Вы находитесь на странице: 1из 16

Research in Multi Level Issues

Emerald Book Chapter: HOW? AND WHY?: THEORY EMERGENCE AND USING THE GROUNDED THEORY METHOD TO DETERMINE LEVELS OF ANALYSIS Ken W Parry

Article information:

To cite this document: Ken W Parry, "HOW? AND WHY?: THEORY EMERGENCE AND USING THE GROUNDED THEORY METHOD T ANALYSIS", Fred Dansereau and Francis J. Yammarino, in (ed.) Multi-Level Issues in Organizational Behavior and Strategy (Research in Multi Level Issues, Volume 2), Emerald Group Publishing Limited, pp. 127 - 141 Permanent link to this document: http://dx.doi.org/10.1016/S1475-9144(03)02008-3 Downloaded on: 01-02-2013 References: This document contains references to 4 other documents To copy this document: permissions@emeraldinsight.com This document has been downloaded 222 times since 2008. *

Users who downloaded this Chapter also downloaded: *

Ken W Parry, "HOW? AND WHY?: THEORY EMERGENCE AND USING THE GROUNDED THEORY METHOD TO DETERMINE LEV Dansereau and Francis J. Yammarino, in (ed.) Multi-Level Issues in Organizational Behavior and Strategy (Research in Multi Level Issues, Volume 2), Emerald Group Publishing Limited, pp. 127 - 141 http://dx.doi.org/10.1016/S1475-9144(03)02008-3

Ken W Parry, "HOW? AND WHY?: THEORY EMERGENCE AND USING THE GROUNDED THEORY METHOD TO DETERMINE LEV Dansereau and Francis J. Yammarino, in (ed.) Multi-Level Issues in Organizational Behavior and Strategy (Research in Multi Level Issues, Volume 2), Emerald Group Publishing Limited, pp. 127 - 141 http://dx.doi.org/10.1016/S1475-9144(03)02008-3

Ken W Parry, "HOW? AND WHY?: THEORY EMERGENCE AND USING THE GROUNDED THEORY METHOD TO DETERMINE LEV Dansereau and Francis J. Yammarino, in (ed.) Multi-Level Issues in Organizational Behavior and Strategy (Research in Multi Level Issues, Volume 2), Emerald Group Publishing Limited, pp. 127 - 141 http://dx.doi.org/10.1016/S1475-9144(03)02008-3

Access to this document was granted through an Emerald subscription provided by Kungliga Tekniska Hgskolan For Authors: If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service. Information about how to choose which publication to write for and submission guidelines are available for all. Please visit www.emeraldinsight.com/authors for more information. About Emerald www.emeraldinsight.com With over forty years' experience, Emerald Group Publishing is a leading independent publisher of global research with impact in business, society, public policy and education. In total, Emerald publishes over 275 journals and more than 130 book series, as well as an extensive range of online products and services. Emerald is both COUNTER 3 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation.
*Related content and download information correct at time of download.

HOW? AND WHY?: THEORY EMERGENCE AND USING THE GROUNDED THEORY METHOD TO DETERMINE LEVELS OF ANALYSIS
Ken W. Parry
ABSTRACT
The grounded theory method is more appropriate than just a grounded theory approach, for teasing out the detail of the level of analysis of constructs. Also, triangulation is important to this kind of research, but the methodological distinctions between qualitative and quantitative data and qualitative and quantitative analysis need to be made clear when mapping out a methodology. The contention here is that the qualitative analysis of quantitative data is more important than the quantitative analysis of qualitative data. Qualitative analysis in line with the full grounded theory method will generate explanations of how and why a construct is represented at various levels of analysis. In turn, such an explanation can illustrate whether the questionnaire instrument is representing the levels of analysis of the construct adequately or not.

INTROUCTION
It is wonderful to see more deliberate attempts to triangulate data and methodology in order to seek a greater understanding of the phenomena that are researched.
Multi-Level Issues in Organizational Behavior and Strategy Research in Multi-Level Issues, Volume 2, 127141 Copyright 2003 by Elsevier Science Ltd. All rights of reproduction in any form reserved ISSN: 1475-9144/doi:10.1016/S1475-9144(03)02008-3

127

128

KEN W. PARRY

Triangulation has been talked and written about for many years with little real success in getting to grips with its implementation in organizational research. What this commentary will attempt is to provide some different insights into triangulation more in line with the grounded theory method. It will elaborate some grounded theory principles that might help make more sense of Berson, Avolio and Kahais proposed approach. It will also introduce the notion of levels of abstraction to complement levels of analysis consideration. By so doing, a more rigorous application of qualitative method will evolve which will give better insights into how levels of analysis are manifested and measured. To start, it is necessary to make some comment about the grounded theory method as opposed to a grounded theory approach.

