The Methodology Center's New Podcast Series - Methodology Minutes!

The Methodology Center's New Podcast Series - Methodology Minutes!

with Linda Collins

April 5, 2010

Introduction to The Methodology Center - who are we, what is our mission.

Hosted byMichael ClevelandandLinda Collins

Speaker 1:Methodology Minutes is brought to you by The Methodology Center at Penn State, your source for cutting edge research methodology in the social, behavioral and health sciences.

Linda:Hello. I'm Linda Collins from The Methodology Center at Penn State.

Michael:Hello. I'm Michael Cleveland, also from The Methodology Center at Penn State and I'd like to welcome you to Methodology Minutes. First I want to give the listener just a little bit of an overview of what we're going to be talking about today. I should mention first that this podcast series is a companion to our newsletter which is published every Fall and Spring semester and a very popular component of our Methodology Center outreach.

First we'll be talking a little bit about the background of The Methodology Center, its mission and a little bit of the history of the center. We'll also talk about what types of research we do. We'll talk a little bit about the training and the mentoring that center scientists do.

Finishing up with some discussion of our dissemination activities at The Methodology Center. One of those main activities and the reason that we're here today is that we have decided to introduce the podcast series as a kind of a new and fun way to introduce ourselves to a listening audience. With that I will turn it over to Linda again.

Linda:Okay, thank you, Michael. A little bit of background about The Methodology Center. We are at Penn State University in State College Pennsylvania and also the University of Michigan. Here at Penn State we're in the College of Health and Human Development, although we want to stress that we're an interdisciplinary center. We also have involvement from scientists in other colleges. Our over-arching mission is the advancement and dissemination of methodology.

Michael:That brings up a good question, I think Linda. What do you mean by methodology?

Linda:Yeah, that is a great question. By methodology, we mean every quantitative aspect of research methods. This includes data analysis and research design which are usually included under the general heading of statistics, but also some areas that are a little bit outside of traditional statistics, such as measurement, latent variable models and cost-effectiveness analysis.

Michael:How does that make us different from a staff department would you say?

Linda:We're really pretty different from a staff department. First of all, we're interdisciplinary. Statistics, the discipline of Statistics has an important presence in the Center. A very important presence. There's other disciplines present too.

If you take a look at our website and our publication list, you'll see that there's a lot of cross disciplinary collaboration. The interdisciplinary group of scientists that makes up the center meets as a group roughly 30 times a year to discuss research. Even those of us who aren't directly collaborating with each other, influence each other's work all the time.

I think another point to make, another reason why we're different from a traditional Statistics Department is we focus on methodology for applications in the social, behavioral and health sciences. In fact there are people here doing theoretical work in statistics, but even that more theoretical work is inspired by these applications.

In most Statistics Departments there are other application areas besides the social, behavioral and health sciences. Areas such as Environmental Statistics, Spatial Statistics, Astro Statistics, Engineering Statistics. We frequently borrow from these other areas within statistics, but we don't really work on applications for those domains. We're really focused on applications in the social, behavioral and health sciences.

I think another difference is in where we publish. Although we do publish in statistics journals such as the Journal of the American Statistical Association and the Annals of Statistics, we also disseminate our work in other ways by publishing in methodological journals aimed at the social, behavioral and health sciences.

For example Psychological Methods, Statistics In Medicine, those are examples of more methodologically focused as a opposed to purely statistically focused journals that are outlets for us.

We also publish in more content-oriented journals. Journals like Prevention Science, Drug and Alcohol Dependents, Neuro-Psycho Pharmacology. Those are examples of journals where we've published methodological pieces that are aimed primarily at a substantive audience.

Michael:When you mention all of these articles that have been published in these various journals, I was kind of curious what you mean by methodological research as opposed to methodological supports.

Linda:Yeah, when a lot of people, when they hear the name Methodology Center automatically assume that we offer methodological support. For example, walk up consulting hours or the possibility of writing us into a grant to handle data collection and data analysis for a big project.

That kind of support is really important but there are other units at Penn State who provide it and provide it very, very well. That's not what we do. We do original research and methodology, developing new methods, improving existing methods, importing and adapting methods from other fields that can be useful in the behavioral sciences.

