About

Kenneth A. Bollen is the Henry Rudolph Immerwahr Distinguished Professor in the Department of Psychology and Neuroscience and the Department of Sociology at the University of North Carolina at Chapel Hill. He is also a faculty member in the Quantitative Psychology Program within the Thurstone Psychometric Laboratory and a Fellow of the Carolina Population Center. Bollen's primary areas of statistical research focus on structural equation models, longitudinal methods, and model implied instrumental variables. His work contributes to advancing quantitative methodologies in social sciences, particularly through the development and application of sophisticated statistical models to analyze complex data structures over time. More detailed information about his academic background and contributions is available in his curriculum vitae.

Research topics

• Psychology
• Computer Science
• Psychiatry
• Applied mathematics
• Statistics
• Mathematics
• Medicine
• Econometrics
• Thermodynamics
• Emergency medicine
• Engineering
• Neuroscience
• Internal medicine
• Medical emergency
• Statistical physics
• Operations management
• Mathematical analysis
• Physics

Selected publications

• In Defense of Causal-Formative Indicators: A Minority Report

  UNC Libraries · 2025-01-10

  articleOpen accessSenior author

  Causal-formative indicators directly affect their corresponding latent variable. They run counter to the predominant view that indicators depend on latent variables and are thus often controversial. If present, such indicators have serious implications for factor analysis, reliability theory, item response theory, structural equation models, and most measurement approaches that are based on reflective or effect indicators. Psychological Methods has published a number of influential articles on causal and formative indicators as well as launching the first major backlash against them. This article examines 7 common criticisms of these indicators distilled from the literature: (a) A construct measured with "formative" indicators does not exist independently of its indicators; (b) Such indicators are causes rather than measures; (c) They imply multiple dimensions to a construct and this is a liability; (d) They are assumed to be error-free, which is unrealistic; (e) They are inherently subject to interpretational confounding; (f) They fail proportionality constraints; and (g) Their coefficients should be set in advance and not estimated. We summarize each of these criticisms and point out the flaws in the logic and evidence marshaled in their support. The most common problems are not distinguishing between what we call causal-formative and composite-formative indicators, tautological fallacies, and highlighting issues that are common to all indicators, but presenting them as special problems of causal-formative indicators. We conclude that measurement theory needs (a) to incorporate these types of indicators, and (b) to better understand their similarities to and differences from traditional indicators.

  PublisherDOI

• “I still feel so lost”: experiences of women receiving SANE care during the year after sexual assault

  UNC Libraries · 2025-01-10

  articleOpen access

  OBJECTIVE: Emergency caregivers provide initial care to women sexual assault (SA) survivors. An improved understanding of the issues facing this population can aide emergency care practitioners in providing high quality care. The goal of this study was to share the experiences of women SA survivors with the emergency care practitioners that care for them. METHODS: English-speaking adult women (n = 706) who received SA Nurse Examiner (SANE) evaluation within 72 hours of SA at 1 of 13 geographically distributed sites were enrolled in a prospective, longitudinal multi-site observational study. We qualitatively analyzed responses to the open-ended question: "What do you think is most important for researchers to understand about your experience since the assault?" asked 1 week, 6 weeks, 6 months, and 1 year after enrollment. RESULTS: Themes from responses (n = 1434) from 590 women (84% of study sample) fell into 12 broad categories: daily life, justice, medical, and social services, mental health, physical health, prior trauma, recovery, romantic relationships, safety, self, shame, and social interactions. Responses demonstrated that the assault permeates many aspects of assault survivors' daily lives. CONCLUSIONS: Qualitative analyses of open-ended responses from a large cohort of women SA survivors receiving SANE care highlight the challenges for survivors and can increase understanding among the emergency care practitioners who care for them. The authors propose a brief acronym to help emergency care practitioners recall important messages for SA survivors.

  PublisherDOI

• A Longitudinal Study of the Domain-Generality of African American Students’ Causal Attributions for Academic Success

  UNC Libraries · 2025-01-10

  articleOpen access

  Students’ causal attributions about the reasons underlying their academic successes are important because of the influence of those attributions on academic motivation. We investigated whether students’ success attributions tend to be similar across academic subjects versus specific to academic domain, and whether domain-generality or specificity changes with development. African American students (N = 565) reported their causal attributions for math, science, and English successes longitudinally from elementary to high school. Structural equation modeling showed that individual differences in students’ tendencies to attribute successes to ability, effort, or their teachers were domain-general, not differing across academic content areas. Results did not differ by sex. The lack of domain-specificity in attributions suggests that when African American students consider what factors influence their school performance, they view academic outcomes as a single achievement domain rather than differentiating among school subjects.

