesis that gender disparities are explained by women's versus men's greater tendency to draw upon female role-congruent selves when the follow-your-passions ideology is salient compared to when the resources ideology is salient. Drawing upon female role-congruent selves remains a significant mediator even when accounting for alternative mediators (e.g., appropriateness of ideology for one's gender). The follow-your-passions ideology may not seem explicitly gendered, but it causes greater academic and occupational gender disparities compared to some other cultural ideologies. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Selecting a Within- or Between-Subject Design for Mediation: Validity, Causality, and Statistical Power.

Researchers with mediation hypotheses must consider which design to use: within-subject or between-subject? In this paper, I argue that three factors should influence design choice: validity, causality, and statistical power. Threats to validity include carry-over effects, participant awareness, measurement, and more. Causality is a core element of mediation, and the assumptions required for causal inference differ between the two designs. Between-subject designs require more restrictive no-confounder assumptions, but within-subject designs require the assumption of no carry-over effects. Statistical power should be higher in within-subject designs, but the degree and conditions of this advantage are unknown for mediation analysis. A Monte Carlo simulation compares designs under a broad range of sample sizes, effect sizes, and correlations among repeated measurements. The results show within-subject designs require about half the sample size of between-subject designs to detect indirect effects of the same size, but this difference can vary with population parameters. I provide an empirical example and R script for conducting power analysis for within-subject mediation analysis. Researchers interested in conducting mediation analysis should not select within-subject designs merely because of higher power, but they should also consider validity and causality in their decision, both of which can favor between-subject designs.

Characteristics associated with treatment seeking for smoking cessation among heavy-drinking research participants.

Objective: Treatment seeking for smoking cessation has tremendous clinical implications with the potential to reduce tobacco-related morbidity and mortality. The present study seeks to elucidate clinical variables that distinguish treatment seeking versus non-treatment seeking status for smoking cessation in a large sample of heavy drinking smokers using data-driven methods. Materials and methods: This secondary data analysis examines n = 911 (n = 267 female) individuals who were daily smokers and heavy drinkers (≥ 7 drinks per week for women, ≥ 14 for men) that were enrolled in either a treatment-seeking study (N = 450) or a non-treatment seeking study (N = 461) using identical pharmacotherapies. Participants completed measures of demographics, alcohol and cigarette use, alcohol craving, the Barratt Impulsiveness Scale (BIS-11), and the Wisconsin Inventory of Smoking Dependence Motives (WISDM-68). These measures were used in a random forest model to identify predictors of treatment seeking status. Results: The top variables of importance in identifying treatment seeking status were: age, drinks per drinking day, cigarettes per smoking day, BIS-11 cognitive impulsivity, WISDM social environmental goads, WISDM loss of control, WISDM craving, and WISDM tolerance. Age and drinks per drinking day were two of the most robust predictors, followed by measures of nicotine craving and tolerance. Conclusion: Individuals who are daily smokers and consume more drinks per drinking day are less likely to belong to the smoking cessationtreatment-seeking group. Targeting heavy drinking smokers, particularly younger individuals, may be necessary to engage this group in smoking cessation efforts and to reduce the burden of disease of nicotine dependence earlier in the lifespan.

Correcting the Bias Correction for the Bootstrap Confidence Interval in Mediation Analysis.

The bias-corrected bootstrap confidence interval (BCBCI) was once the method of choice for conducting inference on the indirect effect in mediation analysis due to its high power in small samples, but now it is criticized by methodologists for its inflated type I error rates. In its place, the percentile bootstrap confidence interval (PBCI), which does not adjust for bias, is currently the recommended inferential method for indirect effects. This study proposes two alternative bias-corrected bootstrap methods for creating confidence intervals around the indirect effect: one originally used by Stine (1989) with the correlation coefficient, and a novel method that implements a reduced version of the BCBCI's bias correction. Using a Monte Carlo simulation, these methods were compared to the BCBCI, PBCI, and Chen and Fritz (2021)'s 30% Winsorized BCBCI. The results showed that the methods perform on a continuum, where the BCBCI has the best balance (i.e., having closest to an equal proportion of CIs falling above and below the true effect), highest power, and highest type I error rate; the PBCI has the worst balance, lowest power, and lowest type I error rate; and the alternative bias-corrected methods fall between these two methods on all three performance criteria. An extension of the original simulation that compared the bias-corrected methods to the PBCI after controlling for type I error rate inflation suggests that the increased power of these methods might only be due to their higher type I error rates. Thus, if control over the type I error rate is desired, the PBCI is still the recommended method for use with the indirect effect. Future research should examine the performance of these methods in the presence of missing data, confounding variables, and other real-world complications to enhance the generalizability of these results.

