the act frequency approach to personality. psychological review

Psychological ReviewVOLUME 90 NUMBER 2 APRIL 1983

The Act Frequency Approach to Personality

David M. BussHarvard University

Kenneth H. CraikUniversity of California, Berkeley

The act frequency approach to personality is advanced in this article. Dispositionsare viewed as summaries of act frequencies that, in themselves, possess no ex-planatory status. As sociocultural emergents, dispositions function as naturalcognitive categories with acts as members. Category boundaries are fuzzy, andacts within each category differ in their prototypicality of membership. A seriesof studies focusing on indices of act trends and on a comparative analysis of theinternal structure of dispositions illustrates this basic formulation. The act fre-quency approach is then placed within a taxonomic framework of the relationsamong act categories (horizontal dimension) and hierarchic classification (verticaldimension). Theoretical implications of the act frequency approach are examined.Dispositional consistency is distinguished from behavioral consistency and severalact frequency indices (e.g., dispositional versatility, situational scope) are defined.Situational analysis and personality coherence are then viewed from the act fre-quency perspective. Discussion focuses on the possible origins and developmentof dispositional categories and implications of alternative middle-level constructsfor act categorization and personality theory.

The concept of disposition has occupieda central place in personality theory and re-search. Most major efforts have been directedat determining the external relations amongdispositions (Cattell, 1957; Eysenck, 1953;Leary, 1957; Wiggins, 1979). In contrast, theinternal structure of dispositions has receivedremarkably little theoretical treatment. Atthe Ninth International Congress of Psy-chology in 1929, Allport (1931) addressed the

This formulation has benefited from the commentsand critiques of J. Block, R. Brown, H. G. Gough, R.Hogan, B. R. Little, S. R. Maddi, W. Mischel, L. A.Pervin, C. Phinney, D. J. Ozer, J. A. Russell, N. D. Sund-berg, and J. S. Wiggins. An early version was presentedat the Stanley House Conference on Interactional As-sessment, New Richmond, Gaspe Peninsula, Canada,June 29-July 3, 1981.

Requests for reprints should be sent to David M. Buss,Department of Psychology and Social Relations, Har-vard University, 33 Kirkland Street, Cambridge, Mas-sachusetts 02138 or to Kenneth H. Craik, Institute ofPersonality Assessmment and Research, 3657 TolmanHall, University of California, Berkeley, California 94720.

question: What is a trait? On returning tothat basic question 35 years later, Allport(1966) found that no extensive literature ofclose conceptual analysis of the concept oftrait or disposition existed.

Once Allport's own early considerationsof the concept of disposition (Allport, 1921,1927, 1931; Allport & Allport, 1921) hadculminated in his classic volume (Allport,1937), the field of personality appears to haveset its theoretical gears into neutral and tohave coasted with his formulation. Duringthe decades since the 1930s, important phi-losophical analyses of the concept of dispo-sition appeared (Hampshire, 1953; Ryle,1949) that might have sparked renewed con-ceptual discourse, but they failed to do so atthe time. In a recent compelling advocacy,Maddi (1980) has argued for the advantagesof vigorous theorizing for the field of person-ality. In that spirit, an act frequency analysisof dispositions is advanced here, and its im-plications for an approach to personality arereviewed.

Copyright 1983 by the American Psychological Association, Inc. 0033-295X/83/9000-0105$00.75

105

106 DAVID M. BUSS AND KENNETH H. CRAIK

2 14 Person A

Person B

Figure 1. Dominant acts over 1-week period of monitoring.

The Act Frequency Approach:Basic Orientation

Dispositional Assertions asSummarizing Statements

Hampshire (1953) asserts that disposi-tional attributions function to summarize thetrend of someone's behavior, thoughts, andfeelings. In saying that a person is generous,Hampshire claims that "the word 'generous'is so far the right word to summarize thegeneral trend or tendency of his conduct andcalculations" (p. 35). To warrant the claim,one must engage in prolonged and continu-ous study of an individual's conduct. Actualincidents, dispersed over time, must be man-ifested. Lapses are possible; to attribute adisposition to someone is not to preclude thathe or she may on some occasion have acteduncharacteristically (Brandt, 1970; Powell,1959). When someone's disposition is in dis-pute, "the final and conclusive argumentmust be a balancing of one set of actual in-cidents against another set of actual inci-dents" (Hampshire, 1953, p. 35).

Dispositional assertions are summarystatements about behavior up to the present;they are not predictions, although they carry"the normal implications that [the individ-ual's] character is so far continuing the same"(Hampshire, 1953, p. 39). Dispositional as-sertions, in this view, serve descriptive andforecasting functions, but they do not dealwith causal properties nor provide a causalaccount of the behavior at issue.

With various modifications and exten-sions, this philosophical analysis of disposi-tional assertions has guided the developmentof an act frequency conception of dispositionthat can be offered as an approach to per-sonality research.

The Frequency Concept of Disposition

The frequency analysis of dispositionalconstructs focuses on specifying the relativeincidence of acts within circumscribed cat-egories or domains (Buss & Craik, 1980,1981). From a frequency perspective, thestatement "Mary is arrogant" means that,over a period of observation, she has dis-played a high frequency of arrogant acts, rel-ative to a norm for that category of acts. Actswithin a given category may be topographi-cally dissimilar, but they are still consideredto be manifestations of a given disposition.To say that Mary is arrogant one must beable to marshal evidence of her manifesta-tions drawn from the category of arrogantacts over a delimited period of observation.Act frequency tallies from dispositional cat-egories provide not only summary interpre-tations of past conduct but also, on actuarialgrounds, a basis for predicting future trendsin behavior.

Within this approach, the fundamentalmeasure of an individual's disposition is amultiple-act composite index, provided byfrequency summary across a specified periodof observation. It follows that in predictingfuture standing regarding a disposition, theappropriate criterion measure is also a mul-tiple-act composite index, based on the fre-quency tally for the period of observationabout which the prediction is made. Acttrends, operationalized as multiple-act com-posite indices, become fundamental units ofanalysis in personality research.

A paradigmatic assessment is illustrated inFigure 1. Persons A and B have been trackedand their conduct has been monitored overa 1-week period of observation. The entriesindicate the occurrence of dominant acts.

ACT FREQUENCY APPROACH 107

The act frequency approach would assessPerson A as more dominant than Person B,based upon the tally of observed dominantacts and, on actuarial grounds, would fore-cast a continued higher base rate of domi-nance for Person A over Person B.

Temporal Reliability and Prediction

The act frequency approach incorporatesseveral previous recognitions in psychologicalresearch. First, the notion that composite in-dices based upon multiple observations aremore reliable than single observations is notnew. It is the basis for the Spearman-Brownformula (Wiggins, 1981) and is widely usedin scale construction (see Wiggins, 1973) andobserver ratings (e.g., Block, 1961; Horowitz,Inouye, & Siegelman, 1979). More recently,the use of multiple criteria has been advo-cated for attitude measurement (Fishbein& Ajzen, 1974) and, by extension, for per-sonality measurement (Jaccard, 1974). Also,Epstein (1979, 1980) has argued that im-pressive stability can be demonstrated overa wide range of behavioral variables as longas the behavior is averaged over a sufficientnumber of occurrences.

Because an individual's acts are necessarilydispersed over time, the use of aggregationin composite-act indices specifically ad-dresses the issue of temporal reliability. In hisstudies on the stability of behavior, Epstein(1979, 1980) has demonstrated impressivelevels of temporal reliability for even briefperiods of observation (12-14 days). In theassessment of act trends within the frequencyapproach, this reliability yielded by compos-ite indices is afforded to both predictor andcriterion indices. The use of composite mul-tiple-act indices is not simply a matter ofmeasurement convenience in the act fre-quency approach to personality. Rather, it isat the heart of its formulation of dispositionalconstructs. The summary approach and mul-tiple-act indices are intrinsically related con-ceptually. Temporal stability of personalitydispositions is therefore directly and gener-ally linked to constuct validity.

In summary, the act frequency approachasserts that, for a given disposition, an acttrend, or composite multiple-act index, con-stitutes an appropriate basis for predicting

Table 1Dispositions and Explanations

Element Glass Person

Dispositionalconstruct

brittleness dominance

Manifestation of shattering taking charge afterthe disposition the accident

Causal account molecular genes, rolesstructure

future act trends or multiple-act indices. Themeasurement operations and statistical anal-yses for this fundamental kind of predictionin personality fall within the domain of tem-poral reliability. This approach to personalityprediction acknowledges that the reliabilityoffered by composite indices is necessary forboth predictor and criterion variables, if pre-diction is to be conceptually appropriate andsuccessful. In this basic form of personalityprediction from observed act trends to futureacts trends, full symmetry, except for tem-poral locus, holds for the predictor and cri-terion variables.

Act Frequency Approach and Explanation

A critical question in personality psychol-ogy pertains to the status of dispositions ascasual or explanatory accounts. Recent phi-losophical treatments of this issue center onthe relations among three elements: (a) thedisposition, (b) manifestations of the dispo-sition, and (c) a causal account of the man-ifestations (see Table 1).

The first issue is the relationship betweenthe disposition and its manifestations. Addis(1981), Hampshire (1953), O'Shaughnessy(1970), Squires (1968, 1970), and others ar-gue that a dispositional statement does notoffer a causal explanation of its manifesta-tions. Saying that an individual is dominantdoes not explain the acts of taking chargeafter the accident, deciding which movie thegroup will attend, or commanding someoneto leave the room. The manifestations mustinstead be explained on independent groundsand not with recourse to the dispositionalstatement itself. This position contrastsmarkedly with any use of dispositional state-


ments to explain or account for observedmanifestations (e.g., "She issued the com-mand because she is dominant").

A related issue bears directly upon the re-lationship between dispositional statementsand causal accounts. In differing ways, Arm-strong (1969) and Cummins (1974) identifythe disposition with its causal account (e.g.,"brittleness is a sort of bonding of mole-cules"). In contrast, O'Shaughnessey (1970)argues that whereas dispositions may be as-sociated with causal explanations in somepoorly understood fashion, there is no causalrole for the disposition itself ("it simply fallsoutside the causal schema"). Dispositionalconstructs, in this view, perform a job dif-ferent from that of explanations.

