Reviews

Methodological Approaches to Community-Based Research

Title: Methodological Approaches to Community-Based Research
Author(s) : Edited by Leonard A. Jason and David S. Glenwick

Reviewed by R. Burke Johnson, BJOHNSON@usouthal.edu

Review of Methodological Approaches to Community-Based Research,
Edited by Leonard A. Jason and David S. Glenwick.
Reviewed by R. Burke Johnson, BJOHNSON@usouthal.edu
University of South Alabama
College of Education
Mobile, AL 36688-0002

I thoroughly enjoyed reading Methodological Approaches to Community-Based Research (2012) edited by Jason and Glenwick. The book provides an overview of multiple methods that are not provided in the first two graduate statistics courses but nevertheless are very important for the practice of community psychology. The book is appropriate for a third course in statistics or as a supplemental book for a methods course. It is clearly and simply written and should be popular with students, practitioners, and researchers. The book retails for $37.00, so it is inexpensive. It also is relatively short (260 pages) and can be read quickly. This book has an excellent companion website (link: http://pubs.apa.org/books/supp/jason-glenwick/) that includes additional materials provided by each author such as figures, files explaining how to conduct analyses in the chapters, and data sets.

Methodological Approaches to Community-Based Research is divided into four parts or sections and includes 13 chapters, a Forward by Raymond P. Lorion, and an Afterword by James G. Kelly. Chapter 1 is an overview written by the editors.

The first major section or part of the book, entitled Pluralism and Mixed Methods in Community Research, includes Chapters 2-4. Chapter 2, written by Tebes, provides the author’s view of the philosophical foundations needed for engagement with mixed methods research. This and Chapter 4 are the only chapters with the specific words “mixed methods research” (MMR hereafter) used in the title, but MMR is addressed in several other places in the book. Tebes provides a nice argument that classical pragmatism and Ronald Giere’s scientific perspectivism offer useful philosophies and justification for MMR. Pragmatism addresses person-environment and person-person transactions in which our experiences occur and explains that we are engaged in continual adaptations for the purpose of improving our experiences and environments. Pragmatism and perspectivism provide intersubjectivity as a way to avoid common dualisms such as subjectivism vs. objectivism. Although scientific perspectivism accepts naturalism and realism, it also points out that knowledge is affected by our perceptions, constructions, instrumentation, and various personal and disciplinary perspectives. The point is that multiple perspectives are useful and enable greater understanding about what we study (our objects of study) and how we study it (our methods).

Chapter 3, written by Barker and Pistrang, provides a nice discussion of methodological pluralism and clearly shows its importance for community psychology. Pluralism is the existence of variety and methodological pluralism is the inclusion of a variety of methods and approaches. All of the authors in the book appear to have high regard for methodological pluralism. It is correctly pointed out in Chapter 3 that methods should be driven by our research question(s). Chapter 3 is especially helpful for justifying methodological pluralism and guiding the practice of high quality mixed and multimethod research. The quality criteria discussed by the authors are listed here by area of relevance: Criteria Applicable to All Research (including explication of context and purpose, use of appropriate methods, transparency of procedures, ethical treatment of participants, and importance of findings); Research-Relevant Community Psychology Values and Principles (including sensitivity to people’s contexts, respect for diversity among people and settings, giving voice to traditionally underrepresented populations, addressing competencies, promoting empowerment, promoting social justice, research using multiple methodologies); Criteria Specifically Applicable to Quantitative Research (including statistical conclusion validity, internal validity, construct validity, and external validity); and Criteria Specifically Applicable to Qualitative Research (including disclosure of perspective, grounding interpretations in the data, coherence of interpretive framework, and credibility checks). I fully agree with the authors that these criteria are of critical importance for the effective practice of methodological pluralism.

Chapter 4, written by Campbell et al., provides an overview of mixed methods research along with an extended example from the authors’ research program examining the Sexual Assault Nurse Examiner (SANE) program. The authors explain that MMR (the “third paradigm”) is very helpful for understanding complexity and context. They list many specific uses of MMR in community psychology, they discuss the issue of paradigms in MMR, and they provide a discussion of how to design and conduct a mixed methods research study. This is a useful and practical chapter. 

The second major section of the book, entitled Methods Involving Grouping of Data, includes Chapters 5-7. Chapter 5, written by Dymnicki and Henry, focuses on cluster methods as person-oriented methods. Use of these methods helps one to identify clusters of individuals in one’s data set that were not and usually could not be specified a priori. Cluster methods stand in contrast to variable-oriented methods; the former focuses on individuals and discovering their natural groupings; the latter focuses on interrelated variables (as in traditional factor analysis). Locating groups of people is especially important in community psychology because the data sets are typically heterogeneous in unknown ways that need to be determined. This and other exploratory methods are important in community psychology because researchers need to discover what occurs in the world, in addition to testing theory articulated prior to the collection of data. Specific uses of cluster methods listed by the authors are locating the contextual limits of interventions, making complex interactions interpretable, analyzing data with poor distributions, targeting interventions, detecting multivariate outliers, and interfacing between qualitative and quantitative methods. The authors also show how to conduct a cluster analysis.

