Saturday, April 27, 2013

Focus Groups: Research Design, Limitations and Potential.




Robert K. Merton
The focus group has been employed extensively in market research since the late 1940s, from the 1990s it has been increasingly readopted in social science research as an important qualitative research method (Wilkinson, 1998).  Hyden and Bulow (2003, p. 306), in a database search of ‘Psychinfo’,  found nine hundred articles using the keyword ‘focus group’ and almost a third of the articles were published after 1998 indicating a rapid growth of research utilizing the method.  The increased use of focus groups has been accompanied by the elaboration of methodological concerns unique to focus groups and the proliferation of focus group designs based on the research objective of a specific project.   Focus groups are a qualitative research method, and therefore subject to methodological issues that affect qualitative methodologies in general, however focus groups entail further issues of project level design, group level design and unit of analysis not encountered by other research methods (Morgan, 1996; Hyden and Bulow, 2003).  The limitations of focus group research has been both derided,  on the basis that the data obtained has little external validity or reproduces normative discourses, and valorised for providing new insight into social interaction and opinion formation amongst groups of individuals, thus redefining apparent methodological limitations as potential strengths   (Folch-Lyon and Trost, 1981; Smithson, 2000).  Discussion of focus group methods benefits from defining its relation to qualitative methodology more broadly, and qualitative methodology counterpoised with quantitative methodology to highlight points of contradistinction that inform focus groups alongside other qualitative methods.   Once this has been outlined, the distinctive features of focus groups can be more adequately dealt with and the questions of project-level design, group-level design and unit of analysis can be evaluated for its impact on data collection through to data analysis.


In the social sciences, two broad research traditions can be delineated: qualitative and quantitate methodologies (Bryman, 2008).  Though, research projects can employ research methods from both traditions and utilize mixed-method design that generates both qualitative and quantitative data (Creswell, 2003).  In fact, as Morgan (1996, p. 134) notes, “focus groups and surveys are one of the leading ways of combining qualitative and quantitative methods” but that “such designs also raise a complex set of issues, since the two methods produce such different forms of data”.  Therefore, mixed-method studies have a tension between the status of data collected by qualitative and quantitative methods and the means by which these two datasets are analysed and triangulated.  This tension derives from nature of qualitative and quantitative methodologies that generate different forms of data. Qualitative methodology is often defined in contrast to qualitative methodology that employs research methods to produce numerical data that can form the basis of generalizations about a target population from a representative sample.  The use of statistical techniques in quantitative methodologies, in the case of surveys of representative samples, ensures high degrees of external validity relative to qualitative methods.  However, as Folch-Lyon and Trost (1981, p. 445) argue, “one major problem is that a quantitative survey is a highly structured situation in which the respondent has to adapt his responses to previously determined alternatives”.  Qualitative methodology is less subject to this criticism, given that the qualitative methods, for example participant observation or in-depth interviews, are less structured and allow greater ranges of response from participants.    

Qualitative research methods have a long history in the social sciences, particularly in the disciplines of anthropology and sociology, yet it was not until the 1990s that clarification of the distinct forms of qualitative research strategies became more established (Creswell, 2003; Bryman, 2008).  Focus groups are a research method, that can be differentiated from semi-structured interviews and other qualitative research methods, whilst qualitative research strategies refer to methodological approaches to the use of research techniques (Creswell, 2003).  Focus groups represent one qualitative research method that can be employed in an array of different qualitative research strategies.  Wolcott (2001) identified nineteen distinct qualitative research methodologies, from ethnographies through to narrative research. Each distinct research strategy employed, for example grounded theory or phenomenological research, will influence the process of data collection and analysis (Creswell, 2003).  For example, grounded theory is an inductive research strategy the aim of which is to generate a “theory of a process, action or interaction grounded in the views of the participants” (Creswell, 2003, p. 14).   Focus groups employed within a grounded theory approach will typically be less structured to generate new data and concepts from the participants than deductive research that aims to test a predetermined hypothesis.   In general, qualitative research is attuned to the subjective experience and meanings ascribed to social phenomenon by research participants and it has been criticised on the basis that its data is too “impressionistic” and therefore difficult to replicate or inform generalized understandings of social relationships   (Bryman, 2008, p. 391).  Nevertheless, it is precisely the ‘impressionistic’ nature of qualitative data that provides social researchers with a rich source of understanding social relationships and interactions beyond predetermined categories used in quantitative research. 


Qualitative methodologies, from the criteria of quantitative research, are subject to a series of limitations, not least of which is the external validity of research based non-representative samples and relatively unstructured data collection (Folch-Lyon and Trost, 1981; Bryman, 2008).  Focus groups are subject to similar criticisms levelled at qualitative research methods, however focus groups can be differentiated from other qualitative techniques and have a set of unique methodological concerns not encountered with other methods.  These unique methodological concerns stem from the unique features of focus group methods, which can be differentiated from a number of related research methods. 


