In This Issue:
- Director's Report
- Recent Awards
- New SAS Software for Estimating Latent Class Analysis with a Distal Outcome
- Recent News
- New Article Demonstrates Application of Latent Class Analysis to HIV Research
- MC Friends' Corner
- Ask a Methodologist
Today we say good-bye to our paper newsletter. I will admit I am a little sad to be closing this chapter of our outreach efforts, but I am very excited about where we are heading next!
The feedback we have received is that most of our readers prefer information in electronic form. Plus, we have a serious “green” effort underway at The Methodology Center and we were beginning to feel a bit guilty about sending out a paper newsletter! So, we have switched to an eNews format, and we really like it.
Our eNews is already going strong. Our plan is to issue eNews nine or ten times per year, as warranted. The eNews includes many features you will recognize from our paper format. It has the advantage of being much more immediate than our semiannual newsletter. We particularly like the flexibility this approach gives us and the way it enables us to share timely information about software releases, journal articles, and Center events.
We would love your feedback about this new approach to keeping you up-to-date about what is happening at the Center. Please contact us if you have any questions of comments, and if for any reason you prefer a paper version of our eNews, let us know, and we will make sure you receive our eNews in that format. I hope you enjoy this last issue of The Methodology Center Perspective, where you can read about a new SAS macro for conducting latent class analysis (LCA) with a distal outcome, learn about a study that demonstrates an application of LCA to HIV research, and find out more about how our recent work on adaptive treatment strategies is moving into the field.
Linda M. Collins, Ph.D.
Director, The Methodology Center
Penn State University
To make sure you receive the eNews
- If you have already received one or more issues of the eNews, you will continue to receive it.
- E-mail firstname.lastname@example.org with the subject “enews address” and we will make sure that your email address is current in our files.
If you do not wish to receive the eNews
- E-mail email@example.com with the subject “enews unsubscribe,”
- or, when you receive your first eNews in December, click UNSUBSCRIBE at the bottom.
We would appreciate it if you would take these measures rather than reporting our eNews as spam. Of course, we will NEVER share our list of subscribers with anyone.
Methodology Center Director Dr. Linda Collins is the recipient of the 2011 Evan G. and Helen G. Pattishall Outstanding Research Achievement Award from Penn State’s College of Health and Human Development. The award exists to “recognize and reward outstanding researchers for advancing the frontiers of knowledge.” Dr. Collins is being acknowledged for her development of new research methods and contributions to the prevention of public health problems. Specifically noted were the Multiphase Optimization Strategy (MOST) and latent transition analysis.
Methodology Center Principal Investigator Dr. Runze Li was named a fellow of the American Statistical Association (ASA) at the group’s annual meeting on August 2. The ASA supports excellence in the development, application, and dissemination of statistical science through meetings, publications, membership services, education, accreditation, and advocacy. Dr. Li’s publications on statistical methods have been cited by researchers thousands of times.
Also, the United Nations’ World Meteorological Organization has awarded the 2012 Norbert Gerbier-MUMM International Award to a paper coauthored by Dr. Li. In the paper, “Climate control of terrestrial carbon exchange across biomes and continents,” the authors examine relationships between the climate and the carbon exchange of land-based ecosystems to predict future levels of carbon dioxide in the atmosphere. The award is presented annually to “an original scientific paper on the influence of meteorology in a particular field of the physical, natural or human sciences, or on the influence of one of these sciences on meteorology.” Dr. Li was one of 151 authors from 116 academic institutions; he was the only statistician on the project.
The Methodology Center is pleased to offer the SAS macro, %LCA_Distal, to estimate the effect of latent class membership on a distal outcome. This macro, which is available as a free download, uses a model-based approach that can accommodate binary, count, continuous, and categorical distal outcomes.
Applications of the latent class analysis (LCA) abound in the substance use literature, including models of drug use patterns, alcohol use motivations, stages of change in smoking cessation, and marijuana use and attitudes. In LCA, a categorical latent (i.e., unobserved) variable (C) is measured by multiple observed variables (X1, X2, …, Xj). The figure to the right depicts this measurement model. Very often, in order for researchers to understand the public health implications of latent class membership, links must be drawn to a later outcome (Z) such as nicotine dependence or binge drinking.
Because true latent class membership is unknown, researchers have been forced to either ignore this uncertainty or use pseudo-class draws, an approach similar to multiple imputation for missing data. However, simulation studies have shown that these approaches produce biased results. This new SAS macro uses a model-based approach, essentially eliminating bias in estimates of the effect of latent class membership on a distal outcome. Details about this approach can be found in a users’ guide (Lanza, Tan, & Wagner, 2011) and technical report by Dr. Lanza and her colleagues (Lanza, Tan, & Bray, 2011); both are available for download from the Methodology Center website.
