About

Cynthia Lakon is a Professor in the Joe C. Wen School of Population & Public Health at the University of California, Irvine, where she also serves as Interim Chair. She earned her Ph.D. in Health Behavior and Health Education from the University of North Carolina at Chapel Hill in 2004. Her research focuses on the intersection of social networks and health, with particular emphasis on social support, social influence, adolescent social networks, injection drug user networks, ecological models of health, and systems science. Lakon's work explores how social structures and relationships impact health behaviors and outcomes, especially among youth and high-risk populations. She has held a postdoctoral scholar appointment at the University of Southern California's Department of Preventive Medicine and Institute for Preventative Medicine from 2004 to 2006. Throughout her career, she has been recognized for excellence in undergraduate education and is a member of several honor societies, including the Royster Society of Fellows and Delta Omega National Scholastic Honor Society in Public Health.

Research topics

• Psychology
• Machine Learning
• Medicine
• Mathematics
• Computer Science
• Data Mining
• Environmental health
• Social psychology
• Nursing
• Psychiatry
• Statistics
• Applied psychology
• Econometrics
• Developmental psychology

Selected publications

• The Decay of Impact with Network Distance in Linear Diffusion Processes

  ArXiv.org · 2026-04-24

  articleOpen access

  Many processes related to status, power, and influence within social networks have been modeled using forced linear diffusion models; examples include the highly successful Friedkin-Johnsen model of social influence, the status/power scores of Katz and Bonacich, and the widely used network autocorrelation model. While a basic assumption of such models is that the impact of one individual on another through any given path falls exponentially with path length, the total impact of the first individual on the second involves contributions from walks of all lengths; thus, while total impact is expected to decline with network distance, the relationship is not trivial. Here, we provide an approximate solution for the total impact of one node on another as a function of network distance, showing that the total impact is given to first order by a product of eigenvector centrality scores together with an expression in terms of the graph spectrum (eigenvalues of the adjacency matrix) that falls exponentially with distance. We also show how this solution can be refined using higher-order eigenvectors of the adjacency matrix. A numerical study on interpersonal networks drawn from educational settings verifies an average exponential decline in impact strength under the linear diffusion model, and shows that the first-order eigenvector approximation can often be a good proxy for total impact as obtained from the exact solution. This suggests a simple model that can be used to approximate total impact for social influence or status processes in a range of settings.

  PublisherOA PDF

• The Decay of Impact with Network Distance in Linear Diffusion Processes

  arXiv (Cornell University) · 2026-04-24

  preprintOpen access

  Many processes related to status, power, and influence within social networks have been modeled using forced linear diffusion models; examples include the highly successful Friedkin-Johnsen model of social influence, the status/power scores of Katz and Bonacich, and the widely used network autocorrelation model. While a basic assumption of such models is that the impact of one individual on another through any given path falls exponentially with path length, the total impact of the first individual on the second involves contributions from walks of all lengths; thus, while total impact is expected to decline with network distance, the relationship is not trivial. Here, we provide an approximate solution for the total impact of one node on another as a function of network distance, showing that the total impact is given to first order by a product of eigenvector centrality scores together with an expression in terms of the graph spectrum (eigenvalues of the adjacency matrix) that falls exponentially with distance. We also show how this solution can be refined using higher-order eigenvectors of the adjacency matrix. A numerical study on interpersonal networks drawn from educational settings verifies an average exponential decline in impact strength under the linear diffusion model, and shows that the first-order eigenvector approximation can often be a good proxy for total impact as obtained from the exact solution. This suggests a simple model that can be used to approximate total impact for social influence or status processes in a range of settings.

  PublisherDOI

• Peer influence decay and behavioral diffusion in adolescent networks: A simulation approach

  Science · 2026-04-30

  articleSenior author

  How far does peer influence spread through social networks before dissipating? This study investigates the diffusion of smoking behavior in adolescent friendship networks using longitudinal data from two schools ( n = 3154 students) in the National Longitudinal Study of Adolescent to Adult Health. Using Stochastic Actor–Oriented Models, we simulate interventions targeting heavy smokers using various strategies (random, in-degree, eigenvector centrality) and coverage (10 to 100%). A new exponential decay model quantifies influence attenuation, revealing indirect peer influences, or spillover effects, up to three steps from targets. Targeting 10 to 30% of central individuals maximizes smoking reductions, but gains plateau beyond 40 to 50% owing to network saturation. In our analyses, the denser network exhibits broader diffusion and slower decay than the larger, sparser network. This decay metric optimizes intervention design across diverse network structures.