SOME COMMENTARY ABOUT THE METHOD


The article seems to reect some confusion about the discussion of methods. We all need to be more specic about when and where we are talking about data and analysis, rather than just method. To use qualitative methods does not automatically mean one is using qualitative data and qualitative analysis of those data. Being specic about the distinction between data and analysis makes the method clearer. It is fair to say that discussion about the need to integrate or concurrently conduct qualitative and quantitative methods is almost pass today. However, we e really need to realize when and where it is appropriate to integrate qualitative and quantitative data with qualitative and quantitative analysis. Table 1 illustrates the choices available when researchers demonstrate methodological triangulation. Having qualitative and quantitative data, which can be analyzed qualitatively and quantitatively, gives the researcher four general strategies to follow. The traditional full grounded theory method requires the qualitative analysis of qualitative data (Quadrant A). Traditional mainstream organizational research normally utilizes quantitative data and quantitative analysis thereof (Quadrant C). What Berson, Avolio and Kahai seem to be advocating is the traditional grounded Table 1. Methodological Choices Available to Triangulate Research.
Data Qualitative Qualitative Analysis Traditional grounded theory method A Integrated into grounded theory method B Quantitative Analysis Posited inter alia by Berson et al. includes content analysis D Traditional mainstream quantitative method C

Quantitative

Theory Emergence and Using the Grounded Theory Method

129

theory method and some quantitative analysis of qualitative data (Quadrant D). The grounded theory method does not discount the qualitative analysis of quantitative data (Quadrant B). In fact, it has been advocated strongly in recent years (Conger & Toegel, 2002) and is providing intriguing insights into organizational phenomena, especially when accompanied by the traditional method represented in Quadrant A (Kan, 2002). I will advocate that Berson et al. can generate greater insights into level specication by utilizing more than just a grounded theory approach to construct validation and level of analysis specication. They can do this by adopting more of the procedures associated with the grounded theory method. Note that in Quadrant B of Table 1, some use of a questionnaire instrument, analyzed to the extent of undertaking internal reliability analysis, factor analysis, correlation and regression analysis, even analysis of variance and conrmatory factor analysis, would still be classied as data rather than as analysis. The ndings from these short-term analyses can be analyzed along with all the other qualitative data concurrently. Explaining the similarities and differences between the various forms of data would be the challenge of the researcher. Kan (2002) has provided some insights into how this can be done, and how the explanations of phenomena can be richer as a result. I will say more on this later. It is important to reinforce here that Berson et al. are writing generally about a grounded theory approach, rather than the grounded theory method. There is some loose interpretation of grounded theory in the article. Specically, the full grounded theory method has more theoretical sampling than the authors indicated, more iterations for theoretically sampled data, and the theory emerges rather than is tested in the way that they suggest. Because of this, I would advocate that the grounded theory method, as opposed to the grounded theory approach, generates greater explanatory insight into construct validation and level specication. So much qualitative research draws upon the grounded theory procedures and philosophies that qualitative research generally is often referred to incorrectly as grounded theory. Theory testing is not a characteristic of the grounded theory method. Theory emergence is. However, it is perfectly appropriate to qualitatively test theory it just cannot be called grounded theory. What the authors were actually doing was partial grounded theory for reasons that I will continue to elaborate. However, having said that, the title indicates correctly that they used the grounded theory approach rather than the grounded theory method.

GENERAL QUALITATIVE vs. QUANTITATIVE ISSUES


My interpretation of Berson et al.s argument is that quantitative method, generally meaning quantitative analysis (Quadrants C and D in Table 1), asks the question,

130

KEN W. PARRY

Does this construct t this level of analysis? On the other hand, qualitative method (Quadrants A and B in Table 1) asks the question, What is the construct that does t this level of analysis? Much of the level of analysis work thus far has been in answer to the former question. However, much effort now needs to be directed toward the latter question. After all, there is so much variation inherent within the construction of quantitative measures, as Berson et al. correctly point out, that it is difcult to obtain a measure that will accurately reect the construct at the intended level. Having thought through these general methodological issues, the question becomes one of how this relates to level of analysis consideration.