The %LCA_distal macro works with PROC LCA, also developed by The Methodology Center and available as a free download. The macro relies on simple SAS syntax, and researchers who do not have experience using SAS macros can view a four-minute tutorial video available on the Methodology Center homepage.
Lanza, S.T, Tan, X., & Bray, B.C. (2011). A technical introduction: A model-based approach to latent class analysis with distal outcomes. (Technical Report No. 11-116). University Park, PA: The Methodology Center, Penn State. Retrieved from http://methodology.psu.edu
Tan, X., Lanza, S. T., & Wagner, A. T. (2011). Lca distal SAS macro users’ guide (Version 1.1.0). University Park: The Methodology Center, Penn State. Retrieved from http://methodology.psu.edu
The Methodology Center welcomes new research associate, Dr. Michael Yang, and new Prevention and Methodology Training (PAMT) postdoctoral fellows Drs. Keri Jowers and Sara Vasilenko.
The Summer Institute on Innovative Methods was held this September in Bethesda, Maryland. Dr. Linda Collins presented the two-day workshop “The Multiphase Optimization Strategy (MOST).”
Three new “Methodology Minutes” podcasts available for download at methodology.psu.edu/podcast. Episode 8 is an interview with Dr. Mark Stemmler from Bielefeld University, Germany, on configural frequency analysis; Episode 9 is an interview with Drs. Linda Caldwell, Ed Smith, and Linda Collins about HealthWise South Africa, a risk-reduction, life-skills curriculum for adolescents in South Africa; and Episode 10 features the instructors and participants from the 2011 Summer Institute with new host Aaron Wagner.
Thirteen percent of adults in Namibia are infected with HIV; this is among the highest infection rates in the world (UNAIDS, 2010). To combat the transmission of HIV in the region, some prevention scientists distribute health information through the social networks that exist within local populations. Methodology Center researchers Drs. Rachel Smith and Stephanie Lanza examined data from 400 individuals in Namibia to determine whether latent class analysis (LCA) could be useful in designing social network interventions. Their article, “Testing Theoretical Network Classes and HIV-related Correlates with Latent Class Analysis,” recently appeared in the journal AIDS Care.
The article focuses on the type of social network intervention known as an opinion-leader intervention. In opinion-leader interventions, scientists deliver information to influential members of an at-risk community with the expectation that these community members will distribute that information throughout their social network.
One possible way to improve the effectiveness of these interventions is to empirically validate which individuals are labeled “opinion leaders.” Toward that end, Drs. Smith and Lanza used LCA to discern whether the groups typically described in social network theory actually exist in the population. In the LCA, information about group memberships (including churches, sports clubs, and professional organizations) and health behaviors (including sexual activity, HIV testing, and alcohol use) was analyzed to identify subgroups of individuals with shared characteristics. By identifying these groups within the population, prevention scientists can better identify potential opinion leaders and see if targeted opinion leaders display behaviors that would enable them to distribute information effectively.
LCA revealed the existence of four classes within the population: Connectors, Selective Connectors, Single-Group Members, and Single-Group Loyalists. These classes did not correspond with the groups described in social network theories. In fact, the characteristics of the identified classes revealed potential complications with opinion-leader interventions.
Members of two classes, Connectors and Selective Connectors, have connections to other people that would make them likely candidates to serve as opinion leaders in an intervention. Compared to all other classes, however, Connectors were less likely to be tested for HIV or to perceive themselves as at risk for contracting HIV. Since members of this class are less likely to believe they are at risk for HIV infection, they may be ineffective at disseminating information about HIV prevention. Furthermore, Selective Connectors are likely to be targeted as opinion leaders, but membership in this class is characterized by rejection of other groups. If they do not share information with all members of the community, Selective Connectors who serve as opinion leaders could also impede dissemination efforts.
Effective opinion-leader interventions can improve the health of at-risk populations. LCA can empirically validate the identification of roles in social networks and can reveal obstacles to an intervention’s success.
Note: The interviews used in this research were part of a larger study conducted by the Johns Hopkins University Center for Communications Programs. This research was supported by the Center for Communications Programs and the National Institute on Drug Abuse.
Smith, R.A., & Lanza, S.T. (2011) Testing theoretical network classes and HIV-related correlates with latent class analysis. AIDS Care, 23(10), 1274-1281.
UNAIDS. (2010). Namibia. Retrieved from http://www.unaids.org/en/regionscountries/countries/namibia/#1
Dr. Hendrée Jones studies treatments for substance use disorders in pregnant women and in utero exposure to abused substances. Her work focuses on developing and scaling up drug-addiction interventions for women. She is Co-Principal Investigator on a Sequential, Multiple-Assignment, Randomized Trials (SMART) conducted at the Center for Addiction and Pregnancy at Johns Hopkins Bayview Medical Center, Baltimore, MD.