The origin of the focus group method is often attributed to Merton’s (1987, p. 551) “focused interview” technique developed in the 1940s at the Bureau of Applied Social Research.  Merton and Kendall (1947, p. 541) outlined several criteria to delineate focused interview from other group techniques,   the participants were homogenous in respect to having experienced a “particular concrete situation” - i.e. having watched a television program or read a magazine article,  the concrete situation was analysed prior to the group session to generate a hypothesis and interview guide,  that would then be tested by the “subjective experience of persons exposed to the pre-analysed situation”.   Thus, Merton and Kendell’s (1947) conception of the focus interview is an example of deductive research, given that the study is used to validate or modify pre-established hypotheses and theory,  which can be counterpoised to focus groups informed by grounded theory.  Focus groups that approximate Merton and Kendell’s (1947) original design are still employed in social science research; however there has been a proliferation of distinct application in recent years and deductive research with structured interview guides represents one pole of a spectrum of research designs (Morgan, 1998; Wilkinson, 1998). 


Across the spectrum of focus groups there remains an emphasis on the subjective experience of participants, Wilkinson (1998, p. 187) defined the purpose of focus groups to “elicit people’s understandings, opinions and views, or to explore how these are advanced, elaborated and negotiated in social context”.    Though the emphasis on subjective experience is shared with qualitative methods in general, the focus group can be differentiated from other qualitative methods on a number of bases.   The most salient feature of focus group methods is that it is a group interview, thus Morgan (1998, p. 130) defined focus groups as a “research technique that collects data through group interaction on a topic determined by the researcher”.   Moreover, Morgan (1998, p. 130) emphasized three essential components:  “first, it clearly states that focus groups are a research method devoted to data collection. Second, it locates the interaction in group discussion as the source of the data.  Third, it acknowledges the researcher’s active role in creating the group discussion for data collection purposes”.  Frey and Fontana (1991) developed a typology of groups and highlighted the importance of directive interviewing and structured interview guides for focus group sessions.  In this respect, focus groups can be differentiated from dyadic interviews and structured group formats in which participants interact with moderators and not directly with other participants, in the case of nominal and delphi groups structured to generate consensus among expert analysts (Morgan, 1998; Steward and Shamdasani, 1990).  Therefore, in focus groups the discussion is the data, but the contour of this discussion is influenced by a number of design issues (Folch-Lyon and Trost, 1981).  Morgan (1998, p. 142, 144, 149) identifies “project-level design”, “group-level design” and “unit of analysis” issues in the development of focus group based research,  that affect how the method is employed in data collection and analysis.


Morgan (1998, p. 141) identified the importance of “project-level design” for focus group research, that “apply to the research project as a whole”.    On this level of design, Morgan (1998) outlined issues of standardization, sampling and the number of groups in the study that affect data collection, quality and analysis.   Standardization of project-level design refers to the extent that all groups in the study draw on the same set of questions and procedures.   On one end of the spectrum of standardization,  focus group design that emphasises “emergence” in which questions and procedures shift in reaction to the themes generated by discussion and of the other end are fixed questions and procedures across focus groups to assist hypothesis testing and replication of the study’s results (Morgan, 1998, p. 142).  The difference between emergence models of focus group design and structured questions and procedures is echoed in the broader discussion of qualitative methods and quantitative methods.  Standardized project-level design allows for replication and cross-comparison of data generated in a study, yet Morgan (1998, p. 142) notes the criticism that “this aspect of focus groups is inconsistent with many of the key tenets of qualitative research”.   While emergence models, utilized by grounded theory research or exploratory research, allow for participants to determine the salient themes that emerge from discussion of a given topic.  However, project-level design of focus group research can combine elements of low and high standardization to respective limitations of each design type.   Project-level design can be broken into two phases, with a relatively unstandardized exploratory phase that generates significant themes that can be used to produce a more standardized series of focus groups sessions that can be used to test hypotheses and more easily compare data from group  (Morgan, 1998).
 
The second project-level design issue is sampling.   There is a historical association of focus groups with market research, which has led to a practice of “segmentation” in sampling practices (Morgan, 1998, p. 143).  Folch-Lyon and Trost (1981, p. 446) argue that focus groups should have a high degree of homogenisation, “since it is not desirable to conduct group sessions with people of different backgrounds”.   In studies that involve multiple target populations,   segmentation and a series of homogenous group session are recommended by Folch-Lyon and Trost (1981).  This project-level design feature is advanced for a number of reasons.   Morgan (1998) outlines two advantages of segmentation, firstly it allows for comparison between different groups of participants and secondly, segmentation is said to facilitate discussion because of the similarity between participants.  The third issue of project-level design, the number of groups in a study is directly related to the issue of sampling and segmentation.   Generally, the number of focus groups is determined by issues of data saturation (Wilkinson, 1998; Bryman, 2008; Morgan; 1998),   diversity of participants and segmentation will increase the number of focus groups needed to adequately fulfil the requirements of data collection.