LEVEL CONSIDERATIONS
It is of course refreshing to see acknowledgement that the level at which a particular construct exists can be open to interpretation. More importantly, the level at which a construct operates can be the individual level, the group level and the organizational level concurrently. It is also gratifying to see that the identication of the level(s) at which the construct operates requires more than agreement on a survey. However, conducting the questionnaire data gathering and the qualitative data gathering concurrently will allow the researcher to see whether or not, or to what extent, questionnaire data at the individual level can be aggregated to generate a group level measurement of the construct. I would contend that the real methodological advance that Berson et al. can offer is not to sequentially conduct questionnaire data gathering and analysis followed by qualitative data gathering and analysis, but rather to conduct them together. The iterations that should be followed are those to achieve theoretical coding and theoretical sampling rather than to achieve deduction followed by verication. If the deduction cannot be veried, the research has failed. But, if deduction in the form of working hypotheses and verication can be carried out continuously with several iterations, the research will be able to determine what form the construct actually takes. In addition, the authors seem to be falling for the old trap of dening the construct and specifying the level of analysis before they investigate it. By so doing, they are working within the parameters of the epistemology they are eschewing and attempting to move on from. Rather than dening a construct, they should be probing a phenomenon, determining what denition emerges and the levels at which the phenomenon manifests itself. This follows the epistemologies of phenomenology and symbolic interactionism rather than of theory testing. In other words, it is far more protable for the denition to emerge from the analysis

Theory Emergence and Using the Grounded Theory Method

131

rather than for it to be dened pre hoc or a priori. In this way, the reality of the construct will emerge, rather than an assessment of whether the a priori denition does or does not t. The even greater risk is that the a priori denition will be forced into the randomly-selected and representative data that are eventually gathered.

THEORETICAL CODING
By concurrently investigating levels of analysis, the researcher becomes more aware of the ways in which the construct might manifest itself differently or similarly at different levels. It is also an effective theoretical coding question to ask of the data. With the grounded theory method, the researcher will ask questions of the data, as opposed to asking questions of interviewees. One can ask of the data, At what level is this emerging construct being manifest? The researcher can ask this of the properties of the categories that have emerged, of the interview transcripts, of the explanatory memos and of the quantitative questionnaire data that might have been gathered. Asking questions of the data is central to the theoretical coding schemas or families proposed by Glaser (1978). Glaser proposed 18 general coding families, but the researcher will ask all sorts of questions of the data when probing for the validity of the working hypotheses that will emerge. Another question that can be asked is, What processes might explain the similarities or differences being found between the qualitative data and the quantitative data? Berson et al. discussed the example of organizational climate. Some comment needs to be made on that example, and how these methodological issues are represented in relation to climate.

THE CLIMATE CONTROVERSY


Since the 1980s, climate has been identied as operating at any of the various levels of analysis, and of being either a class of organizational variables or a class of psychological variables. Once again, what qualitative analysis of qualitative (and quantitative) data will provide is an insight into the levels at which climate operates and whether it is organizational, psychological, social or some other class of variables. More likely than not, it will identify when and whether climate operates at all levels and when and whether it is all these classes of variables concurrently. Even more importantly, such analysis will identify how and why climate operates as it does, not just when, whether, and where it operates. However, such ndings will be substantive only to the context in which they are undertaken.

132

KEN W. PARRY

Replication will be required to generalize about the nature and manifestation of climate.

The Impact of Context This article engaged in some insightful discussion about the role of context in the manifestation of certain constructs. The observation is accurately made that the nature of context can modify the level at which a construct is viewed as operating. What qualitative analysis of qualitative (and quantitative) data will provide is insight into how and why context has this impact, not just whether, when, where and for whom. For example, Kan (2002) assessed work group supervisory leadership at the individual level with use of the MLQ. Respondents rated the leadership by the individual in their role as supervisor of the work group. She also interviewed for and observed leadership at the individual, group and organizational levels. What became clear was that many group and organizational (contextual) issues affected the perceptions of individuals as they completed the questionnaire about their leader. People were considering group and organizational issues when assessing the leadership of their supervisor. In other words, it was found that leadership at the group and organizational levels explained the variance in leadership at the individual level of analysis. Put another way, the questionnaire was not measuring leadership at just the individual level of analysis. This work also explained how context has an impact upon leadership at the individual and group levels of analysis. The presence of such impact could be determined with questionnaire work alone, but how and why it had the impact could only be determined with qualitative data and certainly only with qualitative analysis. Even more importantly, the uniqueness of this specic substantive situation meant that mere aggregation of the individual level data could not be used as a measure of the group or even organizational level of analysis. In another situation, aggregated individual level data may well be a valid measure of group level constructs, but not in this specic context. The situational uniqueness or substantiveness of this construct could only be uncovered because Kan undertook both qualitative and quantitative data gathering and qualitative analysis concurrently.