By acknowledging the "poweflessness ofdispositions" in explanation (O'Shaughnessey,1970), the act frequency approach separatestwo distinct scientific endeavors: (a) mappingregularities in conduct, and (b) providingcausal or explanatory accounts of them.Once regularities in behavior are identified,the usefulness of concepts drawn from ge-netics and biology (e.g., Buss, 1983; Eysenck,1981), role theory (e.g., Sarbin & Allen,1968), motivational theory (e.g., McClelland,1983), functional analysis (e.g., Skinner,1938), interactional analysis (e.g., Magnus-sen & Endler, 1977), and other explanatoryschemes must be determined. Prior to suchdetermination, however, the act frequencyapproach to personality dispositions pro-vides, an actuarial grounds, useful predic-tions about future trends in conduct andidentifies regularities of act patterns that callfor explanatory accounts.

Dispositions as Natural CognitiveCategories of Acts

The frequency concept of disposition treatsacts as basic units of analysis and seeks tospecify the nature of dispositional categoriesthat encompass these acts. The view of dis-positions as summary statements carries theimplication of multiple-act categories. Thatis, dispositional assertions summarize topo-graphically dissimilar manifestations acrossa variety of situations (Buss & Craik, 1980,1981, in press; Craik, 1976; Fishbein, 1972;Jaccard, 1974; Wiggins, Note 1). The cog-

nitive form of dispositional categories, theircriteria for membership, and their structurepose important and heretofore relativelyunexamined issues in personality theory.

Four different facets of dispositional cat-egories may be examined. The first pertainsto the internal category structure—a facetthat has received almost no attention in per-sonality research. Examination of the inter-nal category structure raises important ques-tions about category boundaries, internal re-lationships among category members, thediffering conceptual status of category mem-bers, and so on. The second facet deals withthe comparative analysis of dispositions interms of their internal and manifested struc-tures.

The third facet of the dispositional cate-gory structure involves examining the exter-nal relationships among dispositional cate-gories that are posited to reside at the samelevel in a given taxonomic scheme. In thecontext of the act frequency approach, thissecond issue yields questions such as (a)What are the empirical relationships betweenfrequency summaries of dispositional cate-gories such as dominance and gregarious-ness? (b) Can a taxonomic model be appliedto describe these external category relation-ships?

The fourth facet of the dispositional cat-egories entails what Rosch (1978) and othershave called the vertical level of categoriza-tion. Analysis of dispositions along the ver-tical dimension involves examining the re-lations between superordinate dispositionalcategories (e.g., the rubric of interpersonaltraits), middle-level dispositional categories(e.g., dominance, gregariousness), and sub-ordinate categories (e.g., specific acts). Be-cause the internal category structure has beenthe most neglected in personality research,and because the act frequency approach af-fords a novel contribution at this level, it willbe considered first and in the greatest detail.

Internal Category Structure

Acts are to the behavioral world what ob-jects are to the inanimate world: basic con-stituent elements. Dispositional constructsoffer a fundamental system for the categori-zation of acts. Dispositional constructs can


be analyzed as natural cognitive categories(Rosch, 1975a, 1978; Rosch & Mervis, 1975)or fuzzy sets (Zadeh, Fu, Tanaka, & Shimura,1975) in that act categories for specific dis-positions are assumed to be cognitively struc-tured around prototype or central members,with nonprototype members becoming pro-gressively more peripheral to the category. Atthe borders, the array of peripheral acts fora given dispositional category blends into ad-jacent act categories.

Conjoining the summary view of person-ality dispositions with the cognitive analysisof natural categories generates a program ofpersonality research. Thus far, the acts sub-sumed within six interpersonal dispositions(agreeable, aloof, dominant, gregarious,quarrelsome, and submissive) have been ex-plored and their internal structure examined(Buss&Craik, 1980,1981; Buss, 198 la). Theprocedure used entails two steps: act nomi-nations and prototypicality ratings.

Procedures

Act nominations. For each of the six dis-positional constructs, undergraduates wereasked to nominate acts that would count asmanifestations of the disposition. The basicinstructional set (e.g., for dominance) was"Think of the three most dominant females[males] you know. With these individuals inmind, write down five acts or behaviors theyhave performed that reflect or exemplify theirdominance." The instructions were then re-peated, with sex of actor altered. The aim ofthis procedure was to secure for each dis-position 100 acts that could reasonably beconsidered to fall somewhere within the dis-positional act category.

Systematic analysis of this act-nominationprocedure has not yet been undertaken butis warranted. Despite the aim of the proce-dure and its instructions, many of the nom-inations did not constitute reports of occur-rences, in Ryle's term (1949), or accounts ofepisodes (e.g., "issued orders to the group")that had happened. Some of the nominationswere phrased in general terms (e.g., "givesout orders") and often included a frequencyterm (e.g., "constantly, forever, sometimes,rarely, never") in the act description (e.g.,"always issuing orders"). Nominations of thiskind can be readily converted to occurrence

statements. A substantial portion of the nom-inations, however, missed the point of the in-structions and offered nonact terms, often inthe form of trait adjectives (e.g., regardingdominance: "argumentative, talkative, stub-born"). The central focus of the instructionsupon specific persons may have shifted thepsychological set. Variations in instructionsand provision of examples offer a basis forsystematic examination of the act-nomina-tion procedure.

For the purposes of the initial series of em-pirical studies, the lists of acts generated foreach disposition were subsequently reducedby eliminating redundancies, nonact state-ments, general tendency statements, fre-quency statements, and statements that wereconsidered too vague to constitute an ob-servable act. Grammatical errors were cor-rected, and each selected act statement wasphrased in a way suitable for performance byeither sex. A list of 100 acts was derived foreach dispositional construct in this way. Inthe case of aloofness, 11 acts generated froman expert panel were used to supplement the89 acts generated by the undergraduate pan-els. Apparently, acts of aloofness are lessreadily summoned up than are acts for thecategories of dominance, gregariousness,submissiveness, quarrelsomeness, or agreea-bleness.

Prototypicality ratings. For each of the sixact lists, panels of judges rated the prototyp-icality of each of the 100 acts for the dispo-sitional construct at issue. Instructions in-cluded this adaptation from the Rosch andMervis (1975) procedure for judging the pro-totypicality of colors:

Close your eyes and imagine a true red. Now imaginean orangish red ... imagine a purple-red. Althoughyou might still name the orange-red or the purple-redwith the term red, they are not as good examples of red(as clear cases of what red refers to) as the clear "true"red. In short, some reds are redder than others.

Judges then rated on a 7-point scale how goodan example each act was of the dispositionalcategory at issue.

The alpha reliabilities of the compositeprototypicality ratings and the average be-tween-rater agreements (panel size in paren-theses) are as follows: aggreeable, .77, .12(31); aloof, .97, .42 (45); dominant, .95, .20(79); gregarious, .95, .31 (42); quarrelsome,


.95, .44 (29); and submissive, .96, .36 (47).Except for agreeable, these indices are high,indicating that each rating panel displayedadequate composite reliability in judgingwhich acts were more or less prototypical ofthe dispositional category, even within setsof acts independently nominated as fallingwithin each category.

The ranked listing of acts on prototypi-cality was partitioned into quartiles, eachsuccessive 25 acts forming an independentlycomposited multiple-act index (from Proto1, the most central acts, to Proto 4, the mostperipheral acts). Table 2 presents illustrativeacts from each multiple-act index for the sixdispositions under study.

Prototypicality ratings afford a convenient

and relatively direct means of examining cat-egory structure. However, the assumptionthat dispositional constructs function throughnatural categories of acts would be strength-ened by convergent evidence from other typ-icality indices. In the domain of concreteobjects, convergent findings have been dem-onstrated for prototypicality ratings, verifi-cation times for category membership, prob-ability of item output in membership-nom-ination tasks, and expectations generated bythe category name (Rosch, Simpson, & Miller,1976). Similar examination of alternativetypicality indices should be undertaken fordispositional categories.

The composite reliabilities of prototypi-cality ratings for act-disposition judgments

Table 2Acts of Varying Prototypicality for Six Dispositional Constructs

Dispositionalconstruct Act

AgreeablenessProto 1Proto 2Proto 3Proto 4

AloofnessProto 1Proto 2Proto 3Proto 4

DominanceProto 1Proto 2Proto 3Proto 4

GregariousnessProto 1Proto 2Proto 3Proto 4

QuarrelsomenessProto IProto 2Proto 3Proto 4

SubmissivenessProto 1Proto 2Proto 3Proto 4

I readily did the dishes after dinner.I forgave my acquaintance after she had spread a false rumor about me.I picked up the tab for lunch.I arrived on time for the meeting.

I offered a monosyllabic response to the conversational overture.While the others chatted, I gazed into the fireplace.I declined the invitation to the large party.I visited a museum alone.

I forbade her to leave the room.I gave advice, although none was requested.I resisted conceding an argument.I walked ahead of everybody else.

I introduced myself to new coworkers without hesitation,told a joke at the dinner party,studied with a group to prepare for the examination,went to the football game.

picked a fight with the stranger at the party,ended the conversation by stalking out of the room,complained about having to do him a favor.

I insisted upon doing the driving on the trip.

I walked out of the store knowing that I had been shortchanged.I continued to apologize for the minor mistake.I let my partner choose which movie we would see.When the three of us set out on the journey, I took the back seat of the car.

Note. Proto 1 = most prototypical; Proto 4 = least prototypical.


are substantial and adequate for the act fre-quency research. It is less clear what minimallevel of between-rater agreement is requiredto justify application of the concept of nat-ural cognitive categories to a judgment do-main. For concrete object categories (e.g.,bird, vehicle, vegetable, clothing), Rosch(1975b) reports split-half reliabilities of .97or higher for a panel of 201 judges, which arecomparable to those obtained for the dis-positional categories. However, Rosch doesnot cite average between-rater agreement,which may be higher for concrete object cat-egories than for dispositional categories(which ranged from .12 to .44). Systematiccomparison of between-rater agreement lev-els for the categories of objects and acts isneeded.

Within the act frequency approach, dis-positional constructs are treated as sociocul-tural products held by members of a culture.Panels offer a direct means of seeking actspecifications for dispositional categories,with individual misinterpretations, transienterrors, and other variations presumably can-celing each other out. Thus, reliability esti-mates for composite indices are appropriatetheoretically and, with the possible exceptionof agreeableness, reach sufficient levels topursue a research program in a manageablefashion (e.g., with panels of 20 or so judges).

