A Cognitive Network Approach to Educing Shared Political Identity: The Case of Macedonia
Philip Murphy Graduate School for Public and International Affairs University of Pittsburgh
Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
Self, Group, and Country Philip Murphy
Abstract Scholars and practitioners from both in- and outside of the region often assume that ethnicity is almost exclusively driving political behavior in the Republic of Macedonia. This research moves beyond the oftentimes simplistic observations of ethnic cleavages to consider more of the richness and complexity that characterizes the region, using constructivist models of individual and collective identity. Constructivists such as Niklas Luhmann (1985) regard identity groups as sharing a common frame of reference that allows group members to interpret the world around them and make decisions based on those interpretations. To better investigate collectivist perception and decision making, it is necessary to investigate how decisions are defined at the level of the individual actor and then extrapolate upward, to the society level. To do this, Luhmann (1995, p. 242) recommended employing George R. Kelly’s (1955) repertory grid methodology to elicit the constructs that comprise an individual’s interpretive framework. The frameworks-in-use for various groups in this population have therefore been elicited through standardized repertory grid interviews. Resulting frameworks have been analyzed to reveal patterns that belie the simplistic models that rely on ethnicity to explain political behavior and begin to distinguish emergent cognitive and other middle ground(s).
Overview I.
Background a. Identity Research and Macedonia b. Inquiry
II.
Methodological Theory Review a. Measuring Perception and Identity b. Analytic Approach
III.
Data Collection a. Sample b. Elicitation Procedure
IV.
Results a. Baseline Consensus Measure b. Stage 1: Imperfect Partitions: Ethnicity and Rural vs. Urban c. Stage 2: Frames of Reference May be Shared Across Ethnicities
V.
Discussion and Context
Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
1
Self, Group, and Country Philip Murphy
I.
Background a. Identity Research and Macedonia
The Republic of Macedonia (hereafter Macedonia) is a small Balkan country, located between Serbia-Kosovo, Greece, Albania, and Bulgaria, that has been subject to international stabilization efforts since its secession from the failing Yugoslav state. Of the approximately two-million inhabitants, roughly sixty-four percent are ethnic Macedonian Christians, twenty-five percent are ethnic Albanian Moslems, and the remaining eleven percent hail from a variety of ethnic backgrounds (ROM 2005). The relative sizes of the ethnic Albanian and ethnic Macedonian populations result in the frequent reference to Macedonia as a country with two majorities. As the country pursues its aspirations of accession into the European Union, national politics tend to figure heavily in the reform and modernization efforts. In fact, politics are so pervasive in every aspect of Macedonian life at the moment that it is not uncommon for its citizens to wryly refer to it as a country of two-million politicians. Political identity in Macedonia has continued to develop since its 1991 secession from Yugoslavia. Over time, conventional wisdom and the works of area studies scholars have characterized Macedonia as a state divided by conflict between two, or at best three competing identity groups. These views give rise to two possible models of Macedonia’s political identity. The first, in following with the theory of “primordial” (Geertz 1963) divisions between ethnicities, depicts a society primarily consisting of two mutually opposed mono-ethnic polities: Albanians and Macedonians. The second model adds the factor of inter-group social communication (Deutsch 1966) with the suggestion that denser urban environments tend to create greater communication and association, giving rise to a third, ethnically mixed identity group: Macedonians, Albanians, and a mixed group of urban dwellers who have advanced education and opportunities for inter-ethnic interaction and communication. This study attempts to test these two- and three-group hypotheses and create a wider and possibly more complex and valid understanding on the basis of empirical evidence.
Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
2
Self, Group, and Country Philip Murphy
The two- and three-group models appear to have informed policy inquiry relating to Macedonia, particularly on behalf of international entities. However, these overly simplified models are likely to misinform and severely restrict the scope of inquiries that are likely to take place, and therefore stunt an overall understanding of how policy is, or can be shaped there. Research into the region is increasingly suggesting that the current store of knowledge in and of Macedonia and the region (Balkan stereotypes and other popular assumptions) is of dubious value for those interested in understanding political development within the region and the resulting policy environments (Wilmer 1997). There is also an increasing call for more generative inquiry that relies less on prior assumptions and western models (Todorova 2004, intro.). Balkan countries have historically been described in terms based more frequently on ‘romantic notions’ and unsupported conjecture, than on verifiable or primary data. In following with this practice, contemporary scholars and practitioners from both in- and outside of the region frequently rely on the assumption that ethnicity is the major driving issue in Balkan political identity. While, superficially, this appears to be a difficult concept to dismiss, newer research into Balkan politics is beginning to move beyond the oftentimes simplistic realm of ethnic cleavages and is beginning to consider more of the actual richness and complexity that characterizes the region. According to Maria Todorova, much of what clouds contemporary perspectives of the Balkans essentially arises from Western bias (1997, 2004) and an intense focus on nationalism (2005).
b. Inquiry A primary challenge of the current study was therefore to minimize the effects of outside influences in the form of researcher bias and the imposition of Western models and stereotypes. The study was additionally designed to move beyond nationalist and similarly motivated discourse through the use of more generative research methods to distinguish identity frameworks. The fundamental task of the study was thereby to probe for the constructs that individuals employ to frame their perceptions of public figures who have had some effect on life and politics in Macedonia. The reference frames elicited in this manner essentially provide a depiction of individual political identities, and comparing these identities across the population allows for extrapolating whether new identity groups may be emerging across disparate cultures Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
3
Self, Group, and Country Philip Murphy
there. In this way, popular assumptions were set aside in favor of inductive inquiry into the state of the political identities developing in this relatively new republic. This approach to investigating political identity in Macedonia is novel, as it presents an opportunity to escape the limitations of what appear to be established misconceptions about political identity and ethnopolitics there, allowing for more contextually relevant policy decisions. This information is worth investigating for its value in generating a richer empirical representation of individual and collective political decision making. Scholars have come to accept that, rather than investigate every possible alternative, individuals simplify a choice or situation into what they consider to be the most likely alternatives or explanations (Simon 1957). Simon’s original assertion to this affect, meant as an improvement to the rational actor model, was directed at individual decision making, but may also be applied to how we perceive collective perceptions and decisions. A number of researchers and theorists have taken an approach similar to that of Simon in examining group perception and decision making (e.g. Cohen 1985; Dunn and Ginsburg 1986; Holzner 1972; Kelly 1955; Luhmann 1985, 1995). Under this paradigm, perception is viewed as being essentially comparative in nature and determined by one’s identity (frame of reference). Further, groups who share identity on some level are similarly assumed to share a general comparison scheme with which members of that group or system make sense of what they perceive in the world around them (Kelly 1955; Luhmann 1985, 1995). One of the goals of this research was to better define the range of contrasts or comparisons that individuals in Macedonia employ in their political decision making in order to render political identity blocks on the basis of overlap within cognitive frameworks. The political decision making frameworks for various groups in this population were elicited using George R. Kelly’s (1955) repertory grid interviews. Each of the completed repertory grid interviews represents a unique cognitive framework reveals an individual’s perception of Macedonia’s political landscape. These individual frameworks have been analyzed both qualitatively and quantitatively. Quantitative analyses employed generalized procrustes analysis (GPA) (Gower 1975, Dijksterhuis and Gower 1992), which mathematically compares respondents by means of their ratings of public figures. Qualitative analyses in the form of semantic comparison of the individual constructs (descriptive terms) were used to compare the public figures. Cluster analysis (multidimensional scaling (MDS)) was employed to analyze how constructs are shared within the population. Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
4
Self, Group, and Country Philip Murphy
Much of the repertory grid method’s richness is derived from comparisons within and across populations that are made possible by means of quantitative measures of individual respondents’ cognitive characteristics. It is possible to discern commonality of perception within or among some social, ethnic, or other groups according to the particular construct or group of constructs that group members tend to include in their perceptual framework. When analyzed through GPA, it is also possible to provide a quantitative measure of group consensus in regard to how the individuals that comprise a group construe similarities and differences within a given set of stimuli. Overall, the combination of quantitative and qualitative measures provides a rich, detailed, and measurable description of identity frameworks that traditional hermeneutic and ethnographic analyses can not replicate.