CONSTRUCT SPECIFICATION AND DEVELOPMENT


Berson et al. also made indirect reference to the notions of theoretical sensitivity and saturation. These grounded theory concepts are implied in the discussion of

Theory Emergence and Using the Grounded Theory Method

133

construct specication and development, although they are not explicitly stated. They need to be claried here. Theoretical sensitivity (Glaser, 1978) involves using a priori observations and the researchers own theoretical background to develop broad notions of a construct. Saturation (Glaser & Strauss, 1967) includes the consistency across situations and the a ha experiences that allows the researcher to conclude that the nature of a construct and its relationship with other constructs is valid and reliable. These are the terms that should be used to explain these methodological phenomena. Theoretical sampling is another allied concept that begs further discussion.

Theoretical Sampling It is slightly concerning that the authors write about collecting data from a range of sources, and then moving to interpretation. The essence of the value of the grounded theory approach is that the collection and the analysis of data are constantly compared over a series of iterations. It is not a two-stage process of: (a) data gathering and then; (b) interpretation. This constant comparative method of qualitative analysis (Glaser & Strauss, 1967) or analytic induction (Silverman, 1993) is the key to eshing out the nature of the emerging construct. It is a multiple- and iterative-stage process, not a two-stage process. Meaning is derived from several iterations, not just one. The basic problem with random sampling is that the researcher may well not get the richness of data that will allow the construct to be elaborated adequately. What is required for richly explanatory data is purposive or theoretical sampling. What the researcher is after is representativeness of the construct or problem, not representativeness of the population. By utilizing random sampling, the researcher might interview the person who wonders why she has been selected, will clam up and provide no insight. In effect, they will be a non-response. However, by working with the key informant to determine who the person might be who is best positioned to comment, and who does not feel threatened and is mature about what has occurred, the researcher will obtain a rich insight into the construct. Of course, speaking with one person will not sufce, but nor will selecting them randomly. Perhaps this dilemma epitomizes the dilemma about non-response bias. If one selects subjects randomly, the problem is that the subjects with the best insights may not be responding. By interviewing and observing one person who is a rich source of data, that person will simultaneously provide data at the individual level, the team level, and the organizational level. Thus, the sampling criterion must be to provide richness of data, not representativeness of population, I would suggest.

134

KEN W. PARRY

When the authors talk of higher-order categories, they are referring to higher levels of analysis (group or organizational levels). What also needs to be recognized are the higher-order levels of abstraction. Grounded theory provides these levels of abstraction as part and parcel of the data-analytic method. Lets probe the concept of level of abstraction further with regard to its relationship to level of analysis.

` LEVELS OF ABSTRACTION VIS-A-VIS LEVELS OF ANALYSIS


Figures 1 and 2 are intended to show how hierarchies of abstraction can be represented. The explanation of constructs at increasingly higher levels of abstraction is axiomatic of the grounded theory method. The gures show how grounded theory leadership research has provided three social processes of leadership (optimizing, resolving uncertainty and enhancing adaptability) that operate concurrently at all levels of analysis (individual, group and organizational). Figure 1 shows the hierarchy of social processes of leadership that were generated from grounded

Fig. 1. Hierarchy of Abstraction Model (H. A. M.): Conceptual Hierarchy of Abstraction, Derived from Grounded Theory Research. Sources: Parry (1999), Irurita (1996).

