Assessment of Self-Regulative Traitspeople.psych.cornell.edu/~pac_lab/pdf/Zayas,Whitsett,etal... ·...

PART IV

Assessment of Self-RegulativeTraits

9781412946520-Ch18 5/23/08 12:40 PM Page 375

9781412946520-Ch18 5/23/08 12:40 PM Page 376

From Situation Assessment toPersonality: Building a Social-Cognitive Model of a Person

Vivian Zayas, Donna D. Whitsett, Jenna J.Y. Lee, NicoleWilson and Yuichi Shoda

INTRODUCTION

Cross-situational variability has posed apuzzle for personality psychology: How doesone reconcile the fact that any given person’sbehavior varies across situations and frommoment-to-moment with a central assump-tion of personality theories that personality isstable and consistent across situations? Inother words, how does one account for the invariance of personality while at thesame time taking into account the dynamicchanges in the cognitions, affects, and behav-iors experienced by the individual?

Take, for example, David Letterman. Hehas been a stand-up comic since 1975 andhas been the host of one of the most watchedlate night shows in America. Each evening,he tells witty jokes, engages in playful banterwith his audience, interviews his guests, andentertains millions of TV viewers with histrademark gap-toothed smile. It is thereforesurprising to learn that in his personal life he is extremely shy and introverted

(Walters, 1992). How does one account forsuch variability?

In the last decade, research has shownthat it is possible to identify predictable pat-terns of variability across situations. Thesepatterns of cross-situational variability canbe seen in stable if ... then ... situation-behavior profiles, or ‘behavioral signatures’that characterize each individual (Shoda etal., 1994). They capture how a person’sbehavior varies reliably as a function of theparticulars of the situation a person encoun-ters. For example, David Letterman’sbehavioral signature could be described as:if David Letterman is performing on stagein front of thousands of people, then he istalkative, extraverted, and outgoing. Ifinstead he is at a small, intimate gathering,then he is reclusive, withdrawn, and shy.But what about these situations is responsi-ble for this pattern? Is it because the TVaudience is anonymous and impersonal, oris it because there are well-defined scriptsand props on the set, for example?

18

9781412946520-Ch18 5/23/08 12:40 PM Page 377

In the present chapter, we propose that akey for understanding stable individual differ-ences in people’s behaviors, as well as thepatterns of intraindividual variability for eachperson, is in knowing the features of situa-tions that are psychologically active for agiven person. Psychologically active featurestrigger particular cognitive and affectiveprocesses that ultimately lead to predictableresponses in feelings, thoughts, and actions.This chapter will summarize and discussrecent conceptual and methodologicaladvances in the identification of situationalfeatures. Three research examples illustratesix steps that can be used to identify (1) a set ofpsychological features that capture importantaspects of a given situation, and (2) the featuresthat are particularly salient for a given individ-ual or groups of individuals. This approach, inturn, allows the assessment of each person’sstable and distinctive ‘behavioral signature’, orif ... then ... situation-behavior profile, relatinga person’s thoughts, feelings, and behaviors tothe situation, both external and internal, thatthe person encountered.

MAKING SENSE OF CROSS-SITUATIONAL VARIABILITY: IF ...THEN ... SITUATION-BEHAVIORPROFILES

Some of the early evidence that cross-situa-tional, intraindividual behavioral variabilityis meaningful (as opposed to error variancethat needs to be removed or controlled) wasobtained in a study of children at a residentialsummer camp (Shoda, 1990; Shoda et al.,1989, 1993a, 1993b, 1994; Wright andMischel, 1987, 1988). Countless hours ofobservations revealed that each child’saggressive behaviors were systematicallyrelated to the features present in his or hersituation. For example, for some children,being teased by peers elicited high levels ofaggressive behavior, but being reprimandedby a counselor did not. Other childrenshowed the opposite pattern; being scolded

by a counselor elicited high levels of aggres-sive behaviors, but being teased by peers didnot. Thus, each child was characterized by aunique ‘behavioral signature’, or if ... then ...situation-behavior profile (e.g., if teased bypeers, then the child is aggressive.). Mostimportantly, these situation-behavior profilesproved to be highly stable across time. Thechild who showed the ‘if teased by peers,then the child is aggressive’ pattern duringone half of the summer session showed asimilar if ... then ... profile in the other half ofthe summer session.

THE MIND AS A COGNITIVE-AFFECTIVEPROCESSING SYSTEM

In response to the discovery of if … then …situation-behavior profiles, Mischel andShoda (1995) proposed a reconceptualizationof personality that could account for both thevariability of a person’s behavior across situ-ations as well as consistency in a person’sbehavior over time. Inspired by cognitive andneural network models of the mind, the cog-nitive-affective processing system (CAPS)approach (Mischel and Shoda, 1995; Shodaand Mischel, 1998) conceptualizes eachperson’s mind as a network of interconnectedcognitions and affects. Figure 18.1 providesschematic representations of the CAPS net-works of two hypothetical individuals.

A key premise of the CAPS approach isthat each person’s network remainsunchanged from situation to situation. Whatchanges from situation to situation are theparticular cognitions and affects that becomeactivated within the network, which in turninfluence the corresponding behaviors thatbecome expressed. Each person’s unique andstable network is responsible for mediatingthe effect of the situation on behavior byguiding how a person encodes and construesa particular situation. The network alsoguides the goals, expectations, values, andstrategies that become activated in the partic-ular situation. The pattern and strength of

378 THE SAGE HANDBOOK OF PERSONALITY THEORY AND ASSESSMENT

9781412946520-Ch18 5/23/08 12:40 PM Page 378

associations among the cognitions andaffects guide the activation of thoughts andfeelings within the network. For example,when two cognitions are strongly associatedwith one another, the activation of one willautomatically spread and activate the other.In contrast, when two cognitions are weaklyassociated or unassociated with one another,the activation of one will not necessarily acti-vate the other. Associations among cogni-tions may also be inhibitory such thatactivation of one concept will make it moredifficult to activate another.

The networks of the two individuals illus-trated in Figure 18.1 are assumed to differ inthe cognitions and affects that are availablewithin each person’s network, the strengthsof associations among those available thatinfluence the accessibility of particular cog-nitions and affects, and the strengths of asso-ciations between external features andcognitions and affects within the network.

The interactions among cognitive-affec-tive processes within a person’s network andthe sensitivity of the network to situational,external or internal, influences give rise to

FROM SITUATION ASSESSMENT TO PERSONALITY: BUILDING A SOCIAL-COGNITIVE MODEL 379

Figure 18.1 Schematic representation of the cognitive-affective processing system (CAPS) fortwo hypothetical individuals (i.e., person A and person B). Each person’s mind is conceptualizedby a stable network of interconnected cognitions and affects that mediates the effect of thesituational features on behavior. Solid lines within and outside of the network represent exci-tatory associations (e.g., activation of one cognition automatically activates associated cogni-tions). Dotted lines within and outside of the network represent inhibitory associations (e.g.,activation of one cognition makes it more difficult to activate associated cognitions). In theabove illustration, each person encounters the same situation that consists of a common setof features (e.g., a through n). Because not all features are meaningful for all people, Person Aand Person B differ in the specific situational features that activate (or inhibit) certain cogni-tions and affects within each person’s network, which in turn lead to a behavioral response.

Person B’s “mind”

Presence of feature g

Person A’s “mind” Features ofsituation Person A’s

Feature-Behavior Profile

Person B’sFeature-Behavior Profile

Beh

avio

rB

ehav

ior

Beh

avio

rB

ehav

ior

Presence of feature g

feature a

feature b

feature c

feature d

feature e

feature f

feature g

feature h

feature i

feature j

feature k

feature l

feature m

feature n

9781412946520-Ch18 5/23/08 12:40 PM Page 379

observable, stable, and predictable patterns ofif ... then ... situation-behavior profiles. Forexample, a situation may elicit an initial fearresponse in most individuals. For one individ-ual the fear response may lead to increasedattention and hypervigilance. For another indi-vidual, however, the fear response may lead toavoidance of the fear-eliciting stimulus. Thefirst individual is characterized by an ‘ifafraid, then hypervigilance’ pattern, whereasthe second individual is characterized by an‘if afraid, then avoidance’ pattern (Miller,1987).

To summarize, at the surface level aperson’s behaviors may vary considerablyfrom moment to moment and across anumber of different situations, and may even appear disparate and inconsistent.Nonetheless, within this variability theremay be stable and meaningful intraindividualpatterns of consistency represented by if ...then ... situation-behavior profiles thatuniquely characterize each individual. The ‘if’is the situation, either external (e.g., a requestfor public speaking) or internal (e.g., fearresponses), and the ‘then’ is the affective, cognitive, and behavioral responses to the ‘if’.

PSYCHOLOGICALLY ACTIVEFEATURES OF SITUATIONS

Because the nature of a person’s network isrevealed in observable if ... then ... situation-behavior profiles, predicting each person’sbehavior requires conceptualizing and assess-ing situations. What are the most importantpsychological features of a given situation?The first step is then to identify, for a givenindividual, those features that are ‘psycholog-ically active’ and that trigger particular cogni-tive-affective processes within a person’snetwork and lead to behavioral responses.

How does one conceptualize and assess asituation, particularly the psychologicallyimportant features of a situation? Initial efforts at classifying situations were based on nominal features, such as whether a behav-ior occurred in a playground or cafeteria

(Hartshorne and May, 1928; Newcomb,1929). A limitation of this approach is that it isunclear how to generalize the results to othersituations. For example, if a child behavesaggressively at the playground, but is friendlyand agreeable in the cafeteria, what would onepredict her behavior to be in a classroom?When situations are defined nominally,knowledge of past behavior (e.g., aggressive atthe playground) is useful in predicting behav-ior only in that exact situation, but shows littlegeneralizability to other situations (e.g.,aggressive when working on homework withpeers; Hartshorne and May, 1928).

