New Learning Resource for Latent Class Analysis

June 25, 2020:

Blank LCA diagram: an oval represents the latent construct. 4 arrows point from the oval to rectangles that represent the manifest variables that are used to measure the latent construct.Our newest resource helps researchers teach themselves latent class analysis. LCA allows researchers to identify unobservable, or latent, subgroups within a population. The LCA Learning Path is designed to allow SAS users to efficiently teach themselves how to plan, run, and interpret an LCA in PROC LCA for SAS.

The Learning Path allows users to select from a variety of educational resources including videos, presentation slides, web pages, and hands-on SAS exercises. This format allows researchers to access the specific content they need in the format they desire to develop their skills as quickly as possible. Content is divided into the following sections:

  • conceptual introduction,
  • detailed introduction,
  • LCA and latent transition analysis (LTA) software,
  • LCA with a grouping variable,
  • LCA with covariates, and
  • LTA.

We hope this will be a valuable tool for students and teachers alike. We also offer a TVEM Learning Path that uses a similar format.

Open the LCA Learning Path.

Principal Investigators’ New Websites

May 6, 2020:

Stephanie Lanza, Bethany Bray, Linda Collins, Susan Murphy, Runze LiAs previously announced, substantial changes lie ahead for The Methodology Center. Over the coming months, we will stop updating this website. Resources will remain available for at least a year, but in order to provide the latest information, each of our investigators will maintain her or his own website. These sites will include content developed at The Methodology Center and new resources related to the researcher’s future work.

Stephanie Lanza and Ashley Linden-Carmichael built a website for the Addictions and Innovative Methods (AIM) lab. Their great new site,, describes their research and includes the content about time-varying effect modeling (TVEM) from The Methodology Center’s website.

Susan Murphy has incorporated The Methodology Center’s content on just-in-time adaptive interventions in her website, The site also includes workshop materials and other resources.

Bethany Bray‘s new site at will include The Methodology Center’s resources for latent class analysis (LCA) and latent transition analysis (LTA). Bethany has concrete plans for new LCA and LTA resources, so stay tuned.

Runze Li will update his page at to incorporate Methodology Center resources on variable screening and variable selection for high-dimensional data analysis.

Linda Collins will build a new website to house The Methodology Center’s content on the multiphase optimization strategy (MOST) for optimizing interventions after she moves to New York University. In the meantime, follow Linda on Twitter, @collins_most.

Daniel Almirall and Inbal “Billie” Nahum-Shani’s informative website,, will soon incorporate The Methodology Center’s resources for the sequential, multiple assignment, randomized trial (SMART).

More information will follow in June or July. Thank you for staying connected to our research! We are all proud of our time at The Methodology Center and very excited about the future.

New SAS Software for LCA With Covariates

May 4, 2020:

blank LCA diagram: a latent variable represented by an oval and four arrows pointing towards manifest variables represented by rectangles

PROC LCA is our free add-on to SAS statistical software for estimating latent class models. We are pleased to announce the release of the LCA Covariates 3-Step SAS macro, which supplements PROC LCA’s functionality. The macro estimates the association between covariates and latent class membership using the approach of Bolck, Croon, and Hagenaars (2004), as adapted by Vermunt (2010) and Vermunt and Magidson (2015). It is a “three-step” (noninclusive) approach, which is a flexible and robust alternative to the “one-step” (inclusive) approach implemented via the COVARIATES statement in PROC LCA.

Note: This does not mean that models using the COVARIATES statement are invalid. It does mean that covariates will not affect estimation of the measurement model; when assumptions of the “one-step” approach are met, results using the new macro will be very similar to those with the COVARIATES statement.

Read more or download the macro.

Join Us at SPR!

April 30, 2019:

Join us at the Society for Prevention Research (SPR) 2019 Annual Meeting, May 28 through 31 in San Francisco. Methodology Center researchers will present symposia, talks, posters, a technical demonstration, and participate in the SPR Cup. We hope to see you there! Below is a list of the places where you can find us.