The effect of neuroimmune modulation on subjective response to alcohol in the natural environment.

BACKGROUND: Despite the promising implications for novel immune therapeutics, few clinical trials have tested these therapies to date. An understanding of how immune pharmacotherapies influence complex alcohol use disorder (AUD) profiles, including subjective response to alcohol, is very limited. Initial findings show that ibudilast, a neuroimmune modulator, reduces rates of heavy drinking and measures of alcohol craving. METHODS: This study is a secondary analysis of a 2-week clinical trial of ibudilast that enrolled a nontreatment-seeking sample with AUD. Eligible participants (N = 52) were randomized to receive ibudilast or matched placebo and completed daily diary assessments (DDAs) during the 2-week period. Each morning, participants reported on their mood and craving levels both before and during the previous day's drinking episode, as well as stimulation and sedation levels during the previous day's drinking episode. Multilevel models were used to compare the effects of ibudilast and placebo on subjective alcohol response. Exploratory analyses tested whether ibudilast moderated the relationship between daily stimulation/sedation and alcohol intake and whether withdrawal-related dysphoria moderated ibudilast's effects on subjective response. RESULTS: Ibudilast did not significantly alter mean levels of stimulation or sedation (p's > 0.05). It did, however, moderate the effect of daily stimulation on drinking (p = 0.045). Ibudilast attenuated alcohol-induced increases in craving compared with placebo (p = 0.047), but not other subjective response measures. Ibudilast significantly tempered daily alcohol-induced changes in urge to drink and positive mood only among individuals without withdrawal-related dysphoria. CONCLUSIONS: Ibudilast's effects on subjective alcohol responses appear to be nuanced and perhaps most salient for individuals drinking for positive reinforcement as distinguished from those who drink to feel normal. Consistent with previous findings, reductions in alcohol craving may represent a primary mechanism of ibudilast's effects on drinking. The ecologically valid nature of DDAs provide a clinically useful window into how individuals experience alcohol's effects while taking ibudilast.

Causal Mediation Analysis in Single Case Experimental Designs: Introduction to the Special Issue.

This special issue of Evaluation and the Health Professions is dedicated to methods for causal mediation analysis in Single Case Experimental Designs (SCEDs). Mediation analysis is used to identify intermediate variables that transmit the effect of the independent variable on the outcome. Until recently, mediation analysis was mostly confined to between-subjects designs and panel studies with few exceptions. Consequently, most of the developments in causal mediation analysis have also been restricted to such designs. In applied health research, SCEDs have been used to evaluate total effects of treatments on outcomes of interest. Providing researchers with the methods for evaluating causal indirect effects for individual participants can lead to important improvements in diagnosis, treatment, and prevention. This special issue includes articles that describe advanced quantitative methods for testing mediators in SCEDs, propose and test approaches that allow for relaxing statistical assumptions that may not hold in real data, and illustrate mediation analysis for a single participant in real and simulated SCEDs data.

The relationship between drinking and smoking in a clinical trial for smoking cessation and drinking reduction.

Heavy drinking smokers experience poorer smoking cessation outcomes. Less is known about the relationship between drinking and smoking among those who are trying to reduce or abstain from both substances. The present study used data from 115 heavy drinking smokers who completed a 12-week clinical trial comparing varenicline alone (1 mg/bid) versus varenicline (1 mg/bid) plus naltrexone (50 mg/day) for smoking cessation and drinking reduction. We tested whether drinking outcomes mediated the relationship between medication and cigarettes per smoking day (CPSD) during the active medication phase (Week 4, 8, and 12) and follow-up phase (Week 16 and 26). CPSD and drinking variables predicted respective use at subsequent time points (p's < .0001). Results revealed a nonsignificant mediation effect of our primary mediator drinks per drinking day (DPDD) at Week 12: 95% CI = [-1.03, .58] and Week 26: 95% CI = [-.09, .51], and our secondary mediators of percent heavy drinking days (PHDDs) and percent days abstinent (PDA) at Week 12: 95% CI = [-.14, .35] and Week 26: 95% CI = [-.15, .41]. Cross-lagged effects (e.g., Week 4 drinking predicting Week 8 smoking) were nonsignificant between DPDD and CPSD (p's ≥ .07), and PHDD and PDA and CPSD that met our a priori cutoff (p's ≥ .02). There was a significant relationship between drinking and smoking concurrently indicated by fixed error covariances (CPSD and DPDD: p < .01; CPSD and PDA p = .01). Our findings highlight an association between drinking and smoking behaviors, respectively, across the span of 6 months. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Socioemotional mechanisms of loneliness in subclinical psychosis.

