emocion1988 borg facets

7/27/2019 Emocion1988 Borg Facets

1/20

4 ON THE SYMBOLICBASIS OF SHAM EIngwer Borg t.",,:,j,'~lfiO1'r\iJ'University of Giessen. f,)'(fti,~i,Qt~t'"~X$~.1{t'!!':~

.",.. ~,"...o f, ,.;.".."'.",.' t.""Thomas Staufenbiel ."..". ,r'..'" \"""""T-c"".'J,\1University of Giessen \1l!~I',;11:1)'fff}"'rtWf:~*~


2/20

j80 BORG, STAUFENBIEL, CHERERwe do have a methodology to assess ognitive pro cesses ndependent ofverbal report (Scherer, 1986b; Wallbott & Scherer, 1985a; see Overviewof Part 1 in fuis volume). Two major approacheshave been used in theresearch literature concerned with the study of cognitive dimensions ofemotion elicitation: (1) asking subjects about emotion-eliciting events thatreally happened to them and attempting to obtain as much informationas possible about the nature of the evaluation process (Averill, 1982;Scherer, Wallbott, & Summerfield, 1986; Smith & Ellsworth, 1985) (thisapproach is illustrated in further detail in Chapter 3 in fuis volume); (2)presenting subjects with stories or vignettes and asking them to indicatewhat a person to whom the descFibed event would happen would feel,that is, which type of emotion is most likely to be evoked by the type ofsituation described (Roseman,1984; Schwartz& Weinberger, 1980;Weiner,1985).ln fuis chapter we are primarily concemed with the second approach.The advantage of using a vignette method consists in the fact that theemotion-eliciting situations are standardized, that is, ail subjects are ex-posed to exactly the same situation, which is, of course, Dot the case whenactual events experienced by different individuals are studied. Further-more, the vignette approach allows us to systematically vary different fac-tors of the situations and thereby the likely cognitive dimensions or criteriathat will be involved in evaluating the described events. For example, inthe studies by Weiner and bis group (Weiner, 1985),variables such as abili-ty and effort or controllability are systematically varied in order to studythe effect of different causal attributions on the emotions elicited. ln thisway,an experimental approach to what is otherwise a highly individualisticphenomenon becomes possible.

One critique that is often leveled against the vignette method is thatsubjects are asked to put themselves nto a situation and to imagine howthey themselves or another person would feel. The argument is that fuisprocedure is likely to result in stereotypical responses at the very best andpure artifact at the worst. Furthermore, it is argued that the imaginaryresponses described by the subjects are unlikely to have any relation tothe responses subjects would show if they actually found themselves inthe situation. This point ofview seems o imply that spontaneous evalua-tions of emotion-eliciting events are somehow microgenetically pure andunique, affected only by the impressions of the moment. It can be shownfor many domains of psychology, however, that perception, cognition, andbehavior are to a large extent affected by stored symbolic representationsof knowledge, prior experiences, social norms, and other schemataof fuissort. For example, even something as fleeting as the experience of rimeseems o be affected by some sort of symbolic representation of experi-enced rime (Fraisse,1981; Galinat & Borg, 1987). A distinction between the