Tuesday, May 28

5:30 – 7:00 p.m. Poster Session I

  • “Heavy drinking and academics: Daily-level associations, or do less serious students just drink more?” Hannah Allen
  • “Profiles of dysregulation moderate the impact of preschool teacher-student relationships on later school functioning” Benjamin Bayly
  • “Identifying substance use disorders among individuals with spinal cord injury: Using big data Sources via electronic health records” Scott Graupensperger
  • “Effects of a mindfulness training intervention on alcohol use in public school teachers” Natalia Van Doren

Wednesday, May 29

1:15 – 2:45 p.m. Roundtable: Enhancing the reach and impact of drug abuse and behavioral health preventive interventions: Mining existing data for bold new discoveries Stephanie Lanza, Discussant

5:45 – 7:00 p.m. Poster Session II

  • “Approaches to characterizing drinking episodes in college students from wearable alcohol sensors” John Felt
  • “Gender differences in the time-varying association between cigarette use and weight concerns across adolescence” Anna Hochgraf
  • “Drug use patterns among young men of color who have sex with men” Eric Layland

7:00 –8:30 MOST-ly Mingling Join Kate Guastaferro in the Eclipse Kitchen & Bar, located in the lobby of the Hyatt Regency San Francisco, to socialize and discuss issues related to the optimization of interventions.

Thursday, May 30

10:15 – 11:45 a.m. Organized Paper Symposium: Opioid and other nonmedical prescription drug use in the United States: Contemporary trends in use, co-use, and correlates to identify opportunities for prevention Stephanie Lanza, organizer

  • “Contemporary trends in nonmedical prescription drug use as a function of individual and sociodemographic characteristics: Ages 12 to 90” Stephanie Lanza
  • “Age-varying trends in co-use of marijuana and heavy episodic drinking: Implications for nonmedical prescription drug use” Ashley Linden-Carmichael

10:15 – 11:45 a.m. Sloboda and Bukoski Cup Team:  Hannah Allen, Andrew Dismukes, John Felt, Natalia Van Doren, and Adrienne Woods

10:15 – 11:45 a.m. Roundtable Discussion: SPR task force on reducing health disparities and improving equity through prevention Bethany Bray, Discussant

3:00 – 4:30 p.m. Individual paper presentations: Prevention related to drug abuse across developmental stage Bethany Bray, Moderator

3:00 – 4:30 p.m. Individual paper presentations:Family, individual, and neighborhood risk factors as predictors of long-term behavior and mental health problems 

  • “Constellations of family risk and long-term adolescent antisocial behavior: A latent profile analysis” Emily LoBracio

6:40 – 7:55 p.m. Poster Session III

  • Technology Demonstration: Software, instructional materials, videos, and other resources from The Methodology Center at Penn State Bethany Bray

Friday May 31

8:30 – 10:00 a.m. Organized Paper Symposium: Applying latent class models in prevention science: Practical solutions to everyday problems Bethany Bray, Organizer

  • “Multiple imputation of missing covariate information in latent class analysis: evaluation of a step-by-step approach” John Dziak
  • “Multilevel latent profile analysis for daily diary data: Understanding triadic family dynamics” Mengya Xia
  • “Combining latent class analysis and time-varying effect modeling: Understanding the epidemiology of alcohol use” Bethany Bray

8:30 – 10:00 a.m. Individual Paper Presentations: using mobile health techniques to understand and prevent substance use

  • “Day and within-day trends of drug cravings: Ecological momentary assessment among a sample of patients with prescription opiate dependence” Jamie Gajos

10:15 – 11:45 a.m. Plenary Session III, Mobile health (mHealth) in prevention science: Assessment, intervention, and analysis Stephanie Lanza, Chair

1:00 – 2:30 p.m. Plenary Session III Roundtable: Mobile health (mHealth) in prevention science: Assessment, intervention, and analysis Stephanie Lanza, Chair

2:45 – 4:15 p.m. Organized Paper Symposium: Using time-varying effect models to understand predictors of substance use and depression within-days and across developmental periods Benjamin Bayly, Organizer

  • “Age-varying association between childhood maltreatment and depression and substance use” Yuen Wai Hung
  • “Age-varying effects of parental warmth and closeness on adolescent and young adult substance use and depression” Benjamin Bayly

Video: Two-Hour Webinar on Latent Class Analysis (LCA)

March 19, 2019:


Thanks to all who participated in our 1 & 1 workshop on latent class analysis (LCA). This is a video of the webinar that Methodology Center Associate Director Bethany Bray presented on Thursday, February 26, 2019. The video includes both the one-hour presentation and the one-hour question-and-answer session that followed. This recording is a great way to learn the basics of LCA or to use as a refresher.