Going beyond this limitation requiresunderstanding, for a given person, the mostinfluential psychological features of the situ-ation. What feature or features of the situa-tion most strongly affect that individual’sbehavior? For example, imagine one knowsthat the reason why the child is aggressive at the playground is because she becomesinsecure when she feels that she is beingexcluded by her classmates. Then one maypredict that even in a cafeteria, which showslittle physical resemblance to the play-ground, she will become aggressive if herpeers start talking about their weekend plansto spend time together but do not invite her.

Social psychology has uncovered situa-tional determinants of various behaviors (e.g.,aggression, conformity, affiliation) and oftenfocuses on the most important psychologicalfeatures of situations, such as the effects ofpower, fear, authority, and group pressure.Despite the vast research on situational influ-ences on social behavior, a systematic effortto develop a taxonomy of situations has onlybegun (Kelley et al., 2003). Furthermore, it isquite possible that a ‘one-size-fits-all’ taxon-omy may not be adequate. That is, the featuresof psychological situations serving as the ‘ifs’in individuals’ characteristic if ... then situation-behavior profiles may depend on the particularindividual as well as the particular situationand behavior in question. For example, for thechild who becomes aggressive when she feels excluded by her peers, the feature of the situation involves interpersonal rejection,


9781412946520-Ch18 5/23/08 12:40 PM Page 380

perceived or actual, by her peers. This feature,however, may be of lesser or no importance toanother child who instead becomes aggres-sive and belligerent when teachers give himpoor marks on assignments. For the firstchild, the psychologically active feature ispeer rejection. For the second child, it is neg-ative evaluations from an authority figure.

METHODS FOR IDENTIFYINGPSYCHOLOGICAL FEATURES OFSITUATIONS

Consider the myriad ways in which a situationcan be conceptualized and operationalized. Isit even possible to find a small enough numberof features that can provide a reasonably com-prehensive description of the situations, whichin turn can be used to characterize if ... then ...situation-behavior profiles? Ultimately, this is an empirical question. However, followingthe success in identifying the psychologicalfeatures of situations influencing children’sbehaviors in the summer camp, methods forassessing if ... then ... profiles in a number ofdifferent domains have been developed.

Three examples of research programsaimed at identifying psychological features ofsituations are presented below. In a departurefrom techniques used in the original summercamp study (e.g., Shoda et al., 1994), thethree illustrations introduce a new approachfor assessing feature-behavior profiles.Instead of classifying ‘ifs’ on discrete cate-gories (e.g., peer teasing vs. teacher praise),the latest techniques attempt to assess on continuous scales the degree to which eachpsychological feature was present in a givensituation. As will be described, the use of con-tinuous indices of situational features allowsthe use of multi-level analysis to simultane-ously estimate (i) the expected effect of a fea-ture on the average individual’s behavioralresponse (conceptually similar to the maineffect of the feature), (ii) the effect of the feature on a given individual’s behavioralresponse, and (iii) interindividual differencesin behavioral responses to the feature.

One goal of such research is to determinea common set of features influencing a par-ticular response. For example, as illustratedin Figure 18.1, a situation may consist of fea-tures a, b, c, d, e, and so on. In addition, thistype of research aims to identify those fea-tures that are particularly salient for a givenindividual or groups of individuals. Such fea-tures are considered ‘psychologically activefeatures’ or ‘psychologically active ingredi-ents’ of situations. For example, as illustratedin Figure 18.1, person A’s behavior may behighly influenced by features a, g, and l.Person B’s behavior may be highly influencedby features g, m, and n. And yet anotherperson’s behavior may not be affected by anyof the features. Thus, the process of featureidentification – both to obtain a common setof features as well as to identify those featuresthat are ‘psychologically active’ to a particu-lar individual or group of individuals – is akinto identifying a set of environmental aller-gens to which some individuals have a strongreaction, while other individuals do not.

Often, the presence of psychologicallyactive features triggers particular cognitive-affective processing dynamics within aperson’s network that leads to an increase inthe expression of a particular behavior. Forexample, someone who enjoys novel experi-ences may express greater positive affect inresponse to novelty-related features. Such fea-ture-behavior links are excitatory. However,psychologically active features can also acti-vate cognitive-affective processing dynamicsthat may lead to a decrease in the behavior ofinterest. Such feature-behavior profiles may be considered inhibitory in nature. For exam-ple, in the classic bystander intervention studies (e.g., Darley and Latané, 1973), thenumber of other people present in a situationcan be considered an important active featureof the situation for some individuals in that thefeature suppresses the helping behavior thatthey may have otherwise engaged in. As willbe discussed in the ‘Further Considerations’section, the relation between the presence of afeature and resulting response may not neces-sarily reflect a one-to-one correspondence.


9781412946520-Ch18 5/23/08 12:40 PM Page 381

Rather, it may be influenced by the occurrenceof other features also present in the situation.However, as a first step in feature identificationand its effect on subsequent responses, thischapter will focus on the identification of fea-tures that have been shown to singularly influ-ence behaviors.

To determine the psychological features ofa situation – both the set of features that cap-tures the most important psychologicallymeaningful aspects of situations as well asthose features that are psychologically activefor a given individual or group of individuals –one method is to use a top-down approach,basing identification of features on preexist-ing theory and research. This entails cullingthe literature for features of situations thatare likely to influence the behavior of interestand developing stimuli that differ on the fea-tures selected. For example, if one wereinterested in the features of situations influ-encing partner preference, one might turn topast research that has identified the charac-teristics of potential dating partners (e.g.,physical appearance, similarity to self) influ-encing perceived attractiveness. Anothermethod for identifying features of situationsinvolves using a bottom-up approach. In thisapproach, people’s responses to a number ofsituations are recorded. One then asks: whichsituations elicit similar responses among dif-ferent people and what do these situationshave in common? Finally, a combination ofthe two approaches may at times be used.

The following sections provide step-by-stepillustrations of three approaches for identify-ing psychological features of situations.Although the three approaches differ procedu-rally, they all employ the following six steps toidentify psychological features of situations:

1 One needs to identify the domain of situations inwhich the behavior of interest occurs (Shoda et al., 1994).

2 The researcher then needs to develop (e.g., byconducting a separate study) or obtain, if possible,(e.g., from prior research) a relatively large numberof stimuli that represent the situations of interest.

3 The next step is to identify a common set of psychological features that capture the most

important aspects of the situations and that arelikely to influence the behavior of interest.

4 Objective indices reflecting the extent to whicheach of the features is present in the stimuli arederived.

5 In a sample of individuals whose behavioral sig-natures the researcher is seeking to understand,a highly repeated within-subjects research design(Shoda, 2003) is used to observe and record par-ticipants’ behavioral responses across the differ-ent situations.

6 Through the use of multi-level analyses, it is thenpossible to determine the effect of individual fea-tures and interactions among features, on a givenindividual’s responses as well as to identify ‘psy-chologically active features’ for a given individ-ual or group of individuals.

Top-down approach to identifyingpsychological features of situationsinfluencing partner preference

For some research questions, it may be usefulto turn to past research and theory to identifycandidate features of situations that are likelyto influence a given behavior. For example,research and theory in social psychology hasextensively examined the situational factorsinfluencing person perception, attribution,impression formation, and spontaneousinference, to name just a few. This body ofknowledge is a good starting point in identi-fying a common set of psychological fea-tures. Similarly, research on response tofailure and experience of threat, as well aswork on stereotypes, is also likely to informthe initial feature identification process.Ultimately, these possible features need to beempirically validated using methods such asthose described in this chapter.

An illustration of a top-down approach is given in a study by Zayas and Shoda(2007), who set out to examine the question:Are women who have been the victim of psychological abuse in the past more likelyto prefer an abusive dating partner in thefuture? This question was motivated by the widespread belief among lay persons,academics, and professionals alike thatsome people may be recreating negative


9781412946520-Ch18 5/23/08 12:40 PM Page 382

relationship experiences through the datingpartners they select, or attract. In a departurefrom traditional conceptualizations of situa-tions, the researchers defined situations asanother person (i.e., a potential dating part-ner). In this context, psychological featurescan be conceptualized as the personality char-acteristics of the potential dating partners(Zayas et al., 2002). In the present example,the features were identified a priori usingexisting research on the characteristics asso-ciated with victimized women and abusivemen. Zayas and Shoda then tested the hypoth-esis that personality characteristics associatedwith abusive individuals (e.g., possessivenessand aggressive behaviors) in male potentialpartners may be the ‘active psychologicalingredients’ that attract some women andrepel others. The following section discussesthe Zayas and Shoda (2007) study as a way of providing a step-by-step illustration of a theory-driven top-down approach foridentifying psychological features.

Step 1: Identify the situations in which thebehavior of interest is likely to take placeAs a first step in identifying the psychologi-cal features of situations, the researchermust identify the situations in which thebehavior of interest is likely to take place. Inthe present research illustration, the behav-ior of interest was college-aged women’spreferences for male college-aged datingpartners who may be potentially psycholog-ically abusive. Thus, in this context, situa-tions were operationalized as descriptions of potential dating partners in the form ofpersonal ads. Features of the situations were the characteristics of the dating part-ners conveyed in the personal ads that werehypothesized to influence behaviors (i.e.,preferences). Ideally, in this initial step, theresearcher aims to identify the universe ofsituations in which the behavior of interest is likely to occur. Given the difficulty of this task, the researcher may choose to refineand narrow his or her definition of thebehavior of interest. For example, in thepresent example the researchers focused on

preferences for college-aged dating partners,rather than dating partners of all ages or non-college students.

