Home / Topics / Research Design / Statistical Analysis / Latent Growth Models

Latent Growth Models

Posted on November 23, 2011

C. Hendricks Brown (bio) talks about the use of latent growth models to examine change over time.


To promote the health equity of Hispanic populations in here, one of the strategies that we've had is to do very careful rigorous evaluations and designs. That is, the same kinds of rigorous evaluations that we require for... that FDA requires for new drugs, for example, but to do them in behavioral interventions and systems... in communities themselves -- schools and other areas like that.

So what we're trying to do is trying to develop a methodology for measuring and modeling and testing these rigorous interventions that we have in the field, so that we can help identify what characteristics of these interventions, how they could be assembled and put together in communities so that they have beneficial impact for our kids and families. So that's the idea of this one.

A major part of trying to do this one is just trying to deal with the methodology. So what we have learned over time is that, and in the prevention field, which is where I do most of my work, what we can do is we can actually change the trajectories of kids over time and help improve their outcomes. We do this by earlier interventions before the time at which they become at risk.

So, if you wait until somebody is already using drugs it's a lot harder to change that behavior than it is to prevent the drugs from being used early on. So, what that means is we have to look at interventions. We have to look at methods that allow us to examine repeated measures over time as we look at the developmental course.

So, one of the methods that we've been working with in here has to do with modeling these trajectories over time and we call those growth models and sometimes people use more technical words of latent growth models and the word "latent" really means that it is unobserved by itself, but it is measured and we take into account the measurement error and the modeling in here. So these are statistical methods that are advanced methods in here and they can be used to evaluate overall changes in the trajectory that occur in an intervention versus a control group or a comparison group.

What we've also found in the prevention field is that when you start talking about these trajectories that people go through, that they change... that not everybody follows the same kind of trajectory. So you can see that in your own life with kids. Some kids are very low risk all the time. Some kids have more elevated risk, as they start to go up, and some kids start off very early in life and they are already at a very high risk period in here.

So we look at multiple trajectories over time and we call these growth mixture models. The mixtures being that a particular population has a mixture of these different kinds of trajectories. So we've analyzed data to... we have methods to analyze data accounting for this kind of variation in these trajectories over time. What we find is that when we do many of these analyses and look at the prevention effects in here, we find a differential effect of the prevention program as a function of level of risk.

So, often what happens is that our interventions seem to work better, the prevention interventions seem to work better for those at higher risk, which is good news, because we can change those trajectories. They're not immutable. They can be changed over time.

The growth models themselves are put together at the individual level, so these are individual level measures across time, but then you can combine them together and say what does the overall group do and so you could look at an average slope for the trajectory of a group, for example, and compare it to another group as well.

Viewing Preferences




Excerpted from an interview with the researcher conducted at the 2011 NHSN Conference held in Miami, FL.


Please note that the feedback is viewed only by 4researchers staff and is not intended for communication with individual contributors.


Use the form below to submit feedback about this article. If you would like a response, please be sure to include your e-mail address.

More About "Statistical Analysis"


Introduction to Moderators and Mediators (2007)


More From C. Hendricks Brown (bio)


A Varied Career


Combining Treatment and Prevention

Show All...