Download the presentation slides.

Watch the video on YouTube.

Free, Two-Hour, Webinar on Latent Class Analysis

Bethany BrayJanuary 22, 2019:

For our next 1 & 1 webinar, Methodology Center Associate Director Bethany Bray will present an introduction to latent class analysis (LCA). 1 & 1 workshops consist of a one-hour live video presentation on a method followed by a one-hour question-and-answer session with the presenter. The workshop will be held on Tuesday, February 26, from 3:00 to 5:00 p.m. ET. To join the webinar, click this link when the webinar is starting. Registration in advance is not necessary.

Latent class analysis (LCA) allows researchers to identify unobservable (latent) subgroups within a population based on individuals’ responses to multiple observed variables. LCA can be used to understand the impact of exposure to patterns of multiple risks, as well as the antecedents and consequences of complex behaviors, so that interventions can be tailored to target the subgroups that will benefit most. Bethany’s talk will be introductory in nature, but questions of all levels of complexity are welcome during the Q & A.

The webinar can accommodate up to 500 participants, so all interested people should be able to attend. If you try to attend the workshop but are unable to log on for any reason, please send an email to We hope you will join us.

Featured Article: LCA on Trends in Teen Marijuana Use

October 17, 2018:jlb

Over the last several years, laws about marijuana use have been changing across the United States. Methodology Center researchers Jessica Braymiller, Ashley Linden-Carmichael, and Stephanie Lanza wanted to know how marijuana use and attitudes about marijuana use might be changing in the face of those legal changes. In a recent article in Journal of Adolescent Health, the authors examined these questions using data from the 2010-2016 waves of the Monitoring the Future study.

The authors applied latent class analysis (LCA) to reveal patterns in marijuana use and attitudes among high school seniors in the United States. The analysis revealed that beginning in 2014, increases were observed in two subgroups: nonusers who are tolerant of marijuana use and marijuana users who generally approve of marijuana use at any level (i.e., experimentation, occasional use, and regular use).

Jessica Braymiller, graduate student at The Methodology Center and lead author on this paper, explained the implications of these findings. “Our findings indicate that many high school seniors in the United States have used marijuana recently and/or have approving attitudes regarding marijuana use at any frequency. Interestingly, the prevalence of these subgroups have increased in recent years, and males are most likely to belong to these subgroups. Since many adolescents are tolerant of marijuana use by the time they reach 12th grade, prevention and intervention efforts should address marijuana use behaviors and related attitudes early on.”

She went on to describe the value of applying latent class models to mature panel data like Monitoring the Future. “LCA allows researchers to categorize and describe individuals based on variety of shared characteristics. This approach is particularly useful in the context of nationally representative data sets, as we are able to comprehensively examine patterns of multiple behaviors and attitudes within a given population. Further comparing these patterns based on demographic characteristics enables researchers to identify population subgroups who may be at risk, having important implications for prevention and intervention efforts.”

Open the article.


Braymiller, J. L., Masters, L. D., Linden-Carmichael, A. N., & Lanza S. T. (2018) Contemporary patterns of marijuana use and attitudes among high school seniors: 2010–2016. Journal of Adolescent Health, 63(4), 394-400.

Join Us at SBM

February 27, 2018:sbm

Are you planning to attend the Society for Behavioral Medicine’s 39th Annual Meeting & Scientific Sessions in New Orleans on April 11-14, 2018? If so, schedule some time for some of our many talks, posters, workshops, and more.

