Multilevel Modeling. Douglas A. Luke

approach and a structural equation modeling (SEM) approach. This book will focus on the mixed-effects regression-based multilevel modeling. Generally, I will use “multilevel” or “longitudinal” models when I am talking about the purposes of a model; I will tend to used “mixed-effects” models when I am talking about the broad class of statistical models that can be used for multilevel or longitudinal analyses. For a good introduction to the SEM-based approach, see Chapter 6 in Heck and Thomas (2009).

      Multilevel and longitudinal models are starting to appear more frequently in every area of the social and health sciences, as the techniques have become more widely known and integrated into the major statistical packages. There are as many specific types of multilevel models as there are scientific questions. However, there are certain types of overarching models that can be seen across the different research disciplines, as indicated in Table 1.2. This table is not meant to be exhaustive but to provide a catalyst for the reader and indicate the extremely wide applicability of multilevel methods. In particular, it is not immediately obvious to the new analyst that mixed-effects models can be extremely useful for modeling longitudinal data (where multiple observations are nested within an individual) and also for meta-analytic studies (where multiple effect statistics are nested within individual studies).

Table 1.2

      Online Book Resources

      Finally, a variety of resources are available online to support the material covered in this book. These resources include all the data sets used for the examples, R and Stata code to reproduce most of the models, results, and figures, and an occasionally updated mixed-effects modeling resource list. These materials can be found at https://www.douglasluke.com.

      Descriptions of Images and Figures

      Back to Figure

      To the left is an upward pointing arrow indicating level of analysis, which goes from micro, psychological, and biological at the bottom to mezzo, macro, and global at the top. Micro includes Professional or personal social networks, e.g. organizational staff or clients. Psychological has Program-relevant knowledge, attitudes and beliefs and biological has Program-relevant physiological, biological and genetic characteristics. Mezzo has Setting of program implementation, e.g. organization, hospital, neighborhood. Macro has National, state, and large area dynamics. Global has Political/economic climate and dynamics. At the center, the D&I timeline is indicated by a right-pointing arrow. The D&I timeline represents the program, policy or practice being implemented and has translation, dissemination, implementation, and sustainability along its length.

      I should venture to assert that the most pervasive fallacy of philosophic thinking goes back to neglect of context.

      —John Dewey, 1931