Step 2: Gather specific stimuli representingthe situations of interestTo develop stimuli for the study, 112 malecollege students wrote short descriptions ofthemselves in the form of personal ads.Because the researchers were specificallyinterested in whether characteristics associ-ated with an abusive personality were psy-chological features of situations influencingwomen’s preferences in potential partners, itwas necessary to assess whether the stimuligenerated (i.e., 112 personal ads) possessedthese characteristics. Three female under-graduate research assistants, who wereunaware of the research hypotheses, pro-vided initial evaluations of all 112 personalads with regard to the degree to which themen described in the ads would be perceivedby women, on average, as being (1) poten-tially abusive, and (2) desirable as a datingpartner. Coders’ evaluations, however, indi-cated that none of the ads were perceived asboth potentially abusive and highly desirable,and that only 4 of the 112 ads were consensu-ally evaluated as potentially abusive. Becauseit was not feasible to have participants in astudy read all 112 ads, the following threetypes of dating partners were selected for thestudy: (1) four ads that described potentialdating partners that were rated high on abu-siveness and low on desirability (abusive); (2) eight ads that described potential datingpartners that were rated low on abusivenessand low on desirability (undesirable); and (3) four ads that described potential datingpartners that were rated low on abusivenessand high on desirability (desirable).

Step 3: Identify a common set of featuresthat could be used to capture the mostpsychologically important aspects of thesituation of interestThe next step is to identify features of the situation of interest. Past research and theory were used to identify the relevant


9781412946520-Ch18 5/23/08 12:40 PM Page 383

features a priori. In particular, past researchhas linked characteristics such as jealousy,impulsivity, dependence, and violence withan abusive personality in men (e.g., Dutton et al., 1996; Dutton and Browning, 1988;Murphy et al., 1994; Walker, 1979). Thus, atop-down approach involves validating theextent to which the a priori features werepresent in the stimuli.

Step 4: Code the stimulus set according to a list of relevant features and deriveindices reflecting presence of features Construct validation and continuous indicesof the features were obtained by having sep-arate samples of female college studentsevaluate the descriptions. Specifically, somewomen (n = 24) rated the extent to whichthey perceived the man described in each ofthe 16 personal ads to be potentially abusive,psychologically and physically, with aromantic partner. A different sample offemale college students (n = 22) rated theextent to which the man described in each adwas a desirable dating partner and someonethey would be interested in dating. (Thesetwo ratings were averaged within each raterto index the desirability associated with eachmale dating partner.) In addition, both sam-ples of female raters indicated the extent towhich each male dating partner possessedpersonality characteristics that past researchhas linked to an abusive personality in men,including jealousy, impulsivity, dependence,and violence.

To create continuous indices that charac-terized each of the 16 male dating partners interms of the psychological features (e.g.,potential for abusiveness, aggressiveness,desirability, impulsivity, possessiveness), allthe ratings were averaged across raters.Cronbach’s alpha estimate (α) of the reliabil-ity of the index was computed to estimate thereliability of each index. Cronbach’s alphaswere 0.93 and 0.89 for the potential for abu-siveness and desirability indices, respec-tively, and ranged from 0.72 to 0.87 for thespecific personality characteristics (e.g.,aggressiveness, jealousy, impulsiveness).

Not only did the women’s ratings provide acontinuous measure of the features present ineach ad, they also provided support for theinitial classification of the ads. Specifically,the abusive dating partners were rated as sig-nificantly more likely to inflict psychologicalas well as physical abuse than the undesirableand desirable dating partners. The desirabledating partners were rated as significantlymore desirable than the undesirable and abu-sive men, which were rated as approximatelyequally undesirable. Adding further validityto the ads used as stimuli, abusive dating part-ners were rated as more impulsive, jealous,possessive, dependent, clingy, aggressive,hostile, violent, and angry than the undesir-able and desirable dating partners.Undesirable and desirable ads did not differfrom each other on any of these characteris-tics. As an additional step in the construct val-idation of the 16 personal ads, the relationbetween female raters’ judgments of the adsand the male ad writers’ own self-reportedcharacteristics was examined. Correlationsbetween men’s self-reports and the femaleraters’ judgments of the ads showed moderateto high convergence. Specifically, women’sratings of each ad writer’s potential for psy-chological abuse were positively correlated inthe expected direction with ad writer’s self-reported hostility, impulsivity, jealousy, andpast experiences behaving in psychologicallyabusive ways (e.g., controlling, jealous, ver-bally abusive).

Step 5: Use a highly repeated within-subjects design to observe and recordresponses to stimuli In this study, the research employed a highlyrepeated, multi-level approach (Shoda, 2003)in which a person’s preference was assessedacross 16 potential dating partners, whovaried systematically on key personal char-acteristics or ‘psychological ingredients’(Zayas et al., 2002). Given that the behaviorof interest was preferences for dating partners,the situations involved providing individualswith the opportunity to express their prefer-ences. Therefore, the study procedures


9781412946520-Ch18 5/23/08 12:40 PM Page 384

employed an Internet dating service para-digm in which participants read about thedifferent potential dating partners and indi-cated through a four-round selection processwho they wanted to get to know better.

Step 6: Analyze each participant’sresponses as a function of the situationThe focus on the analysis of psychologicalfeatures of situations, and identifying indi-viduals’ behavioral signatures with regard tosuch features, calls for a shift in the generalparadigms used for social and personalitypsychology research. Specifically, the tradi-tional ‘one-shot’ data collection approach,which examines individual differences at onetime in one situation, has intrinsic limitationsfor discovering patterns of psychologicalregularities within each person. An alterna-tive approach, which might be called a‘highly repeated, within-subjects design’(Shoda, 2003) allows a systematic and quan-titative characterization of if … then … situa-tion-behavior profiles, for each person, bywithin-subject regression analyses. Theseindividual-level characterizations can in turnbe predicted from individual difference vari-ables, using a multi-level modeling approach(e.g., HLM, hierarchical linear modeling).The set of level-1 coefficients from suchanalyses, describes the effect of each situa-tion feature within a given individual. Thus,it is a behavioral signature, specifically, a fea-ture-behavior profile, that can be computedwith continuous, rather than categorical,characterization of situations with regard to

the degree to which each psychological fea-ture is present.

To illustrate, Zayas and Shoda (2007) usedmulti-level modeling to predict, for eachwoman, the effects of psychological featureson partner preference.1 Specifically, asshown in equation (18.1), the level-1 modelexamined, for each woman, the relationbetween the presence of particular features(e.g., aggressiveness and desirability) thatcharacterized each potential dating partnerand the behavior of interest (i.e., preferencefor each dating partner).2 The level-1 modelis as follows:

[Preference for partner i]i

= baggressiveness j [partner i’s aggressiveness]+ bdesirability j [partner i’s desirability] + rij

(18.1)

Because each woman can now be charac-terized by a baggressiveness j and bdesirability j, thesecoefficients can be used to represent her feature-behavior profile. For example,Table 18.1 reports the level-1 regressionequations for five participants. As shown, forparticipant 308 there is an inverse relationbetween the aggressiveness of the potentialdating partner and her preference for thedating partner (baggressiveness j = −0.80); she wasless likely to prefer a potential dating partnerwho was perceived by other women to beaggressive. This woman also shows a slightpositive relation between the desirability of thedating partner and her preference (bdesirability j =0.13), such that the more desirable the


Table 18.1 Level-1 regression equations predicting for each participant the effect of partner aggressiveness and desirability on her partner preferenceParticipant PMI Level-1 regression equation308 0.02 [Preference for partner i ]j = −0.80[partner i ’s aggressiveness] + 0.13 [Partner i ’s desirability] + rij

241 0.05 [Preference for partner i ]j = −0.04[partner i ’s aggressiveness] + 1.38[partner i ’s desirability] + rij

293 0.08 [Preference for partner i ]j = −0.53[partner i ’s aggressiveness] + −0.02[partner i ’s desirability] + rij

287 2.13 [Preference for partner i ]j = 0.41 [partner i ’s aggressiveness] + 0.61[partner i ’s desirability] + rij

289 2.24 [Preference for partner i ]j = 0.89[partner i ’s aggressiveness] + 1.58[partner i ’s desirability] + rij

Notes: Each participant’s regression equation predicts her preferences for each dating partner as a function of the datingpartner’s aggressiveness and desirability. Each woman’s score on the self-report measure assessing frequency of experienc-ing psychological abuse in her most recent romantic relationship is reported in the second column (range of possible scores= 0–4). PMI = Psychological Maltreatment Inventory.

9781412946520-Ch18 5/23/08 12:40 PM Page 385

partner, the more likely she was to selecthim. Visually scanning the coefficients forthe different women in the study, one seesthat the pattern differs for each one. For partic-ipant 287, aggressiveness is positively relatedto her partner preference (baggressiveness j = 0.41)and so is desirability (bdesirability j = 0.61). It isimportant to keep in mind that features thatdo not appear to have an effect on responses,as indexed by regression coefficients that arenot statistically different from zero, may stillbe psychologically meaningful, but may notlead to observable responses.

Are the differences in women’s profiles, asrepresented by the level-1 regression coeffi-cients meaningful? To the extent that theanswer is yes, one would expect that these coef-ficients could be predicted by individual differ-ence measures. For example, do women whohave experienced more instances of psycholog-ical abuse show a preference for potentiallyabusive dating partners? To identify the featuresof situations that may be particularly salient(i.e., ‘active’) for women with a higher inci-dence of experiencing psychological abuse in apast romantic relationship, the set of level-1coefficients can, in turn, be predicted by indi-vidual difference variables, such as a woman’spast relationship experiences. Specifically, thelevel-2 model is a between-subjects analysisthat predicts each woman’s preference forpotentially abusive dating partners, as reflectedin baggressiveness j, from the degree to which shehad experienced psychologically abusivebehaviors in her past romantic relationship asassessed by the Psychological MaltreatmentInventory (PMI; Kasian and Painter, 1992).The level-2 model is as follows:

baggressiveness j = γaggressiveness 0 + γaggressiveness 1

[participant j’s past experiences of psychological abuse] + µaggressiveness j

(18.2)

bdesirability j = γdesirability 0 + γdesirability 1 [participantj’s past experiences of psychologicalabuse] + µdesirability j

(18.3)

In these equations, the level-2 intercepts,γdesirability 0 and γaggressiveness 0, can be interpretedas the expected average effect of partners’desirability and potential for aggressiveness,respectively, on partner preference, from aparticipant who was average on the level-2predictor (because all level-2 variables werecentered, level-2 intercepts predict the level-1 coefficients for participants at the mean ofthe distribution).