3/20

4. ON THE SYMBOLIC BASIS OF SHAME 81

spontaneously perceived or cognized information and a symbolicrepresentation of information pertaining to the abject cognized seemsto be useful (Galinat & Borg, 1987). Equally, in social psychology therehas been increasing interest in "social representation," a notion that goesback to Drkheim, since it can be easily shawn that social behavior isstrongly influenced by shared, symbolically represented social knowledge(Moscovici & Farr, 1986).We argue that emotional experience is similarly affected by symbolicrepresentation of emotion-producing eventsand the dimensions or criteriaused in evaluation. As in other domains, it is highly likely that these sym-bolic representations will influence the evaluation process in an actualemotion-eliciting situation. It would seem hat this assumption can be par-ticularly weIl defended in the case of emotion-producing cognitive ap-praisals, inasmuch as the values or criteria against which events arejudged, such as the relationship of an event to an important goal or socialvalue of the persan, self-images,etc., are at least in part stable referencepoints and must therefore be stored in different parts of the central ner-vous system (see Leventhal & Scherer, 1987, for a discussion of the dif-ferent levels that are likely to be involved). At least hose emotion represen-tations that are stored on the conceptual or symbolic level should beamenable to verbal report. Given the involvement of emotion represen-tation in the actual emotion-producing process, t would seem easonableto assume that judgments based on the symbolic representation bear arather strong resemblance to actual evaluation judgments that would beoccurring in natural situations. It is of course possible that these udgmentsare somewhat "stereotypical" in nature, due to the fact that it is indeedsharedknowledge that operates in a culture which influences aIl ourbehavior, including emotion (seeGordon, 1981,1984;Levy, 1984,on socialand cultural effects on emotion).While the method of obtainingjudgments on imagined situations withsystematically varied features seems highly promising, then, one of theshortcomings in the literature is that the complete domain of potentialcriteria or dimensions is rarely studied in a systematicand principled man-fier. We have suggested using facet theory (Borg, 1979; Guttman, 1959)as the basis for a more complete theoretical model of emotion-elicitingvariables (Borg, Scherer, & Staufenbiel, 1986; Scherer, 1983). This ap-proach is not only useful for a systematic description of the potentiallyemotion-inducing events, it also provides the basis for the developmentof a theory about the empirical structure of observations in the frameworkof the definitional system. n this charter we use the emotions of embar-rassment and shame as an example to demonstrate the use offacet theoryin the study of cognitive processes underlying emotional experience.Before turning to the descriptions of the major assumptions of this ap-


4/20

82 BORG,STAUFENBIEL,CHERERproach, we briefly survey some of the factors that have been proposedin the literature as being responsible for shame induction.Antecedent Conditions for Embarrassmentand ShameShame is generally considered to be one of the most important "social"emotions because t tends to assure he adherence to social norms withoutrequiring the use of extemal sanctions. This seems o be achieved by anintemalization of the social norms and values that are linked to self-esteem.Because self-esteem depends to a large degree on the evaluation byreference groups (see the important contributions of symbolic interac-tionist theories, (Cooley, 1922; Mead, 1934),shameseems o imply a dangerto our self-esteem ased on a feared negative evaluation by others becauseof one's own shortcomings relative to important social values and norms.This essential COTe f the shame concept, ofwhich embarrassment seemsto be a mild form, bas been expressed very cogently in Aristotle's Rhetoric(Aristoteles, 1935, p. 138). Much of what bas been written about shameseems o confirm this COTetatement. Most often the discrepancy betweena certain type of individual and the demands of the ideal self, resultingin a lowering of self-esteem,are underlined (Gaylin, 1981; Izard, 1977;Lynd, 1961; Piers & Singer, 1971; Solomon, 1976; Wicker, Payne, & Mor-gan, 1983). ln terms of the symbolic representation mentioned previous-ly, we cao assume he self ideal to be represented by conceptual cognitiveelements, and the presumed evaluation by a reference group as therepresentation of the meaning of one's own behavior for reference groupmembers in relation to a set of shared social values. A comparison bet-ween the conceptual elements of the self ide al and the other-evaluationrepresentation would field slight embarrassment or shame, dependingon the degree of discrepancy, the strength of the ego ideal, the impor-tance of the values affected, and the importance of the reference group.

Self-esteemdepends on the degree to which these values are realizedin the behavior of a person, and one can argue that pride will result ifthe degree of realization exceeds some means, shame when it falls shortof what cao be expected minimally of a member of the reference group.As mentioned, it would seem that values cao differ in importance andthat the degree to which self-esteem s affected would depend on therelative importance of the respective value.

Reviews of shame-inducing experiences (see for example Izard, 1977,p. 397-399) show that many different types of shortcomings can produce~e experience of shame, for example, clumsy behavior, moral failliTe,failliTe at tasks.Presumably, these different types of inadequate behaviorare linked to different types of values toward which a member of a


5/20


6/20

84 BORG, STAUFENBIEL, CHERERMapping Sentence and Itemsln this section, we discuss the facets that determine beliefs on the elicita-tion of embarrassment and shame. That is, given a set of different situa-tions as described in the form of written vignettes, s it possible to predicthow embarrassed or ashamed subjects believe they would feel in thesesituations?