  PublisherDOI

• Simulating social network-based interventions for adolescent cigarette smoking

  Social Science & Medicine · 2025-05-15 · 3 citations

  articleOpen access1st authorCorresponding

  Social network-based adolescent substance use interventions have demonstrated potential for reducing adolescent cigarette smoking. This approach is premised upon leveraging youths' social networks for the diffusion of peer influence. Determining which adolescents to select in network interventions for reducing smoking is a major consideration. We utilize a simulation approach that first estimates Stochastic Actor-Oriented models (SAOM) of adolescent smoking using data from two of the largest schools from the longitudinal saturation sample of the National Study of Adolescent to Adult Health (Add Health) (n = 3,154). We then conduct Agent-Based Simulation models which mimic the consequences of intervention strategies selecting adolescents in network positions and structures that are salient for smoking and the diffusion of peer influence within school-based networks, and we select adolescents smoking at different levels. Our findings indicate that selecting adolescents occupying central network positions yielded the greatest reductions in the number of smokers in a school, one year post intervention. Moreover, our findings indicate that in the school with the higher smoking prevalence, there was a beneficial network multiplier effect one year later, which resulted in more non-smokers than those smokers initially intervened upon. When examining the effects of varying the magnitude of peer influence, we find that targeting central positions in networks led to even greater decreases in smoking in schools with higher levels of peer influence. Our findings highlight interdependence and sensitivity of peer influence to network position and have implications for informing school-based network interventions for adolescent smoking.

  PublisherDOI

• Simulating Social Network-Based Interventions for Adolescent Cigarette Smoking

  CrimRxiv · 2025-05-29

  preprintOpen access1st authorCorresponding

  Social network-based adolescent substance use interventions have demonstrated potential for reducing adolescent cigarette smoking. This approach is premised upon leveraging youths’ social networks for the diffusion of peer influences. Determining which adolescents to select in network interventions for reducing smoking is a major consideration. We utilize a simulation approach that first estimates Stochastic Actor-Oriented models (SAOM) of adolescent smoking using data from two of the largest schools from the longitudinal saturation sample of the National Study of Adolescent to Adult Health (Add Health) (n= 3,154). We then conduct Agent-Based Simulation models which mimic the consequences of intervention strategies selecting adolescents in network positions and structures that are salient for smoking and the diffusion of peer influence within school-based networks, and we select adolescents smoking at different levels. Our findings indicate that selecting adolescents occupying central network positions yielded the greatest reductions in the number of smokers in a school, one year post intervention. Moreover, our findings indicate that in the school with the higher smoking prevalence, there was a beneficial network multiplier effect one year later, which resulted in more non-smokers than those smokers initially intervened upon. When examining the effects of varying the magnitude of peer influence, we find that targeting central positions in networks led to even greater decreases in smoking in schools with higher levels of peer influence. Our findings highlight interdependence and sensitivity of peer influence to network position and have implications for informing school-based network interventions for adolescent smoking.

  PublisherOA PDFDOI

• Social network tie functions of social support and social influence and adult smoking abstinence

  PLoS ONE · 2024-03-07 · 3 citations

  articleOpen access1st authorCorresponding

  Adults' social network ties serve multiple functions and play prominently in quitting smoking. We examined three types of adults' egocentric social networks, including family, friends, and friends online to investigate how two network characteristics with major relevance to health behavior, network size and tie closeness, related to the emotional and confidant support and to pro- and anti-smoking social influence these ties may transmit. We also examine whether the social support and social influence constructs related to smoking abstinence. We utilized baseline and 7-day abstinence survey data from 123 adult current smokers attempting to quit prior to the start of a randomized controlled quit-smoking trial of a social support intervention for quitting smoking on Twitter. To examine study relationships, we estimated Negative Binomial Regression models and Logistic Regression models. For all networks, network size and tie closeness related positively to most of the social support and social influence constructs, with tie closeness related most strongly, especially for online friends. Family pro-smoking social influence related negatively to smoking abstinence, and there were marginally negative relationships for family emotional support and family confidant support. Online friend emotional support had a marginally positive relationship with smoking abstinence. Overall, our findings indicated the importance of the social support and social influence functions of each type of network tie, with larger networks and closer ties related to higher levels of social support and social influence. Moreover, family network pro-smoking social influence may compromise abstinence while emotional support from online friend network ties may reinforce it.

  PublisherOA PDFDOI

• Sociodemographic Predictors of Viral Load Testing Among Hepatitis C Antibody-Positive Patients in a Large Southern California County: An Example of Health Care Inequalities

  2023-08-10

  book-chapterSenior author

  Abstract Purpose The study examines health care inequities in viral load testing among hepatitis C (HCV) antibody-positive patients. The analysis predicts whether individual and census tract sociodemographic characteristics impact the likelihood of viral load testing. Methodology/Approach This a study of 26,218 HCV antibody-positive patients in Orange County, California, from 2010 to 2020. The case data were matched with the 2017 American Community Survey to help understand the role of neighborhood socioeconomic characteristics in testing for viral load. Multivariable logistic regression was used to predict the probability of ever testing for HCV viral load. Findings Thirty-six percent of antibody-positive persons were never viral load tested. The results show inequalities in viral load testing by sociodemographic factors. The following groups were less likely to ever test for viral load than their counterparts: (1) individuals under 65 years old, (2) females, (3) residents of census tracts with lower levels of health insurance enrollment, (4) residents of census tracts with lower levels of government health insurance, and (5) residents of census tracts with a higher proportion of non-white residents. Research Limitations/Implications This is a secondary database from public health department reports. Using census tract data raises the issue of the ecological fallacy. Detailed medical records were not available. The results of this study emphasize the social inequality in viral load testing for HCV. These groups are less likely to be treated and cured, and may spread the disease to others. Originality/Value This chapter is unique as it combines routinely collected public health department data with census tract level data to examine social inequities associated with lower rates of HCV viral load testing.