The results of the Zayas and Shoda studyshowed that the frequency with whichwomen reported experiencing psychologi-cally abusive relationships in their past waspositively related to preferences for aggres-sive dating partners, even though the effect ofdesirability was statistically controlled in thelevel-1 model (γaggressiveness1 = 0.28, p <0.001). In addition, when the HLM analysiswas repeated in order to examine the effect ofdifferent features on preference, the resultsshowed that women who reported more psy-chological victimization in their most recentromantic relationship showed a strongerpreference for dating partners who had beenjudged to be impulsive, jealous, possessive,hostile, and violent. All of these characteris-tics have been linked to personality charac-teristics of abusive men (e.g., Dutton et al.,1996). Thus, these characteristics are the fea-tures of situations or ‘psychological ingredi-ents’ that differentially affected women'spartner preferences.

In addition to identifying features of situations that have high functional signifi-cance for a given person, the analysis also allowed one to see how a person’s part-ner preference varies as a function of acommon set of ‘psychological ingredients’present in the situation. For this analysis, situations were classified into ‘abusive’,‘undesirable’, and ‘desirable’, allowing each woman’s ‘behavioral signature’ to beassessed using the more categorical approachto situation assessment used in earlier work(Shoda et al., 1994). For example, as shownin Figure 18.2, the ‘behavioral signature’ forwomen who had experienced psychologicalabuse differed meaningfully from the


9781412946520-Ch18 5/23/08 12:40 PM Page 386

if ... then ... ‘behavioral signature’ for womenwho had not.

BOTTOM-UP APPROACH TOIDENTIFYING PSYCHOLOGICALFEATURES OF SOCIAL SUPPORTSITUATIONS

Given that the endeavor of identifying fea-tures of situations is relatively new, even inwell-developed research areas, it is possiblethat not all the relevant psychological fea-tures affecting individual differences inbehavior have been identified. Thus, forsome research questions, it may be useful toutilize bottom-up approaches in which iden-tifying the psychologically active features ofsituations is based on responses providedfrom individual participants rather than fromexisting theory and research.

The research below illustrates a bottom-upapproach to answer the question: What arethe features of situations that influencewhether one person perceives that anotherneeds help, specifically social support?Certainly, it is not simply a matter of hearingthe person say ‘I need help’, because dis-tressed individuals may have good reason tonot express an explicit desire for help.

A direct request for help involves the possi-bility of rejection or stigmatization, can beassociated with feelings of loss of independ-ence, and comes with the knowledge that onecould be burdening the listener (Goldsmithand Parks, 1990). If people in distress do notdirectly ask for support, what situational fea-tures do potential support providers (PSPs)rely on to determine whether someone needshelp? Is a person who clearly and freelyexpresses nonverbal signals of distress (e.g.,a large variety and high frequency of nega-tive facial expressions) more likely to be per-ceived as needing help compared to someonewho attempts to manage their distress (e.g.,attempting to suppress the expression of neg-ative emotions)? The research describedbelow examined the support seeker’s verbaland nonverbal communication as possiblekey ingredients of the situations that PSPsencounter.

Step 1: Identify the situations in which the behavior of interest is likely to takeplace The present research example focuses on situations in which one person is distressedand another person has the opportunity toprovide support. Distress signals may differdepending on the nature of the distress-eliciting event and context in which they occur.


Figure 18.2 The characteristics of the potential dating partners (i.e., desirability and poten-tial for abusiveness) are the ‘psychological ingredients’ of the situation, and partner prefer-ence (‘thens’) varied systematically as a function of the dating partner’s characteristics (‘ifs’).Women who had been the victim of psychological abuse (i.e., scoring above the median on thePMI) showed if ... then ... (situation-behavior) profiles that were distinct from the profiles thatcharacterized women who had experienced less abuse (i.e., scoring below the median on thePMI). Figure is reproduced from Zayas and Shoda (2007).

0.0

0.5

1.0

1.5

2.0

Abusive

Dating partner’s characteristics (“ifs”)

Undesirable Desirable

Wom

en’s

Par

tner

Pre

fere

nce

(“th

ens”

)

Below the Median

Above the Median

Women’s Score on the PsychologicalMaltreatment Inventory (PMI) scale

9781412946520-Ch18 5/23/08 12:40 PM Page 387

For example, it may be perceived as moreacceptable to freely express sadness and griefin response to a family member’s death, butless so in response to a loss at an athleticevent or competition. As a first effort inunderstanding the features of situations thatare ‘active’ and addressing the question,‘What features of situations do PSPs rely onto determine whether someone needs help?’,the researchers deliberately limited theirfocus to situations in which individualsbelieved that they would have to engage in anaversive behavior (i.e., eating dried worms).By equating the nature of the distressing sit-uation encountered, the range of psychologi-cal features that could play a role indetermining potential support providers’responses (e.g., perceptions that a discloserneeds help) was restricted. Constraining therange of psychological features affects thegeneralizability to other kinds of distressingsituations. Nonetheless, it is a first step indetermining the effectiveness and feasibilityof such an approach.

Step 2: Gather specific stimuli representingthe situations of interest The next step is to identify a set of stimuli thatcapture the most psychologically importantaspects of the situation of interest. For this pur-pose, a sample of situations was developed byasking 65 different people to serve as dis-closers. Upon arriving in the laboratory, the dis-closers were led to believe they were in ataste-testing study in which they would beeating dried worms. Past research has shownthat individuals react negatively in response tothis procedure (Johnson, 2006). Prior to per-forming the presumed dried worm taste-test, avideo recording was made of each of these 65disclosers while they were interviewed, ‘so thatfuture participants would know what the studywas like.’ Disclosers were asked to describetheir experience, say how they felt about it(positive, negative, or neutral) and to elaborateon their response. Each interview lastedapproximately five minutes.

Videotaped interviews were edited inorder to delete the interviewer’s questions

and focus solely on disclosers’ responses.Any responses not having to do with the dis-closer’s response to the worms (e.g., dis-closers’ descriptions of the study procedures)were also deleted. The final clips werebetween 20 and 25 seconds in length.Therefore, after completing this step in theprocess, there were 65 clips that could bepresented to a new group of participants.

Step 3: Identify a common set of featuresthat could be used to capture the mostpsychologically important aspects of thesituation of interest All the video clips developed from the firststudy were then presented to a differentsample (from the same population of inter-est) to identify the psychological features ofthe situation; namely, the features of situa-tions to which people respond when decidingwhether or not another person needs help(Whitsett and Shoda, 2007). Participants (N = 58) viewed randomly assigned pairs ofclips, with each video depicting a differentperson describing his or her response to thedistressing event.3 Participants selected oneclip from each pair in terms of who theythought needed more help, which wasdefined to participants in the following way:‘By “needs help” we mean that the personwould benefit from someone providing somecomforting words of support. An example ofthis type of help is someone saying, “I knowhow you feel” to the person’. To identify thefeatures that affected their judgment of who needed more help, they were asked toreply to an open-ended question asking whythey chose one individual over the other.Participants were asked to type a few sen-tences explaining their choice for each pair.In a variation of this method, participantscould be asked to rate the extent to whicheach person needed help instead of making achoice between two disclosers, and thenexplain why they gave each person the ratingthey did.

The result of this step was a large qualita-tive dataset, which consisted of reasons participants spontaneously provided for


9781412946520-Ch18 5/23/08 12:40 PM Page 388

perceiving one discloser as needing morehelp compared to another discloser (32 videopairs × 58 participants = 1,856 responses).Several steps were taken to extract theimportant psychologically meaningful fea-tures from these data, which entailed creat-ing a bottom-up coding system (Lampertand Ervin-Tripp, 1993). First, two codersread through all responses, noting any fea-tures that were consistently mentioned by anumber of participants. Features were quitediverse, including demographics (e.g.,gender), personality characteristics (e.g., shyness), and specific behaviors (e.g., fid-geting). A list of 99 themes was developed,providing a preliminary list of psychologicalfeatures.

The open-ended responses were then readthrough a second time, with the goal ofcoding every response using the preliminarylist of 99 features. In cases when codersencountered a response that could not becoded using the preliminary list of features, anew code was added. As a result, there were120 features in the preliminary list. Next,each rater coded 60% of the responses anddetermined how frequently each of the 120 features was mentioned in each partici-pant’s response. Because there was a 20%overlap between coders, it was possible tocompute inter-rater reliabilities, usingCronbach’s alpha, for each feature. To reduce

the list of 120 features, only the 30 of the 120features that were reported by at least 10% ofparticipants and had a Cronbach’s alpha ofgreater than 0.70 were retained. These fea-tures are listed in Table 18.2. Given that par-ticipants’ responses were obtained inopen-ended format questions, 10% of partic-ipants spontaneously reporting the same fea-ture is considerable.4 To summarize, thecoding procedure reduced the preliminarylist of 120 features to a more manageable listof 30 features.

