Facet Theory Basic Concepts Erik H. Cohen Bar Ilan University Rome July 2005 3rd International Lab Meeting – Summer session 2005 11th Edition of the International Summer School of the European Ph.D. 1 • Submit a small matrix to convert in hand SSA, according to two rules: inverse and direct • Submit a small number of profiles to be transformed by hand in POSAC 2 Table of Contents The Who Population The What Population The Concept of Mapping Sentence What is a Facet? The Range Definition of theory by Guttman The various components of the research process according to Facet Theory Regionality hypothesis Theoretical Regionalization (I) What is the meaning of each geometrical figure? What is good manager? Our findings Interpretation 3 Extension of the research A new mapping sentence Hypothesis The Findings relating to the second facet Interpretation Conclusion Theoretical Regionalization (II) The axial model The modular model The polar model What is the meaning of each of these three geometrical figures? The axial model The modular model The polar model 4 An Exploratory Mapping Sentence for Assessing Jewish Education Factorial design (R.A. Fisher) Cumulative Knowledge Another example Israel Experience Programs mapping sentence The Monotonicity The research process The Smallest Space Analysis (I, II, III) Bibliography 5 Facet Theory Basic Concepts The Who population We know how to cover a population in such a way that the surveyed sample represents the referenced population. p represents P 6 For this purpose, one may use many methods of sampling : - Quota method - Random - Snowball technique - etc. 7 According to certain rules, we may be even capable of estimating the degree of error in the sampling. For example : + or – 2,5% is a quite good such estimate for surveys in elections. 8 The population may be everything in the world: people of a certain country, trees of a forest, articles published in international conferences, stamps of a certain year / country / language, etc. 9 The What population But there is another kind of population, which is generally less systematically analyzed: the population / universe of contents. How can we know - as we can do it in the Who population - that the survey will represent the What population? What kind of tool could help us in this aim? 10 The bibliography literature survey is obviously the first thing to do. A state of the art survey is surely helpful. It may even lead to discover uncovered issues in the research. 11 The Concept of Mapping Sentence As Louis Guttman claims : “the problem of proposing fruitful facets to the design of the content for a system of variables has generally been found to be a far more difficult one than for that of proposing background variables or experimental conditions for populations” (Guttman, 1992 in Levy 1994). 12 In order to allow a systematic design of a field under investigation, Guttman introduced the concept of a mapping sentence. A mapping sentence, which is a basic device of facet theory, contains a variety of facets. Each facet is one way of classifying the research content. (Guttman 1982) 13 What is a Facet? A facet is one way of classifying the research content. Each facet contains elements. It is a set with a rule in such a way that its elements are both exclusive and exhaustive 14 The Range When we survey a field, any field, we apply some range in our observation. In esthetic judgment, we refer to the extent of beauty in some object. What may define an item as value, is its range from “important” to “not important”. (See Levy 1985 in Canter) 15 The most basic abstract symbolic mapping sentence has the following format: see Levy 1976 P {A} R Population Content facet Range 16 Though the mapping sentence introduces formality, it is a flexible device as it can easily, but systematically, be enlarged or condensed by adding or reducing facets or elements within the facets. 17 Definition of theory by Guttman As some basic concepts are now understood, we may approach the definition of Theory. This definition will transform these concepts into an operational and integrative process of research. 18 "A theory is an hypothesis of a correspondence between a definitional system for a universe of observations and an aspect of the empirical structure of those observations, together with a rationale for such an hypotheses" (Guttman 1973, 1982; in Levy 1994) 19 The various components of the research process according to Facet Theory (I) Definitional system for an Universe of observations Rationale Regionalization An aspect of the empirical structure 20 The various components of the research process according to Facet Theory (II) Definitional system for an Universe of observations = mapping sentence Rationale = facet role Regionalization An aspect of the empirical structure = intercorrelations 21 Regionality hypothesis According to the regionality hypothesis, to each element of the facet considered, there will be a specific and continuous region in a geometric representation of the various items analyzed. There will be a distinction among the various items according to the facet definition of each item. 22 Regionality hypothesis (2) A facet with n elements A map with n regions With a rationale 23 Regionality Theoretical Regionalization (I) Let first consider the smallest facet existing, i.e. a facet with only 2 ordered elements. This facet may be represented in two and only two different geometric ways, according to the regionality hypothesis. 