April 11

8:30 a.m.—6:00 p.m. Pre-Conference Course
“Novel experimental approaches to designing effective multi-component interventions” Daniel Almirall, Linda Collins, Susan Murphy, Inbal “Billie” Nahum-Shani

8:30 a.m.—11:15 a.m. Pre-Conference Seminar
“Analysis of ambulatory assessment data in behavioral medicine” Stephanie Lanza, Michael Russell

6:30 p.m.—7:30 p.m. Poster Session A
“High-intensity drinking: Prevalence rates across adulthood by gender and race/ethnicity” Ashley Linden-Carmichael

“Biopsychosocial correlates of discrimination in daily life” Lindsey Potter

April 12

10:45 a.m.—11:45 a.m. Midday Meeting
“Lightning rounds with optimization of behavioral and biobehavioral interventions experts: MOST, SMART and MRTs” Daniel Almirall, Linda Collins, Thelma Milenz, Susan Murphy, Inbal “Billie”Nahum-Shani

2:00 p.m.—3:15 p.m. Symposium
“Physical activity and optimization of behavioral and biobehavioral interventions SIGs present: Optimization experiments in the field: The MOST framework through 3 clinical trials” Bonnie Spring, Linda Collins

6:15 p.m.—7:15 p.m. Poster Session B
“Contemporary patterns of marijuana use and attitudes among high school seniors: 2010-2014” Jessica Braymiller

April 13

6:15 p.m.—7:15 p.m. Poster Session C
“Latent Classes of Discrimination among Sexual Minority Adults: Associations with Substance Use Disorders” Cara Rice

April 14

10:00 a.m.—11:00 a.m. Poster Session D
“Exploring quantitative approaches to examining the effect of multiple disadvantaged social status indicators on health” Lindsey Potter

Free, Two-Hour, Online Workshop on Latent Transition Analysis

February 9, 2018:online workshop lta

For our next 1 & 1 workshop, Methodology Center Associate Director Bethany Bray will present an introduction to latent transition analysis (LTA). 1 & 1 workshops consist of a one-hour live video presentation on a method followed by a one-hour question-and-answer session with the presenter. After the presentation, Bethany will accept questions via instant message and answer them live. The workshop will be held on Thursday, March 29, from 3:00 to 5:00 p.m. ET.

LTA and latent class analysis (LCA) are closely related methods. LCA identifies unobservable (latent) subgroups within a population based on individuals’ responses to multiple observed variables. LTA is an extension of LCA that uses longitudinal data to identify movement between the subgroups over time. Workshop participants should possess basic familiarity with LCA before attending. For an introduction to LCA, see Collins and Lanza (2010) or Lanza, Bray, and Collins (2013).

The 1 & 1 will be hosted via Zoom webinar at The workshop can accommodate up to 500 participants, so all interested people should be able to attend. If you try to attend the workshop but are unable to log on for any reason, please send an email to We hope you will join us.


Collins, L. M., & Lanza, S. T. (2010). Latent class and latent transition analysis: With applications in the social, behavioral, and health sciences. New York: Wiley.

Lanza, S. T., Bray, B. C., & Collins, L. M. (2013). An introduction to latent class and latent transition analysis. In J. A. Schinka, W. F. Velicer, & I. B. Weiner (Eds.), Handbook of psychology (2nd ed.,Vol. 2, pp. 691-716). Hoboken, NJ: Wiley.

Teachers Corner: New Resource for Instructors

August 17, 2017:bbray lca

Attention, instructors of graduate-level methods courses: The Methodology Center’s new Teachers’ Corners will provide resources for instructors who want to incorporate instruction on innovative methods into their teaching. Teachers’ Corners include PowerPoint presentations, introductory articles for instructors, SAS exercises, reading lists for students, and other items designed to make comprehending and incorporating a method as easy as possible. Our first Teachers’ Corner is for latent class analysis (LCA). LCA is a broadly applied method for identifying hidden subgroups within a population.

Download or read more.

New Software for LCA with a Distal Outcome

June 16, 2017:

Our LCA distal outcome estimation software allows users to estimate the association between a latent class variable and a distal outcome in either Stata or SAS. Our previous LCA distal outcome releases have employed a model-based approach known as the LTB approach. As research on LCA with a distal outcome advanced, it became clear that a different method, one developed by Bolck, Croon and Hagenaars (2004), produces more accurate estimates in certain cases. We have implemented this method, known as the BCH approach, in a new SAS macro and Stata function. We recommend the new LCA_Distal_BCH SAS macro and LCA_Distal_BCH Stata function for estimating LCA with a distal outcome.