Step 4: Code the stimulus set according toa list of relevant features and deriveindices reflecting presence of features After establishing a list of features relevant forthe situations, the next step was to rate eachvideo, on a 7-point Likert scale, according tothe degree to which each feature was presentin the clips. A new group of individuals, sim-ilar in characteristics to those participating inthe final study of interest, rated the videoswith regard to the list of features. In this way,the presence of psychological features of sit-uations in the set of stimuli is rated from theperspective of the individuals who are similarto the eventual test participants. In addition,given that these judgments will be used todevelop a consensually agreed-upon index tocapture the extent to which each feature ispresent in a situation, the greater the number


Table 18.2 List of 30 features identified as describing the most psychologically importantaspects of one type of distressing situation1. Negative expectations about what is to come next 16. Tense body2. Disgust 17. Nervous tics (playing with earlobe, hair, etc.)3. Lack of confidence 18. Looks at worms4. Distressed/upset 19. Looks down5. Angry 20. Voice in general indicates a negative state6. Scared 21. Quiet voice7. Anxious 22. Words that are used indicate a negative state8. Surprised 23. Relays a negative personal memory9. Negative attitude 24. Focuses on negative aspects of the task

10. ‘Playing it cool’/covering up true feelings 25. Is willing to try the worms/open-minded11. Convincing self 26. Task seems difficult for the discloser12. Conveys a forced/ fake smile 27. Willing to try the worms, but with reservations13. Conveys nervous laughter 28. Not open to experience/opposed to trying worms14. Body movements in general indicate a negative state 29. Has no previous experience with worms15. Engages in fidgeting/squirming 30. Participant can relate to person/understand point-of-view

9781412946520-Ch18 5/23/08 12:40 PM Page 389

of raters, the more reliable the indexbecomes.

In order to make the task more manage-able for raters, each rater was responsible foronly one feature. Ten raters were assigned toeach feature, and their responses were aver-aged for each video clip. By having multipleraters per feature, idiosyncratic responsescancel each other out when averaged with theresponses of other raters. After averagingacross raters, each video has a value assignedto it for each feature, indicating the level ofeach feature contained in each of the 65video clips. The reliability of these indicescan be computed using inter-rater correla-tions, Cronbach’s alpha, or both.

Step 5: Use a highly repeated within-subjects design to observe and recordresponses to stimuli Now that the important features and the levelof presence in each of the videos are known, itis possible to examine how responses to theclips are based on features contained in theclips. A new group of participants served as thePSPs and were asked to view the videos andrespond, for each one, with how much help (ona Likert scale) they think each person needs.

Step 6: Analyze each participant’sresponses as a function of the features ofthe situation For each individual PSP, his or her percep-tion of disclosers’ need for help can be pre-dicted as a function of the various features ofthe film clips. Specifically, multi-level

modeling can be used to predict, for eachpotential support provider, the effects of psy-chological features on perceived need forhelp. As shown in equation (18.4), the level-1 model predicts, for each PSP, the extent towhich disclosers’ negative expectations andexpressions of disgust influence his or herperception that a discloser needs help. Thelevel-1 model is as follows:

[Perceived need for help i]j

= b0 + bneg.expectations j [discloser i’s negative expectations] + bdisgust j

[discloser i’s disgust] + rij

(18.4)

Table 18.3 reports the regression equationsfor five participants. As shown, for PSP 1, themore negative expectations and disgust that adiscloser displays, the more likely she is toperceive that the discloser needs help. Thefeature-behavior profile for PSP 2 is some-what different. As for PSP 1, the more a dis-closer displays negative expectations, themore PSP 2 is likely to help. However, unlikePSP 1, disgust does not appear to be a rele-vant feature for PSP 2.

It can also be determined if there are sys-tematic differences between PSPs in whatfeatures of the disclosers’ expressions of dis-tress prompt the perception that help isneeded. Recall that individual level charac-terizations that result from the level-1 modelcan be predicted from individual differencemeasures. In this particular study, the focuswas on examining individual differences in


Table 18.3 Level-1 regression equations predicting for each participant the effect of thediscloser’s negative expectations and disgust on each participant’s perception of whetherthe discloser needs helpParticipant ECR Level-1 regression equation1 2.22 [Perceived need for help i ]j = b0 + 0.36(negative expectations) + 0.58 (disgust) + rij

2 1.83 [Perceived need for help i ]j = b0 + 0.97(negative expectations) − 0.07(disgust) + rij

3 1.50 [Perceived need for help i ]j = b0 − 0.12(negative expectations) + 0.92(disgust) + rij

4 4.11 [Perceived need for help i ]j = b0 + 0.47(negative expectations) + 0.35(disgust) + rij

5 5.33 [Perceived need for help i ]j = b0 + 0.79(negative expectations) + 0.06(disgust) + rij

Notes: Each participant’s regression equation predicts his or her perceptions of whether the discloser needs help as a func-tion of the discloser’s negative expectations and disgust. Each participant’s score on the avoidance dimension of the experi-ences in close relationship (ECR) scale is reported in the second column (range of possible scores = 1 to 7).

9781412946520-Ch18 5/23/08 12:40 PM Page 390

attachment avoidance. The level-2 model forthe slopes predicts the magnitude and direc-tion of the association between features ofdisclosers’ communication and perceivedneed for help as a function of the PSP’sattachment avoidance score as measured bythe Experiences in Close RelationshipsQuestionnaire (ECR; Brennan et al., 1998).In other words, the level-2 model examineswhether attachment avoidance moderatesthe relation between features of disclosers(i.e., whether disclosers were perceived ashaving negative expectations and expressingdisgust) and perceived need for help. Thelevel-2 model is as follows:

bneg.expectations j = γneg.expectations 0 + γneg.expectations 1

[PSP j’s avoidance score] + µneg.expectations j

(18.5)

bdisgust j = γdisgust 0 + γdisgust 1 [PSP j’s avoidance score] + µdisgust j

(18.6)

The level-2 intercepts, γneg.expectations 0 andγdisgust 0, can be used to examine perceivedneed for help in relation to negative expecta-tions and disgust. In general, results showedthat the more the discloser in a video clip dis-played negative expectations, the greater theperceived need for help (γneg.expectations 0 = 0.50,t(39) = 11.31, p < 0.0001). In addition, themore the discloser in a video clip displayeddisgust, the greater the perceived need for help(γdisgust 0 = 0.47, t(39) = 10.91, p < 0.0001).

Rather than assessing the moderatingeffect of attachment avoidance via productterm interactions, multi-level modelingdetermines the extent to which the slopes inthe level-1 model for situation features (i.e.,disclosers’ negative expectations and disgust,represented by bneg.expectations j and bdisgust j,respectively) varied as a function of partici-pants’ attachment avoidance. Attachmentavoidance did not appear to moderate therelationship between negative expectationsand perceived need for help (γneg.expectations 1 =0.06, t(39) = 1.52, p = 0.14), but it was found

to be an important moderator of the relationbetween disgust and perceived need for help(γdisgust 1 = −0.10, t(39) = −2.80, p = 0.008).Participants with low attachment avoidance,compared to those with high attachmentavoidance, showed a stronger relationbetween disclosers’ expressions of disgustand their perception that help was needed.

COMBINED (TOP-DOWN ANDBOTTOM-UP) APPROACH TOIDENTIFYING PSYCHOLOGICALFEATURES INFLUENCING SELF-RELEVANT EVALUATIONS

In some cases, using a bottom-up approach –in which individuals describe the features ofsituations that are most psychologicallymeaningful – may result in more features thanone may wish to, or is able to, address withina study. In such cases, it may be useful toidentify potentially important psychologicalfeatures by combining a bottom-up approachwith a top-down approach. Pre-existingtheory-driven concepts are used to pare downthe number of features studied. Thisapproach is illustrated by the followingresearch example identifying those psycho-logical features of situations that are likely todifferentially influence how people feelabout themselves, that is, people’s affectiveself-evaluation. This study further demon-strates the stability of individuals’ feature-behavior profiles by using the profiles fromone time point to predict affective self-eval-uative responses to situations at a secondtime point more than one week later.

Step 1: Identify the situations in which thebehavior of interest is likely to take place As mentioned in the first two research exam-ples, the researcher must identify where thebehavior of interest is likely to take place.The behavior of interest in this study wasaffective self-evaluation. Therefore, theresearchers aimed to find situations in whichindividuals would feel positively or nega-tively about themselves.


9781412946520-Ch18 5/23/08 12:40 PM Page 391

Step 2: Gather specific stimuli representingthe situations of interestIn contrast to the previous two illustrations, thepresent study used existing stimuli. Kitayamaand colleagues (Kitayama et al., 1992, 1997)had participants describe situations in whichtheir self-esteem increased or decreased. Thisresulted in 200 positive and 200 negative situ-ations. Each situation described a single socialepisode and was usually one sentence in length(e.g., ‘In an attempt to get a person’s attention,I poked my nose into that person’s affair andhe/she became upset at me’).

Step 3: Identify a common set of featuresthat could be used to capture the mostpsychologically important aspects of thesituation of interest Next, a bottom-up strategy was used to iden-tify features that might differentially predicthow individuals will feel about themselves.Participants (N = 194) first completed a situa-tion-self-evaluation task. This task consisted oftwo ten-minute blocks. In one block, negativesituations from the stimulus set described inStep 2 were presented in random order one ata time on a computer screen. In the otherblock, positive situations were presented inrandom order. After reading each situation,participants rated how positively they wouldfeel about themselves in that situation, hownegatively they would feel about themselves inthat situation, and how frequently they encoun-tered each situation. A 5-point scale (0 = never;4 = frequently) was used to assess the fre-quency with which participants experiencedeach situation. Situations with mean frequencyratings less than 1.05, indicating that they wereinfrequently experienced by the majority ofparticipants, were excluded from the stimulusset of future studies (i.e., those described inSteps 4 and 5). This reduced the list to 139negative and 168 positive situations. On aver-age, participants rated 38 different situations(SD = 13.56) per block.

