Personal Need for Structure: Individual Differences in

1 downloads 0 Views 4MB Size Report
fices, wearing headphones in the streets). People may choose ... ing toward him or her across the room—smiling, with arm outstretched—is a politician, one is ...
Copyright 1993 by the American Psychological Association, Inc. 0022-3514/93/S3.00

Journal of Personality and Social Psycholc 1993, Vol.65, No. 1.113-131

Personal Need for Structure: Individual Differences in the Desire for Simple Structure Steven L. Neuberg and Jason T. Newsom Individual differences in the desire for simple structure may influence how people understand, experience, and interact with their worlds. Studies 1 and 2 revealed that the Personal Need for Structure (PNS) scale (M. Thompson, M. Naccarato, & K. Parker, 1989,1992) possesses sufficient reliability and convergent and discriminant validity. In Studies 3-5, Ss high in PNS were especially likely to organize social and nonsocial information in less complex ways, stereotype others, and complete their research requirements on time. These data suggest that people differ in their chronic desire for simple structure and that this difference can have important social-cognitive and behavioral implications. A consideration of chronic information-processing motives may facilitate the theoretical integration of social cognition, affect, motivation, and personality.

It is nearly trite to make the point that, as social beings, people live in an unimaginably complex, information-rich world. Given the vast amount of information impinging on the senses and the well-documented limits in attentional capacity (e.g., Kahneman, 1973; Norman & Bobrow, 1975; Pashler, 1992), the information-processing task is quite formidable. Thus, people look for ways to reduce the information load. People lessen their cognitive load in two fundamental ways. First, through avoidance strategies, they limit the amount of information to which they are exposed. People may create barriers that restrict the likelihood that social and environmental information will intrude unexpectedly on their lives (e.g., by placing walls around their yards, closing the doors of their offices, wearing headphones in the streets). People may choose not to venture beyond their homes and workplaces, thus reducing exposure to unwanted information. Also, when compelled to confront and interact with others, people may actively ignore

Steven L. Neuberg and Jason T. Newsom, Department of Psychology, Arizona State University. Some of this work was presented at the 98th Annual Convention of the American Psychological Association, August 1990, Boston, at a symposium on "Affect and Self-Complexity." Portions of this research were supported by the North Atlantic Treaty Organization under a grant awarded to Steven L. Neuberg in 1987. We appreciate the help of Cheriese Chambers, Tiffany Dixson, Frank Russell, and Miguel Velez, who served as able experimenters. We additionally thank Dick Sorrentino and Steve Hanna for their help with the Uncertainty Orientation measure and, for their useful comments at various stages of this project, we thank Bob Cialdini, Susan Fiske, Doug Kenrick, Mike Naccarato, Morris Okun, John Reich, Erika Schupak-Neuberg, Megan Thompson, Steve West, Mark Zanna, the participants of the 1989 Nags Head Conference on Self-Control of Thought and Emotion, the graduate students in Steven L. Neuberg's stereotyping seminar, and several anonymous reviewers. Correspondence concerning this article should be addressed to Steven L. Neuberg, Department of Psychology, Arizona State University, Tempe, Arizona 85287. 113

potentially available information (e.g., by avoiding eye contact and diminishing the length of social encounters). The task of information reduction is also assumed by attempts to structure the world into a simplified, more manageable form. Behaviorally, people may establish routines and rely on formalized social scripts in their encounters with others, thus reducing the amount of information to which they must attend. Our present focus is on a second structural means of reducing informational quantity and complexity—cognitive structuring. When we speak of cognitive structuring, we refer to the creation and use of abstract mental representations (e.g., schemata, prototypes, scripts, attitudes; and stereotypes)—representations that are simplified generalizations of previous experiences (e.g, see Abelson, 1981; Allport, 1954; Bruner, 1957; Fiske & Taylor, 1991; Harvey, Hunt, & Schroder, 1961; Markus & Zajonc, 1985). Such knowledge structures represent related classes of experience, abstracting from these experiences features that differentiate them from others. For example, even though one's cognitive representations of politicians may contain detailed information about particular individuals, they importantly also contain one's beliefs regarding what politicians in general—as a class—tend to be like. Although these generalizations may not always accurately apply to certain instances (e.g., there clearly exist politicians who fail to fit held stereotypes), they serve people's cognitive efficiency needs quite well, enabling them to draw inferences about new events without having to expend a great deal of cognitive resources trying to understand each individual event in its particulars. Thus, when one discovers that the person walking toward him or her across the room—smiling, with arm outstretched—is a politician, one is able to rapidly draw stereotypical inferences regarding the person's intentions, characteristics, and so forth and respond accordingly. Thus, the creation and use of cognitive structures is a means of understanding one's world with a relatively minimal expenditure of cognitive resources. Note that such structures are best able to reduce one's cognitive load to the extent that they are "simple"—relatively homogeneous, well defined, and distinct