To facilitate methodological research, we have also repackaged the previous LCA_Distal SAS macro (v. 3.0.2) and the previous LCA_Distal Stata function as the LCA_Distal_LTB SAS macro and the LCA_Distal_LTB Stata function, respectively.

Open the SAS distal outcomes software page.

Open the Stata distal outcomes software page.

Featured Article: LCA of Why Young Adults Use Marijuana and Why Their Reasons Matters

March 9, 2017:smoke--marijuana

As state laws regarding marijuana change around the nation, legislators and the public need information about the impacts of marijuana use. Research has shown that smoking marijuana in order to cope with problems is associated with later marijuana-related problems (Fox et al., 2011). In a recent article in Journal of Studies on Alcohol and Drugs, a team of researchers including Methodology Center Associate Director Bethany Bray examined data on self-reported motives for using marijuana during young adulthood and then determined which motivational profiles were associated with later marijuana use and problems.

In the article, “Reasons for marijuana use among young adults and long-term associations with marijuana use and problems,” the authors applied latent class analysis (LCA) to a sample from the long-running Monitoring the Future study. The sub-sample included participants who provided marijuana use data at age 19 or 20 and again at age 35. The analysis revealed five classes of motives for using marijuana in young adulthood: Experimental Reasons, Get High and Relax Reasons, Typical Reasons, Typical and Escape Reasons, and Coping and Drug Effect Reasons. “Typical” reasons included “to feel good or get high,” “to have a good time with my friends,” “to experiment,” and “to relax.” When looking at the association between class membership in young adulthood and later problems, the authors found that members of the Experimental Reasons class were the least likely to use marijuana at age 35. Members of the Get High and Relax Reasons and Coping and Drug Effect Reasons classes were most likely to use marijuana and have problems with marijuana use at age 35.

Co-author Bethany Bray said, “Because individuals may hold multiple reasons for marijuana use simultaneously, methods like LCA can provide unique information about how reasons cluster within individuals and which patterns confer the greatest risk for later problems. Our findings support the need for motivation-based interventions that can be targeted and adapted based on salient reasons for use among young adults.”

Open the article (Journal access required)


Patrick, M. E., Bray, B. C., & Berglund, P. A. (2016). Reasons for marijuana use among young adults and long-term associations with marijuana use and problems. Journal of Studies on Alcohol and Drugs, 77(6), 881-888.

Fox, C. L., Towe, S. L., Stephens, R. S., Walker, D. D., & Roffman, R. A. (2011). Motives for cannabis use in high-risk adolescent users. Psychology of Addictive Behaviors, 25, 492–500. doi:10.1037/a0024331

Podcast: Practical Advice on LCA

December 2, 2016:jjd-atw

Latent class analysis (LCA) is a widely used tool for identifying subgroups in a population. Many researchers have questions about how to conduct an LCA as responsibly and accurately as possible. In our latest podcast, John Dziak discusses important points to consider when conducting an LCA, like how to tell when an analysis is successful and how to make sure your model is properly identified. John is a Methodology Center research associate who studies LCA, and he is the lead developer of our LCA software, including PROC LCA. Note: this podcast is a companion piece to podcasts 15 and 16 with Stephanie Lanza and Bethany Bray. If you are new to LCA, you may want to start with Podcast 15.

Podcast Timeline:

00:30—What is LCA for?
01:15—Why would someone use LCA?
02:27—How does LCA work?
04:20—How do I select a model?
07:39—How do I know if my LCA worked?
13:45—How do I select items for my model?
18:20—What “percent identified” of random starts is high enough?
19:23—When should I use a higher value in NSTARTS?
20:13—What should I do if my model won’t converge?
23:00—When should I use the RESTRICT option?

Download Podcast 26

Article: Do Social Networks Change After Quitting Smoking?

December 1, 2016:

A common fear of many smokers who want to quit is that they will lose many people in their social network―family, friends or co-workers―when they quit.Kids-smoking In a recent article in the journal Nicotine and Tobacco Research, researchers applied latent transition analysis to examine the changes in the social networks of smokers who are quitting. The authors identified five types of networks and found that people who successfully quit are actually likely to increase the size of their social networks.