II.
Methodological Theory Review a. Measuring Perception and Identity
Rather than begin to define the constructs and comparisons in use from the state or community level and extrapolate downward, this research instead begins at the individual level in order to elicit the components of the comparison structures that comprise individual actors’ frameworks and then extrapolate upward, to the society level. These comparison structures are essentially communicable maps that depict some aspect of how individual observers experience reality (Holzner 1972). Identity groups therefore make sense of the world around them through comparison structures that tend to have a good deal of overlap, originate through experience, and are reinforced through communication (Luhmann 1995). Such identity groups are not static and are not monolithic, which is to say that any given individual may have a number of levels of identity (e.g., parent, citizen, employee). It should be additionally noted that, because individuals acquire experiences differently, even individuals who superficially belong to the same community may still perceive symbols and events in completely different ways (Cohen 1985, Kelly 1955). In essence, group identity is constituted, at least in part, by a shared frame of reference and reinforced through dialogue. For some, such as Niklas Luhmann (1995) the most important factor in determining one’s belonging in an identity group was a shared frame of reference. These referential frameworks allow members of identity groups to interpret the world around them and make decisions based on those interpretations. In order to better investigate this sort of Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
5
Self, Group, and Country Philip Murphy
collectivist perception and decision making, it is necessary to investigate how decisions are defined at the level of the individual actor. To accomplish this, Luhmann (1995, p. 242) recommended employing George R. Kelly’s (1955) personal construct psychology as a well suited and appropriate method of eliciting the constructs that comprise an individual’s interpretive framework. Kelly’s repertory grid method is a semi-structured psychological interview technique that is specifically designed to elicit culturally relevant frames of reference in a manner that minimizes the impact of the researcher’s own cultural biases. Whereas the grid technique can provide fairly sophisticated data on individual perceptions, its application is not overly complex. Variants of the method have been standardized in the form of computer-based applications, making interviews conducted in this manner much less labor-intensive and costly than similar numbers of open-ended or unstructured interviews. When combined with more traditional ethnographic techniques, the repertory grid method allows a researcher to inductively draw out many of the characteristics that constitute a given level of identity and its reinforcing mechanisms. It additionally minimizes the impact of respondents who offer only what they want the researcher to hear, a common concern of Balkan political research. As a result, the repertory grid offers a relatively resilient method that is less subject to concerns of interpretation and reliability than many other survey or interviewing techniques. Further, when used in combination with clustering algorithms, such as MDS, multiple individual frames of reference can be compared for overlap/similarity in order to provide a representation of what are essentially collective frames of reference. Such a societal ‘cognitive map’ may then be used as a starting point from which to test assumptions that are often made about the perceptions a particular group or society. To this end, individual repertory grid interviews were conducted to elicit the cognitive constructs that are employed when individual respondents make comparisons between public figures. While most contemporary research into identity and political decision making tends to take the form of surveys that rely on researcher-defined constructs or hermeneutic evaluations of what subjects are willing to state publicly, the repertory grid method (Kelly 1955) provides respondents with an opportunity make evaluations on their own terms, using their own comparative constructs, and differentiating quantitatively between the elements they are evaluating. The advantage that the repertory grid offers over the aforementioned others is that it does not impose the researcher’s expectations upon respondents. Rather than ask a respondent to Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
6
Self, Group, and Country Philip Murphy
rate decision elements in terms that are meaningful to the researcher’s theoretical framework, the respondent is instead asked to provide his or her own theoretical framework and use that to differentiate between the events or personalities in question. Repertory grid methodology is part of a family of constructivist methods that includes Q sort (Stephenson 1953) and semantic differential (Osgood, Suci, and Tannenbaum 1957). Though originally introduced for use in psychotherapy, variants of the repertory grid have since been employed in marketing research (Marsden and Littler 1998; Steenkamp et al 1987) and the food industry (Deliza et al 1999; González-Tomás and Costell 2006; Meilgaard et al 2001) to elicit consumer perceptions of products; in occupational research (Crump et al 1980; Eden and Jones 1984) to identify organizational stresspoints and problem construction; in management research (Calori et al 1994; Kearns 1995; Oppenheim et al 2003) to investigate decision making and information use among managers; in organizational research (Dunn and Ginsberg 1986; Goodwin and Ziegler 1998; Locatelli and West 1996) to asses culture and cognition in organizations; in public opinion research (Frewer et al 1997); in research into identity formation (Stojonov et al 1997); and in other applications where a comprehensive understanding of an individual’s or group’s conceptualization of a topic is deemed important (see Fransella et al 2004). Although a repertory grid interview can be administered in a wide variety of variations (Fransella et al 2004), all types consist of three main components: (1) elements to be compared, which can take the form of people, events, places, or other concrete and discrete stimuli; (2) bipolar constructs, used to differentiate between elements; and (3) ratings, to indicate to which pole and/or by what degree an element is characterized by a construct. A popular form of the repertory grid, as employed in the current study, consists of presenting the respondent with a group of elements (usually 10-15) three at a time and asking them “in what way are two of these [public figures] the same and different from the third?” The respondent will then supply their construct (e.g. corrupt…honest, or for a mono-ethnic state…for a multiethnic state) and rate each element from the entire group according to the comparison structure they have supplied. This process is subsequently repeated with other triads of elements until the respondent can supply no new constructs. The data that results from a repertory grid interview is organized in matrix format (Figure 1), providing a graphic representation of an individual’s perceptual framework as it relates to a given set of stimuli. The data consist of a list of elicited, qualitative constructs that frame an Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
7
Self, Group, and Country Philip Murphy
individual’s perceptions and provide quantitative measurement of how elements (i.e., public figures) are perceived in light of the respondent’s comparative constructs, and in relation to each other. Each respondent’s matrix of constructs and comparison measurements is idiosyncratic, as it is inductively constructed from that individual’s experiences and observations and is used by that individual to model what they would expect to occur, given an encounter with similar personalities, events, or situations in the future. It has also been argued that these systems are influenced or colored by the communities to which an individual belongs, even while they determine the communities to which an individual may desire to belong (Shaw 1985).