Loneliness is an important predictor of physical and mental health in the general population and in individuals across the psychosis spectrum, including those experiencing subclinical psychotic-like experiences (PLEs). However, the mechanisms underlying loneliness in the psychosis spectrum are not well understood. Emotion processing deficits are well described across the psychosis spectrum, and socioemotional processing biases are critical for the development and maintenance of loneliness through altered social appraisal, including judgements of rejection. Therefore, we propose that PLEs are associated with increased loneliness, and the relationship is mediated by alterations in socioemotional processing. We also explored how this pathway may be affected by mood and anxiety symptoms, which have been associated with loneliness across the psychosis spectrum. As part of the Human Connectome Project, generally healthy adults (n = 1180) reported symptomatology and social functioning and completed the Penn Emotion Recognition Task to assess efficiency in identifying emotions. We found that higher reported PLEs were associated with elevated levels of loneliness and perceived rejection and that these factors were linked by multiple independent pathways. First, anxiety/depression and emotion processing efficiency independently mediated the PLE-loneliness relationship. Second, we found that the association between PLEs and loneliness was serially mediated through inefficient emotion recognition then higher levels of perceived rejection. These separable mechanisms of increased loneliness in subclinical psychosis have implications for treatment and continued study of social functioning in the psychosis spectrum.

Preliminary study of alcohol problem severity and response to brief intervention.

BACKGROUND: Findings have been mixed as to whether brief intervention (BI) is appropriate and effective for individuals with more severe alcohol use problems. Motivation to change drinking has been supported as a mechanism of behavior change for BI. This exploratory study examined aspects of motivation as mechanisms of clinical response to BI and alcohol problem severity as a moderator of treatment effects. METHODS: Non-treatment-seeking heavy drinkers (average age = 35 years; 57% male) were randomized to receive BI (n = 27) or attention-matched control (n = 24). Three indices of motivation to change were assessed at baseline and post-intervention: importance, confidence, and readiness. Moderated mediation analyses were implemented with treatment condition as the focal predictor, changes in motivation as mediator, 1-month follow-up drinks per day as the outcome, and an alcohol severity factor as second-stage moderator. RESULTS: Analysis of importance displayed a significant effect of intervention condition on importance (p < 0.003) and yielded a significant index of moderated mediation (CI - 0.79, - 0.02), indicating that the conditional indirect effect of treatment condition on drinking through importance was stronger for those with higher alcohol severity. For all motivation indices, alcohol severity moderated the effect of post-intervention motivation levels on drinking (p's < 0.05). The direct effect of treatment condition on drinking was not significant in any model. CONCLUSIONS: Findings highlight the relevance of considering one's degree of alcohol problem severity in BI and alcohol screening efforts among non-treatment seeking heavy drinkers. These nuanced effects elucidate both potential mechanisms and moderators of BI response. Trial registration Clinicaltrials.gov: NCT04710095. Registered January 14, 2021-retrospectively registered, https://clinicaltrials.gov/ct2/show/study/NCT04710095 .

The Poor Fit of Model Fit for Selecting Number of Factors in Exploratory Factor Analysis for Scale Evaluation.

Model fit indices are being increasingly recommended and used to select the number of factors in an exploratory factor analysis. Growing evidence suggests that the recommended cutoff values for common model fit indices are not appropriate for use in an exploratory factor analysis context. A particularly prominent problem in scale evaluation is the ubiquity of correlated residuals and imperfect model specification. Our research focuses on a scale evaluation context and the performance of four standard model fit indices: root mean square error of approximate (RMSEA), standardized root mean square residual (SRMR), comparative fit index (CFI), and Tucker-Lewis index (TLI), and two equivalence test-based model fit indices: RMSEAt and CFIt. We use Monte Carlo simulation to generate and analyze data based on a substantive example using the positive and negative affective schedule (N = 1,000). We systematically vary the number and magnitude of correlated residuals as well as nonspecific misspecification, to evaluate the impact on model fit indices in fitting a two-factor exploratory factor analysis. Our results show that all fit indices, except SRMR, are overly sensitive to correlated residuals and nonspecific error, resulting in solutions that are overfactored. SRMR performed well, consistently selecting the correct number of factors; however, previous research suggests it does not perform well with categorical data. In general, we do not recommend using model fit indices to select number of factors in a scale evaluation framework.

