3 Research design
One of the most difficult parts of doing research is designing the research process so that the thinking element melts naturally into the practical element or how to address the overall mix of theory and practice in the research process. One difficult issue is how to move from a big question or issue to be able to design a piece of doable research which we discussed above. Usually a range of methods can be tailored to a particular research question - a choice of methods is available for each research question. This is explored in the next reading which is an extract from a paper on a clothing cluster in New York City. The research is about exploring the concept of embeddedness and how that changes economic relations and outcomes, or to put it another way: how social relations change market transactions. The network studied is not highly innovative, but joint problem solving arrangements and complex adaptation are present as foci for knowledge production. The concept of embeddedness helps explain how trust develops in networks and is relevant to study of the way clusters operate as you may recall from your study of clusters in unit 3. The paper, however, adopts quite a narrow view of embeddedness based only on structural relations between economic actors, and ignoring (deliberately) other dimensions of embeddedness such as institutions and culture, which was embraced by the theoretical framework in the unit 3 paper on north Italian clusters. Nonetheless this simplifying of the research field allows a very focused and fine-grained analysis through an ethnographic approach, with clear and important findings (including the risks of over-embeddedness, leading to reduced exposure to information external to the network).
The extract is particularly interesting because of its detailed exposition of the methodology used, the boundaries set around the investigation and the rationale for doing what was done. The appendix is also worth looking at as it lists the questions that informed the open-ended interviews and provides a useful example of how the big question of how embeddedness changes economic relations and outcomes was deconstructed and reframed to provide a manageable framework for research.
Read pp. 38-42 (starting Research methodology to include Appendix) of Brian Uzzi (1997) ‘Social structure and competition in interfirm networks: the paradox of embeddedness’, along with the appendix on p.67.
How does the author justify his choice of field?
Do you have any comments on the choice of firms and interviewees? Can you suggest any variations on this aspect of research design that might have been worth considering?
How were interviews and field observations used to improve the validity of the findings?
What was the purpose of the ‘cross site display’?
Note the different phases in the study, and the gradual development of a framework for interpreting the data, and that this was initially based on theory. How was ‘construct validity’ tested, and do you have any comment on this approach to validity?
The questions at the end, clearly relate to complex theoretical terminology, but Uzzi has been quite successful at pitching them at a level that the ordinary senior manager would understand. How does Uzzi guard against response bias?
Classic ‘arms length’ market relations ought to apply in this type of highly competitive field; an ethnographic approach provides rich data for theorising and analysis, but with 23 cases generalisability is only moderate.
Considerable attention has been paid to the issue of representativeness, and to the potential influence of size and ethnicity, and firms were not chosen to be part of the same trading network. This suggests some design variations: either examining a trading network, or examining pairs (firms that trade with each other) – this would allow triangulation on economic transactions from the perspective of each partner pair. However this would have required a larger sample to address the issue of representativeness. There is considerable variation in size, from 2-182 employees. This does raise questions about the interview strategy in the larger firms, so that relevant social networks are examined, by selecting appropriate interviewees.
Note that the unit of analysis was the interfirm relationship, and that various sources of information were used to gather background statistics and provide a sampling base.
Ethnographic interview accounts were compared with field observations to check for differences.
It is often the case in ethnographic studies that researchers quantify their data in some respects to estimate frequency and weighting of data across cases (as here).
Construct validity has been tested by getting industry experts to check if the interpretation (based on the conceptual framework) is plausible. There is considerable debate amongst some ethnographic researchers regarding validity (with some arguing there is a crisis of representation, based on the perspective that interpretations are just constructs of the ethnographers mind).
Uzzi partly addresses response bias by non-directive questioning; there may be a tendency for respondents to cast themselves in a good light; also his triangulation of field observation and ethnographic data would help remedy this.
2 Designing research questions