That's what our funding is for. Not for providing statistical and methodological support. Now, although we do not provide ad hock consulting, we love to engage in collaborations with behavioral sciences on projects that make both a substantive and a methodological contribution ideally.

Michael:I think I'd like to move next and talk a little bit about the history of The Methodology Center. Real briefly mention some important highlights that have taken place in the past several years, starting first with the idea that the first P50 Center Grant was funded by the National Institutes on, National Institute on Drug Abuse in 1996.

Although the personnel have changed over the years, Linda Collins, you have always been the center director. Another important fact in regard to the history is that Susan Murphy at the University of Michigan was one of the first or was one of the original principle investigators for that first center grant.

Linda:Right. Susan was at Penn State at that time. She's at the University of Michigan now and she remains one of the principle investigators on our center grant. Still very involved in the center and in fact the activity at the University of Michigan in regard to this center has grown over the years.

Michael:Since that time, since 1996, the center has had two successful renewals of that grant and right now, currently undergoing or has in the past I guess semester, undergone a third competitive renewal which as we all are aware, was a very intensive process.

Linda:Yes. Yes, center grants are a really big, intensive process. That's a good way to put it. That was submitted and we're awaiting review now.

I want to mention our program officials over the years. Oh and I should say The National Institute on Drug Abuse, that's a bit of a mouthful, so most people call it NIDA. I think we can refer to it as NIDA from now on. We're funded by the Prevention Research branch of the Division of Epidemiology Services and Prevention Research.

We have been so lucky over the years with the program officials we've been able to work with. Wilson Compton currently is the Division Director. Liz Jenecksi is our current program official. We had previously Larry Sights as a program official. He's now retired. Hewas really great and Liz Robertson was our program official for a while. She is now the branch Chief and handed us over to the able hands of Liz Jenecksi.

Most of the support for The Methodology Center comes from the P50 grant from NIDA that Michael referred to a minute ago. There are other grants in the center. Other grants from NIDA, some grants from the National Science Foundation and there's a couple of grants from the NIH Roadmap Initiative. These grants are all to support our original research and methodology.

We also get some funding for kind of our basic infrastructure from the College of Health and Human Development. The Dean of our college is currently Nan Crowder. The Associate Dean for Research is Neil Sharkey and they've been wonderfully supportive of the center. It's really just a great environment to be in.

I also would like to mention our partnership with the Prevention Research Center which is directed by Mark Greenberg. The two centers do a lot of things together. We have a training grant together, which we'll talk about in a little bit and there's a lot of projects that involve personnel from both centers. It's just been a very, very productive and enjoyable collaborative process over the years.

Michael:I think that's a nice segue to start talking a little bit about who actually works here in the center. Currently The Methodology Center is home to more than 20 PhD investigators and 12 graduate students. All these scientists come from a variety of academic backgrounds. Those include Statistics, Human Development and Family Studies, Communication Arts and Sciences, Health Policy and Administration and Psychology.

Linda:Now it would be great I think to give some examples of the research that people do and Michael, why don't you start with yours? That would be a great place to start.

Michael:Very good. My research, I'm interested in identifying the mechanisms through which social contextual conditions influence adolescents' substance use and abuse. For example, I am very interested in how neighborhood processes such as social cohesion and social disorganization interact with more proximal factors like interpersonal relationships between family members and other close ties to produce differential trajectories of risk for adolescents.

Methodologically, these types of questions obviously require multi-level approaches that can account for that clustering or nested structure of the data. Recently I have become interested in integrating my interests in multi-level modeling and epidemiologic methods with my interest in subgroup analytic approaches such as latent class analysis.

Of course my ultimate goal is to understand why certain communities are characterized by different levels of risk and protection and then use this information to help guide the development of more efficient and integrated interventions that ultimately will be the most efficient ways to intervene and prevent substance use.

Linda:That is so important. I think that multi-level approach is getting a lot of attention now and it's about time. I think you're in the right place at the right time.

Michael:Thanks.

Linda:I guess I'll jump in now and talk about what I've been doing. I'm involved in a number of different funded research projects. I have a component in our current P50 grant where I'm working on developing a phased experimental approach for optimizing behavioral interventions.

You'll hear a little bit more about this in a future podcast, but just for right now I'll say that we usually evaluate interventions as a package and we rarely stop to think okay, maybe this intervention has been shown to have a statistically significant effect. Maybe it hasn't, but is the intervention's effect as large as it can be?