Figure 1: An example of a repertory grid matrix
The elicited constructs that comprise a repertory grid may be analyzed semantically at the level of the individual, to give a representation of the qualitative comparisons that comprise a respondent’s political identity. Also, given the assumption that identity groups share similar cognitive characteristics (Kelly 1955, Luhmann 1995), these data can also be analyzed at the society level by comparing individuals’ construct sets (Shaw 1985) with clustering algorithms (e.g., MDS) to delineate such groups. Identity groups may then be further assessed to discern to what extent individuals in a group actually differ or agree in their perceptions of the same stimuli. A measure of shared perception may serve as an indicator as to whether emergent Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
8
Self, Group, and Country Philip Murphy
groups, or clusters, are viable representations of shared identity, defined here as groups of individuals whose frames of reference yield similar perceptions.
b. Analytic Approach The analyses in this study are conducted in two stages. The first stage of analysis tests established theories of identity and decision making in Macedonia, and the second stage tests a more generative approach to delineating identity groups. Both stages of analysis are therefore predicated on establishing a baseline consensus measure by using Generalized Procrustes Analysis (GPA; Gower 1975) to provide a measure of within-group consensus for the entire sample (CP0). The establishment of a baseline measure provides a reference point, from which to qualitatively compare all further partitioning approaches. GPA is a technique that is particularly well suited to the task of measuring the level cognitive agreement that is present within groups, as indicated by repertory grid data. GPA was developed as a mathematical technique for assessing similarities and differences between individuals who are employing varied sets of evaluative criteria in differentiating between elements in a given set of stimuli (Djiksterhuis and Gower 1992). Any number of matrices (i.e., repertory grids) may be analyzed collectively, and the dimensions (i.e., the number of rows and columns) of individual matrices can vary, providing they all share at least one dimension (either the same number of rows or columns; Grice 2006). In terms of this study, GPA can be employed to analyze the repertory grid matrices of any number of individual respondents together as a group, and each individual in the group can provide any number of constructs, provided that all respondents are assessing identical elements (i.e., public figures), and that the elements are arranged identically from matrix to matrix. In order to measure the communality between a set of grids, the GPA algorithm involves minimizing spurious differences in rating style or the range and number of constructs used from individual to individual before averaging the grids overall to reach a consensus configuration (i.e., a centroid grid that summarizes the average responses within the group). Differences are minimized through a standardization process of iteratively scaling, rotating, and centering individual grids until arbitrary differences are minimized without affecting the relative differences among the elements (in this case, public figures). During this process, the algorithm will dimensionally scale grids by appending columns of zeros to resize grids with fewer Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
9
Self, Group, and Country Philip Murphy
constructs to the same dimensions as the largest grids (i.e., those with the most constructs). The scaling process prevents larger grids from unduly influencing the results while offering no harmful side effects to the overall analysis (Djiksterhuis and Gower 1992). Grids are additionally rotated to account for respondents using different constructs or groups of constructs to describe the same assessment, and the scores in individual grids are centered to account for variation in scoring styles. (Steenkamp et al 1987; Grice 2006) It is important to note that GPA does not provide an indication of how much the members of a group agree in their preferences, but merely provides a summary of how the respondents in a group perceive the relationships between elements. What GPA can indicate is to what degree individual grids in a set agree on the relative amounts of similarity and difference between each of the public figures, as revealed by the consensus configuration. High amounts of agreement may be interpreted as suggesting that respondents’ assessment criteria are based on similar standards. A principal components analysis can additionally be carried out on the consensus configuration in order to graphically depict relative distances between elements, or relative distances between respondents. For example, individual respondents may agree that public figure A and public figure C are similar, and that both are very different from public figure B, but that is no indication that the respondents agree in their preference for said figures. Certain respondents may prefer figures A and C, while viewing B as an unappealing opposite, whereas others may well view the situation in converse and both would still be in agreement with the consensus configuration. Agreement between grids in a set is indicated by the consensus proportion (CP), which may take a value ranging between 0 and 1.0, with 1.0 indicating perfect agreement. The larger the consensus proportion, the less respondents vary in regard to their representation of the relative similarities and differences between public figures. Another way to view the consensus proportion is therefore as an indicator of the amount of variance between grids that is accounted for by the consensus grid. Lastly, because the standardization process of scaling, rotating, and centering matrices is an iterative optimization-based procedure, the significance of the consensus proportion can also be tested under the null hypothesis that the consensus configuration is only an artifact of the statistical technique. Because GPA is so efficient in reducing differences between grids, it is important to check whether an observed CP is actually outside the bounds of what may be observed from a randomly generated set of grids. The hypothesis is tested by means of a Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
10
Self, Group, and Country Philip Murphy
randomized permutation test, where data in the set are randomly reassigned and reanalyzed over a large number of iterations. Stage 1 of analyses tests established hypotheses that assume that political identity in Macedonia can be characterized in terms of either ethnicity or rural/urban designations. If the ethnicity hypotheses are correct, then ethnic groups think alike more often than the population at large. Therefore, the measure of within-group agreement (i.e., consensus proportion) for the ethnic Macedonian partition of the sample (CPM) and that of the ethnic Albanian partition of the sample (CPA) should each be greater than that of the entire sample (CP0) when the population is divided by ethnicity (CPM>CP0 and CPA>CP0). Similarly, the rural/urban hypothesis assumes that urban areas create greater opportunity for cross cultural or interethnic social communication, resulting in greater consensus among ethnically disparate urban dwellers than among those residing in rural areas. Therefore, consensus proportion for urban populations (CPU)1 would be expected to be greater than that of the population at large, and those of rural populations (CPR) would be expected to be less than those of urban populations (CPU>CP0 and CPU>CPR). Stage 2 breaks away from established models in order to take an inductive approach to delineating political identity in Macedonia. To do so, the sample is partitioned according to shared frames of reference and the resulting partitions are analyzed in the same manner as those in Stage 1. Further analyses are then conducted in order to provide a more detailed portrayal of the groups identified through this procedure. In following with the idea of shared frames of reference being a primary determinant of some level of shared identity (Cohen 1985; Holzner 1972; Kelly 1955; Luhmann 1985, 1995), individuals who share a preponderance of constructs are interpreted as sharing a frame of reference. To that end, construct data from all individuals in the sample were standardized in a process that involved translating each construct from Albanian and Macedonain into English and cross-checking for equivalence of meaning before organizing them into construct classes. An association matrix was then assembled, with respondents and emergent construct classes as the axes. The matrix of respondents and constructs was assessed using MDS in order to discern clustering patterns. As with the partitions in Stage 1, the matrices within each cluster were 1
“Urban” dwellers were defined as those respondents who listed their primary residence as being Skopje, by far the largest and most cosmopolitan population center in the Republic of Macedonia.
Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
11
Self, Group, and Country Philip Murphy
analyzed using GPA to assess within-group perceptual consensus. For the emergent clusters to be considered viable representations of identity groups that share frames of reference, it was expected that each each of n cluster partitions would have a greater proportion of within-group agreement than what is noted when similarly assessing the entire sample. Specifically, the measure of consensus for each of n coherent clusters (CPx, where x = {1, 2, …, n}) is expected be greater than that of the entire sample population (CPx>CP0, where x = {1, 2, …, n}). Provided that emergent clusters were judged to be viable representations of identity groups, GPA output and frequency data were employed to further characterize each. The examination of GPA residuals for each cluster reveals which public figures were the subject of the greatest or least amount of disagreement within the partitioned group. It is possible to see where members of a particular cluster disagree most strongly in relation to specific public figures by identifying the highest value(s) among GPA residuals for each group. In addition, each partitioned cluster was conservatively described according to which constructs were employed by at least twenty-five percent of the respondents in a given cluster. While any actual correlation between a cluster’s relative unity or disunity over a particular public figure and the framework in use by that cluster is difficult to ascribe from the above approach, it is possible to make some casual, qualitative observations. In light of this, relative similarities and differences between individual respondents in each cluster may also be discerned. To do so, a principal components analysis was run on the procrustes statistics for all individuals in each set and the results were graphed to provide a visual depiction of the relative differences in the preferences among individuals in each group.
III.
Data Collection a. Sample
Macedonia’s university students were selected as the study population. As a group that is in the process of transitioning into the country’s educated elite, they are representative of the future of their country’s public- and private-sector elite while they are still subject to the current opinions of the wider population. The students of Macedonia’s universities are far more likely to participate at higher levels in government and society than their peers who do not attend a university. As such, they may provide indications of the future of political cognition in Macedonia and of future trends for political identities. Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
12
Self, Group, and Country Philip Murphy
The cognitive frameworks for political decision making were elicited from 109 respondents, 39 females and 59 males (mean age = 20.7 years, SD = ± 2.2 years), recruited from Macedonia’s four state-accredited undergraduate universities. Of this group, 54 were selfreported as ethnically Albanian, 54 ethnic Macedonian, and one respondent fell outside both categories. Upon completion of the repertory grid interview, respondents were given a token compensation for their participation. Respondents were recruited at random in common areas and cafeterias on university campuses. To ensure sampling diversity and to proactively curtail the practice of focusing their recruitment efforts on their own friendship groups, interviewers were required to recruit respondents according to a randomly assigned shoe color while on or nearby to their home campus.
b. Elicitation Procedure For the repertory grid interviews, respondents were asked to make comparisons between ten public figures (Table 1) that were easily recognizable to Macedonian citizens for their effect on Macedonia in the national or international arenas. Although repertory grid interviews can be conducted using a wide range of different types of elements, provided that they are concrete, discrete, and readily familiar to the individuals being interviewed; it is also imperative for the investigator to choose elements that are somehow representative of the phenomenon being investigated (Fransella et al 2004). For this reason, various symbols of politics and society in Macedonia were investigated for their suitability in this investigation (e.g., laws, public and private institutions, institutional reforms), but none were as easily recognized and familiar to society at large in Macedonia as the public figures themselves. In keeping with this line of reasoning, the public figures selected for use as elements in the repertory grid interviews were chosen because they have each had some effect on Macedonia and were judged by informal focus groups as most likely to be familiar to the average citizen there. To enhance respondents’ recognition of these individuals, a standard set of cards was also prepared and presented as visual stimuli during interviews. Each card bore the picture, name, and brief title of a public figure, translated into both Albanian and Macedonian.
Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
13
Self, Group, and Country Philip Murphy
Public Figures Arben Jaferi Ali Ahmeti Kiro Gligorov Branko Crvenkovski Vlado Buckovski Lupco Georgievski Nikola Gruevski Havier Solana Mother Teresa George Bush
Leader of the ethnic Albanian DPA party Leader of the ethnic Albanian DUI party First President of the Republic of Macedonia President of the Republic of Macedonia Prime Minister of the Republic of Macedonia Leader of ethnic Macedonian VMRO party Leader of ethnic Macedonian VMRO-DPMNE party Chief of External Policy in the European Union Roman Catholic Nun President of the United States of America
Table 1: Grid Elements: Public figures who have had an effect on Macedonia and are likely to be familiar to the average citizen there.
Repertory grid interviews were conducted by trained student interviewers, using an interactive, computer-based elicitation format under the supervision of the primary investigator. The Rep IV interview program (Gaines and Shaw 2005) was reconfigured, with all written prompts and sub-routines translated into both Albanian and Macedonian (Cyrillic font) so that respondents would be able to follow along without need for translation. To the extent that it was possible, every effort was made to ensure respondents’ privacy and reduce extraneous distractions. Respondents and interviewers were seated side-by-side in front of a computer, with the ten cards bearing likeness of the public figures placed directly in front of the respondent. Respondents were then shown the names of three public figures and were prompted to choose two who are alike and yet different from the third. Once the respondent made their selection, they were then prompted to consider how they were comparing the three public figures and type in the word or words that describe each pole of that comparison (e.g., nationalist/non nationalistic). After identifying and entering their bipolar constructs, the respondent was then prompted to rate each of the ten public figures according to the construct they had supplied. Ratings were made on a scale of one to seven, with one being the first of their construct poles and seven being the opposite pole (i.e., nationalist = 1, not nationalist = 7). Once they had had an opportunity to rate all ten public figures, respondents were requested to review their ratings of Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
14
Self, Group, and Country Philip Murphy
each public figure and make any changes that they felt were necessary. Upon completion of the rating phase, the process would begin another iteration with a different triad of public figures. This continued until the respondent stated that they could no longer supply new constructs. In this manner, respondents were able to supply anywhere from 3 to 15 constructs, with a mean number of 8 constructs.
IV.
Results a. Baseline Consensus Measure
The 109 resulting grids were analyzed as a group using GPA to evaluate the degree of similarity in how respondents preceived the ten public figures across the entire sample. The consensus proportion (CP), serves as a baseline, by which to compare any further partitions of the sample. As indicated in the first column of Table 2, the baseline measure (CP0) is 0.63, a permutation test indicates that the value is statistically significant (p<0.01), and the minimum and maximum values that resulted over the 100 iterations of the permutation test were 0.56 and 0.57, respectively. Given the values listed above, it may be conceptually easier to think of the baseline CP as indicating that the amount of within-group consensus in regard to the relative distances between public figures appearing in the study is 0.63, or perhaps that the respondents ‘agree’ 63% of the time.
b. Stage 1: Imperfect Partitions: Ethnicity and Rural vs. Urban In addressing the original hypotheses that Macedonian identity can be divided in terms of ethnicity or by rural/urban designations, the results, presented in Table 2, do not offer a validation of either hypothesis. Only the ethnic Albanian group displayed quantitatively greater within-group agreement in the consensus proportion (CPA: 0.71 vs. CP0: 0.63), whereas evaluating ethnic Macedonians as a group did not offer any apparent difference from the baseline consensus proportion (CPM: 0.63 vs. CP0: 0.63). A similar finding is noted when viewing the consensus proportions of rural (CPR) and urban (CPU) dwellers. Whereas a greater proportion of consensus was noted among rural respondents relative to the baseline (CPR: 0.66 vs. CP0: 0.63),
Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
15
Self, Group, and Country Philip Murphy
counter to expectations, urbanites appear to be less similar in their construal that the population at large (CPU: 0.62 vs. CP0: 0.63).2 University affiliation was also assessed for possible confounding affects. Table 3 presents the consensus proportions for when groups are partitioned according to university to evaluate whether or not the university attended was the strongest factor in shaping cognition within the sample. There do appear to be higher levels of perceptual agreement in University 1 (CPS1) and University 2 (CPS2) relative to the baseline consensus proportion. There is, however, negligible qualitative difference between the baseline CP and the CP for University 3 and University 4. Of the two universities that demonstrated higher levels of agreement, one is a multicultural university located in a predominantly ethnic Albanian area (University 1/CPS1: 0.68), and the other is located nearby and caters primarily to ethnic Albanian students and faculty (University 2/CPS2: CP=0.73). The remaining two universities are primarily attended by ethnic Macedonian students, but have faculty from a variety of ethnicities.