114

STEVEN L. NEUBERG AND JASON T. NEWSOM

from other structures—thus enabling clean, clear interpretations of new events. The more heterogeneous the structure and the more blurred its boundaries, the less useful it is for efficient characterization of new events and situations. When we discuss, then, the cognitive efficiency inherent in the creation and use of cognitive structures, we are referring to the creation and use of such simple, well-bounded representations. Much research supports the significance of such simple structures. Not only do they aid in the efficient processing of environmental, social, and personal information, but also a wealth offindingsdemonstrate structure-consistency biases in attention, interpretation, memory, inference, and impression formation and change (for reviews, see Fiske & Taylor, 1991; Higgins & Bargh, 1987; Markus & Zajonc, 1985; Miller & Turnbull, 1986), suggesting that people also possess a tendency to maintain such structures. Indeed, some researchers have explicitly suggested the existence of a general need for structure. For example. Harvey and Schroder (1963) wrote that One might go so far as to postulate a need for structure. . . Without such an ability survival itself would in most cases be rendered impossible or at least markedly impaired. Concepts contribute to this assessment of the world; indeed, they are perhaps the sole embodiment of it. As such, they provide one with a means of fate control, a vehicle through which some sort of adaptive equilibrium is maintained between the inside and outside, (p. 99)

More recently, Kruglanski and colleagues (e.g., Kruglanski & Freund, 1983; see also Kruglanski, 1989) have conceptualized the need for structure as a situationally induced motive activated under such acute circumstances as time pressure to reach a decision (Bechtold, Naccarato, & Zanna. 1986; Freund, Kruglanski, & Shpitzajzen, 1985; Jamieson & Zanna. 1989; Kruglanski & Freund, 1983). Clearly, the creation and use of simple cognitive structures have their function in the simplification of a complicated world, in both a general and more circumscribed sense. The purpose of the present article is to assess the usefulness of the need for simple structure construct as an individual-difference variable. That is, we argue that people meaningfully differ in the extent to which they are dispositionally motivated to cognitively structure their worlds in simple, unambiguous ways. We are interested in the dispositional approach for several reasons. First, it is now clear that the use of simple cognitive structures often underlies many interesting and important psychological events—stereotyping and prejudice, affective extremity, certain psychopathologies including depression, and self-fulfilling prophecies, to name just a few. Thus, to identify individuals for whom simple structuring processes are particularly desirable is to identify those individualsespecially susceptible to these phenomena. Hence, in addition to helping researchers better understand the nature of thought processes, the articulation of a dispositional need for structure construct may also have significant theoretical and practical implications for many domains of broader psychological interest. Second, one result of the "cognitive revolution" in psychology has been the assumption of a ubiquitous need to simplify one's environment. Although recent research has begun to emphasize the social perceiver's flexibility in response to alternative

social-cognitive motivations (for reviews, see Fiske & Neuberg. 1990; Higgins & Sorrentino, 1990; Hilton & Darley, 1991: Kunda, 1990; Miller & Turnbull, 1986; Neuberg, in press; Pittman & Heller, 1987; Pyszczynski & Greenberg, 1987; Showers & Cantor. 1985: Sorrentino & Higgins, 1986; Srull & Wyer, 1986), much of this work still assumes that simplification motives serve as the default, with alternative motives (e.g., accurate social perception) arising primarily when activated by extraindividual forces (e.g., social interdependence). An exploration of individual differences in the desire for simple structure enables us to begin addressing the ubiquity of the simplification motive. Moreover, evidence supporting such a motive-based individual difference would not only further imply the flexibility of the social perceiver but also would place the source of at least some of this flexibility within stable aspects of the perceiver as opposed to within the perceiver's social environment. Finally, we believe that an investigation of individual differences in the preference for simple structure could serve an important integrative function. Although it is the accepted wisdom that the way people think about their world is determined jointly by both perceiver and environment (e.g., Cantor, 1981; Mischel. 1979; Neisser, 1976), the relevant research has focused primarily on environmental factors (e.g., the nature of available information and the environmental context) and on short-term, unstable perceiver factors (e.g., primed categories and mood). The present conceptualization thus joins other attempts—for example, the constructs of chronically accessible categories (e.g., Bargh & Pratto, 1986; Higgins & King, 1981), self-monitoring processes (Snyder. 1974. 1979), self-consciousness (Fenigstein. Scheier, & Buss, 1975), and need for cognition (Cacioppo & Petty, 1982)—to link notions of motivation and social cognition with personality. We feel, then, that an investigation of individual differences in the desire for simple structure may be quite important. Our strategy here is as follows: First, seeking a dispositional operational ization of the desire-for-simple-structure construct, we explore the theoretical and psychometric characteristics of several potentially relevant existing scales. To preview our findings, we conclude that the Personal Need for Structure (PNS) scale (M. Thompson et al., 1989,1992) is best equipped for this task. Second, we assess the convergent and discriminant relations between PNS and relevant personality measures. Third, we present a series of studies investigating the predictive validity of the PNS operational ization. Finally, we expand our focus to discuss the conceptual and practical importance of PNS for additional aspects of social functioning. Study 1 The purpose of thisfirstinvestigation is to operational ize the desire for simple structure construct. A person high in desire for simple structure might be conceptualized as one leading a simple, tightly organized life, both cognitively and behaviorally. The contents of their cognitive structures should be relatively homogeneous, each structure should be well bounded, relatively distinct from others, and such individuals should be relatively uninhibited in their use of their simple structures to interpret new events. These individuals should be especially likely to