In other words, have we taken steps to optimize the intervention? Another project that I'm involved in, I'm very excited about is the optimization of a smoking cessation intervention. I'm one of the investigators on a center grant that's funded by the National Cancer Institute, NCI.

The principle investigators are Tim Baker and Mike Fiore of the Center for Tobacco Research and intervention at the University of Wisconsin in Madison. The study is taking place there.

The idea is to use a phased experimental approach to optimize their clinic-based intervention for smoking cessation.

Michael:Another center scientists that we want to just briefly introduce among the many other scientists here is Donna Coffman. Donna is another research associate in the Methodology Center who like me, was a PAMT post-doctoral trainee, which we will talk a little bit about the PAMT Program in just a little bit.

Donna recently received funding from NIDA for an RO3 Project to investigate methods for accessing causal inference in causal mediational processes in prevention programs, particularly when individuals are not randomly assigned to treatment or they're only pseudo randomly assigned.

This is the case in many prevention programs which are usually designed to target mediating variables such as the perceptions of alcohol use or perceptions of alcohol use norms, rather than the ultimate outcome such as substance use itself.

Donna is one of the project directors for the most recent P50 application that you just mentioned and she will continue this line of research by taking recent developments from the theoretical statistics literature concerning these causal inferences and adapting them for use in substance use prevention and treatment research.

Linda:That is so important because a lot of the literature on causal inferences is very technical.

Michael:Right.

Linda:It just isn't getting out there to people who can really use it.

Michael:Yeah and I think that's one of the main points of Donna's research is to ultimately develop these methods and bring them into the hands of substance use and intervention scientists.

Linda:Yes, very important. The next scientist that we want to talk about is Dr. Runze Li who's a professor of Statistics and one of the Project Directors on our P50. Runze, I want to apologize in advance. I know I'm not going to do justice to your work in this little summary, but I did want people to get a little sense of what you're doing.

Runze works in a number of different areas and here I'm going to talk briefly about his work in functional data analysis for modeling intensive longitudinal data.

Intensive longitudinal data is longitudinal data with many observations. For example, nowadays it's getting very easy to collect such data with wearable devices that might, for example, give an almost constant feed of an individual's activity level.

There are also diary methods of data collection using handheld computers. People, especially with college students it seems, are collecting data more and more using the internet. Also there are more and more what we've started calling mature longitudinal studies. These are sort of standard longitudinal studies that have been out in the field long enough, but they might have 12, 15, 20 observations in time.

These kinds of data collection approaches often require new ways of analyzing data. In fact it can be very challenging to analyze this kind of data. Runze has been a pioneer in the use of functional data analysis for modeling intensive longitudinal data. By using functional data analysis, you can model data in which change is not a simple functional form.

Maybe it's not just, well, not a straight line, but also maybe not even quadratic or cubic. Something that has a lot of ups and downs in it.

In Runze's component of our current P50, he's working on models that include not only time varying co-variants, but time varying effects, so that the relation between two variables itself can change over time.

For example, he's shown that the relation between mood and urge to smoke varies as a function of number of days post-quit in smokers. Runze also has a R21 funded by the NIH Roadmap and there are two principle investigators on that. One is Runze, the other is Dr. Lisa Dierker who is at Wesleyan University.

They're working on using functional data analysis combined with mixture model approaches to identify subgroups of individuals showing different complex patterns of change over time.

Now there's other people doing lots of interesting stuff in The Methodology Center and we hate to leave them out, but we could go on and on and this podcast would potentially get very long. In future podcasts we hope to be talking with some of these folks. Stephanie Lanza, Susan Murphy and others who are doing really interesting work in the center.

Michael:I want to next talk a little bit about the training that is done here at the center. One of our things we've mentioned a few times is that one of the primary missions of The Methodology Center is to grow the next generation of prevention and treatment methodologists. To date The Methodology Center has trained more than 30 pre-doc and post-docs who have worked closely with center scientists on a variety, a wide range of methodological research questions.

Many of these trainees have been hired to work directly with the center scientists on specific research projects, however a significant number of these trainees have come to work specifically or be involved with the NIDA funded PAMT Program. PAMT is an acronym that we use that stands for Prevention and Methodology Training Program.