Theory Emergence and Using the Grounded Theory Method

135

Fig. 2. Hierarchy of Abstraction Model (H. A. M.): Psychometric Validation of the Grounded Theory Hierarchy of Abstraction. Source: Parry (2002).

theory research. Figure 2 shows the hierarchy of abstraction model generated from questionnaire-based validation of the model. Figure 1 represents the results of two research projects. Figure 2 reects a third research project. As one moves up through the models, the level of abstraction of the constructs increases, as does the explanatory power of those constructs. What the fully qualitative method will provide is more than just a more detailed operationalization of the construct and its level of analysis. It will position the construct within its hierarchy of abstraction. When Berson et al. posit the existence of the construct of collective trust they seem to be positing the presence of a construct at a higher order of abstraction. Figure 1 shows that there are six manifest variables, derived principally from interview and observation, that represent the higher-order latent construct of leader adaptability. Leader adaptability and follower adaptability together represent the (even) higher order latent construct of enhancing adaptability. Enhancing adaptability is one of two basic social processes of leadership. The other is optimizing. This Hierarchy of Abstraction Model (HAM) adds explanatory power to the constructs that emerged and must be part of any grounded theory analysis. Leader adaptability and follower adaptability are manifest mainly at the individual level. Enhancing adaptability is manifest mainly at the team and organization

136

KEN W. PARRY

levels. What is disconcerting about Berson et al.s research is that it is never conducted at a level of abstraction any higher than that of the manifest variable. Thinking in terms of hierarchy of abstraction as well as level of analysis will add greatly to the explanatory power of their ndings. When there is disagreement between the aggregation tests and the triangulation procedure, it is less likely to be a call for a return back to the deduction and verication stages, than a call for an explanation at a higher level of abstraction. Disagreement between ndings probably does not mean that the research has been conducted improperly and has to be replicated. What it most likely means is that there is an explanation that has to be made apparent which explains and incorporates the different and apparently paradoxical ndings. Collective trust could well be such a construct at a higher level of abstraction which explains the apparent ambiguity between team members asking for reassurances and other team members no longer asking for reassurances. Figure 3 is another hierarchy of abstraction model taken from research at the individual level of analysis. Managers from around the country were asked to rate one colleague according to the multifactor leadership questionnaire. A large number of possible explanatory models were tested, but Fig. 3 represents the model of

Fig. 3. Hierarchy of Abstraction Model (H. A. M.) of the Full Range of Transformational Leadership. Source: Parry & Meindl, 2002.

Theory Emergence and Using the Grounded Theory Method

137

best t. The gure shows how individualized consideration consists of four items (manifest variables) that the MLQ uses to operationalize the construct. Berson et al. contend correctly that two of those manifest variables represent coaching and mentoring and two represent meeting specic needs of followers. The relationship with contingent reward is represented by the latent construct of developmental exchange at a higher level of abstraction. Avolio, Bass and Jung (1999) also proposed developmental exchange as a higher-order factor, generated from conrmatory factor analysis. In turn, management-by-exception-active has more in common with developmental exchange than with passive management. Hence, this commonality is reected in the presence of the higher order latent construct of active management. The above comments are not to say that levels of abstraction are superior or preferable to levels of analysis. Rather, it is to say that levels of abstraction are a crucial part of qualitative research, especially that related in some way to the grounded theory method. Researchers following this line of research will nd it very hard to get away from level of abstraction thinking as they pursue level of analysis understanding.

THE MATTER OF TRIANGULATION


Triangulation is a cornerstone of the arguments put forward by Berson et al. However, the value of triangulation to the proposed research appears not to be fully appreciated by the authors. Of course, triangulation is not a method, as Berson et al. seem to claim. However, the triangulation of methods is extremely fruitful, quite clearly. Triangulation can be of data, theory, researcher and method method means specically analysis in this case (Denzin, 1978; Guba & Lincoln, 1994). The triangulation of data that Berson et al. engage in is quite clear, and extremely productive. They propose the use of a range of sources of data. In particular, the concurrent use of qualitative data with quantitative data is a future direction that levels of analysis research must pursue. Theoretical triangulation involves the use of multiple perspectives for analyzing and interpreting data. This has been incorporated into consideration very well through discussion of levels of analysis, the role of context, and so on. Researcher triangulation is alluded to by the authors, and also has enormous potential to provide detail and richness to the explanation of the construct at all levels. Egri and Herman (2000; and Herman & Egri, 2002) have demonstrated the richness of explanatory power that can come from researcher triangulation when using the grounded theory method. Methodological triangulation involves the use of multiple methods of data gathering and analysis, and it is this area that the