Comparative Analysis of the InternalStructure of Personality Dispositions

The present framework provides two kindsof structural comparisons among personalitydispositions. The procedures of act nomi-nation and prototypicality rating suggest po-tentially useful attributes for comparing theinternal cognitive structures of dispositionalconcepts. Additional and quite distinct com-parisons among dispositions in terms of theirmanifested structures can be generated fromthe analysis of overt act performance, derivedfrom self- or observer-based field monitor-ings.

The internal structures of dispositionalcategories can be compared according totheir category volume, the composite reli-ability of prototypicality ratings, and therange and central tendency of prototypicalityratings for act members of a category.

First, the difficulty in soliciting 100 acts ofaloofness from undergraduate panels (11 actshad to be derived from expert panel nomi-nations) suggests that the volume of acts cog-nitively available for specific dispositionalcategories may vary. More systematic actnomination procedures can provide usefulprobes concerning the variation among dis-positions in category volume. For example,examining the number of acts nominated perunit of time would provide an index of rel-ative differences among dispositions in cat-egory volume. The main implication is thatthe total act membership of each disposi-tional category can be estimated, with an ac-tual size and specifiable distribution along theprototypicality continuum.

Category volume may be related to Hamp-son's (1982) notion of the imaginability oftrait categories, that is, to how easy it is "toimagine a behavior that would be describedspecifically" by the trait term (p. 5). The gen-eration of instances of traits rated high onimaginability (e.g., helpful, clumsy) was foundby Hampson to be more subjectively difficultthan for traits rated low (e.g., important, sin-cere). Behavioral instances nominated forhighly imaginable traits were also judged tobe more prototypical than those for lessimaginable traits.

A second comparative attribute is the re-liability of prototypicality ratings. For ex-ample, the composite reliability for ratingsof agreeable acts is somewhat lower than forour other five categories. Agreeable acts maynot vary along the prototypicality continuumas much as acts in other dispositional cate-gories. This restriction of range would makedifferentiation among agreeable acts difficultfor judges, thus reducing composite reliabil-ity. The findings in Table 3 support this sug-gestion: The variance in the mean prototyp-icality ratings for aggreeable acts is lower thanfor those in the other five categories. Exam-ination of the prototypicality ratings for the100 aggreeable acts shows that they are con-centrated within the middle range, with nei-ther extremely central nor extremely periph-eral acts appearing. Whether this distributioncharacterizes the actual category of agreeableacts or merely represents an artifact of ournomination and selection procedure will re-quire more systematic examination.

112 DAVID M. BUSS AND KENNETH'H. CRAIK

Table 3Comparative Analysis of the Internal Structure of Dispositional Categories

Prototypicality ratings

Reliability

Category

AgreeablenessAloofnessDominanceGregariousnessQuarrelsomenessSubmissiveness

Alpha

.77

.97

.95

.95

.95

.96

Between rater

.12

.42

.20

.31

.44

.36

M

4.623.384.033.964.074.25

SD

0.581.220.780.991.241.02

Assessed act performance

Base rate(%)

795366674148

Manifested structure

Alpha

.93

.89

.94

.93

.93

.91

Interact r

.12

.07

.14

.12

.13

.09

Note. Each category includes 100 acts. M = prototypicality ratings across the 100 acts; SD = standard deviationsof the means across the 100 acts; and base rates = average percentage across the 100 acts within each category ofthose individuals reporting act performance.

Third, in addition to range, the central ten-dency of the distribution of acts on the pro-totypicality continuum may vary from dis-position to disposition. As shown in Table 3,the mean prototypicality ratings for each ofthe six sets of 100 acts show the highest meanprototypicality for agreeable acts and the low-est for aloof acts. The interesting question iswhether the disposition of aloofness is ac-curately characterized as relatively lacking incore acts or whether this result is a deficiencyof our current act-nomination and ratingprocedures.

Systematic monitoring of individuals' con-duct in everyday settings over standard pe-riods of observation will eventually providea basis for analyzing the manifested structureof dispositions and for making comparisonsbetween them. For a preliminary examina-tion of the issues encountered in these anal-yses, the 100 acts for each disposition understudy were rephrased as first-person state-ments that yielded six act reports, one foreach dispositional construct (i.e., agreeable-ness, aloofness, dominance, gregariousness,quarrelsomeness, and Submissiveness). Asample of 100 university students completedthe act reports, providing a dichotomous(yes/no) report for each act and, for thosethey had performed, a frequency rating(rarely, sometimes, often).

Despite the retrospective self-report natureof the act reports, findings from the reportsserve to illustrate two ways of comparing dis-positions that are based upon the assessed

performance of acts, in contrast to the anal-ysis of the internal cognitive structures. Thesefeatures of dispositional categories are baserates of occurrence and tightness of mani-fested structure.

Using the present method of self-reportedassessment of act performance, base ratesaveraged across the 100 acts within each cat-egory yielded a range from 41% for quarrel-someness to 79% for agreeableness, with sub-missiveness (48%), aloofness (53%), domi-nance (66%), and gregariousness (67%) fallingin between. If confirmed by converging meth-ods of assessing act performance, these dif-ferences in category base rates provide a cen-tral, but as yet relatively unexamined, issuefor personality theory and actuarial predic-tion.

Finally, dispositions may vary in the tight-ness of their manifested structure, gauged bythe empirical intercorrelations of acts withineach category. An extremely tight empiricalstructure (very high correlations among acts)may suggest an undifferentiated style of con-duct; a looser structure indicates a potentialfor differentiating styles in manifesting a dis-position. In the present studies, the alphacoefficients and the mean between-act cor-relations for the six act reports were as fol-lows: agreeableness, .93, .12; aloofness, .89,.07; dominance, .94, .14; gregariousness, .93,.12; quarrelsomeness, .93, .13; and Submis-siveness, .91, .09. Thus, this set of disposi-tions displays only modest variation in thetightness-looseness of manifested structure.


Loevinger (1957) referred to the level of tight-ness of a disposition's manifest structure asits characteristic intercorrelation and drewimplications for the structural component ofpersonality scale validity. It is noteworthythat a disposition can serve as a relativelyunitary cognitive and conceptual act categoryand still possess either a tight or loose man-ifested structure.

The Horizontal Dimension: RelationsAmong Act Categories at the Same Level

In addition to a close analysis of the in-ternal and manifested structure of act cate-gories, a second critical issue pertains to therelations among act categories that are pos-ited to reside at the same level in a giventaxonomic framework. A simple list of suchcategories, each analyzed separately, is un-desirable on both heuristic and aestheticgrounds. A taxonomic model that specifiesor posits the relations between each act cat-egory and every other act category offers amore useful guide to conducting personalityresearch and generating theory. Such a modelis offered by the Wiggins circumplex (Wig-gins, 1979, 1980).

Briefly, the Wiggins circumplex model isa two-dimensional taxonomy consisting of 16interpersonal dispositional categories (e.g.,dominant, arrogant, calculating, cold) ar-rayed in a circular fashion. The two majordimensions that define the circumplex, dom-inant-submissive and agreeable-quarrel-some, are orthogonal to each other in themodel. Each of the remaining dispositionalcategories is posited to possess varying de-grees or facets of these two dimensions. A

distinct advantage of the Wiggins circumplexis that it offers a basis for making predictionsabout the relations among dispositional cat-egories: Orthogonal variables are predictedto be uncorrelated, adjacent variables posi-tively correlated, and opposing variables neg-atively correlated.

The most direct test of the circumplexmodel within the act-frequency approach issimply to examine the correlation matrix ofact-frequency summaries across all studiedcategories and then to examine its corre-spondence to the predicted correlation ma-trix. The fairest test entails using compositesof the most prototypical acts from each cat-egory—in this case, the top quartiles or com-posite-act indices based on the most proto-typical 25 acts from each dispositional cat-egory. Table 4 shows the correlation matrixof these prototypic composites for the six actcategories (above the diagonal) and the cor-relations predicted on the basis of the cir-cumplex model (below the diagonal). Thecorrelation between the predicted and ob-tained correlations (N = 15) for the six cat-egories is .89 (calculated via Spearman's rho).This finding indicates that, overall, the pat-tern of predicted correlations correspondswell to those that were obtained.

The absolute magnitudes of the correla-tions are, however, discrepant from those pre-dicted by the circumplex model, in manycases. This finding points to a key feature ofthe horizontal level of dispositional catego-rization within the act frequency approach:Performance of many acts within one dis-positional category does not preclude perfor-mance of many acts within other disposi-tional categories, even if they are concep-

Table 4Correlations of Multiple-Act Criteria With Each Other

Disposition 1. 2. 3. 4. 6.

1. Aloof2. Gregarious3. Dominant4. Submissive5. Hostile6. Agreeable

_-1.00-.25

.25

.75-.75

-.13—.25

-.25-.75

.75

.22*

.55**

—-1.00.00.00

.32**-.07-.13

—.00.00

.46**

.11

.36**

.00

—-1.00

.12

.45**

.23*

.36**-.11

—

Note. Correlations to the left of the diagonal are those predicted by the circumflex model; those to the right are theobtained correlations.* p < .05, two-tailed; ** p < .001, two-tailed.


tually or semantically opposite to each other.The correlations among act-frequency sum-maries may not correspond to semantic re-lations inherent in trait-descriptive vocabu-laries.

The present data set cannot rule out thehypothesis that methodological artifacts (e.g.,general acquiescence or endorsement tenden-cies) may have unduly inflated the magni-tudes of the correlations observed in Table4 or the possibility that, were they absent, theabsolute magnitudes of the correlations wouldcorrespond more closely to those posited bythe Wiggins circumplex model. But if we holdthis potential confound in abeyance for themoment, the present data may indicate theabsence of true bipolarity in the act frequen-cies studied; those who perform many dom-inant acts, for example, may be neither morenor less apt to perform many submissive acts.If absence of bipolarity among semanticallyopposed act categories is confirmed by fieldstudies, a different taxonomic frameworkmay be called for by the act frequency ap-proach.