ln order to study this question systematically, one first needs a defini-tion that specifies the bounds of the universe of situations to be studied.This definition reads: "A situation in which a person p is involved belongsto the universe of embarrassmentlshame-inducing situations of p if andonly if a systemof values ofsome reference group is violated and the viola-tion is or can be perceived by at least one member ofthis reference group."The logical next step s to structure this universe by defining facets hatconceptually distinguish different types of situations. Many such facetsare possible if one is interested only in systematicclassification. However,in the context offacet theory, any such facet not only introduces concep-tuai categories, it is also an empirical hypothesis. That is, of ail possiblefacets that could be selected we choose only those where we have reasonto hypothesize that the conceptual structure they induce in the universeof situations is, in some way or another, mirrored in the empiricaljudgments of the subjects on how much embarrassment or shame thesesituations are likely to elicit.

A very simple correspondence between the definitional and the em-pirical structure of the situations would be, for example, if the situationsdefined to be of type X lead to less embarrassment han those in categoryY.Many other possibilities exist, and some of them will become relevant ater.One possibility for structuring the universe of shame-eliciting situationsis to sim pl Y ist a number of facetsas in Borg and colleagues (1986),wherethe facets "Who is the actor in the situation?", "content of the violatedvalues," and "actor's responsibility for violation" were distinguished. Yet,although these facets proved useful, to some extent, in organizing the em-pirical observations in a nontrivial way, a simple listing of facets s subop-timal for design, hypothesis development, and empirical testing. A betterway is to incorporate such facets into a mapping sentence, which makesit clearer how the facets are related to each other. The mapping sentencein Figure 4.1 builds on the findings by Borg and colleagues (1986).The mapping sentence distinguishes the universes of persons, situa-tions, and responses. The person universe is facetted only in a mostprimitive way (male-female) to illustrate the general principle to facetizethis universe. The responseuniverse is the range of the mapping sentence,that is, what follows after the mapping arrow.The content universe comprises those 10 facets that we thought most


7/20

(male)The ( ) persan (p) believes that, in a situation( female )

AB( very comprehensive ) ( focused at time t )where a ( to ) system of ( to )

( very narrow) ( not focused at time t )

C( good, not mean and bad )( physically skillful, not clumsy )( pretty, not ugly )( attractive, not repulsive )( powerful, not a pushover )

values on being (civilized, not a hick )( intelligent, not stupid )( gutsy, not a wimp )( )( combinations of the above )

0( very important )

of a reference group (g) that is ( to ) to (p)( not very important)

-E-(strongly ) -F-is ( to ) (violated ), and where (p) believes that(weakly )

the (violated) value system is-G- -H-(certainly) (very important)( to ) ( to ) to (g) and( possibly) (not important)

1( absolutely sure)

(p) is ( to ) that the (violation)( not so sure)

-J-( strongly )

is ( to ) attributed to him/her, feels that s/he would feel( weakly )

R( very embarrassed ) "( to )( not embarrassed )

FIGURE 4.1. Mapping sentence for beliefs on embarrassment.

J


8/20


9/20

4. ON THE SYMBOLIC BASIS OF SHAME 87(1986) and from an unpublished data set collected by the authors as afollow-up to the former (1986) study. The item construction was clone inthe following way: For each structuple, a group of researchers thoughtof realistic situations that would satisfy the respective facet definition asclearly as possible. Because in fuis study the items were given, there wasno possibility of systematically controlling aIl the facets. It was possible,however, to reliably classify most of the items with respect to those facetsin which we are interested here. Table 4.1lists these items for the Borgand colleagues (1986) study, together with their structuples. Each structu-pIe shows the respective item's definition on facets A, E, and], with 1 =strong and 0 = weak in the senseof the common order of the facets. Forexample, item 5 bas structuple 010, because it was defined as weak onfacet A (i.e., narrow value system), strong on] (i.e., strong attribution),and weak on E (i.e., weak violation of value system). n checking throughthe items, please note that what we present here are translations fromGerman. Because of the different cultural context, the reader may feelthat some of the structuples are odd definitions. Moreover, the structupleswere assigned to the items according to what the investigators felt wouldbe most appropriate for the subjects. These subjects were psychologyundergraduates ofthe]utus-Liebig-Universitat in Giessen,Germany. Forailier subjects,such as psychologyundergraduates at Michigan, ailier struc-tuples might be in order. The assignment of structuples is always basedon subjective udgment, which, however, can be checked by empiricallystudying interjudge agreement.The items of the second study were similar in nature to those in Table4.1. ln fuis set of items a particular effort was made to specify the kindof observers or reference group (g) present.Parenthetically, it may be noted that a slight modification of the do-main part of fuis mapping sentence generates a definitional system forboth shame and pride: AlI that needs to be clone is to differentiate facetF into the elements "violation" and "realization" of relevant values andadd a second range for pride responses.HypothesesBecause acets A, E, and] are ordered from 0 = weakly to 1 = stronglyembarrassing, a partial-order hypothesis on the embarrassment scoresfollows automatically. That is, the situations with structuple III shouldbe udged as most embarrassing, hose with 000 as east embarrassing. AlIailier items raIl, by definition, in between these bounds and hence, f thedefinitions are indeed empirically useful, should be seen as ntermediatein embarrassment. However, not aIl items are comparable. For example,nothing can be derived about the order of 101 and 010, even though the '