  PublisherDOI

• Redefine statistical significance

  Artefactual Field Experiments · 2025-01-10 · 21 citations

  articleOpen access
  PublisherOA PDFDOI

• Racial differences in presentations and predictors of acute pain after motor vehicle collision

  Carolina Digital Repository (University of North Carolina at Chapel Hill) · 2025-01-10 · 1 citations

  articleOpen access

  African Americans experience a greater burden of acute pain than non-Hispanic white individuals across of variety of acute medical conditions, but it is unknown whether this is the case after trauma. We evaluated pain, pain-related characteristics (eg, peritraumatic distress), and analgesic treatment in 2 cohorts of individuals (African American [n = 931] and non-Hispanic white [n = 948]) presenting to the emergency department (ED) after a motor vehicle collision. We performed a propensity-matched analysis (n = 796 in each group) to assess racial differences in acute pain in the ED. In multivariable models conducted within the matched sample, race was associated with moderate to severe axial pain (odds ratio [OR] 3.2; 95% confidence interval [CI]: 2.1-5.0, P < 0.001) and higher average numerical rating scale scores (1.3; 95% CI: 1.1-1.6; P < 0.001). After adjustment for pain and other covariates, non-Hispanic white patients were more likely to receive an opioid analgesic in the ED (OR 2.0; 95% CI: 1.4-3.0, P < 0.001) or at discharge (OR 4.9; 95% CI: 3.4-7.1, P < 0.001) and also less likely to receive an NSAID in the ED (OR 0.54; 95% CI: 0.38-0.78; P = 0.001) or at discharge (0.31; 95% CI: 0.43-0.84). Racial differences in the severity of acute posttraumatic pain after a motor vehicle collision are not explained by factors such as socioeconomic status or crash characteristics. Despite a higher burden of acute pain, African Americans were less likely to receive opioid analgesics and more likely to receive NSAIDs. Further work is needed to understand the relationship between pain severity, disparities in analgesic treatment, and longer term outcomes, such as post-motor vehicle collision chronic pain.

  PublisherOA PDFDOI

• Association of PFAS and Metals with Cardiovascular Disease Risk: Exploring the Mediating Effect of Diet

  Environments · 2025-05-28 · 3 citations

  articleOpen access

  Background: Cardiovascular disease (CVD) is a major global health burden influenced by genetic, behavioral, and environmental factors. Among these, exposure to per- and poly-fluoroalkyl substances (PFASs) and toxic metals has been increasingly implicated in adverse cardiovascular outcomes. However, the mediating role of dietary inflammation in these associations remains unclear. Objective: This study investigates the relationship between PFAS and metal exposures and CVD risk, focusing on the potential mediating role of diet, operationalized through the Dietary Inflammatory Index (DII). Additionally, this study examines age as an effect modifier in these associations. Methods: = 660), we assessed environmental exposures (lead, cadmium, mercury, perfluorooctanoic acid-PFOA, perfluorooctane sulfonate-PFOS), dietary inflammatory potential (DII), and cardiovascular markers (blood pressure, lipid profile, C-reactive protein). Statistical analyses included linear regression and Bayesian Kernel Machine Regression-Causal Mediation Analysis (BKMR-CMA) to estimate the direct, indirect (through DII), and total effects of exposure on CVD risk biomarkers. Results: = 0.006). Mediation analysis suggested dietary inflammation may play a role, though estimates were imprecise. Conclusions: PFAS and metals may influence CVD risk through inflammatory pathways, with potential age-related differences. Future longitudinal studies are needed to clarify these complex interactions, reduce measurement error, and guide age-specific exposure regulations.

  PublisherOA PDFDOI

• Fifty years of structural equation modeling: A history of generalization, unification, and diffusion

  UNC Libraries · 2025-01-10

  articleOpen access
  PublisherDOI

• Association of PFAS and Metals with Cardiovascular Disease Risk: Exploring the Mediating Effect of Diet