24 First theoretical regionalization of a Facet with two elements All the items of element a All the items of element b 25 Second theoretical regionalization of a Facet with two elements All the items of element a All the items of element b 26 What is the meaning of each geometrical figure? In the first one, the two elements have an equal role / status In the second one, the two elements have a different role / status. 27 What is good manager? A very simple example of mapping sentence.and structural verification through SSA and regionality hypothesis assessment 28 In their work, “The ‘good manager’: masculine or androgynous” (1979), Powell and Butterfield were the first researchers to specifically apply the concept of androgyny to the work setting. Based on findings in non-organizational settings, they hypothesized that the “good manager” would be perceived as androgynous in sex-role identification. 29 This hypothesis was not supported; both male and female business students described a good manager most of all in masculine terms and least of all in feminine terms. A study, twenty years later (Cohen & Bloom 2000), re-explores the same research questions. 30 79 Israeli students completed the Bem Sex Role Inventory (BSRI) (Bem, 1974), a standard tool for measuring femininity, masculinity, and androgyny, both for him/herself and for a good manager. 31 A first very simple mapping sentence has the following format: 32 masculine characteristics of Israeli students consider the feminine management as almost never true to almost always true 33 In the facet content, “masculine” and “feminine” are called facet elements 34 Our findings 35 Interpretation The previous geometric representation of the intercorrelations between the 80 various items shows clearly (there is only one “error”) that the respondents distinguish between male and female characteristics of the good manager. The regions corresponding to the two elements of the Facet are specific and contiguous, without a “mixed” region. 36 Extension of the research As we already said, the mapping sentence is a formal but flexible way to formulate the content of a domain. Now that we have established a first Facet, we may be conducted to enlarge the mapping sentence by a new Facet. As the respondents were asked to designate the characteristics of a good manager in general, but also if they consider themselves as fitting these characteristics, we can add a new Facet, which formulates the person targeted by the definition. 37 A New Mapping Sentence masculine characteristics of Israeli students consider the feminine management regarding himself / herself a manager in general almost never true to almost always true 38 Hypothesis As the first Facet, the new Facet includes only 2 elements. We may expect a modular geometric representation, with the himself/herself items in the center. 39 Findings re to the second facet 40 Interpretation The previous geometric representation of the intercorrelations between the various items shows that the respondents DO NOT distinguish between himself/herself characteristics and the characteristics of a good manager in general. There are not regions corresponding to the two elements of the Facet! The various items belonging to the two elements of the new Facet are “mixed”. 41 Conclusion The interviewees do clearly distinguish gender characteristics. The interviewees do not clearly distinguish between themselves and the good manager. 42 Theoretical Regionalization (II) Let now consider a more complex facet, i.e. a facet with 3 elements. This facet may be represented in three and only three different geometric ways, according to the regionality hypothesis. 43 The axial model First theoretical regionalization of a Facet with three elements All the items of element a All the items of element b All the items of element c 44 The modular model Second theoretical regionalization of a Facet with three elements All the items of element a All the items of element b All the items of element c 45 The polar model Third theoretical regionalization of a Facet with three elements All the items of element a All the items of element b All the items of element c 46 What is the meaning of each of these three geometrical figures? 