 
The second level of design outlined by Morgan (1998) is the question of group-level design.   In this sphere, the two issues that confront focus group design is the level of moderator involvement and group size.  These issues can be relayed back to broader project level design issues, high moderator involved could be indicative of highly standardized project-level designs.  Nevertheless, the level of moderator involvement is an important feature of group-level design.   Hyden and Bulow (2003, p. 307) argue that “in structured focus groups, the moderator takes an active role in controlling not only the topic but also the group dynamics.  In less structured groups the participants are encouraged to pursue their own interests in relation to the topic being focused on”.    This can be contrasted with Morgan’s (1998) two axis model of control in group-level design, differentiating between focus groups that have structured interview guides, tightly controlling questions and topics,  and focus groups were the moderator directs group dynamics by encouraging high levels of participant response.   Groups can be structured on one axis, and unstructured on the other.

 
High levels of moderator involvement in terms of the questions presented to the group are indicative of hypothesis testing research, which provides the basis for easy comparison between groups in the data analysis phrase of research (Wilkinson, 1998).   Moderator involvement to encourage higher levels of participant engagement is one means to overcome the problem of “dominate voices” that where “one or several group member(s) dominating the discussion so that theirs the only opinion clearly articulated” which is said to skewer data generated by focus groups (Smithson, 2000, p. 107).  Of course, as Morgan (1998) argued, these design features can be moderated independently of each other.  The second group-level design feature outlined by Morgan (1998) is the issue of group size.  On a pragmatic level, smaller groups are easier for moderators to manage.   However, the most salient determinate of group size is the topic to be discussed.   It is recommended that smaller groups of participants be used when the topic being discussed are controversial and emotionally charged, given that these topics generate high levels of participant response (Morgan, 1998).   For less controversial subject,   larger groups are recommended where participant engagement tends to be lower on the basis that multiple participants can contribute information to the data set.    The synchronization of topic and appropriate group size is an important design feature; miscalculation at this point can lead to failure to attain quality data at the level of implementation.  Research design, at the project-level and group-level have important implications for the status of the data collected.   Moreover,   focus groups have a unique issue surrounding the issue of units of analysis and whether the data analysis should centre on “the groups, the participants, or the participants’ utterances” (Morgan, 1998, p. 149). 


Hyden and Bulow (2000, p . 305) argue that “a central methodological question in analysing focus group material is who’s talking – that is, in what way are the utterances of individual members of the focus group to be interpreted”.   This can be related to what Smithson (2000, p. 107, 112) described as the problem of “dominate voices” and the problem of reproducing “normative discourse” that can be linked to the discussion of “group effects” that engender conformity to the majority opinion among participants (Asch, 1951; Carey and Smith, 1994).   These problems are distinctive to focus group analysis,   in-depth interview between an interviewer and participant do not occasion group effects because there is no group directly present to exert its influence.  Thus, dyadic interviews are touted for allowing participants space to express controversial opinions that would not be expressed in focus groups (Smithson, 2000).    Folch-Lyon and Trost (1981, p. 445) dissent from this position, arguing that in focus groups “paradoxically, there is a greater feeling of anonymity in a group than in a personal interview” that yields richer data.   However, if it found that focus groups are subject to group effects and problems of dominate voice and normative discourse; this does not necessarily undermine the quality of data generated but needs to be taken account of in the process of data analysis (Smithson, 2000).    


Smithson (2000) argues that it precisely these supposed limitations, group effect, dominate force and normative discourse, which are the strength of focus group methods by providing insight into the processes of group interaction and opinion formation.   Individuals that dominate group opinion formation are a feature of opinion formation that focus group methods highlight, the sidelining of controversial and marginal viewpoints in group interaction can provide insight into the hegemony of normative discourse, and coupled with interviews, focus groups can be used to analyse the relationship between “public” and “private” viewpoints (Smithson, 2000).  To overcome these issues in data analysis, Hyden and Bulow (2003) argue that attention needs to be paid to ‘who’s talking’. 