SMART designs allow researchers to construct adaptive treatments (interventions that alter in response to patients’ needs). Methodology Center researcher Dr. Susan Murphy led the development of methodology for SMART and is actively working to advance and disseminate SMART.
Dr. Jones and her collaborators’ SMART design tests how to create the most effective and efficient drugtreatment program for pregnant women who abuse substances. Dr. Jones has already tested a program that uses Reinforcement Based Treatment (RBT), which provides regular counseling along with social and financial incentives to pregnant women who stay off drugs. This treatment works in experimental settings, but it only works for some participants and it is expensive and time consuming to run. In order to achieve the best results at the lowest cost, the current SMART seeks to determine rules for when and whether participants (a) should continue to receive the current level of RBT to maintain success, (b) can maintain success on a reduced level of RBT, or (c) require enhanced RBT.
A SMART design is highly efficient and allows the researchers to answer questions about when, how and for whom interventions should adapt over time. Dr. Jones recommends SMART design to “any researcher who needs to know which treatment works best for which patients.” We wish Dr. Jones and her colleagues the best of luck on this exciting and vital work.
Dr. Hendrée Jones is a Senior Research Psychologist at RTI International and an Adjunct Professor in the Department of Psychiatry and Behavioral Sciences and the Department of Obstetrics and Gynecology, Johns Hopkins University. She leads two ongoing NIDA-funded international projects to develop and test women-specific interventions aimed at reducing HIV risk behaviors and increasing drug abstinence (one project is in the Republic of Georgia and the other —just funded this month—is in South Africa).
I designed a study to assess 50 college students’ motivations to use alcohol and its correlates during their first semester. The most innovative part of this study was that I collected data with smart phones that beeped at several random times on every Thursday, Friday, and Saturday throughout the semester. Now that I’ve collected the data, I’m overwhelmed by how rich the data are and don’t know where to start! My first thought is to collapse the data to weekly summary scores and model those using growth curve analysis. Is there anything more I can do with the data? — Signed, Swimming in Data
You did indeed collect an amazing dataset! With technological advances, the collection of intensive longitudinal data, such as ecological momentary assessments (EMA), is becoming popular among researchers hoping to better understand dynamic processes related to mood, cigarette or alcohol use, physical activity, and many other states or behaviors. Some of the most compelling research questions in these studies often have to do with effects of time-varying predictors.
One familiar way to approach the analysis of EMA data is to reduce the data, summarizing within-day or within-week assessments to a single measure, so that growth curve models may be fit to estimate an average trend and predictors of the intercept and slope. However, this approach disregards the richness of the data that were so carefully collected. Further, EMA studies are typically designed in order to capture something more dynamic than what could be captured as a linear function of time.
A more common approach to the analysis of EMA data is multilevel models, where within- and between-individual variability can be separated. This approach is helpful for understanding, for example, the degree of stability of processes. However, these methods typically impose important constraints, such as the assumption that the effects of covariates on an outcome are stable over time.
New methods for the analysis of intensive longitudinal data have been proposed in the statistical literature, and hold immense promise for addressing important questions about dynamic processes such as the factors driving alcohol use during the freshman year of college. For example, the time-varying effect model (TVEM) is a flexible approach that allows the effects of covariates to vary with time. A detailed introduction to time-varying effect models for audiences in psychological science will appear in Tan, Shiyko, Li, Li, & Dierker (in press).
A demonstration of this approach will appear in an article by Methodology Center researchers and colleagues (Shiyko et al., in press). The authors analyzed data collected as part of a smoking-cessation trial and found that individuals with a successful quit attempt had a rapid decrease in craving within the first few days of quitting, whereas those who eventually relapsed did not experience this decrease. Eventual relapsers had low levels of confidence in their abilities to abstain on cravings early in their quit attempt, but among successful quitters the association with confidence in ability to abstain was significantly weaker.
Any researcher with access to EMA data can fit a TVEM using the %TVEM SAS macro, which is freely available at /downloads/tvem. Give it a try so that you can explore the timevarying effects of individual factors, such as residing in a dorm, and contextual factors, such as excitement about an upcoming sporting event, on motivations to use alcohol.
Shiyko, M.P., Lanza, S.T., Tan, X., Li, R., & Shiffman, S. (in press). Using the time-varying effect model (TVEM) to examine dynamic associations between negative affect and self-confidence on smoking urges: Differences between successful quitters and relapsers. Prevention Science.
Tan, X., Shiyko, M.P., Li, R., Li, Y., & Dierker, L. (in press). A time-varying effect model for intensive longitudinal data. Psychological Methods.