Key dimensions of post-traumatic stress disorder and endothelial dysfunction: a protocol for a mechanism-focused cohort study.

INTRODUCTION: Both trauma exposure and post-traumatic stress disorder (PTSD) are associated with increased risk of cardiovascular disease (CVD), the leading cause of death in the USA. Endothelial dysfunction, a modifiable, early marker of CVD risk, may represent a physiological mechanism underlying this association. This mechanism-focused cohort study aims to investigate the relationship between PTSD (both in terms of diagnosis and underlying symptom dimensions) and endothelial dysfunction in a diverse, community-based sample of adult men and women. METHODS AND ANALYSIS: Using a cohort design, 160 trauma-exposed participants without a history of CVD are designated to the PTSD group (n=80) or trauma-exposed matched control group (n=80) after a baseline diagnostic interview assessment. Participants in the PTSD group have a current (past month) diagnosis of PTSD, whereas those in the control group have a history of trauma but no current or past psychiatric diagnoses. Endothelial dysfunction is assessed via flow-mediated vasodilation of the brachial artery and circulating levels of endothelial cell-derived microparticles. Two higher order symptom dimensions of PTSD-fear and dysphoria-are measured objectively with a fear conditioning paradigm and attention allocation task, respectively. Autonomic imbalance, inflammation, and oxidative stress are additionally assessed and will be examined as potential pathway variables linking PTSD and its dimensions with endothelial dysfunction. Participants are invited to return for a 2-year follow-up visit to reassess PTSD and its dimensions and endothelial dysfunction in order to investigate longitudinal associations. ETHICS AND DISSEMINATION: This study is conducted in compliance with the Helsinki Declaration and University of California, Los Angeles Institutional Review Board. The results of this study will be disseminated via articles in peer-reviewed journals and presentations at academic conferences and to community partners. TRIAL REGISTRATION NUMBER: NCT03778307; pre-results.

Modeling motivation for alcohol in humans using traditional and machine learning approaches.

Given the significant cost of alcohol use disorder (AUD), identifying risk factors for alcohol seeking represents a research priority. Prominent addiction theories emphasize the role of motivation in the alcohol seeking process, which has largely been studied using preclinical models. In order to bridge the gap between preclinical and clinical studies, this study examined predictors of motivation for alcohol self-administration using a novel paradigm. Heavy drinkers (n = 67) completed an alcohol infusion consisting of an alcohol challenge (target breath alcohol = 60 mg%) and a progressive-ratio alcohol self-administration paradigm (maximum breath alcohol 120 mg%; ratio requirements range = 20-3 139 response). Growth curve modeling was used to predict breath alcohol trajectories during alcohol self-administration. K-means clustering was used to identify motivated (n = 41) and unmotivated (n = 26) self-administration trajectories. The data were analyzed using two approaches: a theory-driven test of a-priori predictors and a data-driven, machine learning model. In both approaches, steeper delay discounting, indicating a preference for smaller, sooner rewards, predicted motivated alcohol seeking. The data-driven approach further identified phasic alcohol craving as a predictor of motivated alcohol self-administration. Additional application of this model to AUD translational science and treatment development appear warranted.

MIMIC Models for Uniform and Nonuniform DIF as Moderated Mediation Models.

In this article, the authors describe how multiple indicators multiple cause (MIMIC) models for studying uniform and nonuniform differential item functioning (DIF) can be conceptualized as mediation and moderated mediation models. Conceptualizing DIF within the context of a moderated mediation model helps to understand DIF as the effect of some variable on measurements that is not accounted for by the latent variable of interest. In addition, useful concepts and ideas from the mediation and moderation literature can be applied to DIF analysis: (a) improving the understanding of uniform and nonuniform DIF as direct effects and interactions, (b) understanding the implication of indirect effects in DIF analysis, (c) clarifying the interpretation of the "uniform DIF parameter" in the presence of nonuniform DIF, and (d) probing interactions and using the concept of "conditional effects" to better understand the patterns of DIF across the range of the latent variable.

Fitness, Sleep-Disordered Breathing, Symptoms of Depression, and Cognition in Inactive Overweight Children: Mediation Models.