Ethnic M (CPM)
Baseline (CP0) Consensus proportion Permutation test: *p-value (100 reps.) **min/max
0.63 p<0.01 0.56/0.57
0.63 p<0.01 0.52/0.53
Ethnic A (CPA)
Urban (CPU)
0.71 p<0.01 0.62/0.63
Rural (CPR)
0.62 p<0.01 0.54/0.57
0.66 p<0.01 0.60/0.62
Table 2: GPA consensus proportions for baseline (entire sample), ethnic groups, and rural and urban designations * reflects the frequency with which the CP of randomly reassigned data exceeds the observed CP over 100 iterations of the permutation test ** minimum and maximum CPs of randomized data over 100 iterations
Baseline (CP0) Consensus proportion (CP) Permutation test: *p-value (100 reps.) **min/max
0.63 p<0.01 0.56/0.57
Univ. 1 (CPS1) 0.68 p<0.01 0.62/0.63
Univ. 2 (CPS2)
Univ. 3 (CPS3)
0.73 p<0.01 0.61/0.64
Univ. 4 (CPS4)
0.63 p<0.01 0.51/0.54
0.64 p<0.01 0.52/0.56
Table 3: GPA consensus proportions for university groups * reflects the frequency with which the CP of randomly reassigned data exceeds the observed CP over 100 iterations of the permutation test ** minimum and maximum CPs of randomized data over 100 iterations
2
One caveat to keep in mind in regard to the use of the consensus proportion for purposes of comparing two populations is that it is not currently possible to test the statistical significance of the difference between two groups. It is therefore advisable, for inferential purposes, to construct a survey that is grounded in the constructs and findings of the repertory grid interviews, as was done in later sections of this research study (Murphy 2007).
Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
16
Self, Group, and Country Philip Murphy
c. Stage 2: Frames of Reference that May be Shared Across Ethnicities The 109 repertory grids were partitioned according to shared frames of reference. In so doing, grid data was pooled and the resulting 877 constructs were cross-translated and organized into seventy-seven construct classes according to equivalence of meaning (Table 4). Data from this procedure was organized into a respondent : construct association matrix and was then visualized using MDS (Figure 2). The partitions that emerged from this process consisted of four clusters of respondents that each appear to share a characteristic group of constructs and an outlier group that bore very little commonality in their use of constructs.
formation of the state of Macedonia
identity issues
violence manipulates diplomacy
M28M3
M59F4 M08M1
foreign vs. domestic issues A08F1
M29F3 A55M2
A26M1 A04M1
A01F1
development
A32M2
M47M4
Religion multiethnic
A57F2A50F2
stability reforms influence
A52M2
A11M1
M58F4 A34M2
A54M2
A31M2
A36F1 A22M1 M07M1 A21M1 A12M1A58F2 A29F1 A27M1 A17F1 M15F1 A33M2 M38F3 A09F1 M31F3 M44F3 A10F1 A24M1 M41F3A46M2 M13F1 M02M1 M12F1 A07M1 A02F1 A13M1 A35F1 administrative reforms M42F3
education
M17M4 transparency ability / effectiveness
equality / proportionality results international relations
humanism
cooperative tendencies
self-interest
Bureaucratic politics
issues pertaining to "nations" M32F3
war / conflict
M56F4 A39M1
A40M1
Region
loyal
M16F1
liberal / conservative
employment M27M3
M25M3
A41M1
A45M2
M21M4
M06M1
A38M1
corruption
M19M4
experience
M09M1
local / global power nationalist nepotism communist leadership skills special vs. general interest
balance
Effort M48M4 Western interests radical
A49M1
Promises
population in general
M10M1
M57F4
M26M3
A03M1
problem solving
Grass roots
A51F1 M11F1
Indirect / direct
M40F3
A20F1 A47M2
A28F1 A14M1 A15M1
M55F4
M01F4 A56F2
M45F3
A43M2 economics Macedonia in general For/against Albanians
State University of Tetovo Ohrid agreement democracy
minorities
A37F2
M50F4
A30M2 A19F1
peace Issues uniting or dividingEU Macedonia
Kosovo
A18F1
A44M1
A23M1
M35F3
M37F3
M20M4
A53M2
nation vs. people
M22F4
bravery Anticipation
A42M1
qualifications
promising M33F3
communicator Strategy
Dependent/Independent
political / non-political T01M1
M52M4
popular support M43F3 M36F3
M49M4 M24M3 M30F3M23M3
M39F3
legalGovernment aspects leadership / opposition
M04F1
social issues old style / new style
A48M1
M18M4
M60F4
similarity
autocratic tendencies
patriot
Charisma
M03F1
M46M4
Intolerance/bias
Figure 1: Visualization of respondent data after MDS
Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
17
Self, Group, and Country Philip Murphy
Construct Category Humanism Nationalist For/against Albanians Economics Self-interest Corruption EU Issues Local / global War / conflict Influence Experience Macedonia in general Liberal / conservative Equality / proportionality Stability Ohrid agreement Qualifications Uniting or dividing Macedonia Issues pertaining to "nations" Leadership skills Radical Peace Development Multiethnic Reforms Results International relations Power Special vs. general interest Nepotism Promises Democracy Ability / effectiveness Communist Effort Region Strategy Communicator Grass roots
freq. 38 35 33 29 29 26 25 24 20 19 18 18 17 16 16 15 15 15 14 14 14 13 12 12 12 12 11 11 11 10 9 8 7 7 7 7 7 6 6
Construct Category Indirect / direct Political / non-political Western interests Administrative reforms Government leadership / opposition Nation vs. people Popular support Population in general Charisma Legal aspects Minorities Old style / new style Patriot Problem solving Religion State University of Tetovo Transparency Autocratic tendencies Cooperative tendencies Dependent/Independent Diplomacy Education Foreign vs. domestic issues Manipulates Similarity Social issues Anticipation Balance Bravery Bureaucratic politics Employment Formation of the state of Macedonia Identity issues Intolerance/bias Kosovo Loyal Promising Violence
freq. 6 6 6 5 5 5 5 5 4 4 4 4 4 4 4 4 4 3 3 3 3 3 3 3 3 3 2 2 2 2 2 2 2 2 2 2 2 2
Table 4: List of construct categories and frequency of occurrence
Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
18
Self, Group, and Country Philip Murphy
Entire Sample (CP0) Number of grids Ethnicity: Macedonian Albanian Other Hometown Urban Rural University: SEEU (1) SUT (2) UKIM (3) SKO (4) Gender: Male Female Consensus proportion (CP) Permutation test: *p-value (100 reps.) **min/max
Cluster 1 (CP1)
Cluster 2 (CP2)
Cluster 3 (CP3)
Cluster 4 (CP4)
Outliers (CPOL)
109
41
24
13
17
14
54 54 1
13 28 0
7 17 0
13 0 0
11 6 0
10 3 1
70 39
24 17
16 8
11 2
7 10
12 2
50 18 19 22
27 7 7 0
10 9 2 4
2 0 6 5
7 1 4 5
3 2 3 6
59 39
22 19
11 13
6 7
12 5
7 7
0.63
0.69
0.70
0.71
0.70
0.48