Next, participants completed a situationcomparison task in which they were promptedto identify the features of the situations thatled them to feel positively or negatively about themselves. In one of the four blocks

of this task, participants saw two lists of situa-tions, one on either side of their computerscreen. For each participant, one list consistedof ten negative situations that he or she evalu-ated most negatively. The other list consistedof ten negative situations that he or she evalu-ated least negatively.5 The two lists were ran-domly assigned to either side of the computerscreen. Participants were instructed to write alist of characteristics that the situations withineach list all shared and the characteristics onwhich the two lists differed from one another.The remaining three blocks of the situationcomparison task followed the same proce-dure. The only difference was that pairs oflists varied such that participants comparednegative situations that they evaluated mostand least positively, positive situations thatthey evaluated most and least positively, andpositive situations that they evaluated mostand least negatively. The order of these blockswas counterbalanced between participants.After the situation comparison task, all partic-ipants were asked to list any additional situa-tion characteristics that also may have ledthem to feel positively or negatively aboutthemselves. The situation comparison taskand the follow-up question resulted in a list of approximately 2,700 free-response items.Using the instructions for constructing abottom-up coding system outlined by Lampertand Ervin-Tripp (1993), the free-response datawere segmented into potential coding cate-gories (i.e., situation features). This stepyielded a large and diverse set of approxi-mately 85 potential situation features.

As a first pass, a top-down approach wasapplied to reduce this list of situation featuresto those considered to have theoretical valuein the field of social psychology. Tworesearchers independently analyzed socialpsychology textbooks (Brown, 2006; Sears et al., 1991) and identified topics from thetextbooks that were also reflected in thebottom-up coding categories derived from thefree-response data. Those categories identi-fied by both researchers as being representedin the free-response data and in at least one of the textbooks were selected as fea-tures to be used for this study. For example,


9781412946520-Ch18 5/23/08 12:40 PM Page 392

one of the categories identified with thisapproach was ‘conformity due to presence of an authority figure’. Using this procedure,46 features were selected.

Step 4: Code the stimulus set according toa list of relevant features and deriveindices reflecting presence of features A separate sample of participants (N = 318)rated the extent to which each of the situa-tions in the stimulus set contained the situa-tion features identified from Step 3.Situations were randomly selected withoutreplacement from the stimulus set of 307 sit-uations and presented one at a time on a com-puter screen. Participants were assigned oneof the 46 situation features, and they rated theextent to which their assigned feature waspresent in each situation.6

Inter-rater reliabilities for each of thesefeatures were indexed by Cronbach’s alpha (α).Eleven of the features had alphas greater than0.80, and they were selected for the presentstudy (see Table 18.4 for the 11 features).The degree to which a situation reflected thepresence of a feature was indexed by themean rating of these participants.

Step 5: Use a highly repeated within-subjects design to observe andrecord responses to stimuliA third set of participants (N = 37) completed atwo-session study. In each session, participants

read situations one at a time on the computerscreen and rated, on two separate scales, hownegatively and how positively they wouldfeel about themselves in each situation. Thescale consisted of 26 vertically aligned pointsanchored at the first and twenty-sixth points(e.g., ‘I would not feel negative aboutmyself’ and ‘I would feel extremely negativeabout myself’, respectively.) In both ses-sions, participants evaluated one block of 40negative situations randomly selected with-out replacement from the negative stimulusset and one block of 40 positive situationsrandomly selected without replacement fromthe positive stimulus set. Both positive andnegative self-evaluations were measured inorder to evaluate individual differences inwithin-person relations between negative andpositive self-evaluation (Wang et al., 2007).For the purpose of illustration, this examplewill focus only on negative self-evaluationratings in negative situations.

Step 6: Analyze each participant’sresponses as a function of the features of the situationAs discussed in the previous two researchexamples, participants’ feature-behaviorprofiles were indexed by intraindividualslope coefficients predicting negative self-evaluation to situations from the degree towhich features were present in the situations.Within each session, a standardized slope


Table 18.4 The stability of individual feature-behavior profiles for situation featuresbetween time 1 and time 2Situation feature rank-order correlation pShared success 0.67 9.03 × 10−6

Receipt of social support 0.44 0.007Academics 0.38 0.02Receipt of negative feedback from others 0.38 0.02Being alone 0.37 0.03Poor performance due to the presence of others 0.35 0.03Competition 0.35 0.04Achieving success alone 0.29 0.08Being misunderstood or misperceived 0.27 0.1Physical activity (e.g., exercise) 0.26 0.1Receipt of positive feedback 0.13 0.4

Notes: The stability of the profiles for each feature was indexed by Spearman’s rank-order correlation coefficient predicting individuals’ time 2 feature-behavior profile signatures from their time 1 signatures. Only cues with standardized Cronbach’salphas (α) greater than 0.80 were included in this table.

9781412946520-Ch18 5/23/08 12:40 PM Page 393

coefficient was computed for each individualfor each feature. Specifically, the level-1model was as follows:

[Negative self-evaluation in situation i]j

= β0j + βneg.feedback j [negative feedback in situation i] + rij

(18.7)

The top panel of Figure 18.3 illustrates the data for 37 participants’ negative self-evaluative responses to negative situations as a function of receiving negative feedback(e.g., criticism) at time 1. Each scatterplotrepresents the data for one participant. The inset to the right of Figure 18.3 presentsthe data of participants 33, 34, and 35.


Figure 18.3 Top panel (panel A) depicts scatterplots representing the relation between thefeature of the situation (i.e., receipt of negative feedback) and negative self-evaluation, foreach participant at time 1. In each graph, the x-axis represents the degree to which the situa-tions involved receiving negative feedback. The y-axis represents participants’ negative self-evaluation ratings in response to each situation. The inset to the right illustrates that, forparticipant 33, the greater the negative feedback present in the situation, the more hereported negative self-evaluations. The same feature had almost no effect on participant 34and slightly an inverse effect on participant 35. Lower panel (panel B) shows scatterplots rep-resenting the relation between the same feature of the situation (i.e., receipt of negative feed-back) and negative self-evaluation, for each participant at time 2.

A

B

Subj 33 Subj 34 Subj 35

Subj 33 Subj 34 Subj 35

Receipt of negative feedback (‘lfs’)




Neg

ativ

e se

lf-ev

alua

tion

(’The

ns’)

Neg

ativ

e se

lf-ev

alua

tion

(‘The

ns’)

Neg

ativ

e se

lf-ev

alua

tion

(‘The

ns’)

Neg

ativ

e se

lf-ev

alua

tion

(’The

ns’)

Subj 36

Subj 29

Subj 22

Subj 15

Subj 8

Subj 1 Subj 2 Subj 3 Subj 4 Subj 5 Subj 6 Subj 7

Subj 9 Subj 10 Subj 11 Subj 12 Subj 13 Subj 14




Subj 37

β1=.24 β1=.05 β1=.11

β1=.31 β1=.03 β1=.09

Subj 36

Subj 29

Subj 22

Subj 15

Subj 8

Subj 1 Subj 2 Subj 3 Subj 4 Subj 5 Subj 6 Subj 7





Subj 37Self-evaluative response signatures (time 1)

Self-evaluative response signatures (time 2)

1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4

1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4

1 2 3 4 1 2 3 4 1 2 3 4

1 2 3 4 1 2 3 4 1 2 3 4

0

5

10

15

20

0

5

10

15

20

9781412946520-Ch18 5/23/08 12:40 PM Page 394

For participant 33, if the situation involvesreceiving negative feedback, then he is morelikely to feel negatively about himself. This is reflected by a positive beta-coefficient of 0.31. In contrast, for partici-pant 35, the same feature predicted less negative self-evaluations (βneg.feedback = −0.09). For participant 34, this feature was not predictive of negative self-evaluations(βneg.feedback = 0.03).

The present study also looked at the stabil-ity, or predictive utility, of participants’feature-behavior profiles. The bottom panelof Figure 18.3 illustrates data for the same 37participants at time 2. The stability of theprofiles for each feature was indexed by aSpearman’s rank-order correlation coefficientpredicting individuals’ time 2 feature-behavior profile from their time 1 feature -behavior profile. Table 18.4 illustrates thestability of the feature-behavior profiles for all the features. For any given feature, the more positive a correlation coefficient,the greater the stability of participants’feature-behavior profiles between sessions 1 and 2.

This research provides an illustration of amethod used to identify potentially impor-tant psychological features by combining abottom-up approach – in which individuals’self-reports of what influences their feelingsabout themselves were solicited – with atop-down approach in which pre-existingconcepts within the field of social psychol-ogy were used to pare down the featuresresulting from the bottom-up approach.Variability in individuals’ feature-behaviorprofiles suggests that they may trigger cognitive-affective processes that lead to abehavioral response, in this case, negativefeeling about the self. This research also pro-vides an example of how to identify featuresthat yield temporally stable feature-behaviorprofiles. The Spearman’s rank-order correla-tion coefficients in Table 18.4 suggest thatthe stability of feature-behavior profiles arestrong for some features and weak forothers.

Strengths and limitations

Research strategies such as the onesdescribed in this chapter that use top-down orbottom-up approaches, or a combination ofthe two, allow identification of features ofsituations. Each approach has its ownstrengths and limitations and it is up to theresearcher to determine the most suitablestrategy for addressing a specific researchquestion. An obvious strength of top-downapproaches is that identification of features is supported by existing evidence collectedfrom various researchers. These previousresearch findings can be used to identify features of situations hypothesized to affectbehavior and to test specific hypothesesabout the role of specific features on behav-ior. A weakness of top-down approaches is that they are limited to those features that have been studied in the literature. In addition, those psychological features thathave been identified may be specified atvarying levels of abstraction (see section on‘Feature identification and level of analy-ses’). Moreover, it is highly likely that notall the relevant psychological features affect-ing individual differences in behavior havebeen identified. In contrast, bottom-upapproaches are based on individuals’responses, and therefore are not limited byprevious research and theory. However, alimitation of bottom-up approaches is thatthey require that at least some participantsare aware and willing to reveal the factorsthat affected their behaviors. Given thatmany processes involved in judgment, deci-sion-making, and producing a behavioralresponse may be automatic and uncon-scious, individuals may not be aware of allthe factors influencing their decision-making and behaviors. Social desirabilityprocesses, whether to present one’s self in apositive light to others or to oneself, mayalso likely to affect the responses that parti-cipants provide. Finally, in some cases, afeature may be important, but may not beone that spontaneously comes to mind.