  UNC Libraries · 2025-06-04

  articleOpen access

  Background: Cardiovascular disease (CVD) is a major global health burden influenced by genetic, behavioral, and environmental factors. Among these, exposure to per- and poly-fluoroalkyl substances (PFASs) and toxic metals has been increasingly implicated in adverse cardiovascular outcomes. However, the mediating role of dietary inflammation in these associations remains unclear. Objective: This study investigates the relationship between PFAS and metal exposures and CVD risk, focusing on the potential mediating role of diet, operationalized through the Dietary Inflammatory Index (DII). Additionally, this study examines age as an effect modifier in these associations. Methods: Utilizing data from the National Health and Nutrition Examination Survey (NHANES) 2017–2018 cycle (n = 660), we assessed environmental exposures (lead, cadmium, mercury, perfluorooctanoic acid-PFOA, perfluorooctane sulfonate-PFOS), dietary inflammatory potential (DII), and cardiovascular markers (blood pressure, lipid profile, C-reactive protein). Statistical analyses included linear regression and Bayesian Kernel Machine Regression-Causal Mediation Analysis (BKMR-CMA) to estimate the direct, indirect (through DII), and total effects of exposure on CVD risk biomarkers. Results: Linear regression revealed significant associations between mercury and reduced systolic blood pressure (SBP) (p = 0.017) and cadmium with increased C-reactive protein (CRP) (p = 0.006). Mediation analysis suggested dietary inflammation may play a role, though estimates were imprecise. Conclusions: PFAS and metals may influence CVD risk through inflammatory pathways, with potential age-related differences. Future longitudinal studies are needed to clarify these complex interactions, reduce measurement error, and guide age-specific exposure regulations.

  PublisherDOI

• Nonlinear Autoregressive Latent Trajectory Models

  UNC Libraries · 2025-01-10

  articleOpen accessSenior author

  Autoregressive latent trajectory (ALT) models combine features of latent growth curve models and autoregressive models into a single modeling framework. The development of ALT models has focused primarily on models with linear growth components, but some social processes follow nonlinear trajectories. Although it is straightforward to extend ALT models to allow for some forms of nonlinear trajectories, the identification status of such models, approaches to comparing them with alternative models, and the interpretation of parameters have not been systematically assessed. In this paper we focus on two forms of nonlinear autoregressive latent trajectory (NLALT) models. The first form allows for a quadratic growth trajectory, a popular form of nonlinear latent growth curve models. The second form derives from latent basis models, or freed loading models, that allow for arbitrary growth processes. We discuss details concerning parameterization, model identification, estimation, and testing for the two forms of NLALT models. We include a simulation study that illustrates potential biases that may arise from fitting alternative models to data derived from an autoregressive process and individual-specific nonlinear trajectories. In addition, we include an extended empirical example modeling growth trajectories of weight from birth through age 2.

  PublisherDOI

• Experiences of Women Sexual Assault Survivors With Police in the Early Aftermath of Assault: Results From a Large-scale Prospective Study

  Journal of Forensic Nursing · 2025-09-23

  article

  BACKGROUND: Over 100,000 women present for emergency care after sexual assault (SA) annually in the United States. No large-scale prospective studies have assessed SA survivor experiences with police. AIMS: To evaluate SA survivor experiences with the police. METHODS: A large-scale longitudinal study was conducted of women who received SA nurse examiner examinations after presenting to the emergency department following SA at 12 sites ( n = 706). RESULTS: At six weeks ( n = 630), most women were interested in speaking with the police (75%), spoke with police, and reported positive experiences. Latinas and women with lower income were less likely to speak with police ( χ2 = 4.76, p = 0.0370; χ2 = 11.37, p = 0.0008). Survivors with greater posttraumatic stress and previous trauma report worse experiences with police. Qualitative comments provide key points for police to consider. DISCUSSION: This study demonstrates high contact and overall satisfaction with police. However, disparities remain among Latinas, survivors with lower socioeconomic status, and survivors with previous life trauma.

  PublisherDOI

Recent grants

• Collaborative Research: Structural Misspecification in Latent Variable Models: Symptoms, Consequences, and Diagnostic Tests

  NSF · $178k · 2006–2011

Frequent coauthors

• Francesca L. Beaudoin

  Brown University

  91 shared

• Paul I. Musey

  Indiana University School of Medicine

  81 shared

• Phyllis L. Hendry

  University of Florida

  80 shared

• Niels K. Rathlev

  University of Massachusetts Chan Medical School

  79 shared

• Samuel A. McLean

  75 shared

• Sophia Sheikh

  Florida College

  73 shared

• Robert M. Domeier

  72 shared

• León D. Sánchez

  Brigham and Women's Faulkner Hospital

  69 shared

Education

• PhD, Sociology

  Brown University

  1977

• Masters, Sociology

  Brown University

  1975

• BA, Sociology

  Drew University

  1973

• none, Mathematics

  Stetson University

  1970