Chapter 6, written by Bogat et al., provides an introduction and overview of the person-oriented approach and its place in community research.  The person-oriented approach (POA) examines interindividual differences for information and patterns and does not assume that aggregate group statistics based on the full sample will generalize in a simple way to a population. Our samples often are more complex that we think because they include subgroups, many of which we are not aware.  The authors list the six classic tenets of the POA and add three tenets specifically for community psychology. Here are the three new tenets: “(1) The structure and dynamics of individual behavior are, at least in part, specific to the environment in which the individual lives and works; (2) There is lawfulness and structure to (i) intra-systemic constancy and change and (ii) inter-systemic differences in constancy and change. These processes can be described as patterns of the involved factors; (3) Validity is specific to individuals and environments.” (p.92) Techniques used to identify and study these kinds of patterns include cluster analysis, latent class analysis, exploratory factor analysis, configural frequency analysis, data mining methods, comparative method, symbolic data analysis, and several traditional statistical methods used in a “person-oriented way” (e.g., latent growth curve modeling, and polynomial decomposition of repeated measures ANOVA results). A useful example is provided where three latent classes of individuals (adolescents) were located using four variables: college attendance, youth motivation, contextual support, and positive activity involvement. The resulting latent classes included bright underachievers, superachievers, and slackers. Unpacking aggregate data this way provides a more complex and nuanced understanding of one’s data and can help explain ambiguous bivariate relationships found in the literature based on traditional statistical methods that failed to examine latent subgroups.

The last data-grouping technique, meta-analysis, is explained by Durlak and Pachan in Chapter 7. The authors show how to conduct a traditional meta-analysis and point out that cluster analysis can be used as a summarizing technique in one’s literature review of quantitative findings about effects of independent variables. Meta-analyses can also explain (through moderators) variability in the results found in the literature. Moderators might include context, type of participant, type of research, specific outcome, and type of program procedures and strategies. A simple strategy for identifying moderators is to divide the data into subsamples based on the moderator variable and check for differences.    

The third section of the book, entitled Methods Involving Change Over Time, includes Chapters 8-9. In Chapter 8, Hoeppner and Proeschold-Bell explain the use of classical time-series and interrupted time-series in community research. The former is based on long strings of data collected periodically and systematically across time; the latter is based on shorter time-series of baseline and during/after onset of treatment data. Interrupted times-series looks for slope and/or intercept change to determine program impact. The authors point out that when using time-series data, one needs to model and remove serial dependence or autocorrelation found in the data (e.g., where previous time points are correlated with later time points, especially points that are near one another). The authors provide a nice application example where HIV-hepatitis C and social worker training data are examined at multiple socioecological framework levels in the evaluation and understanding of a community intervention. Five major levels of ecological interest to community psychology are intrapersonal/individual, interpersonal, organizational, community, and social structure or public policy.

The last chapter in section three, Chapter 9, is written by Connell and demonstrates how to use survival analysis in community prevention and intervention programs. Survival analysis provides estimates of the probability of occurrence of events and their timing (e.g., recidivism, births, deaths, marriages, divorces, taking one’s first alcoholic drink, first use of illicit drugs), and it can be used to answer questions such as “What is the likelihood that a given event will occur among individuals or program participants? When is the event most likely to occur? [and] What factors, such as intervention condition or participant characteristics, alter the likelihood or timing of the event for individuals?” (p.147) Survival analysis helps community psychologists understand when outcomes occur in the community and can aids in the development of effective programs for different groups and predicting time-to-event duration of important events and program outcomes.

The fourth section of the book, Methods Involving Contextual Factors, includes Chapters 10-13. In Chapter 10, Todd, Allen, and Javdani examine the use of multilevel modeling (MLM) (also called hierarchical linear modeling or HLM). This statistical tool is important because the human world is naturally organized/structured into different levels, with lower level units nested in higher level units. Examples of individual level variables are individual SES, individual- achievement scores, satisfaction with community services, attitudes, and behaviors. Examples of aggregated and group level variables are average SES of a group (aggregated), classrooms, schools, neighborhoods, prisons, departments in an organization, and religious organizations. Examples of setting-level or contextual variables are type of school climate, type of organizational culture, and type of community. MLM can be used with nonexperimental, experimental/intervention, cross-sectional, longitudinal data, and, especially, nested structure data. For example, individuals are nested in churches, neighborhoods, and treatment programs, and these units are nested in yet larger units such as cities and states. Using MLM, community researchers can test several kinds of hypotheses, including within-group (e.g., check to see if individual self-efficacy is associated with their work productivity), between-groups (e.g., check to see if type of organizational climate is associated with average productivity of groups), contextual (e.g., check to see if a group level variable such as type of organizational climate is related to individual job satisfaction), and cross-level phenomena (e.g., check to see if an observed individual level association between relationship quality and personal distress varies according to a group level variable such as neighborhood quality). Interpretation of a variable will depend on its level and the type of mean centering used. In sum, MLM allows fruitful examination of many kinds of hypotheses that have been traditionally overlooked or analyzed using inappropriate statistical methods in the past.