One final conjecture may be advanced:The positive correlations among most mul-tiple-act criteria stem not from methodolog-ical artifacts but from a general activity org factor in the act domain. In this case, in-dividuals may reliably differ in the numberof acts performed, regardless of the act cat-egory within which they are classified. To theextent that such a g factor exists, it wouldhave to be partialed out from any assessmentdevice in order to obtain circumplexity in theact domain. Russell (1979) has suggested ananalogous partialing of a general factor in theemotional domain; however, in his view thegeneral factor is a purely methodological ar-tifact. On the other hand, a g factor in theact domain, if not artifactual, would requirea new conceptualization of personality, justas the discovery of a g factor in intelligence(Spearman, 1904) required a new concep-tualization of intelligence.

The Vertical Dimension: HierarchicTaxonomic Classification

Beyond the internal structure of disposi-tional categories and issues surrounding therelations among different categories at the

same level, another important issue pertainsto vertical or hierarchic relations among dis-positional categories. Broadly speaking, twosuch taxonomic approaches have been pur-sued. First, lexical-conceptual analyses ofdispositional terms within the natural lan-guage have been conducted (e.g., Allport &Odbert, 1936;Cattell, 1957; Goldberg, 1982;Wiggins, 1979). Wiggins (1979), for example,partitions the universe of trait-descriptiveterms into seven superordinate categories(e.g., interpersonal, temperamental, mate-rial). Within each superordinate category arespecific dispositions (e.g., the dispositions of"stingy" and "generous" fall within the su-perordinate category of material traits). Al-though they are not explicit in Wiggins'staxonomy, subordinate categories are pre-sumably specific acts that fall within eachmiddle-level dispositional category.

An alternative strategy has been to develophierarchic classifications based in part on theinterrelations among dispositional measures,through factor analysis and related tech-niques (e.g., Eysenck, 1953;Guilford, 1959).Eysenck's theory of vertical classification, forexample, involves extraversion as a super-ordinate category, with the specific disposi-tions of liveliness, excitability, sociability, andimpulsivity at the second level. Specific habitsand responses define subordinate-level cate-gories.

From the vantage point of the act fre-quency approach, the point of all taxonomichierarchies is that they must ultimately dealwith the categorization of acts. This basicgoal is often missed by taxonomic ap-proaches that stop at the level of lexical-con-ceptual analysis of trait terms or at primaryand second-order factor structures capturingthe covariation among personality inventoryitems or scales. Attention to the categoriza-tion of acts would grant such taxonomicschemes greater ultimate significance.

A comparison of the hierarchic taxonomyof objects and the taxonomy of acts is in-structive (see Table 5). For object categories,Rosch and her associates (1978; Rosch, Mer-vis et al., 1976) have identified three levelsof abstraction: superordinate (e.g., furniture),basic (e.g., chair, table), and subordinate (e.g.,kitchen chair, livingroom table). For dispo-sitional terms, Wiggins (1979, 1981) has sug-


gested groupings that might indicate usefulsuperordinate categories (e.g., interpersonalstyle, temperament, character, mental pred-icate). Presumably, specific dispositional con-structs (e.g., dominance, agreeableness) func-tion as basic level categories. Subordinatecategories would consist of single acts (e.g.,tallied across single or multiple situations).In this scheme, ordinary dispositional con-structs emerge at the basic level within thecategorization system for acts, parallelingbasic categories of objects.

Dispositional categories sort together actsthat are dispersed across time and situationthroughout the individual's stream of behav-ior. In addition to temporal dispersal, the to-pographic distinctiveness of similarly cate-gorized acts suggests a complexity inherentin dispositional categories that may set themapart from other categorization schemes.What common attributes of prototypicallydominant acts warrant the generation of thecategory "dominance"? Such attributes ofacts and their relation to common member-ship in dispositional categories have receivedlittle attention. One possible basis of com-monality rests upon similar effects or impactson the environment (see, e.g., Jones & Davis,1965; Wiggins, Note 1). But whether that fac-tor would hold across the entire range of spe-cific dispositional categories remains to bedetermined.

Theoretical Implications of theAct Frequency Approach

Although founded primarily upon a dis-tinctive formulation of the concept of dis-position, the act frequency approach carriesimplications for a broad range of theoreticalissues in personality research, including thosebearing upon personality consistency, situa-tional analysis, and personality coherence.

Personality Consistency

The issue of consistency and ways of for-mulating the notion have held a central placein discourse about the nature of personality.The act frequency approach draws a funda-mental distinction between behavioral con-sistency and dispositional consistency. Be-havioral consistency refers to indices derived

Table 5Taxonomies of Objects and Acts

Level

Superordinate Basic Subordinate

Concrete object categories

furnituretree

chairmaple

kitchen chairsugar maple

Act-dispositional categories

interpersonal dominance taking charge at thestyle meeting

temperament obstinacy ignoring the associate'ssuggestions

from the molecular level of single-act anal-ysis. Dispositional consistency refers to mo-lar-level multiple-act indices derived fromanalyzing the cognitive structure of disposi-tional categories of acts.

Indices of Behavioral Consistency

Consider the data yielded by monitoringthe total behavioral output of a sample ofpersons during two extended periods (Time1 and Time 2). Prior to any dispositional cat-egorization of acts, several indices of behav-ioral or act-level consistency can be com-puted.

Single-act consistency. The temporal con-sistency of single-act categories (e.g., takingcharge of a meeting; attending psychologyclass) can be gauged. For example, adequatelevels of temporal stability have been re-ported for single-act categories within thedomain of conscientiousness (Mischel &Peake, 1982).

Consistency of overall act output. Personsmay demonstrate reliable individual differ-ences in aspects of the total number of actsperformed during a period of observation,without regard to any dispositional catego-rization of acts. Two additional consistencymeasures of overall act output are (a) con-sistency in overall act versatility—the relativeposition maintenance of individuals acrossperiods of observation on the number of dif-ferent acts performed, with same-act repeti-tions not counted and (b) consistency in over-all situational scope—the relative position


maintenance of individuals across periods ofobservation on the total number of differentsituations in which acts are performed. Notethat these indices of behavioral consistencymay be calculated prior to any dispositionalcategorization and thus are less relevant topersonality consistency.

Indices of Dispositional Consistency

Within the act frequency approach, single-act consistency and the intercorrelationsamong single-act categories are properly con-sidered to be predispositional matters. Thus,the question of why some single acts correlatewith certain others, whereas some do not, isprimarily a behavioral rather than a dispo-sitional issue. Personality research emergesmore clearly when single acts are categorizedand analysis moves from a molecular to amolar level. A fundamental conceptual con-tribution of dispositional constructs is to of-fer a system for categorizing single-act unitsinto middle-level conceptual units.

Basic dispositional consistency: Act-trendconsistency. An act trend is the tally of allacts falling within the boundaries of a mul-tiple-act dispositional category that are per-formed by an individual during a period ofobservation. Act-trend consistency refers torelative position maintenance of individualson act trends for a specific disposition acrossperiods of observation. It should be notedthat act-trend consistency can be high evenif individuals do not display the same specificacts across time periods. Consistency of fre-quency within category rather than same-actrepetition becomes the critical index of con-sistency.

Two additional measures of dispositionalconsistency are (a) consistency in disposi-tional versatility—the relative positionmaintenance of individuals across periods ofobservation on the number of different actsperformed within a specific dispositional cat-egory, excluding same-act repetitions, and(b) Consistency in dispositional-situationalscope—relative position maintenance of in-dividuals across periods of observation on thenumber of different situations within whichprototypical acts for a specific dispositionalcategory are performed. In the first, two in-dividuals may obtain the same act-trend in-

dex for dominance, for example, but differmarkedly in the range and versatility of actsthey display in mainfesting their dominance.In the second, two individuals may obtainthe same act-trend index for dominance, butone may manifest dominance in delimitedkinds of settings (e.g., only at home) whereasothers may display the same absolute numberof dominant acts across the same time period,but disperse them across a wider range of set-tings (e.g., at home, at work, in leisure set-tings). Thus, the act frequency approach of-fers several novel consistency indices thatelude more traditional approaches that assessscale or rating consistency across two or moretime periods. A fuller treatment of these con-sistency issues in personality psychology canbe found in Ozer's (1982) formulation.

Ipsative measures of dispositional consis-tency. Unlike strategies that employ trait-rating indices, Q sorts, or ranking measures,the act frequency approach provides a truezero point: when no acts within a given dis-positional category are exhibited. Therefore,ratio measures can be developed by whichact frequencies in one dispositional categoryare compared with act frequencies for an in-dividual in all other categories. Thus, thoseresearchers persuaded by arguments in favorof idiographic analysis can derive consistencyindices for three ipsative measures: (a) idio-graphic act trend—the act tally for a givendispositional category divided by the person'sindex of overall act output, (b) idiographicdispositional versatility—the number of dif-ferent acts performed within a given dispo-sitional category divided by the person's in-dex of overall act versatility, and (c) idio-graphic dispositional-situational scope—thenumber of different situations in which actsfor a given dispositional category are per-formed divided by the person's index of over-all situational scope.

Behavioral Consistency andDispositional Breadth

The breadth of dispositions, as Mischeland Peake (1982) have rightly noted, is oneof the central issues in continuing contro-versies about the adequacy of traditional for-mulations of personality. They have recentlyreported a study of dispositional breadth that


can be usefully viewed from the vantage pointof the act frequency approach.

For a sample of 63 college students, Mis-chel and Peake analyzed 19 measures fromthe domain of conscientiousness/studious-ness (e.g., attending psychology class, readingreserve library materials punctually, keepingthe dorm room tidy). Adequate levels of sin-gle-act consistency were found, with a meancomposite reliability, averaged across the 19single-act categories, of .66. The 19 measuresof single-act categories were intercorrelated,yielding an average between-act r of .13 (or.20 when attenuation due to unreliability ofmeasures is taken into account). These in-tercorrelations between single acts, termedcross-situational consistency coefficients aretaken as measures of dispositional breadthfor the construct of conscientiousness/stu-diousness. They conclude that the average rof. 13 (or .20) offers evidence for a relativelystable mean level of individual differences butalso reflects behavioral discriminativeness(sensitivity to situational cues). They suggest(a) that we seek to understand when and whythese obtained coherences among single-cat-egory acts emerge and when and why theyfail to emerge, and (b) that we search for con-sistency at different levels of abstraction—from subordinate, molecular levels to molarand superordinate levels.