The results are based on a sample from a smoking-cessation trial conducted by researchers at the University of Wisconsin Center for Tobacco Research (UW-CTRI). In the study, UW-CTRI researchers interviewed 691 adult smokers about their friends and associates.

This analysis was led by Methodology Center Associate Director Bethany Bray. The authors analyzed how participants’ social networks changed during the three years after quitting. Smokers who quit were more likely to transition to larger social networks, especially amongst participants who had the highest levels of exposure to other smokers before quitting. In other words, quitting was associated with an increase in the number of people, especially non-smokers, in people’s social networks.

“Clinicians can use results from this study to reassure smokers that quitting tends to increase, not decrease, the size of social networks,” said co-author and smoking expert Megan Piper. “Many smokers tell us cigarettes are their best friend, and that doesn’t have to be the case. We found our patients who quit expanded their social networks and developed meaningful relationships after they quit smoking.”

Open the article. (Journal access required.)

Latent Classes of Quitting Smokers: Who Will Relapse?

stl_savOctober 25, 2016:

Research indicates that withdrawal is one of the primary reasons that people do not quit smoking (Piper, 2015). Improving our understanding of withdrawal may allow us to better support people who wish to quit smoking. In a new article in Addiction, “What a difference a day makes:  Differences in initial abstinence response during a smoking cessation attempt,” the authors present a latent class analysis (LCA) that identifies four types of smokers based on their withdrawal symptoms on the day they quit. They found that a subset of quitting smokers reported extreme craving or extreme negative affect, and that this predicted earlier relapse.

The authors analyzed data from 1236 participants in a smoking cessation trial who provided ecological momentary assessments (EMA) of their withdrawal symptoms on the day that they quit smoking.  The LCA model incorporated participant reports of hunger, poor concentration, negative mood, cigarette craving, and anhedonia (the inability to feel pleasure). The authors identified four classes of smokers: High-Craving Anhedonia, Moderate Withdrawal, Affective Withdrawal, and Hunger. The Moderate Withdrawal class comprised 64% of the sample and was characterized by lower symptom levels. The High-Craving Anhedonia class comprised 8% of the sample and was characterized by high levels of craving and anhedonia. The Affective Withdrawal class comprised 13% of the sample and was characterized by high levels of poor concentration and negative mood. The Hunger class comprised 15% of the sample and was characterized by high levels of hunger but low levels of the other measures.

Among other results, the authors found that members of the High-Craving Anhedonia and Affective Withdrawal classes returned to regular smoking sooner than members of the Moderate Withdrawal class.

Methodology Center investigators Sara Vasilenko and Stephanie Lanza conducted this research with Megan Piper and Jessica Cook of The University of Wisconsin’s Center for Tobacco Research and Intervention. This LCA in this article is not especially complex and could be understood by people who are relatively new to the method.  The article does not, however, describe LCA extensively and is not intended as an introduction to the method. For an introduction to LCA, please see our recommended reading list.

Read the article. (Journal access required.)



Piper, M. E., Vasilenko, S. A., Cook, J. W., & Lanza, S. T. (2016). What a difference a day makes: Differences in initial abstinence response during a smoking cessation attempt. Addiction. Advance online publication. doi: 10.1111/add.13613

Piper, M.E. (2015). Withdrawal: expanding a key addiction construct. Nicotine and Tobacco Research. 17(12), 1405-15.

New LCA Software: Bootstrap and Distal Outcomes

May 31, 2016:

We are pleased to release two new functions that augment the functionality of the LCA Stata plugin. The LCA Bootstrap Stata function allows users to perform the bootstrap likelihood ratio test in order to choose the number of classes for their latent class model. The LCA Distal Stata function allows users to estimate the association between a latent class variable and a binary distal outcome using a model-based approach. Both functions require the LCA Stata plugin version 1.2.1 or higher and Stata version 11 or higher. For SAS users, the LCA Distal SAS macro and LCA Bootstrap SAS macro have both been updated with small functional tweaks or bug fixes.

Download the Bootstrap function

Download the Distal function