  PublisherDOI

• Insurance Status, Comorbidity Diagnosis, and Hepatitis C Diagnosis Among Antibody-Positive Patients: A Retrospective Cohort Study

  Health Services Research and Managerial Epidemiology · 2023-01-01

  articleOpen accessSenior author

  Background In California, laboratories report all hepatitis C (HCV)-positive antibody tests to the state; however, that does not accurately reflect active infection among those patients without a viral load test confirming a patient's HCV diagnosis. These public health surveillance disease incident records do not include patient details such as comorbidities or insurance status found in electronic medical records (EMRs). Objective This research seeks to understand how insurance type, insurance status, patient comorbidities, and other sociodemographic factors related to HCV diagnosis as defined by a positive viral load test among HCV antibody-positive persons from January 1, 2010 to March 1, 2020. Methods HCV antibody-positive individuals reported to the California Reportable Disease Information Exchange (CalREDIE), with a medical record number associated with the University of California, Irvine Medical Center, and an unrestricted EMR (n = 521) were extracted using manual chart review. Main Outcomes and measures HCV diagnosis as indicated in a patient's EMR in the problem list or disease registry. Results Less than a quarter of patients in this sample were diagnosed as having HCV in their EMR, with 0.4% of those diagnosed (5/116) patients with indicated HCV treatment in the medication field of their charts. After adjusting for multiple comorbidities, a multinomial logistic regression found that the relative risk ratios (RRRs) of HCV diagnosis found that patients with insurance were more likely to be diagnosed compared to those without insurance. When comparing uninsured patients to those with government insurance at the P < .05 level (RRR = 10.61 (95% confidence interval (CI): 4.14-27.22)) and those uninsured to private insurance (RRR = 6.79 (95% CI: 2.31-19.92). Conclusions These low frequencies of HCV diagnosis among the study population, particularly among the uninsured, indicate a need for increased viral load testing and linkage to care. Reflex testing on existing samples and improving HCV screening and diagnosis can help increase linkage to care and work towards eliminating this disease.

  PublisherOA PDFDOI

• Socio-spatial health disparities in Covid-19 cases and deaths in United States skilled nursing facilities over 30 months

  American Journal of Infection Control · 2023-08-08 · 2 citations

  article1st author
  PublisherDOI

• The Moderating Role of Context: Relationships between Individual Behaviors and Social Networks

  Sociological Focus · 2022-04-03 · 1 citations

  articleOpen accessSenior author

  A social context can be viewed as an entity or unit around which a group of individuals organize their activities and interactions. Social contexts take such diverse forms as families, dwelling places, neighborhoods, classrooms, schools, workplaces, voluntary organizations, and sociocultural events or milieus. Understanding social contexts is essential for the study of individual behaviors, social networks, and the relationships between the two. Contexts shape individual behaviors by providing an avenue for non-dyadic conformity and socialization processes. The co-participation within a context affects personal relationships by acting as a focus for tie formation. Where participation in particular contexts confers status, this effect may also lead to differences in popularity within interpersonal networks. Social contexts may further play a moderating role in within-network influence and selection processes, providing circumstances that either amplify or suppress these effects. In this paper we investigate the joint role of co-participation via social contexts and dyadic interaction in shaping and being shaped by individual behaviors with the context of a U.S. high school. Implications for future study of social contexts are suggested.

  PublisherOA PDFDOI

Recent grants

• NIH Grant R21DA031152

  NIH · $411k · 2015

• NIH Grant F31DA014153

  NIH · $54k

Frequent coauthors

• Cheng Wang

  28 shared

• John R. Hipp

  University of California, Irvine

  25 shared

• Carter T. Butts

  University of California, Irvine

  17 shared

• Rupa Jose

  University of Pennsylvania

  14 shared

• Cornelia Pechmann

  University of California, Irvine

  12 shared

• Judith J. Prochaska

  Stanford University

  12 shared

• Kevin Delucchi

  University of California, San Francisco

  11 shared

• Bernadette Boden‐Albala

  University of California, Irvine

  10 shared

Labs

• Network Cascades LabPI

Education

• Ph.D., Public Health

  University of California, Los Angeles

  2005

• Other, Public Health

  University of California, Los Angeles

  2001

• B.A., International Development Studies

  University of California, Los Angeles

  1998

Awards & honors

• School of Health Science Honoree for Excellence in Undergrad…
• Royster Society of Fellows, UNC-Chapel Hill, 2004- present
• Delta Omega National Scholastic Honor Society in Public Heal…