Mischel and Peake's (1982) use of the av-erage intercorrelation among single acts mostclosely resembles the tightness-looseness ofmanifested dispositional structure. That is,tightly-structured dispositions are consideredto be broad or global. Note that a trade-offexists between the tightness of a dispositionalcategory and the number of observationsneeded for an adequate act trend: The tighterthe manifested disposition, the fewer thenumber of observations required.

Within the act frequency approach, dis-positional breadth takes on at least four ad-ditional meanings. In each case, dispositionalbreadth is an empirical matter and not fun-damentally a conceptual issue. For a givendisposition, versatility and situational scoperefer to the breadth of an individual's rep-ertoire of prototypical acts and the breadthof contexts in which they are performed, re-spectively. These two measures of disposi-tional breadth generate person variables. In

contrast, two measures of breadth gaugeproperties of the dispositional construct it-self. Category volume refers to variationsamong dispositions in the number of actsconsidered to be prototypical members, rang-ing from few to many. Finally, category struc-ture refers to the tightness or looseness of thecorrelational matrix for manifested acts fall-ing within the category boundaries.

Situational Analysis

In its most general form, the act frequencyapproach grants little place to situationalanalysis. In assessing individuals on a per-sonality disposition, act frequency analysissums displays of prototypical acts withoutregard for attributions of causality to personor situation. At this level, it remains strictlydescriptive, entailing situational considera-tions only as qualifications in the descriptionof prototypical acts. For example, an act ofdisplaying little emotion when meeting an oldfriend at the airport constitutes an aloof act,but displaying little emotion at a formal cer-emony or while reading the newspaper prob-ably does not. Beyond that, the acts, oncespecified in this manner, are credited to theindividual's account independently of anyinferences of causal attribution to person orsituation and without any effort to match orcontrol for situational and related factors.

It follows from this approach that in fieldmonitoring of persons for the purpose of dis-positional assessment, the basic unit of com-parative measurement is temporal. That is,two persons are deemed similarly dominantif they have achieved equivalent tallies of pro-totypically dominant acts over an equal pe-riod of observation, regardless of whetherthey differ in age, for example, or in the kindof social ecologies within which they function(e.g., one may be a young bus driver and theother an elderly business executive).

However, the act frequency approach canbe conjoined with situational analysis by sub-aggregating act trends according to specificcontexts or categories of situations. This pro-cedure would entail developing multiple-actindices of a given dispositional construct(within specified subsets of situations) andapplying them to individuals (e.g., at work,at home). Such subaggregation resembles


Table 6Aversive Acts Aggregated by Situation and Period Of Observation (Hypothetical)

Situation Frequency

Person

MotherFatherYounger sisterOlder sisterBrotherGrandmother

Total

Mother

_42

1354

28

Father

17—

31881

47

Youngersister

35

—43

10

25

Oldersister

8119

—64

38

Brother

24131020—

2

69

Grand-mother

1125

127

—

37

Time 1

633529672921

Time 2

673133723414

procedurally Patterson's (1975, 1979) func-tional analysis of coercive behaviors. Patter-son and his associates have observed individ-uals within the context of family interaction.Table 6 presents a hypothetical illustrationof the kinds of frequency tallies yielded bytheir observational strategy. The act categoryfor aversive behaviors, for example, mightinclude crying, ignoring others, noncomply-ing with requests, teasing, whining, yelling,and humiliating others. The Time 1 columnsums the frequency of aversive acts by eachfamily member over the period of observa-tion. The Time 2 column gives a hypotheticalresult for a second period of observation. Thelatter information, typically not presented,serves to demonstrate that by tracking per-sons useful data can be produced for an actfrequency assessment of the dispositionalconstruct of aversiveness.

Table 6 also breaks down the manifesta-tions of aversive acts according to the familymember with whom the person was inter-acting. These units can be treated as situa-tions or, within Patterson's theoretical per-spective, as discriminative stimuli (Patterson& Bechtel, 1977), The primary focus of Pat-terson's analyses has not been upon dispo-sitional constructs, and he has not analyzedthe internal structure of the constructs (e.g.,aversiveness). Nevertheless, the operationsemployed would yield summary indices ofthe kind required by the act frequency ap-proach (e.g., the Time 1 and Time 2 columns)as well as multiple-act indices that are situ-ationally specific (e.g., Mother and Brothercolumns).

Finally, Table 6 illustrates the usefulnessof an index of situational scope in act fre-quency analysis. Situational scope assessesthe range or variety of situational contexts inwhich the individual manifests prototypicalacts from a dispositional category (e.g., di-recting aversive acts to all members of thefamily versus only to the youngest sister).Situational scope is a person variable and ananalogue to dispositional versatility (the rangeor variety of different prototypical acts man-ifested over a period of observation, e.g., yell-ing, teasing, and whining versus just teasing).

Within an act frequency analysis, each in-dividual's conduct can be tallied across allsituations or across specific subclasses of sit-uations. If act-trend indices differ signifi-cantly across certain classes of situations andthe differential is stable over time, then thiscontingent relation could be used in predict-ing future act trends. Detailed knowledge ofsituational contingencies of act trends mightwarrant a shift of dispositional constructsfrom the status of categorical summaries(Hampshire, 1953) to that of hypotheticalpropositions (Ryle, 1949).

In contrast to Hampshire, Ryle and others(Tuomela, 1978) argue that personal dispo-sitions are hypothetical propositions, akin todispositional statements in physics (e.g., "theglass is brittle") and take the form: It is likelyor a good bet that the entity will respond incertain ways (x, y, z; e.g., "shatter") to certaincircumstances (a, b, c; e.g., "being hit by astone"). This strategy of subaggregation ofact trends by types of situations presupposessome advances in formulating effective sys-


terns for the categorization of situations(Magnusson, 1981).

Personality Coherence

As Cattell (1957) and Block (1977) haverecognized, it is useful to distinguish betweenkinds of personality data. Most personalityresearch entails data subsumed by four cat-egories: O data (based upon observer re-ports), L data (based upon personal and so-cietal life outcomes), T data (based upon lab-oratory test situations), and S data (basedupon self-reports). Temporal stability of Odata and S data can be demonstrated (Mis-chel, 1968; Block, 1971). Furthermore, sub-stantial relations that make psychologicalsense can be shown among these variouskinds of personality data, giving evidence forthe orderly and robust phenomena withwhich this field of inquiry deals. For example,in the analysis of longitudinal data, Block(1971) has found impressive coherence be-tween Q-sort descriptions for the adolescentperiod (O data) and personality scale scoresfor the mid-30s (S data), whereas Block andBlock (1980), Buss (1981b), and Buss, Block,and Block (1980) have identified meaningfulpatterns of relations between O data and Tdata during the childhood years. Also, S datahave been related to a host of life outcomes(L data), including school achievement(Gough, 1968) and architectural creativity(Hall & MacKinnon, 1969).

The act frequency approach makes twocontributions to the study of personality co-herence. First, it offers act trends (A data) asa fifth hybrid class of personality data, whichcan be usefully differentiated from each ofthe other four types. Second, it highlights anomission that handicaps any personality-re-search agenda that restricts itself to exam-ining the interrelations within or among typ-ical forms of O data, L data, T data, and Sdata. Consider the example of a researcherwho sets out to study professional musiciansby assessing music-conservatory studentsthrough (a) personality descriptions by peers(O data), (b) performance in laboratory tests(T data), and (c) mail-back personality in-ventories (S data), which in later years are allrelated to personal and professional life out-comes (e.g., marital status, prominence as a

performer: L data). What is lacking in thisresearch design is any monitoring of what theindividuals do all day—a missing link thatcan be supplied in part by the act-frequencyapproach.

Act-Trend Data (A Data)The classification of S, O, L, and T data

is based primarily upon the source of data(i.e., from self, observer, laboratory, or societyand other sciences). In this sense, A data area hybrid species that can derive from themonitoring of acts by either self or observers.But from an alternative perspective, the typ-ical forms of S and O data actually draw upononly limited subclasses of data from their re-spective sources and, in each case, differentsubclasses than those that generate A data.S data are typically based upon self-ratingsof dispositions or upon assessments of dis-positions from personality scales, whereas Adata are based specifically on tallies of self-monitored acts. O data are typically derivedfrom dispositional attributions by observers;A data are based upon tallies of observers'recordings of specific monitored acts.

Before comparing A data with the otherkinds of personality data, several additionalfeatures of A data warrant notice. First, theact remains the primitive term in this ap-proach and requires further explication. Suf-fice it to note here that acts as typically ap-perceived by observers (Murray, 1938) entailnot only the physical movements (or actones,in Murray's terminology) but also ingredientsof style and intensity of the act and of itscontext. All of these elements play a role (notyet fully examined) in the categorization andprototypicality rating of specific acts vis-a-visdispositional constructs. Second, adequateanalysis of the internal structure of disposi-tional-act categories presents a challenge. Is-sues in the act-nomination and prototypi-cality-rating procedures have already beennoted. The phrasing of written accounts ofacts is another issue. Queneau (1947/1981)has demonstrated how a simple social epi-sode can be depicted in 195 different literarystyles (e.g., narrative, insistence, reportedspeech, exclamations). And the gathering ofprototypicality ratings of ongoing or recordedacts (e.g., videotaping) presents its own prob-lems.


Third, the challenge of assessing act trendsof individuals over extended periods of time(required by the act frequency approach)poses formidable problems. A variety ofmethods are available for such assessment,but each method carries serious limitations.Retrospective accounts of act performance,both by actor and observers, offer one method."Beeper" technology (e.g., Czikszentmihalyi,1975; Pawlik & Buse, 1982; Sjoberg, 1981),in which individuals carry devices that signalat random intervals, notifying the subjects torecord their actions at that moment, offersa second method. Contemporaneous ob-server recording of acts offers a third method.Each poses a unique set of problems, andideally, convergence should be obtained acrossmethods.

Barker and his associates at the MidwestPsychological Field Station (Barker & Wright,1955; Barker, 1963) initially sought ways ofsegmenting the stream of behavior beforethey shifted from behavior episodes to be-havior settings as the primary unit of analysisin ecological psychology. The act frequencyapproach requires a return to that original,person-centered, observational mission butpoints beyond One Boy's Day (Barker &Wright, 1951) to, for example, One Domi-nant Boy's Day. This task will require atten-tion to the perceptual units of social action(Collett, 1980; Newtson, Engquist, & Bois,1977), to the role of context, and to pluralisticmeanings in the interpretation of social acts(Rommetveit, 1980). Procedures for self-monitoring and other monitoring of thestream of behavior have received increasingattention in recent research (Epstein, 1979,1980; Forgas, 1976; McGowan & Gormley,1976; Mischel & Peake, 1982; Pervin, 1976).