138

KEN W. PARRY

authors have not realized the full potential of the contributions of the grounded theory method to the levels of analysis imperative. There is a problem with writing that qualitative data are gained from survey instruments about organizational climate or culture. Sure, the content of the questionnaires might be about qualitative constructs, but clearly, survey instruments represent quantitative data, and as such are subject to the reliability problems that the authors correctly identify. The authors then write about using qualitative data to help indicate how lower level entities are affected and to provide richer and thicker description of the context in which a construct is embedded. If the qualitative data represent interviews and observations, then they are absolutely correct in saying that such data will help to achieve what they suggest. However, there is a tone to the propositions put by the authors that this type of research consists of distinct stages the questionnaire stage followed by the interview stage. In fact, the stages of the grounded theory method are many and varied. They are iterations that can take many forms. Qualitative data and analysis will probably start the process, followed by a modest questionnaire instrument, analyzed quantitatively and qualitatively, followed by more qualitative data and analysis, followed perhaps by qualitative analysis of extant quantitative organizational data, followed perhaps by another questionnaire distribution, followed by two or so more iterations of qualitative data and analysis interspersed with qualitative analysis of the questionnaire ndings. In so doing, the method allows the researcher to understand levels of analysis and describe the role of context and elaborate of the nature of the construct concurrently. It also allows the researcher to develop a theoretical explanation of the phenomenon and to determine the nature of the basic social process that represents the phenomenon in action. Much has been said about ways to analyze data. Therefore, some comment is needed on the matter of measurement within the proposals put forward.

THE MATTER OF MEASUREMENT


It is refreshing to note that the authors can measure constructs both quantitatively and qualitatively. Essentially, saturation means that the construct is a signicantly discreet variable. It means that signicance cannot be expressed numerically, but it can be expressed qualitatively by the researcher when explaining how there was no more variance emerging from the analysis to indicate that the construct possessed any other properties apart from those already identied. Similarly qualitative magnitude cannot be identied numerically. However, it can be expressed as very high, or frequently, if not always or moderate agreement, and so on. Lets face it, that is the same terminology used by respondents on questionnaires used

Theory Emergence and Using the Grounded Theory Method

139

for organizational analysis, so qualitative measurement is no less rigorous than quantitative measurement, when identifying the magnitude of the manifestation of the construct. The authors provide the example of counting the number of times a leader expressed concern in electronic mail communications with followers, and coding the expression of concern on an intensity scale. Such a practice represents a quantication of qualitative data according to predetermined criteria. As such, it has more in common with content analysis than with the grounded theory method (Quadrant D in Table 1). What Berson et al. are proposing is to identify a property of individualized consideration (concern) and to dimensionalize that property (intensity of concern). This practice of dimensionalizing the construct does conform with the grounded theory approach. However, what they are doing is testing for the presence of a particular property of individualized consideration. In the true spirit of grounded theory, they should be coding the electronic mail communications to see what range of properties are actually present. It could well be that concern is not an important property of the communications, but that another property like anger or apathy or intolerance, might be present. If concern is present, but there is also a range of intensities of anger and apathy and intolerance, the very nature of the individualized consideration is open to a completely new interpretation. If so, they are perfectly justied in observing face-to-face interactions to augment the electronic mail data. Having said all this, the verication methodology employed by the authors is a very good example of the way in which a factor with similar mean but high variance can be explained qualitatively. Without qualitative analysis of the quantitative data, the nature of that variance would never otherwise be determined. What this tells us is that questionnaire data provide us with sterile or shallow data (vis-` -vis rich data), and that qualitative data can ll in the many gaps a in the questionnaire-derived data. In this case, qualitative data demonstrated that individualized consideration operates concurrently at the individual, group and organizational levels. What it also suggests is that the levels at which the construct operates may not be able to be differentiated psychometrically. The constructs may only be differentiated qualitatively. Perhaps, more importantly, if levels can be differentiated psychometrically, this will only be ascertained qualitatively.

An Important Question This all must raise an obvious question, just how accurate and reliable is all the questionnaire-derived research that has been undertaken thus far if the method is