10/20

TABLE .154 Items, with Means of Empirical Embarrassment Ratingsand Structuples.Item struct. mean

1 1 am visiting acquaintances of mine. My dog does his 101 3.13business in the corner.

2 1 am visiting friends. ln a discussion, my mother rigidly insists 000 1.97she is right, even though she is wrong.3 1 am sitting on the toi let in the train. The door suddenly 111 4.74springs open to the corridor packed with people.4 1 am sitting with my sick grandmother in the doctor's office. 000 2.29Suddenly, she farts loudly. Everybody looks away.5 While waiting 8t the bus stop, 1 am shadow-boxing, when 1 010 1.90notice that some people are watching me from their win-dows and are laughing.

6 My brother insists in a discussion on something that is ob- 000 1.29viously absurdo Many people shake their heads.7 1 am attending a lecture of a colleague at a conference. He 001 1.16constantly mixes things up, because the slides are pro-jected head down.8 A neighbor tells me angrily that a member of my horseback 101 2.16rider's club tortured a horse while completely drunk.

9 My boss tells me to park the car in a place reserved for han- 111 3.74dicapped people. People who passby point at the sign.

10 1 am participating in a panel discussion together with another 111 4.71member of the Green Party, when he makes some blatantlyanti-Semitic remarks.11 1 am ma king an officiai visit, together with an older female 101 1.45colleague of mine from the Union. She is made up to look40 years younger than she is.12 1 lost my hair due to chemotherapy. On the bus, people turn 111 3.90around and look at me.13 At an officiai occasion, 1 notice that my colleague right next 000 0.76to me is wearing a jacket soaked with sweat.14 Our school class participates in a quiz on TV. A classmate of 001 1.68mine can barely answer a single question.15 Concerning sex: 1 wanted to, but 1 was not able to ...011 2.75

16 1 had promised to also boycott the exam. 1 asked secretly to 111 6.05get a take-home exam, because 1 needed the credit. Nowmy name appears on the notice board.17 At a convention 1 notice that my colleague's paper is received 000 1.32with pitying smiles.18 At the evening concert, the conductor of our orchestra slips 000 1.29on the polished floor and falls down.19 ln my office, a colleague repeatedly pays obnoxious com- 101 1.87pliments to our new female boss.20 A member from my bowling club shows around 101 3.50sadomasochistic pictures in my presence.21 A member of our travel group is stared at in the theater lobby 101 1.66because he is extremely short.

(ContinuedJ

88


11/20

TABLE4.1(ContinuedJ

Item struct. mean22 When visiting my sister she tells me that 1 forgot the birthday 111 3.66

of my little godchild.23 On the street people look at my girlfriend, who wears super 100 1.76

heavy makeup.24 Accidently, 1 touch my female boss's bosom. 111 3.1825 While in the cafeteria with a female colleague of mine, her 000 1.08

dress zipper suddenly breaks open.26 While in the Gate with a friend of mine, he accidently spi Ils a 100 1.95

cup of coffee over a lady's dress.27 1 greet somebody extremely cordially, and then 1 notice that 111 2.66

he is a complete stranger that 1 had mistaken for a friend ofmine.