A Data and O Data

O data, as recorded consensual and stableimpressions that observers form about thepersonalities of target individuals, constitutefundamental sources for personality assess-ment (Block, 1961; Wiggins, 1973). The pro-cesses entailed when observers move from theperception of acts to the attribution of dis-positions have also been a major topic in re-

search on person perception (Hastorf,Schneider, & Polefha 1970).

The act frequency approach suggests thatinferences about the dispositions of others arebased largely but not completely on memorytraces of observed act trends. This positionhas wide-ranging implications for the direc-tion and interpretation of attribution re-search. For example, the influential researchprogram of Jones and Davis (1965) is guidedby a notion of the act and a general concep-tion of disposition that are not seriously atodds with the act frequency approach. Yettheir typical research design presents judgeswith a single act of the target person, fromwhich inferences about dispositions are re-quested. Indeed, their construct of the "cor-respondence of inference" specifies that "adisposition is being rather directly reflectedin behavior and that this disposition is un-usual in its strength or intensity" (p. 264). Incontrast, the act frequency approach holdsexplicitly that a single act is an inadequatebasis for dispositional inferences; such infer-ences are more appropriately based upon acttrends over a period of observation. Neitherthe intensity nor the consequences of a singleact offer a strong or sensible foundation fordispositional inference. Thus, much of theattribution-research literature based on sin-gle acts or adjectival descriptions is largelyirrelevant from the standpoint of the act fre-quency approach.

Inferring dispositions from observed-acttrends must be recognized as a complex cog-nitive process. Any act is a potential memberof several dispositional categories, especiallyat the periphery of act categories. The infer-ence process presumably entails the encodingand monitoring of topographically dissimilaracts and requires tracking of a person's acttrends, which are extended over time and in-terspersed among other assortments of actsand act trends. Finally, the internal structureof manifested dispositions may not generallybe very tight. Yet to gain an understandingof act-to-disposition inferences, direct studyof the perceptions of act trends and inferencesabout them is necessary.

The evidence that dispositional inferencesmay be influenced by factors other than theperception of act trends complicates matters


even more. First, the nature of natural cog-nitive categories may lead to asymmetric ex-pectations at the level of act-to-categoryjudgements. If the observed person displaysa highly prototypical act (e.g., of dominance),then the perceived likelihood of his or herperforming a given peripheral act may bedeemed relatively high. But in the reversecase, a person performing a peripheral actmay not be expected to perform a given coreact. Rips (1975) has found similar asym-metries of inference in examining categoriesof natural objects (e.g., if a prototypical bird,such as a robin, is found to have a new con-tagious disease, the judged likelihood is alsohigh for a more peripheral form of bird, suchas a duck, but the reverse does not appear tohold). Asymmetric expectations of this sortcould affect the perception of act trends andwarrant study. Second, other cognitive sche-mata, perhaps based upon empirical regu-larities, are at play in dispositional inferences,apart from the role of observed act trends.These factors include inferences based uponnonact attributes (e.g., clothing style) andthose derived from cross-dispositional impli-cative systems (e.g., evidence for wit leadingto inferences regarding intelligence).

A Data and L Data

Life-outcome data are typically found inindices derived by social institutions andother scientific disciplines. The range in-cludes biological and medical indices (e.g.,gender, diagnosed alcoholism), sociodemo-graphic indices (e.g., social class), institu-tional outcomes (e.g., graduation from lawschool, award for creative performance, pro-motion to higher position), and even certainsocially salient single acts (e.g., of heroism,crime). The relation of act-trend indices forpersonality dispositions to specific forms ofL data is an empirical question and repre-sents an important research agenda linkingpersonality to sociology and the other sci-ences.

A Data and T Data

The relations between T data collectedunder laboratory conditions and A data mon-

itored from daily life depend upon severalfactors. The degree of mundane realism(Aronson & Carlsmith, 1968) of the con-trived laboratory situation has some bearing,but two other considerations are perhapsmore important. First, the act frequency ap-proach offers a procedure for the conceptualanalysis of dispositional constructs. In con-trast, T data are often based upon contrivedacts whose centrality to the disposition at is-sue is not established. Second, the act fre-quency approach considers multiple-act in-dices in the form of act trends as appropriatecriteria for personality prediction, whereasstudies employing T data have often soughtto predict single-act criteria (Epstein, 1980).The degree of coherence to be expected be-tween A data and T data depends upon theadequacy of procedures for generating Tdata, a point that has been amply illustratedelsewhere with regard to the relations be-tween O data and T data (Block & Block,1980).

A Data and S Data

The use of A data as multiple-act-criterionindices for the validation of personality scaleshas been demonstrated (Buss & Craik, 1980,1981, in press). In addition, the act frequencyapproach contributes new facets to the moredetailed conceptual analysis of psychologicalscores and measures (Gough, 1965). Theseconceptual indicators based upon A data in-clude act density (the number of significantact correlates of a scale within the nominallyappropriate act category), act bipolarity (thenumber of significant act correlates of a scalefrom the semantically opposing act category),and act extensity (the number of significantact correlates within act categories other thanthe nominally appropriate and semanticallyopposing act categories) (Buss & Craik, inpress; Buss, 198la, 1981c).

Summary and Discussion

The act frequency approach to personalityadopts the categorical-summary view of per-sonal dispositions. A dispositional assertionrefers to the relative frequency with whichthe individual has displayed acts counting asmembers of that dispositional category, over


a period of observation. These multiple-actindices, or act trends, are a fundamental formof personality data and represent a theoreti-cally sanctioned union of the concept of dis-position with principles of aggregation andreliability of measurement.

This approach to personality treats dis-positional constructs as natural cognitive cat-egories of topographically dissimilar acts,with status of act members varying in pro-totypicality from core to peripheral standing.Recent advances in cognitive psychology pro-vide guidance and procedures for an empir-ical research program examining the internalstructure of dispositional categories of actsas well as the horizontal and vertical dimen-sions of this system of act categorization.

Although founded in behavioral observa-tion, the act frequency approach draws a ba-sic distinction between behavioral consis-tency (covariation among observed acts) anddispositional consistency (covariation amongact-trend indices) and identifies several dis-tinctive facets of dispositional breadth. Theprimary frame of measurement in this ap-proach is the period of observation ratherthan a set of specified situations. In its mostgeneral form, the act frequency approachproceeds without regard for attributions ofcausality to persons or situations. However,for analytic purposes, the approach can bejoined to any system of situational classifi-cation, examining act trends subaggregatedaccording to specific types of situation. Fi-nally, the act-frequency approach contributesact-trend data (A data) as a neglected ingre-dient in the delineation and understandingof coherence in personality, augmenting themore often explored forms of observer-based(O data), self-report (S data), laboratory (Tdata), and life-outcome (L data) measures.

Dispositional constructs as categories ofacts are sociocognitive devices or inven-tions—emergents of sociocultural evolution(Campbell, 1965). An effort to understandthe origins and basis of this system for thecategorization of acts leads to some of themost difficult and problematic questions ofpersonality theory. Dispositional constructsserve the specific function of facilitating thecategorization of trends in the acts of persons.It is reasonable to assume that core or pro-totypical acts served as reference acts around

which dispositional categories emerged(Rosch, 1975a; Rosch, Mervis et al., 1976).Three possible sources of emergence can beidentified.

First, dispositional categories may havedeveloped to capture observed cooccurrencesof acts in the streams of individuals' behavior.Individual differences in frequencies for sin-gle acts may have preceded the discernmentand conceptualization of individual consis-tencies for cooccurrences in multiple-act cat-egories. In both cases, covariations in thestructure of individual action form the basisof dispositional categories, in a way not un-like that hypothesized for categories of ob-jects (Craik, 1981; Rosch & Mervis, 1975;Rosch, Mervis et al., 1976).

Second, dispositional categories may haveformed around reference acts that share asalient or notable attribute, somewhat or en-tirely apart from their cooccurrence withinthe stream of individuals' behavior. Jones andDavis (1965) focus upon the assumed desir-ability of the act's effects or its hedonic rel-evance. Wiggins (Note 1) also points to com-mon consequences of acts as the key attributebut strives for a nonevaluative considerationfrom an institutional or rule-system stand-point (e.g., an act likely to harm or injureanother counts as aggressive in the contextof rules for classifying the social conse-quences of actions). Given this possibility,within the act frequency approach the as-sumption regarding the internal structure ofdispositional categories would take prece-dence over the summary notion, and a looserdepiction of variation in the structure ofmanifested individual acts could be antici-pated.

Third, dispositional constructs could re-flect generative mechanisms of action. In thedomain of colors, human categorization ap-pears to possess a physiological basis (i.e., fo-cal colors correspond to properties of color-vision; Kay & McDaniel, 1978; Rosch, 1975a)from which the cross-cultural generality ofRosch's findings in this domain may follow.The relation of dispositional categories ofacts to generative mechanisms of act trendscannot be established at this time but rep-resents a central issue in personality theory(Wiggins, Note 2). If dispositional categoriesare formed primarily around attributes of


acts that possess cultural salience, then im-portant cross-cultural variations in contentand internal structure are possible and thelikelihood that such categories mirror or pro-vide important clues about generative mech-anisms is diminished. Cross-cultural researchbearing on this topic is appearing (White,1980), but studies dealing specifically withthe internal cognitive structure of disposi-tional categories would speak directly to thisissue. '

Within current personality theory, Alston(1970, 1975, Note 3) has identified a con-ceptual divide between frequency conceptsof disposition and purposive-cognitive con-cepts. He deems this distinction the most fun-damental alternative for personality research.Unlike frequency concepts, purposive-cog-nitive concepts do not entail a category ofoccurrences that can be specified to count asdisplays of a disposition. Instead, they derivetheir meaning and standing from their placein a theoretical framework (typically moti-vational in nature) intended to explain be-havioral occurrences. Purposive-cognitiveconcepts include desires, beliefs, and abilitiesthat function within a field of tendencies be-tween activated desires and guiding beliefs onthe one hand and manifested behavior on theother. Alston points to psychoanalytic theory(Rapaport, 1959) and cognitive social-learn-ing conceptualizations of personality (Mis-chel, 1973) as distinctive examples of thepurposive-cognitive approach.