p<0.01
p<0.01
p<0.01
p<0.01
p<0.01
p<0.07
0.56/0.57
0.62/0.63
0.57/0.60
0.54/0.60
0.63/0.66
0.43/0.51
Table 5: GPA Consensus Proportions and breakdowns for Partitioning Clusters * reflects the frequency with which the CP of randomly reassigned data exceeds the observed CP over 100 iterations of the permutation test ** minimum and maximum CPs of randomized data over 100 iterations
When analyzed using GPA, clusters 1-4 were interpreted as being viable representations of identity groups, as the consensus proportion for each cluster (CPx, where x = {1, 2, 3, 4}) was greater than that of the entire sample (CPx > CP0, see Table 5). The one notable exception to this finding is the outlier group, which, true to its nature, actually demonstrates a subtsantially smaller consensus proportion (CPOL : 0.48), indicating less internal agreement. The frequency data in Table 5 further allows for a qualitative revisitation of the hypotheses that were addressed in Stage 1 of analyses. In reference to the urban/rural hypothesis, the clusters that emerged with MDS do not appear to demonstrate obvious separation accordint to their hometown origins, with the exception of one cluster that was primarily (>80%) urban: Cluster 3, which is mono-ethnic and therefore does not refelct the assumption of an environment
Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
19
Self, Group, and Country Philip Murphy
that “crosses cultures.” As for the hypothesis of stark ethnic divisions, again, with the notable exception of Cluster 3, all clusters are multiethnic. In an effort to further contextualize to the emergent clusters, Table 6 provides a list of constructs that are employed by at least twenty-five percent of the respondents in each cluster. This provides a representation of the shared cognitive framework in use by each cluster, with those construct classes that appear most frequently highlighted in orange. Cluster 1, most strongly characterized by references to humanism, is also dominated by references to nationalism, local vs. global issues, and public figures who are either for or against Albanians. Cluster 2, the most strongly Albanian cluster, is also concerned with who is for or against Albanians and is additionally interested in economic concerns. Cluster 3 was comprised entirely of ethnic Macedonians, had the smallest number of constructs in use, and was overwhelmingly dominated by references to nationalism. The final identity cluster, Cluster 4, is primarily associated with references to experience, as well as corruption, and issues pertaining to
Cluster 1
Cluster 2
Cluster 3
Cluster 4
Outlier Group
41
24
13
17
14
Membership: Constructs used by at least 25% of respondents/cluster:
Overall frequency
%
#
%
#
%
#
%
#
%
#
Reforms
12
11%
4
8
33%
0
0
Peace
13
12%
6
6
25%
0
1
Issues Pertaining to "Nations"
14
13%
3
2
1
7
41%
1
Radical
14
13%
1
4
0
6
35%
3
Ohrid Agreement
15
14%
6
2
1
Qualifications
15
14%
1
2
5
0
Equality / Proportionality
16
15%
7
6
25%
0
3
Stability
16
15%
8
8
33%
0
0
Liberal / Conservative
17
16%
6
2
1
6
35%
2
53%
1
0
1 38%
6
5 35%
0 0
18
17%
6
2
0
9
Influence
19
17%
7
10
42%
0
1
War / Conflict
20
18%
10
7
29%
0
3
0
Local / Global
24
22%
17
41%
0
2
1
EU Issues
25
23%
12
29%
7
Corruption
26
24%
12
29%
4
Economics
29
27%
14
34%
11
46%
0
0
Self-Interest
29
27%
12
29%
6
25%
2
6
For/Against Albanians
33
30%
16
39%
15
63%
0
1
1
0
1
3
1
29%
31%
0
1
3
7
Nationalist
35
32%
24
59%
0
10
Humanism
38
35%
29
71%
5
0
77%
36%
1
Experience
4
%
1
5 41%
4 35%
36%
0 29%
3
Table 6: GPA Consensus Proportions and breakdowns for Partitioning Clusters Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
20
Self, Group, and Country Philip Murphy
“nations.” Lastly, true to its nature, the outlier group has no constructs that are shared by a minimum of forty percent of its members. A look at the residuals in Table 7 reveals distinctive patterns of agreement and disagreement that are unique to each cluster. For respondents in Cluster 1, grid element Arben Jaferi, a prominent Albanian politician, was the subject of the most disagreement, whereas grid elements George Bush and Mother Teresa were the subjects of relatively more unanimity.3 This information can be further supplemented by looking at the list of constructs that were most frequently occurring in each cluster (Table 6). For example, Cluster 1, the only cluster to view Mother Teresa with relatively little disagreement, is most strongly dominated by references to humanism, a term that, contrary to its usage in the US, is employed interchangeably with “humanitariansim” in Macedonia. Conversely, Cluster 4, which is most strongly associated with
Elements Kiro Gligorov Ali Ahmeti Lupco Georgievski George Bush Branko Crvenkovski Arben Jaferi Nikola Gruevski Mother Teresa Havier Solana Vlado Buckovski Consensus proportion (CP) Permutation test: *p-value (100 reps.) **min/max
Sample population 3.85 3.55 3.83 3.35 3.28 3.83 3.79 4.21 3.49 3.46
Cluster 1 2.95 3.29 3.27 2.65 3.06 3.67 3.19 2.79 2.94 2.80
Cluster 2 2.92 2.76 3.70 3.20 2.80 3.34 2.75 3.55 2.77 2.67
Cluster 3 3.05 2.12 2.23 2.06 3.00 3.47 3.19 3.14 3.89 3.07
Cluster 4 3.69 2.45 3.20 3.01 2.32 2.65 2.75 4.27 2.79 2.85
Outliers 5.98 5.47 4.97 4.84 5.28 4.41 5.89 5.77 3.75 5.35
0.63
0.69
0.70
0.71
0.70
0.48
p<0.01
p<0.01
p<0.01
p<0.01
p<0.01
p<0.01
0.56/0.57
0.62/0.63
0.57/0.60
0.54/0.60
0.63/0.66
0.43/0.51
Table 7: GPA residuals from analysis of Partitioning Clusters * reflects the frequency with which the CP of randomly reassigned data exceeds the observed CP over 100 iterations of the permutation test ** minimum and maximum CPs of randomized data over 100 iterations
3
Mother Teresa, an ethnic Albanian christian, was born in what is now the capitol city of Macedonia. To those unfamiliar with Macedonia’s politics, it may at first appear surprising that Mother Teresa seems to be the subject of much of the disagreement found in all except for one of the above clusters. Though, when one considers that ethnic Albanians in Macedonia are primarily Moslem and ethnic Macedonians are primarily Christian, it is perhaps much less surprising that such an extraordinary figure should become the subject of disagreement. When one considers that neither ethnic group can lay complete claim to one who is otherwise universally regarded as a positive figure, the idea that Mother Teresa would inspire mixed feelings may begin to make more sense.
Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
21
Self, Group, and Country Philip Murphy
references to experience evinces a relatively large amount of disagreement over the political outsider among the public figures, again, Mother Teresa. The table of GPA residuals serves to reinforce the important caveat that, although the consensus proportion functions well as an indicator of overall cognitive agreement in a group or cluster as to the relative similarities and differences among the stimuli (i.e., public figures), it remains a very general measure. Individual construal is still very subjective to the likes and dislikes of the individual. Actual preference in particular, is something that should be addressed when assessing a group’s homogeneity. To map relative differences in preference among respondents, a principal components analysis (PCA) was run on the procrustes statistics for each group and the results were plotted in two dimensions. Such “PCA Plots” demonstrate that, while respondents tend to ‘agree’ on the relative distances between the elements (public figures) in question, they do not appear to agree in their preferences thereof. In a typical example (Figure 3), ethnic Albanian respondents (those beginning with the letter ‘A’) tend to be clustered to the left of the y-axis, while ethnic Macedonian respondents (beginning with the letter ‘M’) tend to cluster to the right of the y-axis.
Figure 3: PCA Plot for Cluster 2
Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
22
Self, Group, and Country Philip Murphy
V.
Discussion and Context
This research made it possible to test the two-group hypothesis which assumes that ethnic divisions between the people of Macedonia tend to preclude the formation of common identity there, and the three-group hypothesis that adds dense urban areas as a bridge between cultures. These tests revealed that the two- and three-group approaches do not provide optimal representations of identity groups in Macedonia. Consensus scores resulting from applying GPA to the indicated partitions bore little or no relation to the models hypothesized by either ethnic, rural/urban, or university divisions.
The results from testing rural/urban hypothesis were
particularly notable, in that they ran counter to expectations in both stages of analysis. In the first stage, the group that was expected to demonstrate greater consensus, the urban group, actually produced a consensus proportion that was smaller than that of the entire sample. Once the first stage of analyses was complete, such theories were set aside in order to explore the benefits of tackling the question using a more generative, bottom-up approach. A more meaningful way of partitioning the sample, the constructivist approach, provided better guidance. In the second stage of analyses, the partitioning clusters that were determined through visualizing semantic frames of reference appear to to be much better suited to explaining political identity than the preceding partitioning models. When analyzed using GPA, all four clusters that were revealed through multidimensional scaling displayed consensus proportions that were qualitatively greater than that of the sample at large. The only exception was the outlier group, which can be expected to demonstrate less consensus. The relatively greater consensus may be interpreted as suggesting that, within each cluster, respondents’ assessment criteria are based on similar standards. Such findings cannot be stated without some words of caution. It was possible to further contextualize the emergent clusters using GPA and frequency data. Such inquiry revealed that, although the clusters share similar evaluation criteria in their frames of reference and relatively greater internal consensus than is evident in the sample at large, ethnicity ultimately does play a part in individual preferences relating to specific public figures. The project outlined above was designed as an inductive inquiry into the state of identity in the Republic of Macedonia. Using the thoughts and perceptions of university students throughout Macedonia’s university system, it was possible to measure the similarities and differences in how individuals, regardless of variations in language, culture, or other social and demographic characteristics, perceived public figures that have an affect on their country. Those Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
23
Self, Group, and Country Philip Murphy
who saw the same relative similarities and differences were understood to hold some common frame of reference vis-à-vis public figures. Using the bottom-up approach to ascertaining shared cognition, it was possible to discern groups of individuals who use many of the same criteria to come to very similar conclusions about how public figures differ in respect to each other. It was also apparent that members of any particular cluster may still occupy opposite ends of the spectrum in their preference for a given public figure and that, in this case, ethnicity appears to be a strong influence in regard to ultimate preference. While ethnicity is certainly a strong factor in determining an individual’s preference for a particular set of public figures over another, it is not the only factor that should be taken into consideration when working with this population. There are also groups that cross ethnic lines, whose political interests focus on qualitative aspects of governing that fall outside of ethnic categorization. The proposed analytic procedure provides a much richer and more detailed description of how Macedonia’s university students think when differentiating between public figures than would otherwise be possible through casual observation, hermeneutic evaluation, or researcherdefined inquiry. The subject of Macedonian political culture has suffered thus far from a lack of generative theory-building that is grounded (Glaser and Strauss 1967) in data on how its citizens interpret their surroundings. With all of the changes that have taken place in Macedonia since its independence, it is reasonable to assume that students in today’s Macedonian universities are coming of age in an environment that is sharply different from that of their parents. In addition to the changes in government, the present generation of university students is growing up with unprecedented access to communications and media. What sorts of changes are likely to occur? While the nationalism and ethnic conflict model was not best suited for explaining the differences in cognition above, ethnicity is clearly a factor in decision making throughout the country. Tangible evidence of that is given by the relatively frequent use of the for/against Albanians construct in two identity clusters and the similarly frequent application of references to “nations” and nationalism in the others. Still, it is similarly apparent that ethnic differences in Macedonia are not monolithic and do not comprise ‘primordial’ divisions. Rather, there are more subtle differences among the identifiable identities that are evident there. In fact, all but one of the identity clusters delineated above cross ethnicities, even if few of the members of those identity groups would be likely to realize or
Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
24
Self, Group, and Country Philip Murphy
necessarily be willing to openly acknowledge that they may share characteristics with individuals from ethnicities other than their own. The generation to which the respondents belong had, and to some degree still has, the unfortunate fate of witnessing some of the most virulent internal discord in recent memory. The associations that they carry away from that can only color their perceptions. However, there have also been an enormous number of other events and occurrences taking place in Macedonia that add positively to the character and complexity of the country and its inhabitants. Unifying events such as former President Boris Trajkovski’s death and Macedonia’s tie game against England in football (soccer) were strong reminders that there is still much (both positive and negative) that crosses cultures in Macedonia and maybe only someone who lives or has roots there can share. Research of this sort provides information that is likely to be of value to a policy professional who is interested in crossing ethnic and political lines. But it could also be of benefit to a public figure who desires to make of his- or herself a bridge across the ethnic divide. After all, though the multiethnic clusters may be divided in the sense that they choose their representation by ethnicity, they are also united in the issues that most concern them regarding how those representatives focus their time and energy. Policy practitioners can benefit by knowing more about the knowledge in use by the population (Lindblom and Cohen 1979). Similarly, knowing how those knowledge frameworks can be influenced has strong potential as an effective tool for public policymakers (Lakoff 1987). Macedonia, like much of Eastern Europe, suffers from a dearth of research that is grounded or inductive. While this research is meant as an introductory foray into the depth and complexity of the frameworks in use by Macedonian nationals as they interact and make decisions, it is clear that inquiry of this sort can be a valuable aid to our understanding how identity and perception relate to macro-level behavior in that country and elsewhere.
Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
25
Self, Group, and Country Philip Murphy
Bibliography Borgatti, S.P., M.G. Everett, and L.C. Freeman. (1999) UCINET 6.0 Version 1.00. Natick, MA: Analytic Technologies. Calori, Roland, Gerry Johnson and Philippe Sarnin. (1994) “CEOs’ cognitive maps and the scope of the organization,” Strategic Management Journal. 15(6): 437-457. Cohen, Anthony P. (1985) The Symbolic Construction of Community. New York: Tavistock Publications. Crump, John H., Cary L. Cooper and Mike Smith. (1980) “Investigating occupational stress: A methodological approach,” Journal of Occupational Behaviour. 1(3): 191-204. Deutsch, Karl W. (1966) Nationalism and Social Communication. 2nd ed., Cambridge, MA: MIT Press. Djiksterhuis, G.B. and J.C. Gower. (1992) “The interpretation of Generalized Procrustes Analysis and allied methods,” Food Quality and Preference. 3: 67-87. Dunn, William N. and Ari Ginsberg. (1986) “A sociocognitive approach to organizational analysis,” Human Relations. 40(11): 955-976. Dunn, William N., Kearns, Kevin P., Cahill, Anthony G. (1984) “Application of the Policy Grid to Work-Related Frames of Reference.” Working paper KU203, University of Pittsburgh, Program for the Study of Knowledge Use. Eden, Colin and Sue Jones. (1984) “Using repertory grids for problem construction,” The Journal of the Occupational Research Society. 35(9): 779-790. Fransella, Fay, Richard Bell, and Don Bannister. (2004) A Manual for Repertory Grid Technique. 2nd rev. ed. London: Academic Press. Frewer, Lynn J., Chaya Howard and Richard Shepherd. (1997) “Public concerns in the United Kingdom about general and specific applications of genetic engineering: Risk, benefit, and ethics,” Science, Technology, and Human Values. 22(1): 98-124. Gaines, Brian R. and Mildred L. G. Shaw. (2005) Rep IV Research Version 1.12. Cobble Hill, British Columbia, Canada: Centre for Person-Computer Studies. Geertz, Clifford. (1963) “The Integrative Revolution: Primordial Sentiments and Civil Politics in the New States” in Old Societies and New States, C. Geertz, ed., New York: Free Press, pp. 105-157. Glaser, Barney, and Anselm Strauss. (1967) The Discovery of Grounded Theory. Chicago: Aldine. Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
26
Self, Group, and Country Philip Murphy
González-Tomás, Luis and Elvira Costell. (2006) “Sensory evaluation of vanilla-dairy desserts by repertory grid method and free choice profile,” Journal of Sensory Studies. 21: 20-33. Goodwin, Vicki L. and Laurie Ziegler; (1998) “A test of relationships in a model of organizational cognitive complexity,” Journal of Organizational Behavior. 19(4): 371386. Gower, J.C. (1975) “Generalized Procrustes Analysis,” Psychometrika. 49(1): 33-51. Grice, James. (in press) “Generalized Procrustes Analysis,” in N. Salkind. ed., Encyclopedia of Measurement and Statistics. Thousand Oaks, CA: Sage. Grice, James. (2003) Idiogrid: Idiographic Analysis with Repertory Grids Version 2.3. Stillwater, OK: . Holzner, Burkart. (1972) Reality Construction in Society. revised edition. Cambridge, MA: Schenkman Publishing. Kearns, Kevin P. (1984) Sociocognitive Networks and Local Government Innovation. Dissertation housed at the University of Pittsburgh. Kearns, Kevin P. (1995) “Effective nonprofit board members as seen by executives and board chairs,” Nonprofit Management and Leadership. 5(4): 337-358. Kelly, George R. (1955) The Psychology of Personal Constructs. New York: Norton. Lakoff, George. (1987) Women, Fire, and Dangerous Things: What Categories Reveal About the Mind. Chicago: University of Chicago Press. Lindblom, Charles E. and David K. Cohen. (1979) Usable Knowledge: Social Science and Social Problem Solving. New Haven: Yale. Liotta, P.H. and Cindy R. Jebb. (2004) Mapping Macedonia: Idea and Identity. Westport, CT: Praeger. Locatelli, Veronica and Michael A. West. (1996) “On elephants and blind researchers: Methods for assessing culture in organizations,” Leadership and Organization Development Journal. 17(7): 12-21. Luhmann, Niklas. (1985) Soziale Systeme: Grundriße einer algemeinen Theorie. [Social Systems: Blueprint of a general theory]. Frankfurt, Germany: Suhrkamp. Luhmann, Niklas (1995) Social Systems. trans. John Bednarz, Jr., with Dirk Baecker; foreword by Eva M. Knodt. Stanford, CA: Stanford University Press. Meilgaard, Morten, Sarah Bennett and James Murray. (2001) “Sensory technology – its strategic application to brand management,” Technical Quarterly. 38(4): 219-225. Paper Draft: Please do not cite or reproduce without the expressed, 27 written permission of the author.
Self, Group, and Country Philip Murphy
Murphy, Philip (forthcoming, [2007]) Self, Group, and Country: Searching For a Middle Ground among Macedonia’s Emerging Political Identities. Doctoral dissertation, University of Pittsburgh, Pittsburgh. Oppenheim, Charles, Joan Stenson and Richard M.S. Wilson. (2003) “Studies on information as an asset II: Repertory grid,” Journal of Information Science. 29(5): 419-432. Osgood, Charles E., George J. Suci, and Percy H. Tannenbaum. (1957) The Measurement of Meaning. Urbana: University of Illinois Press. Republic of Macedonia (ROM). (2005) Census of Population, Households and Dwellings in the Republic of Macedonia, 2002 – Book XIII. Skopje: State Statistical Office. Shaw, Mildred L. G. (1985) “Communities of Knowledge,” chapter in Anticipating Personal Psychology. Franz Epting and Alvin Landfield, eds. Lincoln, NE: University of Nebraska Press: 25-35. Simon, Herbert. (1957) Models of Man. New York: John Wiley & Sons. Simoska, Emilija et. al. (2004) “Inter-ethnic and Intra-ethnic Dialogue.” Report unpublished at the time of this writing. Steenkamp, Jan-Benedict E.M., Hans C.M. van Trijp and Theo M.M. Verhallen. (1987) “An integrated procedure for building a common perceptual space based on completely individualized data collection,” Contemporary Research in Marketing: Proceedings of the XVIth Annual Conference of the European Marketing Academy. May 31st – June 3rd, Toronto, Canada: York University. Stephenson, William. (1953) The Study of Behavior. Chicago: University of Chicago Press. Todorova, Maria ed. (2004) Balkan Identities: Nation and Memory. Washington Square: New York University Press. Todorova, Maria. (1997) Imagining the Balkans. New York: Oxford University Press. Todorova, Maria. (2005) “The Trap of backwardness: Modernity, temporality, and the study of Eastern European nationalism,” Slavic Review. (64:1), 140-164. Wilmer, Franke. (1997) “Identity, culture, and historicity: The social construction of ethnicity in the Balkans,” World Affairs. 160(1): 3-16.
Paper Draft: Please do not cite or reproduce without the expressed, written permission of the author.
28