28 At a party, 1 make a mean comment about a mutual acquain- 111 4.71tance, when 1 notice that she is standing right behind me.

29 While struggling to get my suitcase on the upper rack in a full 111 3.45train compartment, the seam of my pants breaks open.

30 While walking down the street, 1 feel nauseated and have to 011 2.71throw up.31 1 am at my relatives' New Year's party. When 1 bend down to 111 3.47

pick something up, 1 fart rather loudly. Everybody laughs.32 At a party, my girlfriend, who had tao much to drink, makes a 101 3.29

loud and very improper remark about the hosto Uneasysilence follows.33 My father is tram Bavaria. The other day, he came to visit me 100 1.29in the dormitory wearing his leather shorts and Tyrolian hat.

34 1 am pretty loaded and complain to the waiter that the bill is 111 2.55wrong. It turns out that he is right.

35 1 am being introduced to my new boss. 1 am so nervous that 1 111 2.81start stuttering.

36 1 raise my hand in the classroom and state, just with other 111 3.82words, what was just pointed out to be wrong. The teachergives it to me!

37 1 am traveling to France with a student group. A participant 111 4.16constantly points out to the hast how much cleaner it is inGermany.38 While getting off the bus, the handle breaks off and 1 fall 110 2.16down. The people around me laugh.

39 My friend shows me a ballet piroutte. Some observers start 000 0.47grinning.40 Our tennis club plays a tournament away tram home. One of 000 0.74our player double-faults every serve. The people laugh.

41 1 am in a sex shop, where 1 fun into my boss. 111 2.6842 ln the youth hostel, a guy tram our hiking group sits down 000 0.79right next to me in shredded jogging shorts. Some girlsgiggle.43 While buying a ring with my fiancee, 1 notice that she has 000 1.5Bdirty fingernails.

(ContinuedJ

89


12/20

90 BORG, STAUFENBIEL, SCHERER

TABLE .1rContinuedJ

Item struct. mean

44 My fiancee and 1 are invited to visit distinguished acquain- 000 0.82tances. She shows up in tennis shoes.45 1 play tennis and 1 keep having problems hitting the bail prop- 011 2.71

erly. A few observers watch with great amusement.46 1 corne home unexpectedly and find my parents having sex. 011 2.9247 ln the bus, everbody stares at my Mongoloid brother. 101 1.6848 ln the train compartment, a kid is making fun of my ears that 110 2.26

stick out.49 A TV reporter interviews me on the street. My answers are 111 3.32

rather stupid. ln addition, 1 also stutter.50 Due to a bet, 1 walk down the main shopping street in a night- 011 2.21

gown.51 1 am sitting with my study group in a cafe. A rather proper 000 0.34lady looks disapprovingly at my buddy's long, greasy hair.

52 1 am invited to a party given by my office superiors. 1 show up 110 2.34wearing much too casual and rundown clothing.

53 1 brought a friend home. As we sit with my family at the din- 111 2.34ner table, he cracks a dirty joke.

54 1 am at the swimming pool with an acquaintance from my 010 0.50fencing club. He practices sudden attacks without a sword.The people standing around laugh.

.Each structuple refers to facets A, J, and E from the mapping sentence in Figure 1.A = comprehensive (1 )/narrow (0) value system is violated. J = attribution of violation tosubject is strong (l)/weak (0). E = violation is strong (l)/weak (0).

former bas two "strongs," the latter only one. The reason is that in 101facets A and E are "strong," while in 010 it is facetJ. Without any notionson the relative importance of the facets, these structuples are thereforeincomparable. Formally, a structuple (aje) is stronger than (aj'e') if andonly if the former is stronger in at least one element and not weaker inany other; otherwise, the structuples are incomparable.The partial-order hypothesis can be made for individual data or fordata aggregated over subjects (such as means). Moreover, it can behypothesized that it should hold for aIl items, or just for ~e averagesofthe items that fall into one structuple category.Apart from such evel or intensity hypotheses, ne can also male predic-tions on the similarities of the items. A hypothesis that seems o suggestitself is to postulate that items that have similar structuples should cor-relate more strongly with each other than those with less similar struc-tuples. Foa (1958)proposed the contiguity rinciple,which simply determinesthe similarity of structuples by counting their common elements.However,in such a general form, fuis principle males little sense, ecause t assumes,


13/20

4. ON THESYMBOLIC BASIS OF SHAME 91among other things, that ail facets have the same weight (see Borg, 1986,for a thorough discussion of the contiguity principle).