9781412946520-Ch18 5/23/08 12:40 PM Page 395

FURTHER CONSIDERATIONS

The research projects described above illus-trate approaches for identifying the psycho-logical features of situations in differentdomains – situations involving preferencesfor and selection of potential dating partners,to those involving deciding whether an indi-vidual needs social support, to situations thatmay influence one’s affective self-evalua-tions. As with any research paradigm, thereare issues that each researcher needs to con-sider in light of his or her specific researchgoal. Below is a brief discussion of consider-ations involved in the process of feature identification.

Feature identification and level ofanalyses

As may be evident from the research illustra-tions presented so far, psychological featurescan be conceptualized at different levels. Inthe Zayas and Shoda (2007) study describedearlier, for example, the level of analysis wasgeneral perceptions of personality character-istics, such as whether a potential datingpartner was aggressive, impulsive, and so on.However, other researchers might havefocused on more micro features such aspotential partner’s tone of voice or how theycope with temptation, to examine how suchmicro features give rise to more general per-ceptions of potential dating partners. A focuson more micro-level features, such as subtlenonverbal behaviors (e.g., eye gaze or vocalintonation) would address a different ques-tion: what are the features of situations thatgive rise to perceptions that another personis, for example, jealous or impulsive or needshelp? Specifying the level of analyses for aparticular project is analogous to the study oflinguistics in which one goal is to understandhow written words are encoded and assignedmeaning. In linguistics, the first levelinvolves orthographic analysis. Successfulanalysis is required for the identification of

morphemes, the smallest unit of meaning, suchas prefixes (un-, ad-) and suffixes (-less, -ness).These in turn may combine with other mor-phemes to form words. Similarly, in person-ality and social psychology, the process offeature identification is likely to involvestages and the researcher must decide whichstage and corresponding level of analyses tofocus on. Given that conceptualization andoperationalization of situational features isstill in the early stages, it is up to futureresearch to derive a vocabulary for describ-ing different levels, how the different levelscorrespond with one another, and the mosteffective demarcation of levels.

Highly repeated within-subjectsapproach

The techniques described in the three illus-trations are useful in identifying feature-behavior profiles that differ in notable waysfrom those used in the original summercamp study (e.g., Shoda et al., 1994). Mostimportant, these techniques no longer clas-sify ‘ifs’ on discrete categories (e.g., peerteasing vs. teacher praise). Rather, continu-ous scales are used to assess the degree towhich each psychological feature is presentin a given situation. The use of continuousindices of situational features has alsochanged the statistical methods available forthe analysis of the effects of situation fea-tures. Specifically, multi-level modelingallows one to simultaneously estimate theexpected effect of a feature on the averageindividual (conceptually similar to the maineffect) as well as the effect of the feature fora given individual. It does so without requir-ing that a situation be classified into mutu-ally exclusive categories. Multi-levelmodeling also allows the analysis of the role of individual difference variables, suchas attachment avoidance, without requir-ing that the sample be divided into discrete groups (e.g., high avoidance vs. low avoidance).


9781412946520-Ch18 5/23/08 12:40 PM Page 396

Predicting future behavior:generalizability of ‘ifs’

Ultimately, most researchers wish to saymore than how a person will respond to thefeatures of the specific stimulus set that waspresented in a particular research paradigm.For example, a researcher may wish to gobeyond the psychologically active featuresof expressions of distress in response to thespecific situation of having to eat driedworms to expressions of distress elicited bya broader class of aversive situations. If oneis successful in identifying psychologicalfeatures that are important, identifying a particular person’s feature-behavior profilein a specific research paradigm shouldenable us to predict what the person will doin new situations that contain the same psychological feature, even if they are nominally different situations. For example,the level-1 coefficients from the Whitsettand Shoda (2007) study, such as those thatindex the effect of negative expectations anddisgust on perceptions of whether anotherperson needs help, may allow one to predicteach person’s perception of an individual’ssuffering under different sources of distress,such as being unprepared for an upcomingexam.

The meaning and impact of particular psy-chological features often depend on the otherfeatures present in the same situation. Forexample, although a situation may possessfear-eliciting features that may in themselveslead to withdrawal, the situation may alsopossess other features, such as a highlydesired goal, that simultaneously activateprocessing dynamics that may lead toapproach, thus overriding the fear–with-drawal response. It is the set of cognitionsand affects activated, rather than a singlecognition that influences a person’s response.This possibility can be addressed empiricallyby testing for interactions among features. Ofcourse with the presence of multiple featuresin a given situation, the number of interac-tions quickly becomes unwieldy. Thus, evenfor bottom-up approaches, past research and

theory may inform which particular interac-tions to examine.

Building a Model of a Person

Identifying meaningful and stable if ... then... situation-behavior profiles, or ‘behavioralsignatures’, is one step in building a model ofa person. Ultimately, however, one needs tounderstand how, within a given individual, aparticular situational feature (e.g., Joe’sfacial features that remind Mary of herfather) activates cognitive (e.g., a memory ofhis disdain for mediocrity) and emotional(e.g., anxiety) reactions that mediate theeffect of situation features on behavior. Buthow would one go about directly assessingsuch internal if ... then ... relations?

Assessing internal if ... then ... relations isa significant challenge because individualsare often not aware of the associationsamong their thoughts. However, there arenow some promising methodologies for theassessment of automatic (i.e., not con-sciously controlled) associations among cog-nitions and affects. Recent research hasshown the importance and feasibility ofassessing links between specific cognitive,affective, and behavioral reactions. Forexample, links between situation featuresand cognitive and emotional reactions haverepeatedly been shown to underlie the phe-nomenon of ‘transference’ (Andersen andChen, 2002).

One approach for assessing strengths ofautomatic associations within each individ-ual’s CAPS network utilizes implicit meas-ures, such as the Implicit Association Test(Greenwald et al., 1998). For example,people with secure attachment styles, com-pared to those with insecure styles, wereshown to have stronger automatic associa-tions between the concept of their currentromantic partner and positive reactions(Zayas and Shoda, 2005). For these securelyattached individuals, thoughts about theirpartner more strongly (compared to inse-curely attached individuals) automatically


9781412946520-Ch18 5/23/08 12:40 PM Page 397

activated positive reactions. Furthermore, thestrength of such associations was found to bemeaningfully related to relationship out-comes, such as greater satisfaction and emo-tional commitment.

Another approach builds on basic findingsin research on brain activities. Specifically, aparticular component of Event RelatedPotentials (ERP) waves in response to anevent, called N400, has been shown to bemagnified when participants analyze thesemantic meaning of words. A typical find-ing using the N400 as an index, for example,is that targets preceded by an unrelated prime are more difficult to process (than targets preceded by a related prime), presumablybecause they require greater semantic analy-sis. Applying this finding and methodologyto the social and personality domain, greaterN400 reactions are observed when womenencounter a negative interpersonal outcome(e.g., partner’s inattention) in response to abid for a partner’s support compared toencountering the same situation in more neu-tral contexts (Zayas et al., under review).Moreover, according to research in adultattachment, the negative interpersonal infor-mation encountered in response to a bid forpartner’s support may be a psychologicallysalient feature of the situation, particularlyfor insecurely attached women (compared tosecurely attached women). However, giventhat women with insecure styles may copewith interpersonal rejection in distinctiveways – women who are considered preoccu-pied turn their attention towards threateninginformation whereas women who are highlyavoidant turn their attention away fromthreatening information – the greatest reac-tion should be observed among women witha preoccupied attachment style. In otherwords, these women show not only a sensi-tivity to threatening information, but onceencountered have a difficult time disengag-ing from it. Consistent with these ideas,women who were anxiously attached andlow in attachment avoidance (also referred toas preoccupied with attachment) showed the

greatest reaction to rejection words asassessed by the N400.

CONCLUSIONS

The finding that people can be characterizedby stable if ... then ... situation-behavior pro-files led to two related developments in thestudy of personality. First, there was a recog-nition that the field needed to reconcile thefact that people’s behavior varies across situ-ations with the core assumption of the fielditself, that people possess a stable and coher-ent personality. This led to a reconceptual-ization of personality. Instead of focusing ontraits, social cognitive approaches assumedthat each person is characterized by a uniquenetwork of interconnected cognitions, whichitself remains constant, even if the specificbehaviors that arise from the network varyfrom one situation to another. This then sug-gested that that if ... then ... profiles provideclues for identifying individuality and per-sonality coherence within individuals’ cross-situational variability. This variability neednot be considered a source of error to beeliminated.

But what are the ‘ifs’? As a first step inconstructing a cognitive social model of personality, this chapter discussed threeapproaches for identifying psychological significant ‘ifs’, or psychological features of situations. They illustrate differentapproaches used to identify the features ofsituations, and the different level of analysisinvolved in the conceptualization of the situ-ation and the psychological features embed-ded within it.

Certainly, the task of identifying a set ofpsychological features that are particularlyrelevant for a given individual is not an easyone. But once a list of features that representthe most important features of the situation isdeveloped, stimuli that differ in the presenceof these features can the be used to examinequestions such as: What features of situationsactivate a given individual’s encodings of


9781412946520-Ch18 5/23/08 12:40 PM Page 398

situations and generate behavior, reflecting a person’s unique and stable cognitive-affective processing system?

NOTES

1 Due to the design of the Internet dating service(IDS) procedure used to assess partner preference(Zayas and Shoda, 2007), all participants (in study 1and 2) obtained the same average preference score.Because the average preference for all participantswas zero after centering, the intercept, b0j, whichreflects each participant’s average preference scorefor the 16 dating partners, was not included in thelevel-1 model. Details regarding model specificationfor the HLM analysis are described in Zayas andShoda (2007).