Alston further argues that in addition toother theoretical jobs, purposive-cognitiveconcepts offer an explanation of frequencydispositions. Between observed acts and pur-posive-cognitive concepts, frequency dispo-sitions form a middle level of analysis. Butit is possible, instead, that the act frequencyand purposive-cognitive concepts may becomponents of incommensurate approaches.Consider Figure 1: The act frequency ap-proach would assess Person A as more dom-inant than Person B and forecast a continuingdifferential manifestation of dominant acts.Marshalling purposive-cognitive conceptsrequires a more indirect procedure, includinga theoretical formulation of the individualand an interpretation of the observed behav-ior. For example, the analysis of mood states,displaced affective outbursts, slips of the

tongue, and dreams might lead to the infer-ence that Person B has a strong need for dom-inance that is inhibited by a stronger fear ofrejection; Person A, with a weak need fordominance, incurs heavy psychological costsin his many acts of dominance, which atsome points are motivated by a fear of failurein an institutional role that requires suchdisplays and at other points by a fear of al-lowing someone else to take control of situ-ations. The contrast in the functions of thefrequency concept of dominance and the pur-posive-cognitive concept of need for domi-nance in the analysis of acts within the streamof behavior is striking. The purposive-cog-nitive explanation for Person A's observedbehavior also requires a subaggregation ofdominant acts not entailed by the act fre-quency approach. The relation of frequencyand purposive-cognitive approaches clearlyrequires further theoretical examination.

In his search for intraindividual measuresof personality, G. W Allport (1937, 1958,1962) often championed F. H. Allport's(1937) concept of teleonomic trend. Like dis-positional act trends, teleonomic trends arederived from the monitoring of an individ-ual's daily acts over a period of observation.Instead of being aggregated into dispositionalcategories, however, acts are ordered accord-ing to what the person is trying to do throughthem (e.g., seeking justice, trying to maintainself-esteem, avoiding responsibility, helpingothers, or gaining the attention of elders).According to F. H. Allport, such concepts canbe used objectively and reliably by observersto order acts in a way that indicates how theperson is trying to make adjustments orchanges in his or her environment througheveryday acts.

A purposive-cognitive variant of teleon-omic-trend analysis can be found in the re-cent formulation of personal-projects analy-sis (Palys, Little, & Baker-Brown, Note 4).This approach examines the acts of personsby means of the self-reported ordering ofthem according to the concept of personalprojects, that is, sequences of goal states andmeans-end, or instrumental, acts. The con-cept can be considered a type of serial pro-ceeding within Murray's (1938) framework.Act frequency analysis of dispositions andpersonal-projects analysis offer two quite dif-


ferent systems for the classification of acts.The act of complaining when served a toughsteak at a restaurant, for example, may beclassified with other topographically dissim-ilar manifestations of dominance in the actfrequency approach but might be groupedinstead with other and different topographi-cally dissimilar acts within the personal-pro-jects category such as getting a promotion(in this instance, by impressing the boss).

Efforts to clarify the conceptual and em-pirical interrelations among various middle-level personality approaches to the categori-zation of acts and to explanatory systemsoffer an important road to the revival of theo-retical discourse that Maddi (1980) has ad-vocated. Indeed, this endeavor is likely tooccupy personality theorists in a profitablefashion during this decade of the 1980s. Inthe meantime, the act frequency approach topersonality poses an extensive research agendaand raises a host of novel questions. Whatare the prototypical act members of impor-tant dispositional categories? How do dis-positional constructs differ in terms of theircategory volume, the range and central ten-dency of act membership, and the tightnessof their manifested structure? What are theeffects of various instructional sets upon actnominations and prototypicality ratings? Howdo alternative indices of act prototypicalityinterrelate? Does the average between-rateragreement for prototypicality ratings of actsdiffer from that for objects? Are there age-specific variations in prototypical acts forsome dispositional constructs? Does act-trendanalysis support a circumplex model of in-terpersonal dispositions? In person percep-tion, are there asymmetries of inference to-ward more prototypical act members of dis-positional categories? What levels and formsof coherence obtain in the relations of act-trend data to observer ratings, to laboratoryand self-report measures, and to significantlife outcomes?

Reference Notes1. Wiggins, J. S. In defense of traits. Invited address to

the Ninth Annual Symposium on Recent Develop-ments in the Use of MMPI, Los Angeles, February28, 1974.

2. Wiggins, J. S. Caltell's system from the perspective ofmainstream personality theory. Unpublished manu-

script, University of British Columbia, Vancouver,British Columbia, Canada, 1979.

3. Alston, W. P. A conceptual divide in personality de-scriptions. Paper presented at the meeting of theAmerican Psychological Association (invited addressto Division 8), September 1973.

4. Palys, T. S., Little, B. R., & Baker-Brown, G. A project-oriented inquiry into community organization, dynam-ics and satisfactions. Unpublished manuscript, Uni-versity of British Columbia, Vancouver, British Co-lumbia, Canada, 1980.

ReferencesAddis, L. Dispositions, explanation and behavior. In-

quiry, 1981, 24, 205-227.Allport, F. H. Teleonomic description in the study of

personality. Character and Personality, 1937, 6, 202-214.

Allport, G. W. Personality and character. PsychologicalBulletin, 1921, 18, 441-455.

Allport, G. W. Concepts of trait and personality. Psy-chological Bulletin, 1927, 24, 284-293.

Allport, G. W. What is a trait of personality? Journal ofAbnormal and Social Psychology, 1931,25, 368-372.

Allport, G. W. Personality: A psychological interpreta-tion. New York: Holt, 1937.

Allport, G. W. What units shall we employ? In G. Lind-zey (Ed.), Assessment of human motives. New York;Rinehart, 1958.

Allport, G. W.The general and the unique in psycholog-ical science. Journal of Personality, 1962,30,405-422.

Allport, G. W. Traits revisited. American Psychologist,1966, 21, 1-10.

Allport, G. W., & Allport, F. H. Personality traits: Theirclassification and measurement. Journal of Abnormaland Social Psychology, 1921, 16, 6-40.

Allport, G. W., & Odbert, H. S. Trait-names: A psycho-lexical study. Psychological Monographs, 1936, 47(1,Whole No. 211).

Alston, W. P. Toward a logical geography of personality:Traits and deeper lying personality characteristics. InH. D. Kiefer & M. K. Munitz (Eds.), Mind, scienceand history. Albany, N.Y.: SUNY Press, 1970.

Alston, W. P. Traits, consistency and conceptual alter-natives for personality theory. Journal for the Theoryof Social Behavior, 1975,5, 17-48.

Armstrong, D. M. Dispositions are causes. Analysis,1969, 30, 23-36.

Aronson, E. & Carlsmith, J. M. Experimentation in so-cial psychology. In G. Lindzey & E. Aronson (Eds.),The handbook of social psychology, (Vol. 2, 2nd ed.).Reading, Mass.: Addison-Wesley, 1968.

Barker, R. G. (Ed.). The stream of behavior. New York:Appleton, 1963.

Barker, R. G., & Wright, H. F. One boy's day. New York:Harper & Row, 1951.

Barker, R. G., & Wright, H. F. Midwest and its children.New York: Harper & Row, 1955.

Block, J. The Q-sort method in personality assessmentand psychiatric research. Springfield, 111.: Charles C.Thomas, 1961.

Block, J. Lives through time. Berkeley, Calif.: BancroftBooks, 1971.


Block, J. Advancing the psychology of personality: Par-adigmatic shift or improving the quality of research?In D. Magnussen & N. S. Endler (Eds.), Personalityat the crossroads: Current issues in interactional psy-chology. Hillsdale, N.J.: Erlbaum, 1977.

Block, J. H., & Block, J. The role of ego-control and ego-resiliency in the organization of behavior. In W. A.Collins (Ed.), Minnesota Symposia on Child Psychol-ogy (Vol. 13). Hillsdale, N.J.: Erlbaum, 1980.

Brandt, R. B. Traits of character: A conceptual analysis.American Philosophical Quarterly, 1970, 7, 23-37.

Buss, D. M. The act frequency analysis of interpersonaldispositions. Unpublished doctoral dissertation, Uni-versity of California, Berkeley, 1981. (a)

Buss, D. M. Predicting parent-child interactions fromchildren's activity level. Developmental Psychology,1981, 17. 59-65. (b)

Buss, D. M. Sex differences in the evaluation and per-formance of dominant acts. Journal of Personality andSocial Psychology, 1981,40, 147-154. (c)

Buss, D. M. Evolutionary biology and personality psy-chology: Implications of genetic variability. Personal-ity and Individual Differences, 1983, 4, 51-63.

Buss, D. M., Block, J. H., & Block, J. Preschool activitylevel: Personality correlates and developmental impli-cations. Child Development, 1980, 57, 401-408.

Buss, D. M., & Craik, K. H. The frequency concept ofdisposition: Dominance and prototypically dominantacts. Journal of Personality, 1980, 48, 379-392.

Buss, D. M., & Craik, K. H. The act frequency analysisof interpersonal dispositions: Aloofness, gregarious-ness, dominance and submissiveness. Journal of Per-sonality, 1981, 49, 174-192.

Buss, D. M., & Craik, K. H. Act prediction and theconceptual analysis of personality scales: Indices of actdensity, bipolarity, and extensity. Journal of Person-ality and Social Psychology, in press.

Campbell, D. T. Variation and selective retention in so-cio-cultural evolution. In H. R. Barringer, G. I. Blank-sten, & R. W. Mack (Eds.), Social change in developingareas. Cambridge, Mass.: Schenkman, 1965.

Cattell, R. B. Personality and motivation structure andmeasurement. Yonkers-on-Hudson, N.Y.: World Book,1957.

Collett, P. Segmenting the behaviour stream. In M. Bren-ner (Ed.), The structure of action. New York: St. Mar-tin's Press, 1980.

Craik, K. H. The personality research paradigm in en-vironmental psychology. In S. Wapner, S. B. Cohen,& B. Kaplan (Eds.), Experiencing the environment.New York: Plenum, 1976.