47 The axial model The three elements are ranged. For instance, from simple to complex (for instance the Bloom taxonomy is expected this model: Cohen, Clifton & Roberts 2001) All the items of element a All the items of element b All the items of element c 48 The modular model The three elements have a different role / status, from a core to the periphery. For instance the intelligence tests analysis shows such a structure, with from the core to the periphery: inference, application, learning Guttman Levy 1991 All the items of element b All the items of element c All the items of element a 49 The polar model The three elements have a partially range format. The life areas appear in such a way (Levy Guttman 1975 Cohen 2000). All the items of element a All the items of element b All the items of element c 50 An axial model example 51 The Bloom Taxonomy Erik H. Cohen University of Bar Ilan, Israel Rodney A. Clifton University of Manitoba, Winnipeg, Canada Lance W. Roberts University of Manitoba, Winnipeg, Canada 52 Previous FA results Clifton, Etcheverry, Hasinoff and Roberts (1996) extracted and identified two dimensions in the cognitive domain, even though they constructed the scale items to include all six of the dimensions in Bloom's taxonomy. The two dimensions that emerged included a Structural dimension and a Functional dimension 53 First SSA of the 36 items Figure 1: Smallest Space Analysis (SSA) of the 36 Items Based on the Pearson Coefficients Matrix Space Diagram for Dimensionality 4. Axis 1 versus Axis 2. 54 MONCO of the 36 items 55 Legend of the 36 items 56 Second SSA of the 30 items 57 An example of exploratory Guttman approach: A Structural Analysis of the Reuven Kahane Code of Informality: Elements Toward a Theory of Informal Education Published in Sociological Inquiry. 2001, Vol. 71, No. 3, Summer, 357-380. 58 1975-1997 After years of reflection and deliberation, Kahane concluded that there are eight structural components of informality. These are: voluntarism, multiplexity, symmetry, dualism, moratorium, modularity, expressive instrumentalism, and symbolic pragmatism. 59 These structural components have not remained constant or consistent throughout Kahane’s career, with both their number and their definitions undergoing modifications over the years 60 The two questions Is the list of 8 components exhaustive? Is there redundancy in the list? 61 R. Kahane’s articles • Structures and Uses of Informal Organizations 1974 • Informal Youth Organizations: A General Model 1975 • Informal Socialization 1988 • Multi-code Organizations 1988 • Tutorial Relations Agencies 1989 • Origins of Post-modern Youth 1997 62 Origins of Post-modern Youth 1997 (1) • Voluntarism a relatively constraint-free pattern of choice (of goals, means, affiliations,) in which the cost of changing one’s mind is minimal • Multiplicity/multiplexity a wide spectrum of activities that are more or less equivalent in value • Symmetry a balanced reciprocal equivalence of principles and expectations in which no party can impose his or her will on another • Dualism the simultaneous existence of different orientations such as ascription and achievement, competition and cooperation 63 Origins of Post-modern Youth 1997 (2) • Moratorium a temporary delay of duties and decisions that allows for trial and error within wide institutional boundaries • Modularity eclectic construction of activity sets according to changing interest and circumstances • Expressive instrumentalism/active expressiveness a combination of activities that are performed both for their own sake and as a means of achieving future goals • Symbolic pragmatism/pragmatic symbolism the attribution of symbolic significance to deeds and/or the conversion of symbols into deeds 64 143212242 243433333 324432234 423224323 542333221 642332222 744444432 811111113 _ 933323311 1022222222 1112111112 1244444223 1315252534 1432343333 1535111123 1613121421 1722434325 1842444444 The data set The observations 1 'tsofe' 2 'wsy' 3 'social' 4 'bne' 5 'tsofe2' 6 'wsy2' 7 'social2' 8 'bne2' 9 'wand' 10 'scout' 11 'komso' 12 'excur' 13 'camp' 14 'polit' 15 'cultur' 16 'sport' 17 'intell' 18 'game' The variables vol mul sym dua mor mod exp pra 65 Basic SSA 66 SSA with lines according to the theoretical/qualitative criteria 67 An Exploratory Mapping Sentence for Assessing Jewish Education Erik H. Cohen Bar Ilan University Shlomit Levy Hebrew University of Jerusalem 68 A educator educatee Population P parent manager community etc ... assesses the effectiveness of the Jewish educational system delivered within a B formal institution, informal general 69 which is C open close to the social environment 70 D children youth to recipients students adults families general by E teacher instructor counselor Rabbi peer parent ... 71 using F cognitive affective instrument al tools 72 G for teaching a specific aspect of Jewish subject matter general H religion heritage history culture society unspecifie d 73 in relation to time period J I ancient Israel concerning modern Diaspora general general 74 in order to K construct improve inspire unspecifie d 75 Jewish L commitment identity identifica tion involvemen t continuity unspecifie d in country (M) 76 R low effectiveness in achieving specified in facets K-L. high educational goals 77 Factorial design (R.A. Fisher) The 11 various content facets can generate no less than 1,679,616 various sentences (as combinations of their various elements). And this is even not the total number as the facet R has also to be considered (after Guttman 1992 in Levy 1994). 78 Carrying out such a total design is generally impossible in practice, and, ways are sought in each case in practice to make only a small sample of observations that will nevertheless suffice to yield essential information desired about the facet. (after Guttman 1992). 