2 To examine the effect of the presence of specificpersonality traits, the HLM models were specifiedexactly as those described in equations (18.1), (18.2),and (18.3), except that the aggressiveness index wasreplaced by each personality index (e.g., impulsivity,jealousy), one at a time, as the level-1 predictor.

3 Each participant viewed 64 of the 65 video clipsrandomly selected. All 65 video clips were presentedan approximately equal number of times.

4 It is not necessarily the case that codes withpoor inter-rater reliabilities are not useful. The lowreliabilities could be due to low frequency of the feature across stimuli or simply the fact that thesecues might be perceived differently by differentpeople. In other words, they could be highly relevantfor most or some people, but the effect of the fea-ture on behavior could vary greatly from individual toindividual.

5 If participants evaluated fewer than 20 negativesituations, the two lists were generated by splittingthe evaluated situations into the most and least neg-atively rated.

6 Participants who evaluated fewer than 224(90%) of the situations were not included in the cal-culation of the presence of features in situations.They did not evaluate enough situations to establishwhether there was high inter-rater reliability regard-ing the presence of features in the situations. For thesame reason of needing to establish consensus, fea-tures were also required to have been assigned to atleast five participants. The reasons why only 25 of the46 features fit the criteria described are two-fold. Themain reason was because the initial computer pro-gram for this study randomly assigned participants to1 of the 46 features. Random assignment led to anuneven distribution such that certain features wereassigned to participants more frequently than others.

Once this problem was identified, block randomassignment was used and participants were nolonger assigned to oversampled situation features.Some features also failed to have a sufficient numberof raters (i.e., five or more) because participantsassigned to these features were unable to rate atleast 224 (90%) of the 307 situations. These criteriaresulted in 25 features.

REFERENCES

Andersen, S.M. and Chen, S. (2002) ‘The rela-tional self: An interpersonal social-cognitivetheory’, Psychological Review, 109(4):619–45.

Brennan, K.A., Clark, C.L. and Shaver, P.R.(1998) ‘Self-report measurement of adultattachment: An integrative overview’, in J.A.Simpson and W.S. Rholes (eds), AttachmentTheory and Close Relationships. New York:Guilford Press, pp. 46–76.

Brown, J.D. (2006) Social Psychology. NewYork: McGraw-Hill.

Darley, J. M. and Latané, B. (1968) ‘Bystanderintervention in emergencies: Diffusion ofresponsibility’, Journal of Personality andSocial Psychology, 8(4): 377–83.

Dutton, D.G. (1994) ‘Patriarchy and wifeassault: The ecological fallacy’, Violence andVictims, 9: 167–82.

Dutton, D.G. and Browning, J.J. (1988)‘Power struggles and intimacy anxieties ascausative factors of violence in intimaterelationships’, in G. Russell (ed.), Violence inIntimate Relationships. New York: PMAPublishing.

Dutton, D.G., Starzomski, A. and Ryan, L. (1996) ‘Antecedents of abusive personalityand abusive behavior in wife assaulters’,Journal of Family Violence, 11(2): 113–32.

Goldsmith, D. and Parks, M.R. (1990)‘Communicative strategies for managing therisks of seeking social support’, in S. Duckand R.C. Silver (eds), Personal Relationshipsand Social Support. Newbury Park, CA:Sage, pp. 104–21.

Greenwald, A.G., McGhee, D.E. and Schwartz,J.K.L. (1998) ‘Measuring individual differ-ences in implicit cognition: The ImplicitAssociation Test’, Journal of Personality andSocial Psychology, 74(6): 1464–80.


9781412946520-Ch18 5/23/08 12:40 PM Page 399

Hartshorne, H. and May, M.A. (1928) Studies inthe Nature of Character: Vol. 1. Studies inDeceit. New York: Macmillan.

Johnson, K.L. (2006) ‘Dreading to a better end:How anticipation affects intertemporal eval-uative judgments’, Unpublished manuscript,New York University.

Kasian, M. and Painter, S.L. (1992) ‘Frequencyand severity of psychological abuse in adating population’, Journal of InterpersonalViolence, 7(3): 350–64.

Kelley, H.H., Holmes, J.G., Kerr, N., Reis, H.,Rusbult, C.E. and Van Lange, P.A. (2003) AnAtlas of Interpersonal Situations. Cambridge:Cambridge University Press.

Kitayama, S., Markus, H.R., Matsumoto, H. andNorasakkunkit, V. (1997) ‘Individual and collective processes in the construction of theself: Self-enhancement in the United Statesand self-criticism in Japan’, Journal ofPersonality and Social Psychology, 72(6):1245–67.

Lampert, M.D. and Ervin-Tripp, S.M. (1993)‘Structured coding for the study of language and social interaction’, in J.A.Edwards and M.D. Lampert (eds), TalkingData: Transcription and Coding in Discourse Research. Hillsdale, NJ: Erlbaum,pp. 169–206.

Miller, S.M. (1987) ‘Monitoring and blunting:Validation of a questionnaire to assess stylesof information seeking under threat’, Journalof Personality and Social Psychology, 52(2):345–53.

Mischel, W. and Shoda, Y. (1995) ‘A cognitive-affective system theory of personality:Reconceptualizing situations, dispositions,dynamics, and invariance in personalitystructure’, Psychological Review, 102(2):246–68.

Murphy, C.M., Meyer, S.L. and O’Leary, K.D.(1994) ‘Dependency characteristics of part-ner assaultive men’, Journal of AbnormalPsychology, 103(4): 729–35.

Newcomb, T.M. (1929) ‘The consistency of cer-tain extrovert-introvert behavior patterns in 51problem boys’, Contributions to Education,No. 382, Columbia University, New York.

Sears, D.O., Peplau, L.A. and Taylor, S.E. (1991)Social Psychology. New Jersey: Prentice Hall.

Shoda, Y. (1990) ‘Conditional analyses of personality coherence and dispositions’, PhD dissertation, Columbia University.

Shoda, Y. (2003) ‘Individual differences insocial psychology: Understanding situationsto understand people, understanding peopleto understand situations’, in C. Sansone,C.C. Morf and Panter, A.T. (eds), The SageHandbook of Methods in Social Psychology.Thousand Oaks, CA: Sage, pp. 117–41.

Shoda, Y. and Mischel, W. (1998) ‘Personalityas a stable cognitive-affective activation net-work: Characteristic patterns of behaviorvariation emerge from a stable personalitystructure’, in S.J. Read and L.C. Miller (eds),Connectionist and PDP Models of SocialReasoning and Social Behavior. New Jersey:Lawrence Erlbaum, pp. 175–208.

Shoda, Y., Mischel, W. and Wright, J.C. (1989)‘Intuitive interactionism in person perception:Effects of situation-behavior relations on dis-positional judgments’, Journal of Personalityand Social Psychology, 56(1): 41–53.

Shoda, Y., Mischel, W. and Wright, J.C. (1993a)‘The role of situational demands and cogni-tive competencies in behavior organizationand personality coherence’, Journal ofPersonality and Social Psychology, 65(5):1023–35.

Shoda, Y., Mischel, W. and Wright, J.C. (1993b)‘Links between personality judgments andcontextualized behavior patterns: Situation-behavior profiles of personality prototypes’,Social Cognition, 4(11): 399–429.

Shoda, Y., Mischel, W. and Wright, J.C. (1994)‘Intra-individual stability in the organizationand patterning of behavior: Incorporatingpsychological situations into the idiographicanalysis of personality’, Journal of Personalityand Social Psychology, 67(4): 674–87.(Article also reprinted as a chapter in Cooper,C.L. and Pervin, L.A. (1998) Personality:Critical Concepts in Psychology. NewYork/London: Routledge.)

Walker, L.E. (1979) The Battered Woman. NewYork: Harper & Row.

Walters, B. (1992) [Interview of DavidLetterman, January 29, 1992].

Wang, J., Lee, J.J., Shoda, Y. and Leu, J. (2007)‘Culture and self evaluation: An idiographicapproach’, Poster presented at the AnnualMeeting of the Society for Personality andSocial Psychology, Memphis.

Whitsett, D.D. and Shoda, Y. (2007) ‘Providingemotional support to individuals in distress:Women’s utilization of nonverbal cues’,


9781412946520-Ch18 5/23/08 12:40 PM Page 400

Poster presented at the Annual Meeting ofthe Society for Personality and SocialPsychology, Memphis.

Wright, J.C. and Mischel, W. (1987) ‘A condi-tional approach to dispositional constructs:The local predictability of social behavior’,Journal of Personality and Social Psychology,53(6): 1159–77.

Wright, J.C. and Mischel, W. (1988)‘Conditional hedges and the intuitive psy-chology of traits’, Journal of Personality andSocial Psychology, 55(3): 454–69.

Zayas, V. and Shoda, Y. (2005) ‘Do automaticreactions elicited by thoughts of romanticpartner, mother, and self relate to adult

romantic attachment?’, Personality andSocial Psychology Bulletin, 31(8): 1011–25.

Zayas, V. and Shoda, Y. (2007) ‘Predicting preferences for dating partners from past experiences of psychological abuse:Identifying the “psychological ingredients”of situations’, Personality and SocialPsychology Bulletin, 33(1): 123–38.

Zayas, V., Shoda, Y. and Ayduk, O.N. (2002)‘Personality in context: An interpersonal sys-tems perspective’, Journal of Personality,70(6): 851–900.

Zayas, V., Shoda, Y., Osterhout, L., Takahashi,M.M. and Mischel, W. (under review) ‘Neuralresponses to partner rejection cues’.


9781412946520-Ch18 5/23/08 12:40 PM Page 401

Date post:	23-Jul-2020
Category:	Documents
Upload:	others
View:	2 times
Download:	0 times

Assessment of Self-Regulative Traitspeople.psych.cornell.edu/~pac_lab/pdf/Zayas,Whitsett,etal... ·...

Documents