Multivariate Applications in Substance Use Research: New Methods for New Questions

Multivariate Applications in Substance Use Research: New Methods for New Questions

Multivariate Applications in Substance Use Research: New Methods for New Questions

Multivariate Applications in Substance Use Research: New Methods for New Questions

Synopsis

This edited volume introduces the latest advances in quantitative methods and illustrates ways to apply these methods to important questions in substance use research. The goal is to provide a forum for dialogue between methodologists developing innovative multivariate statistical methods and substance use researchers who have produced rich data sets. Reflecting current research trends, the book examines the use of longitudinal techniques to measure processes of change over time. Researchers faced with the task of studying the causes, course, treatment, and prevention of substance use and abuse will find this volume helpful for applying these techniques to make optimal use of their data. This innovative volume: *introduces the use of latent curve methods for describing individual trajectories of adolescent substance use over time; *explores methods for analyzing longitudinal data for individuals nested within groups, such as families, classrooms, and treatment groups; *demonstrates how different patterns of missing data influence the interpretation of results; *reports on some recent advances in longitudinal growth modeling; *illustrates methods to assess mediation when there are multiple mediating pathways underlying an intervention effect; *describes methods to identify moderating relations in structural equation models; *demonstrates the use of structural equation models to evaluate a preventive intervention; *applies epidemic modeling techniques to understand the spread of substance use in society; *illustrates the use of latent transition analysis to model substance use as a series of stages; and *applies logistic regression to prospectively predict smoking cessation.

Excerpt

This volume was inspired by a number of motivations. One was to share part of the adventure that we have had on the Indiana Smoking Project over the past 20 years. During this time, we have studied how it is that adolescents come to smoke, and how smoking relates to various aspects of people's lives as they develop through adolescence and enter and establish adult roles. in so doing, we have also learned a great deal about the factors involved in smoking initiation and transitions. in addition, we have learned a lot about large-scale research projects, and have come to appreciate the importance of dialog between researchers in the area of substance use and abuse and researchers who are developing new quantitative techniques. in studying the etiology of tobacco use, the questions that we have asked have turned out to be very complex, and, over the 20-year period of our collaboration, these questions often required us to seek out newly developing quantitative methods. For example, during this period we went from asking whether a particular variable prospectively predicted a change in smoking behavior one year later, to asking about predictors of trajectories of smoking behavior over six waves of data spanning adolescence to adulthood. We went from asking whether parental smoking influenced adolescent smoking to testing complex models of the multiple mechanisms where parental smoking might exert its influence. We went from examining psychosocial influences in general to examining the shared versus nonshared nature of these influences as they affected sibling similarity and dissimilarity within families. As these examples illustrate, the changes in the sophistication and complexity of our research agenda necessitated tremendous changes in the way that we approached and analyzed our data.

As with many researchers in the area of substance use, most of us were trained in research traditions that were simpler and, quantitatively speaking decidedly old-fashioned. Without dating the time of our training, let us simply say the ideal study might well involve a 2 × 2 design, with 8 or 16 sub-

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.