There is, however, another classof hypotheses,called regionalhypotheses(Borg, 1979; Canter, 1985; Levy, 1981), that can be used to derive predic-tions on the similarity structure of the items. If the correlations arerepresented as distances in an SSA space (Borg, & Lingoes, 1987), then itshould be possible to partition this space nto regions such that ail item-points representing category c of facetjfall into the same region (for ailcategories c and ail facets ). This hypothesis is but a generalization ofthe usual discriminant analysis hypothesis, except that it does not imposerestrictions such as linearity on the boundary lines, which do not followfrom the substantive hypotheses.

The three facets A, E, and J are ordered in the same sense. Moreover,it is obvious that ail three facets do, in principle, admit more than justtwo elements. For example, in facet A, one could distinguish situationsinto those that violate "very comprehensive," "comprehensive," "relativelynarrow," and" very narrow" value subsets.But in order to be able to cutan SSA space nto regions such that the points of every possible structu-pie form their own "cell" in the space, he partitioning lines of the threefacets must be independent. Because he facets are ordered, the partition-ing lines should also be ordered. This is the case if these lines form, forexample, a primitive dimension system that splits the space nto boxlikeregions. This type of regionalization is known as a multiplex (Borg, &Lingoes, 1987).Becausewe consider three facets here, we predict a triplex,which requires at least a 3-dimensional SSA space.Subjects and Procedureln the following, we consider two studies on embarrassment and shame.Twenty male and 18 female beginning students of psychology at the JustusLiebig Universitat Giessen participated in the first study, and 39 maleand 52 female students in the second. The average age of the subjects inbath studies was about 24 yeaTs.They were given a questionnaire listing72 and 79 situations, respectively, and were asked o rate how embarrassedor ashamed they thought they would feel in each of these situations. Aliratings were clone in private with no experimenter present. The ratingscale can be translated, roughly, as follows:

slightly somewhat deeplyuneasy uneasy embarrassed ashamed ashamed0 1 2 " 4 5 6 7 8 9

Because the ratings in the first study turned out to be relatively low,


14/20

92 BORG, STAUFENBIEL, CHERER

3.03 ,N.O) 3.56 ,N.O) 0.39 ,N.')J003.35 ,N.3)

FIGURE 4.2. Partial orders of structuples (aie), with empirical means and fre-quencies (N) of respective items. Upper panel representsfirst study; lower panel second study.

we felt that subjcts might have attempted to bide their feelings or givesocially desirable answers. Therefore, subjects in the second study wereexplicitly asked to avoid social contrai strategies such as "playing it cool"and, rather, to indicate how they thought they would really feel. Thisresulted in much higher ratings.Fifty.four of the items used in the first study are presented (translatedinto English) in Table 4.1. They represent the set of items that was culledfor the analyses ollowing. Of the items of the second study, 64 were culled