PERSONAL NEED FOR STRUCTURE

establish and enjoy routines, prefer familiar social situations, and so on. In sum, such individuals are motivated to seek out simply structured ways of dealing with their worlds. Our first study focuses on M. Thompson et al.'s (1989,1992) PNS scale, because it most directly approaches the construct of interest. Of course, it may occur to the reader that the individual-difference notion of desire for simple structure seems somewhat similar to previous attempts to address personality influences in social thought processes. For example, Authoritarianism (Adorno, Frenkel-Brunswick, Levinson, & Sanford, 1950), Dogmatism (Rokeach, 1960), Intolerance of Ambiguity (Eysenck, 1954; Frenkel-Brunswick, 1949), Rigidity (Gough& Sanford, 1952), and Uncertainty Orientation (Sorrentino & Hewitt, 1984; Sorrentino & Short, 1986) all possess as a component of their respective conceptualizations the notion of a preference for cognitive simplicity and structure. As the data and conceptual analyses reported herein suggest, however, the PNS scale is much better suited for the task of operationalizing in a reliable, direct manner the construct of interest. Thus, although these other personality constructs may share some similarities with the PNS construct, they are importantly divergent as well. We address this issue in Study 2. The PNS scale was created specifically in an attempt to capture, as a chronic individual motive, several aspects of the desire for simple structure. As M. Thompson et al. (1992) wrote, "A high PNS individual would prefer structure and clarity in most situations, with ambiguity and "grey areas" proving troublesome and annoying . . . Such people should experience discomfort if they perceive structure and clarity to be missing from situations" (p. 3). Such a conceptualization is quite compatible with our own, although our present analysis differs somewhat in that it focuses more heavily on issues of cognitive structuring. We present the PNS scale in the Appendix. A larger set of items were generated on the basis of their face-valid relationship to the constructs of desire for structure, certainty, and cognitive closure (see M. Thompson et al., 1992). Psychometric analyses across several subject populations reduced the total set to the 12 existing items. Subjects respond to these items on a 6-point scale ranging from strongly disagree to strongly agree. Items 2, 5, 6, and 11 are reverse scored, such that higher scores reflect a greater motive for simple structure. Several features of the scale make it particularly suited for our present objectives. First, the scale items appear to assess individual preferences for structure and organization without assessing attitudes toward social and political issues—a problem characterizing many scales with somewhat related purposes (see Study 2). Second, the scale items speak generally to issues of structure and organization and do not specifically address cognitive structuring; although cognitive structuring is indeed of great interest to us, we consider the M. Thompson et al. (1992) strategy beneficial because it does not require respondents to report on processes to which they may have little or no access (Ericsson & Simon, 1980; Nisbett & Wilson, 1977). Finally, the scale is short and easy to administer. Because the scale and the relevant psychometrics are presently unpublished, however, we briefly describe our own psychometric work with the scale. For both conceptual and empirical reasons, we discarded