140

KEN W. PARRY

so shallow and potentially riddled with explanatory gaps. The authors seem to ask the same question when they say that the use of qualitative methods calls into question the conclusions that were based solely on quantitative methods. This is a big question to ask, but the evidence from the present research, and other research like it, certainly indicates that the question its asking is valid. On the other hand, what the research does bring home to the reader quite clearly is that the concurrent use of quantitative (questionnaire) data and qualitative (interview, observation, participation, document analysis) data is almost mandatory if the generation of meaning is to be maximized. Meaning might mean sensitivity toward level of analysis or it might mean insights into dependence, independence, cause, effect or any of the many other epistemological insights we seek from research. A nal methodological issue is raised by the practical suggestion that an unobtrusive structured content analysis coding scheme can be implemented for electronic mail messages. I have a concern that to do this automatically with computerized content coding programs would severely compromise the accuracy of the analysis. One cannot automatically code messages such as this. The only person who can code is the researcher. This is for two reasons. First, only a person is in a position to determine the emotion that might go into an electronic mail. Second, each actual message is unique, and therefore must be coded uniquely, not automatically. The claim that computer-mediated communication is an alternative to quantitative and qualitative data is a perplexing one. Data are either qualitative or quantitative. They must be one or the other. I think what the authors are claiming is that computer-mediated communication is qualitative data that they believe can be quantied readily. It possibly can be quantied readily, but the great risk is that meaning will be at risk, just as it is with quantitative data. Qualitative analysis of computer-mediated communication is still necessary to maximize the meaning that researchers must derive. A conclusion from this work is that researchers need to be specic about when and where they are gathering qualitative or quantitative data, and when and where the analysis of those data is qualitative or quantitative. It appears that the authors have concluded that discrimination between levels of analysis is difcult to achieve, if not impossible, in many situations. I would add that this conclusion is certainly true without the use of qualitative data and denitely without qualitative analysis. The discriminant validity of levels of analysis for many social constructs like leadership, culture and climate can only be determined accurately with a combination of qualitative and quantitative data, and qualitative analysis, and in any case, discriminant validity may be determined only rarely.

Theory Emergence and Using the Grounded Theory Method

141

REFERENCES
Avolio, B. J., Bass, B. M., & Jung, D. I. (1999). Re-examining the components of transformational and transactional leadership using the multifactor leadership questionnaire. Journal of Occupational and Organizational Psychology, 72, 441462. Conger, J. A., & Toegel, G. (2002). A story of missed opportunities: Qualitative methods for leadership research and practice. In: K. W. Parry & J. R. Meindl (Eds), Grounding Leadership Theory and Research: Issues, Perspectives and Methods (pp. 175197). Greenwich, CT: Information Age Publishing. Denzin, N. K. (1978). The research act: A theoretical introduction to sociological methods (2nd ed.). New York: McGraw-Hill. Egri, C. P., & Herman, S. (2000). Leadership in the environmental sector: Values, leadership styles and contexts of environmental leaders and their organizations. Academy of Management Journal, 43(4), 571604. Glaser, B. G. (1978). Theoretical sensitivity: Advances in the methodology of grounded theory. Mill Valley, CA: Sociology Press. Glaser, B., & Strauss, A. (1967). The discovery of grounded theory. Chicago: Aldine Press. Guba, E. G., & Lincoln, Y. S. (1994). Competing paradigms in qualitative research. In: N. K. Denzin & Y. S. Lincoln (Eds), Handbook of Qualitative Research (pp. 105117). Thousand Oaks: Sage. Herman, S., & Egri, C. P. (2002). Triangulation in action: Integration of qualitative and quantitative methods to research environmental leadership. In: K. W. Parry & J. R. Meindl (Eds), Grounding Leadership Theory and Research: Issues, Perspectives and Methods (pp. 129148). Greenwich, CT: Information Age Publishing. Irurita, V. F. (1996). Optimizing: A leadership process for transforming mediocrity into excellence a study of nursing leadership. In: K. W. Parry (Ed.), Leadership Research and Practice: Emerging Themes and New Challenges (pp. 125138). Melbourne: Pitman Publishing. Kan, M. (2002). Re-interpreting the multifactor leadership questionnaire. In: K. W. Parry & J. R. Meindl (Eds), Grounding Leadership Theory and Research: Issues, Perspectives and Methods (pp. 159173). Greenwich, CT: Information Age Publishing. Parry, K. W. (1999). Enhancing adaptability: Leadership strategies to accommodate change in local government settings. Journal of Organizational Change Management, 12(2), 134156. Parry, K. W. (2002). Hierarchy of abstraction modeling (H.A.M.) and the psychometric validation of grounded theory research. International Journal of Organizational Behaviour, 5(5), 180194. Parry, K. W., & Meindl, J. R. (2002). Models, methods, and triangulation: Researching the social processes in our societies. In: K. W. Parry & J. R. Meindl (Eds), Grounding Leadership Theory and Research: Issues, Perspectives and Methods (pp. 199221). Greenwich, CT: Information Age Publishing.

Вам также может понравиться