15/20

4. ON THE SYMBOLIC BASIS OF SHAME 93

for data analyses.None of the items in the two studies were exactIy iden-tical, and only a few referred to similar situations. However, ail items thatwere culled could be categorized consistentIy into the eight cens formedby the facets A, E, and J.The cuning of the items to be used in the data analyseswas done onthe basis of the definition of embarrassing situations and the mappingsentence in Figure 4.1. Situations that did DOt conform to these defini-tions, or that were too ambiguous for reliable stTUctupleassignments,wereeliminated.ResultsThe partial-order hypotheses are most easily checked. We test them herefor the averagescoresof stTUctuple-equivalenttems only. Figure 4.2 showsthe results. Even though the cuning led to a very uneven distribution ofitems over the eight stTUctuplecens, and hence to parameter estimatesthat may DOtbe too stable, t is obvious that the hypotheses are confirmedalmost perfectIf.We DOW um to our triplex hypotheses for the similarity of the items.Because they ask for at least 3-dimensional SSA representations, we testthem first under this smallest space condition by mapping the #L2orrela-tions (Guttman, 1981b) via the program SSA-I (Lingoes, 1972) into3-dimensional ordinal SSA spaces. Note that using montone correlationcoefficients and an ordinal SSA approach is consistent with the regionalhypotheses.) A 3-dimensional space s sufficientIy large to represent theempirical correlations accurately: The alienation coefficient is K = .225,a low value for 54 points (seeBorg, & Lingoes, 1987). Figures 4.3 and 4.4show two projection planes of the SSA representation for the items fromTable 4.1. These projection planes are the plane spanned by the principalcomponents 1 and 2 (Figure 4.3), and the plane spanned by principal corn-ponents 1 and 3 (Figure 4.4).Looking at such SSA epresentations of the data, one is onfronted withthe question of whether it is possible to partition the point set such thatthe emerging regions comprise only points with common facet definitions.Figures 4.3 and 4.4 show partitioning lines induced by facets A, E, andJ. It is obvious that the roughly verticalline in Figure 4.3 partitions theplane such that those points that represent situations where, by defini-tion, a "comprehensive" value system s violated lie to the right of this line,and ail points representing situations where only a "narrow" value systemis violated lie to the left. At the same time, the more horizontalline in-duces an analogous partitioning with respect to facet J, with the weak-attribution items above this line, and the strong-attribution items belowit. Finally, in Figure 4.4 (showing the plane orthogonal to that in Figure


16/20

~

:

~ N.

.

0 . :~I~ . :CC0-0- c~ ~ N

~ .o

.N~

:

'"M ~:~

QI>"ncI.c.DEC~0

;5"

"U0

~C

"00

ccc~

-0

~

"0

'"05

cC~

0

-

c0

C

0c.Q-

E

0.

-c

"0

0=c

US-~c

-c

'"

cO>o"~

-0

ID'"

u

.

CO>O>

.=oc

U

o~co,~

>.0"00

0

0

C~

-

C"

c

coa0-C'~

"cO>O>

O>cE

Cc

c-c

o

i:c"C)

..~~


17/20

~:

~ ~~-

2 ~ ;~ ~~a. : . ..gND.~

D.n

~.

p~~=n~

OD

~~ .

ODN..~.~

.

0."."

.cl1."

-~0c~l1

,~c

(vec

"C)=

~~,0

~

"C

.00_cc~0

,

C0

c

cCQ

E0,l1

0-J

($.~

-C

.

l10

,

cqw,0L

(

-

cO)O)

'~cc

(

c~1,~

>.oU"CO)

0

0

C-C

,

Cl1'

'O

l1c1-0

~

0

c

,

~

.~O)O)

O)O)EC

E

l1

l1

l1.c

"

:"C~

.~~~


18/20

96 BORG,STAUFENBIEL,CHERER4.3), facet F induces a partion line tllat separates strong-violation itemsfrom weak-violation items. Tht; partition lines for A and E admit only onemajor error each (item 3 and 1, respectively). Otherwise, there are onlya few quantitatively small errors.

Clearly, tllis outcome is exactly what the triplex hypotllesis predicted.It is easy o see that by introducing furtller elements into facets A, E, andJ, one should simply expect partitioning lines roughly parallel to thoseinduced by tlle respective dichotomies. Thus, the structure hypotllesizedand supported by tlle findings Tenders furtller extensions of the defini-tional system feasible.The usual question that emerges in this context is whether such parti-

tionings can result by chance. It is easy o see tllat tllis is virtually impossi-ble, even if we drop the triplex hypotllesis in fayoT of the much simpleThypotllesis tllat "some" partitioning into "relatively simple" regions shouldbe possible. If one tales any 54 points and randomly assigns o them thestructuples given in Table 4.1, and tllen mixes tllem tlloroughly tllroughouttllis space,how likely is it tllat tlley can be partitioned into regions and,indeed, into a triplex? Obviously, this probability is extremely small.However, we have an even better answer to this question, that is, wecan look at a replication. The correlations of the 64 items of tlle secondstudy can also be represented weIl in a 3-dimensional SSA space. Thealienation coefficient is K = .247, which is low considering the large