Hyden and Bulow (2003) conceptualized distinct types of focus group interaction that have an impact on analyses of data.  On one level, a group is an “aggregation of individuals” (Hyden and Bulow, 2003, p. 307, 311)   and two important interactive problems is how individuals contribute to a “common communicative ground” and “add their contributions to the common ground”.   Aggregated individuals without a common communicative ground, argued Hyden and Bulow (2003, p. 311), are “serially organized rather than organised as a group”.   The question for analysis is gauging the interactional dynamics of focus groups and the level at which participants contributes to the discussion.   Thus, Hyden and Bulow (2003,p. 319) conclude,  “it is often unclear whether the focus group is viewed as a collection of individuals or whether the interaction between the participants is intended to result in emergent views that are not reducible to the individual”.   This can only be resolved with careful attention to interactional dynamics in focus group data.It is a methodological issue unique to focus group interactions that is both a hurdle and an potential for greater insight into the formation of group opinions and dynamics. 


The increased use of focus group research has given rise to heightened sensitivity to the unique design features of the method and resultant methodological limitations and possible potential.   Focus groups are a qualitative research technique and are therefore subject to methodological criticisms faced by qualitative methodologies in general.   Quantitative methods and qualitative methods produce distinct datasets with inverted strengths and weaknesses, by the standards of quantitative research, focus groups have been deemed too subjective to provide generalizable data on social relationships and social values.  However, beyond these general features of qualitative research, focus groups have unique design features that impact data collection and analysis in a variety of subtle ways.   Morgan (1998) outlined three design features of focus group studies, project-level design, group-level design and unit of analysis issues.   On the project-level, issues of standardization, sampling and the number of groups impact the quality of data produced by focus group research.  Group-level design entails the level of moderator intervention into group dynamics and adherence to structured interview guides also impact the data produced by focus group methods.  Morgan (1998) introduced the issue of unit of analysis, of whether focus group data should be analysed on the level of groups, participants or participants’ utterances.   In this vein, the issue of interactional dynamics in focus groups highlight important issues for data analysis.  Group effects, problems of normative discourse and dominate forces demonstrate how individual opinions can be marginalized within group dynamics and therefore distort viewpoints generated in the data (Smithson, 2000; Asch, 1951; Carey and Smith, 1994).  However, focus group methods can be used to analyse the processes by which group form opinions and reproduce normative discourse.   In this respect, focus group methods provide an important research technique for a series of question that previous qualitative method had little application.  

Written by Mathew Toll.

Bibliography

Asch, S.E. (1951), “Effect of Group Pressure upon the Modification and Distortion ofJudgments”, Groups, Leadership and Men, Ed. H. Guetzkow, Pittsburg; Carnegie press.
Bryman, A. (2008), Social Research Methods, 3rd Ed, Oxford;  Oxford University Press. 
Carey,  M.A. and Smith,  M. (1994),  Capturing the Group Effect in Focus Groups: A Special Concern inAnalysis”,  Qualitative Health Research,  Vol. 5, pp. 225-2241. 
Creswell, J.W. (2003), Research Design: Qualitative, Quantitative and Mixed Methods Approaches, 2nd Ed, London; Sage Publications. 
Folch-Lyon, E. and Trost, J.F. (1981), “Conducting Focus Group Sessions”, Family Planning, Vol. 12, No. 12, pp. 443-449.
Frey,  J.H. and Fontana, A.  (1991), “The Group Interview in SocialResearch”,   Social Science Journal, Vol. 28, pp. 175-187.
Hyden, L-C. and Bulow, P.H. (2003),  Who’s Talking:Drawing Conclusions from Focus Groups- Some Methodological Considerations”, International Journal of Social Research Methodology, Vol. 6, No. 4, pp. 305-321. 
Merton, R.K. and Kendall, P.L. (1946), “The Focused Interview”, The American Journal of Sociology,  Vol. 51, No. 6., pp. 541-557.
Morgan,  D.L. (1998),  Focus Groups”,  Annual Review of Sociology,  Vol. 22, pp. 129-152.
Smithson, J. (2000), “Using and Analysing Focus Groups: Limitations and Possibilities”,  International Journal of Social Research Methodology,  Vol. 3, No. 2, pp. 103-119. 
Steward, D.W.  and Shamdasani, P.N. (1990)  Focus Groups: Theory and Practice,  Sage;  Thousand Oaks.
Wilkinson, S. (1998) "Focus Group Methodology: A Review", International Journal of Social Research Methodology, Vol. 1, pp. 181-203.
Wolcott, H.T. (2001), Writing Up Qualitative Research, 2nd Ed, Sage; Thousand Oaks.

2 comments:

Mathew Toll said...

This blog post has been cited in an academic paper: https://www.researchgate.net/publication/283569684_Outlines_of_Focus_Group_Exploring_the_potentials_of_focus_group_research_method_perspective_for_practical_research_purposes_whilst_indicating_the_limitations_and_advantages_of_the_method

Mathew Toll said...

This post has been cited in a thesis:
http://etd.cput.ac.za/bitstream/handle/20.500.11838/2394/207096465-Hendrickse-CJ-Mtech-Design-FID-2016.pdf?sequence=1&isAllowed=y

The authorship was misattributed to Robert Merton.