79 The facets of a given design are generally but a sample from a much larger set of possible facets, and some selection rule is needed for them 80 Cumulative Knowledge Using a device such as the mapping sentence and Guttman methods such Smallest Space Analysis, the various researchers in a certain field of research may launch a process of an international cumulative knowledge process. 81 Another example In the next slides, we will present another study conducted among participants in summer and winter educational programs for Jewish youth from Diaspora during the last years (Cohen 1995). 82 Israel Experience Programs mapping sentence: satisfaction The extent to which participant (P) evaluates his/her improvment Jewish specific Israeli character of the educational with respect to a aspect of the general general content programs high facilities domain of his/her visiting student program in Israel staff low peer group in general evaluation. 83 In the following slide, appears the geometrical representation of correlation matrix of the 41 variables of the survey as designed following the previous mapping sentence. Two main geometric shapes appear: a polar one (corresponding to the Facet “domain”) and a modular one (corresponding to the “specific / general” Facet) 84 85 The Linear Model The linear model • There is only one way to get the peak of the mountain • The distance from one point to another one must be equal in x and y 86 The Monotonicity • There are many ways to get the peak of the mountain • The distance from one point to another one must not be equal in x and y. In other terms, when x grows, the only request is that y does not regress. The monotone model 87 The Monotonicity The linear model The monotone model 88 The research process Biblio and other devices survey Exploratory mapping sentence Intercorrelations SSA representation Facet analysis Final mapping sentence 89 The research process Biblio and other devices survey Exploratory mapping sentence Intercorrelations SSA representation Facet analysis Final mapping sentence Iterations 90 The Smallest Space Analysis (1) Based on a correlation matrix, points representing the variables are plotted on a cognitive “map” revealing distinct regions of correlated data (Guttman, 1968, 1982; Levy, 1994). By definition, a structure can be found for any data in n-1 dimensions, where n equals the number of items in the correlation matrix. Therefore, the smaller the number of dimensions necessary to discern a structure, the stronger the significance and credibility of the findings. 91 The Smallest Space Analysis (2) SSA analyzes a matrix of correlations between items by graphically representing them as points in a Euclidean space called the "smallest space." The points are plotted according to the principle: the higher the correlation between two items, the closer they are on the map and, conversely, the lower the correlation, the further apart they are. 92 The Smallest Space Analysis (3) The regionality in an SSA map is defined semantically and not in terms of pure proximity. The region is not necessarily a cluster. 93 Bibliography (1) Bem, S.L. (1974). The Measurement of Psychological Androgyny. Journal of Consulting and Clinical Psychology, 42, 155-162. Cohen, E.H. (1995). Toward a Strategy of Excellence, A Systemic Analysis and Policy Research Based on External Variables in SSA. In Facet Theory Analysis and Design (J.J. Hox, G.J. Mellenbergh, P.G. Swanborn, Eds.), University of Amsterdam, 55-62. Cohen, E.H. (2000). A facet theory approach to examining overall and life facet satisfaction relationships. Social Indicators Research, Vol. 51, No. 2, 223-237. Cohen, E. H., R. A. Clifton and L. W. Roberts (2001). The Cognitive Domain of the Quality of Life of University Students: A Re-analysis of an instrument. Social Indicators Research, Vol. 53, 63-77. Guttman, L. (1968). A General Nonmetric Technique for Finding the Smallest Coordinate Space for a Configuration of Points, Psychometrika, 33, pp. 469-506. Guttman, L. (1973). Quoted in Gratch, H. (ed.). Twenty-Five Years of Social Research in Israel. Jerusalem: Jerusalem Academic Press. Cohen, Erik H and Naomi Bloom (2000). Evaluating the “Good Manager” and the Self According to the BEM Sex Role Inventory, in Work Values and Organizational Behavior Toward the New Millennium, Jerusalem. Pp. 77-82. 94 Bibliography (2) Guttman, L and Levy, S. (1975). On the Mutlivariate Structure of Wellbeing, Social Indicators Research, 2, pp. 361-388. Guttman, L (1992). The Mappimg Sentence for Assessing Values, in Levy, S. (1994). Louis Guttman on Theory and Methodology: Selected Writings. NH: Dartmouth U . Guttman, L and Levy, S. (1991). Two Structural Laws for Intelligence Tests, Intelligence, 15, pp. 79-103. Levy, S. (ed.) (1994). Louis Guttman on Theory and Methodology: Selected Writings. NH: Dartmouth U . Levy, S. and L. Guttman (1975). On the multivariate structure of well-being. Social Indicators Research 2, 361-388 Levy, S. (1985). Lawful Roles of Facets in Social Theories, in Canter, D. Facet Theory: Approaches to Social Research, .New York: Springer-Verlag. Powell, G.N. and Butterfield, D.A. (1979). The “Good Manager”: Masculine or Androgynous? Academy of Management Journal, 22, 395-403. 95 Linearity? The zoology as the science of the nonelephants (Stanislaw Ulam) 96