OBJECTIVES: We used mediation models to examine the mechanisms underlying the relationships among physical fitness, sleep-disordered breathing (SDB), symptoms of depression, and cognitive functioning. METHODS: We conducted a cross-sectional secondary analysis of the cohorts involved in the 2003-2006 project PLAY (a trial of the effects of aerobic exercise on health and cognition) and the 2008-2011 SMART study (a trial of the effects of exercise on cognition). A total of 397 inactive overweight children aged 7-11 received a fitness test, standardized cognitive test (Cognitive Assessment System, yielding Planning, Attention, Simultaneous, Successive, and Full Scale scores), and depression questionnaire. Parents completed a Pediatric Sleep Questionnaire. We used bootstrapped mediation analyses to test whether SDB mediated the relationship between fitness and depression and whether SDB and depression mediated the relationship between fitness and cognition. RESULTS: Fitness was negatively associated with depression ( B = -0.041; 95% CI, -0.06 to -0.02) and SDB ( B = -0.005; 95% CI, -0.01 to -0.001). SDB was positively associated with depression ( B = 0.99; 95% CI, 0.32 to 1.67) after controlling for fitness. The relationship between fitness and depression was mediated by SDB (indirect effect = -0.005; 95% CI, -0.01 to -0.0004). The relationship between fitness and the attention component of cognition was independently mediated by SDB (indirect effect = 0.058; 95% CI, 0.004 to 0.13) and depression (indirect effect = -0.071; 95% CI, -0.01 to -0.17). CONCLUSIONS: SDB mediates the relationship between fitness and depression, and SDB and depression separately mediate the relationship between fitness and the attention component of cognition.

The dark side of fluency: Fluent names increase drug dosing.

Prior research has demonstrated that high processing fluency influences a wide range of evaluations and behaviors in a positive way. But can high processing fluency also lead to potentially hazardous medical behavior? In 2 controlled experiments, we demonstrate that increasing the fluency of pharmaceutical drug names increases drug dosage. Experiment 1 shows that drugs with fluent names are perceived as safer than those with disfluent names and this effect increases drug dosage for both synthetically produced and herbal drugs. Experiment 2 demonstrates that people chose a higher dosage for themselves and for a child if the drug bears a fluent (vs. disfluent) name. Using linear regression based mediation analysis, we investigated the underlying mechanisms for the effect of fluency on risk perception in more detail. Contrary to prior research, we find that affect, but not familiarity, mediates the fluency-risk link. Our findings suggest that a drug name's fluency is a powerful driver of dosing behavior. (PsycINFO Database Record

Two-condition within-participant statistical mediation analysis: A path-analytic framework.

Researchers interested in testing mediation often use designs where participants are measured on a dependent variable Y and a mediator M in both of 2 different circumstances. The dominant approach to assessing mediation in such a design, proposed by Judd, Kenny, and McClelland (2001), relies on a series of hypothesis tests about components of the mediation model and is not based on an estimate of or formal inference about the indirect effect. In this article we recast Judd et al.'s approach in the path-analytic framework that is now commonly used in between-participant mediation analysis. By so doing, it is apparent how to estimate the indirect effect of a within-participant manipulation on some outcome through a mediator as the product of paths of influence. This path-analytic approach eliminates the need for discrete hypothesis tests about components of the model to support a claim of mediation, as Judd et al.'s method requires, because it relies only on an inference about the product of paths-the indirect effect. We generalize methods of inference for the indirect effect widely used in between-participant designs to this within-participant version of mediation analysis, including bootstrap confidence intervals and Monte Carlo confidence intervals. Using this path-analytic approach, we extend the method to models with multiple mediators operating in parallel and serially and discuss the comparison of indirect effects in these more complex models. We offer macros and code for SPSS, SAS, and Mplus that conduct these analyses. (PsycINFO Database Record

Why are some STEM fields more gender balanced than others?

Women obtain more than half of U.S. undergraduate degrees in biology, chemistry, and mathematics, yet they earn less than 20% of computer science, engineering, and physics undergraduate degrees (National Science Foundation, 2014a). Gender differences in interest in computer science, engineering, and physics appear even before college. Why are women represented in some science, technology, engineering, and mathematics (STEM) fields more than others? We conduct a critical review of the most commonly cited factors explaining gender disparities in STEM participation and investigate whether these factors explain differential gender participation across STEM fields. Math performance and discrimination influence who enters STEM, but there is little evidence to date that these factors explain why women's underrepresentation is relatively worse in some STEM fields. We introduce a model with three overarching factors to explain the larger gender gaps in participation in computer science, engineering, and physics than in biology, chemistry, and mathematics: (a) masculine cultures that signal a lower sense of belonging to women than men, (b) a lack of sufficient early experience with computer science, engineering, and physics, and (c) gender gaps in self-efficacy. Efforts to increase women's participation in computer science, engineering, and physics may benefit from changing masculine cultures and providing students with early experiences that signal equally to both girls and boys that they belong and can succeed in these fields. (PsycINFO Database Record