number of points (seeBorg, & Lingoes, 1987). Most importantly, the struc-ture of the point locations is such that tlle same type of partitioning ispossible as before, that is, facets A and J induce a duplex in tlle planeof tlle first two principal components, and facet E splits tlle plane spannedby principal components 1 and 3, so tllat tlle same triplex results as above.DISCUSSIONThe findings reported con cern only three of tlle tell facets of tlle mapp-ing sentence in Figure 4.1. Indeed, as was mentioned, even more facetsthanjust tllese tell maybe studied. How many one selects s largely deter-mined by practical con cerns, but, in any case, t can be assumed tllat thesituation al determinants of embarrassment are quite complex or multi-faceted. It is tllerefore remarkable tllat the tllree facets tllat were studiedled to such clear-cut results. Apparently, tlley are major determinants ofembarrassment.

Starting from these results, one can proceed by introducing morecategories for facets A, E, and J, and/or by turning one's attention to fur-tller facets. Borg and colleagues (1986) studied tlle facet "Who is the ac-


19/20

4. ON THE SYMBOLIC BASIS OF SHAME 97tor?" with elements "subject him/herself," "member of a primary group,"and "~ember of a secondary group." This facet also showed a systematiccorrespondence to the data, but it was dropped here in favor of the "at-tribution" facet,], because this seems to capture a more fundamentalpsychological process. Although it should be most often true that theattribution is highest if the subject is the actor, fuis is by no means al-ways the case. For example, consider item 10 in Table 4.1, where the sub-ject is not the actor him- or herself, but attribution should still be veryhigh by pronounced "continguity" (see ntroduction). On the other band,attribution can be low even though the actor is the subject him- or her-self. An example would be a situation where a brilliant student writes2 + 2 = 3 on the blackboard, and the reference group g attributes fuis viola-tion more to a lack of concentration than ability.Borg and colleagues (1986) studied another, Piaget-like facet thatdistinguishes value violations with respect to whether the violator is"responsible," whether he or she s "responsible, but acted not intentional-Ir," or whether he or she is "responsible and acted fully intentionally." Thisfacet proved empirically useless,however. This supports the suggestionmade by Weiner (1985) according to which shame may be more closelyrelated to ability, whereas guilt may depend more on effort or intention.From a methodological point ofview, fuis shows that not every facet thatleads to conceptually useful distinctions also corresponds to some regulari-ty in the empirical observations. ln other words, theoretical predictionsbased on facet theory can be falsified in this manner.Borg and colleagues (1986) also computed, in addition to the methodspreviously shown, a conventional analysis of variance as a conveniencefor those who are unfamiliar with fuis type of data analysis methods. Forthe data considered above, such an ANOV A would treat the facet designAxEx] as a 2x2x2 factorial design. Even though ANOV A introduces ex-trinsic linearity constraints into the data analysis (Guttman, 1981a), hecell means of the factorial design may provide some nsights into the struc-ture of the data with respect to the facet design.Ifthere were more facets on the persons, an interesting question wouldbe to ask whether they allow one to discriminate among the scores of therespective subjects. For example, it seemspossible that female and malessubjects have different belief systems on embarrassment. A very simplehypothesis might predict that the female scores are ail higher, say, hanthe male scores. A more sophisticated hypothesis would predict wherethe scores are high, and where low, for each group. The best approachto study such questions for ordered facets would be to use partial orderscalogram analysis (Shye, 1985).Finally, from a wider perspective, a number of interesting questionscould be asked on the relationship of the symbolic basis of embarrass-


20/20

98 BORG, STAUFENBIEL, CHERERment/shame to these emotions when generated in concrete situations. Itmay be assumed that the symbolic basis is but a generalization of such"experienced" emotions over a large number of situations. Thus, the beliefsabout emotions should generally correlate positively with actually ex-perienced emotions. ln principle, it would be desirable to study eachsystemby itself first, and then investigate how they are related. However,as soon as beliefs corne into existence, uis may prove a most difficult task,because it is not unlikely that they themselves play a role in how an emo.tion is experienced in a concrete situation, due to the norms and expec-tations they represent. Moreover, becauseeven concrete experiences canonly be reported using language, this in itself requires the use of sym-bolic representation.

emocion1988 borg facets

Documents