In Chapter 11, Jason, Porter, and Rademaker provide some insights and methods found in epidemiology (defined as the study of the distribution and determinants of disease, problematic behaviors, and social conditions in populations). The authors focus on how to obtain unbiased estimates of prevalence of diagnostic conditions in populations. Diagnostic-condition or prevalence-outcome unreliability is found in five kinds of variance: subject variance (e.g., research participants have different conditions at different times); occasion variance (e.g., research participants are in different stages of a condition at different times); information variance (e.g., different professionals have different kinds of information about the research participants); observation variance (e.g., clinical psychologists differ in what they detect in research participants even though they are presented with the same symptoms and signs); and criterion variance (e.g., clinical psychologists use different criteria for their diagnostic conclusions). Of the five variances, criterion variance is the biggest culprit for producing diagnostic unreliability in community based research. In the remainder of this chapter the authors provide strategies borrowed from epidemiology to overcome unreliability using an example of measuring the prevalence of chronic fatigue syndrome. 

Chapter 12, written by Morton, et al., is an excellent overview of geographic information systems (GIS) methods for community psychology. The authors emphasize “thinking spatially.” Using GIS one can learn how outcomes measured at the individual level show patterned relationships with higher level and contextual factors such as physical environment of place, ethnographic history of area, aggregated unemployment rates by neighborhood or zip code, disease diffusion, presence of child abuse or crime, food availability in community, and presence of risk and protective factors. GIS helps community researchers fulfill the following SCRA principle: “Human competencies and problems are best understood by viewing people within their social, cultural, economic, geographic, and historical contexts” (quoted on p. 206). In GIS, data are represented visually (lines, areas, shades of colors, symbols) and are analyzed statistically. Some key questions are “What is it? Where is it? … and What is its relation to other entities?” (p. 210).  Mathematical and Boolean operations help in uncovering patterns. A nice example is provided showing a patterned association among race, average income level, neighborhood population density, and the presence of disease promoting pharmacies (i.e., where alcohol, tobacco, and lottery tickets are sold); this demonstrated that exposure to opportunities for abuse were greater in lower income, minority, high population areas. As more data become available, the authors and I expect that use of GIS will increase significantly.

In next chapter, Chapter 13, Sasso and Jason show the benefits and uses of economic cost analyses in community research.  This type of research, or dimension added to traditional research, is especially useful to inform policy decisions. A key idea in cost analysis is opportunity cost (the cost of the next best alternative), which is different from purchase or accounting costs (e.g., the face value cost of an item or program). Opportunity cost “explicitly measures the value of what society must forgo to devote a resource to a given purpose” (p. 222). The authors explain cost-minimization analysis, cost-benefit analysis, cost-effectiveness analysis, and cost-utility analysis, and they emphasize that these should be used comparatively (i.e., to compare policies, programs, interventions, using one or more of the four analytic methods). Because many costs must be projected into the future, they must be discounted and placed in “present value” units. Sensitivity analysis is conducted to determine the impact of different model assumptions (e.g., different discount rates, different mortality rates, different levels of program impact).

The last chapter in the book is an “afterword” commentary written by James G. Kelly. Kelly and others argued several decades ago for the explicit study of context. He also now explicitly advocates the use of methodological pluralism as a way to illuminate complexity.  Kelly looks back to the early days of community psychology at the Swampscott Conference in 1965, and he notes we have come a long way but we still have a long way to go. He recommends the use of ecological theory, systems theory, examination of complexity and multiple milieus, and he mentions some methods not directly covered in the book such as narrative inquiry, uncovering of stories, and use of videos. Kelly reminds us that science is not just about theory testing; it also is about discovery. In the future, Kelly hopes community psychologists will be open to multiple and new methods and will engage in continual reflective practice as they move thoughtfully into the future.

Summatively, Methodological Approaches to Community-Based Research is an important book because it covers methods that are very important for community psychology but are not typically covered in traditional statistics and methods courses in community psychology graduate programs. The methods covered are equally important for practitioners to enable them to systematically examine a host of contextual factors in relation to the health of groups and individuals they work with. The selection of methods for inclusion in Methodological Approaches to Community-Based Research is excellent, and the writing is consistently good across all chapters. This book reflects the many advances that have been made in quantitative research during the past 25 years that now enable researchers to study the importance of context on individual behavior. I identify only one major weaknesses of this book, and that is the lack of inclusion of qualitative methods that community researchers and practitioners can also add to their repertoire of methods. I recommend that the authors address this weakness in a companion book of the same length as the current book. Once this is done, presentations of mixed methods approaches will be more useful. In short, my only criticism is that another short book is needed to accompany this outstanding book.    
 

Reviewer

R. Burke Johnson, BJOHNSON@usouthal.edu

R. Burke Johnson
University of South Alabama
College of Education
Mobile, AL 36688-0002
BJOHNSON@usouthal.edu

Comments (1)

Add Comment

About this Article