Craik, K. H. Comment on "The psychological represen-tation of molar physical environments" by Ward andRussell. Journal of Experimental Psychology: General,1981, 110, 158-162.

Cummins, R. Dispositions, states and causes. Analysis,1974, 34, 194-204.

Csikszentmihalyi, M. Beyond boredom and anxiety. SanFrancisco: Jossey-Bass, 1975.

Epstein, S. The stability of behavior: I. On predictingmost of the people much of the time. Journal of Per-sonality and Social Psychology, 1979,37, 1097-1126.

Epstein, S. The stability of behavior: II. Implications forpsychological research. American Psychologist, 1980,35, 790-806.

Eysenck, H. J. The structure of human personality. NewYork: Wiley, 1953.

Eysenck, H. J. A model for personality. Berlin: Springer-Verlag, 1981.

Fishbein, M. The prediction of behaviors from attitudevariables. In K. K. Sereno & C. D. Mortensen (Eds.),Advances in communication research. New York: Har-per & Row, 1972.

Fishbein, M., & Ajzen, I. Attitudes toward objects aspredictions of single and multiple behavioral criteria.Psychological Review, 1974, 81, 59-74.

Forgas, J. P. The perceptions of social episodes: Cate-gorical and dimensional representations in two dif-ferent social millieus. Journal of Personality and SocialPsychology, 1976, 34, 199-209.

Goldberg, L. R. From ace to zombie! Some explorationsin the language of personality. In C. O. Spielberger& J. N. Butcher (Eds.), Advances in personality as-sessment (Vol. 1). Hillsdale, N.J.: Erlbaum, 1982.

Gough, H. G. Conceptual analysis of psychological testscores and other diagnostic variables. Journal of Ab-normal Psychology, 1965, 70, 294-320.

Gough, H. G. An interpreter's syllabus for the CaliforniaPsychological Inventory. In P. McReynolds (Ed.), Ad-vances in psychological assessment (Vol. 1). Palo Alto,Calif.: Science and Behavior Books, 1968.

Guilford, J. P. Personality. New York: McGraw-Hill,1959.

Hall, W. B., & MacKinnon, D. W. Personality inventorycorrelates among architects. Journal of Applied Psy-chology, 1969, 53, 322-326.

Hampshire, S. Dispositions. Analysis, 1953, 14, 5-11.Hampson, S. E. Person memory: A semantic category

model of personality traits. British Journal of Psy-chology, 1982, 73, 1-11.

Hastorf, A. H., Schneider, D. J., & Polefka, J. Personperception. Reading, Mass.: Addison-Wesley, 1970.

Horowitz, L., Inouye, D., & Siegelman, E. Y. On aver-aging judges' ratings to increase their correlation withan external criterion. Journal of Consulting and Clin-ical Psychology, 1979, 47, 453-458.

Jaccard, J. J. Predicting social behavior from personalitytraits. Journal of Research in Personality, 1974, 7,358-367.

Jones, E. E., & Davis, K. E. From acts to dispositions:The attribution process in person perception. In L.Berkowitz (Ed.), Advances in experimental social psy-chology (Vol. 2). New York: Academic Press, 1965.

Kay, P., & McDaniel, C. K. The linguistic significanceof the meanings of basic color terms. Language, 1978,57, 383-405.

Leary T. Interpersonal diagnosis of personality. NewYork: Ronald Press, 1957.

Loevinger, J. Objective tests as instruments of psycho-logical theory. Psychological Reports, 1957, 3, 635-694 (Monograph No. 9).

Maddi, S. R. The uses of theorizing in personology. InE. Staub (Ed.), Personality: Basic aspects and currentresearch. Englewood Cliffs, N.J.: Prentice-Hall, 1980.

Magnusson, D. (Ed.). Toward a psychology of situations:An interactional perspective. Hillsdale, N.J.: Erlbaum,1981.

Magnusson, D., & Endler, N. S. (Eds.). Personality at thecross-roads: Current issues in interactional psychology.Hillsdale, N.J.: Erlbaum, 1977.


McClelland, D. C. The nature of human motivation.Glenview, 111.: Scott, Foresman, 1983.

McGowan, J., & Gormly, J. Validation of personalitytraits: A multicriteria approach. Journal of Personalityand Social Psychology, 1976, 34, 791-795.

Mischel, W. Personality and assessment. New York:Wiley, 1968.

Mischel, W. Toward a cognitive social learning recon-ceptualization of personality. Psychological Review,1973, 80, 252-283.

Mischel, W., & Peake, P. In search of consistency: Mea-sure for measure. In M. P. Zanna, E. T. Higgins, &C. P. Herman (Eds.), Consistency in social behavior:The Ontario Symposium of Personality and SocialPsychology (Vol. 2). Hillsdaie, NX: Erlbaum, 1982.

Murray, H. A. Explorations in personality. New York:Oxford,. 1938.

Newtson, D., Engquist, G., & Bois, J. The objective basisof behavior units. Journal of Personality and SocialPsychology, 1977, 35, 847-862.

O'Shaughnessy, B. The powerlessness of dispositions.Analysis, 1970, 31, 1-15.

Ozer, D. J. Consistency, coherence, and convergence: Aframework for the study of personality. Unpublisheddoctoral dissertation, University of California, Ber-keley, 1982.

Patterson, G. R. A three-stage functional analysis forchildren's coercive behaviors: A tactic for developinga performance theory. In B. C. Etzel, J. M. LeBlanc,& D. M. Baer (Eds.), New developments in behavioralresearch: Theory, methods and applications. Hillsdaie,N.J.: Erlbaum, 1975.

Patterson, G. R. A performance theory for coercive fam-ily interaction. In R. Cairns (Ed.), Social interaction:Methods, analysis and illustration. New York: Hal-stead Press, 1979.

Patterson, G. R., & Bechtel, G. G. Formulating the sit-uational environment in relation to states and traits.In R. B. Cattell & D. M. Dreger (Eds.), Handbook ofmodern personality theory. New York: Halstead, 1979.

Pawlik, K., & Buse, L. Rechnergestiitze verhaltensregis-trierung im feld: Beschreibung und erste psychome-trische iiberpriifung einer neuen erhebungsmethode.Zeitschrift fur Differentielle und Diagnostische Psy-chologie, 1982,5, 101-118.

Pervin, L. A. A free-response description approach tothe analysis of personsituation interaction. Journal ofPersonality and Social Psychology, 1976,34,465-474.

Powell, B. Uncharacteristic actions. Mind, 1959, 68,492-509.

Queneau, R. Exercises in style (B. Wright Trans.). NewYork: New Directions, 1981. (Originally published,1947.)

Rapaport, D. The structure of psychoanalytic theory: Asystematizing attempt. In S. Koch (Ed.), Psychology:A study of a science, (Vol. 3). New York: McGraw-Hill, 1959.

Rips, L. J. Inductive judgements about natural catego-ries. Journal of Verbal Learning and Verbal Behavior,1915,14, 665-681.

Rommetveit, R. On "meanings" of acts and what ismeant and made known by what is said in a pluralisticsocial world. In M. Brenner (Ed.), The structure ofaction. New York: St. Martin's Press, 1980.

Rosch, E. Cognitive reference points. Cognitive Psy-chology, 1975, 7, 532-547. (a)

Rosch, E. Cognitive representations of semantic cate-gories. Journal of Experimental Psychology: General,1975, 104, 192-233. (b)

Rosch, E. Principles of categorization. In E. Rosch &B. B. Lloyd (Eds.), Cognition and categorization. Hills-dale, N.J.: Erlbaum, 1978.

Rosch, E., & Mervis, C. B. Family resemblances: Studiesin the internal structure of categories. Cognitive Psy-chology, 1975, 7, 573-605.

Rosch, E., Mervis, C. B., Gray, W. D., Johnson, D. M.,& Boyes-Braem, P. Basic objects in natural categories.Cognitive Psychology, 1976,5, 382-439.

Rosch, E., Simpson, C., & Miller, R. S. Structural basesof typicality effects. Journal of Experimental Psychol-ogy: Human Perception and Performance, 1976, 2,491-502.

Russell, J. A. Affective space is bipolar. Journal of Per-sonality and Social Psychology, 1979, 37, 345-356.

Ryle, G. The concept of mind. New York: Barnes & No-ble, 1949.

Sarbin, T. R., & Allen, V. L. Role theory. In G. Lindzey& E. Aronson (Eds.), The handbook of social psy-chology (Vol. 1, 2nd ed.). Reading, Mass.: Addison-Wesley, 1968.

Skinner, B. F. The behavior of organisms. New York:Appelton-Century-Crofts, 1938.

Spearman, C. "General intelligence" objectively deter-mined and measured. American Journal of Psychology,1904, 15, 201-293.

Squires, R. Are dispositions causes? Analysis, 1968, 29,45-47.

Squires, R. Are dispositions lost causes? Analysis, 1970,31, 15-18.

Sjoberg, L. Life situations and episodes as a basis forsituational influence on action. In D. Magnusson(Ed.), Toward a psychology of situations: An interac-tional perspective. Hillsdaie, N.J.: Erlbaum, 1981.

Tuomela, R. (Ed.). Dispositions. Boston, Mass.: D. Rei-del, 1978.

White, G. M. Conceptual universals in interpersonal lan-guage. American Anthropologist, 1980, 82, 759-781.

Wiggins, J. S. Personality and prediction: Principles ofpersonality assessment. Reading, Mass.: Addison-Wes-ley, 1973.

Wiggins, J. S. A psychological taxonomy of trait descrip-tive terms: I. The interpersonal domain. Journal ofPersonality and Social Psychology, 1979,3 7. 395-412.

Wiggins, J. S. Circumplex models of interpersonal be-havior. In L. Wheeler (Ed.), Review of personality andsocial psychology (Vol. 1). Beverly Hills, Calif.: Sage,1980.

Wiggins, J. S. Clinical and statistical prediction: Whereare we and where do we go from here? Clinical Psy-chology Review, 1981, 1, 3-18.

Zadeh, L. A., Fu, K. S., Tanaka, K., & Shimura (Eds.).Fuzzy sets and their applications to cognitive and de-cision processes. New York: Academic Press, 1975.

Received May 5, 1982Revision received November 1, 1982

the act frequency approach to personality. psychological review

Documents