115

Item 5 ("I enjoy being spontaneous"). Conceptually, we felt that this item assessed something besides desire for simple structure. Moreover, on empirical grounds, preliminary analyses revealed that the item mean was consistently well over the objective scale midpoint, that subject response distributions on this item were highly skewed in the affirmative direction, and that the item had inconsistent factor loadings across several datasets. Such data implied that this item has little ability to differentiate among subjects—most subjects claimed to enjoy being spontaneous—and lacks conceptual coherence with the other items. The analyses reported herein are thus based on a revised 11 -item scale. The remaining 11 items seem to capture two related, but conceptually distinct, elements one might desire in a need for structure scale (see Mikulincer, Yinon, & Kabili, 1991, for a similar interpretation). Some items (proposed Factor 1: Items 3, 4, 6, and 10) appear to assess the extent to which people desire to establish structure in their everyday lives (e.g., "I enjoy having a clear and structured mode of life"; "I find that a consistent routine enables me to enjoy life more"). The remaining items (proposed Factor 2: Items 1, 2, 7, 8, 9, 11, and 12) appear to assess the manner in which people respond to a lack of structure (e.g., "It upsets me to go into a situation without knowing what I can expect from it"; "I enjoy the exhilaration of being in unpredictable situations"). The purpose of our first investigation was threefold: (a) to assess the factor structure of the PNS scale, focusing on a comparison of the proposed two-factor structure and its one-factor alternative, (b) to assess the scale's internal reliability, and (c) to assess the scale's test-retest reliability.

Method The PNS scale was administered to six independent groups of undergraduate male and female students, over 2,900 subjects in all (Group 1, spring 1988, University of Waterloo, Waterloo, Ontario, Canada, n = 205; Group 2, summer 1988, University of Waterloo, n = 254; Group 3, fall 1988, Arizona State University, n = 384; Group 4, fall 1988, Arizona State University, n = 240; Group 5, spring 1989, Arizona State University, n = 969; and Group 6, spring 1990, Arizona State University, n = 914). Because no sex differences were found in any of the following analyses, subject sex is not discussed further. Each subject population completed the PNS scale as part of a battery of questionnaires administered in large group-testing sessions. Cronbach's alphas were calculated, and confirmatory factor analyses (EQS, version 3.0; Bentler, 1989) were used to assess the scale's factor structure. Finally, a subsample of Group 3 (n= 79) completed the PNS scale a second time, 12 weeks later, enabling us to assess test-retest reliability.

Results and Discussion Factor structure. Confirmatory factor analyses testing the predicted two-factor model against the one-factor alternative indicated that, for all six subject populations, the two-factor model better accounted for our subjects' data than did the onefactor model—as indicated by chi-square differences and com-

116

STEVEN L. NEUBERG AND JASON T. NEWSOM

Table 1 Psychometric Properties of the Revised 11-Item Personal Need for Structure Scale for Six Independent Samples Group Measure

1 (n = 205)

2 (n = 254)

3 (n = 384)

4 (n = 240)

5 (n = 969)

6 (n = 914)

120.02

160.62

252.64

1 42.22

654.09

512.59

.82 .85 .88

.75 .75 .80

.76 .74 .79

.78 .79 .84

.69 .63 .70

.80 .77 .81

49.96

113.46

163.09

97.60

343.01

351.58

.93 .99 .99

.82 .84 .88

.85 .85 .88

.85 .88 .91

.84 .81 .86

.86 .84 .88

70.12

47.16

89.55

44.62

311.08

161.01

Test of one-factor model

x\df= 44)

Bentler-Bonett Normed Fit Index Tucker-Lewis Index Comparative Fit Index Test of two-factor model

xV/= 43)

Bentler-Bonett Normed Fit Index Tucker-Lewis Index Comparative Fit Index X2 difference for oneand two-factor models3

' All chi-square differences are significant at p < .0001.

parisons of the Bentler-Bonett Normed Fit Index (Bentler & Bonett, 1980), the Tucker-Lewis Index (Tucker & Lewis, 1973; also see Bentler & Bonett, 1980), and the Comparative Fit Index (Bentler, 1990; see Table 1). Despite the clear advantage of the two-factor model, however, the two factors nonetheless correlated fairly highly (interfactor correlations ranged from .54 to .75 across the six samples; see Table 2). Thus, although the

two-factor measurement model fits better the observed data, the two factors are clearly highly related, as one would expect. Internal reliability. Across the six subject populations, the internal reliabilities of the overall scale and the two predicted constituent factors were adequate (with the median Cronbach a = .11; see Table 2). Test-retest reliability. The Time 1-Time 2 correlations were

Table 2 Item Loadings for Two-Factor Personal Need for Structure Model, Factor 1-Factor 2 Correlations, and Cronbach Alphas Group Item loadings for two-factor model Factor 1: Desire for Structure Item 3 Item 4 Item 6 (reversed) Item 10 Factor 2: Response to Lack of Structure Item 1 Item 2 (reversed) Item 7 Item 8 Item 9 Item 11 (reversed) Item 12 Factor 1 -Factor 2 correlations Cronbach alphas Total 11-item scale Factor 1 items Factor 2 items

1 (n = 205)

2 (n = 254)

3 (n = 384)

4 (n = 240)

5 (n = 969)

6 (n = 914)

.792 .631 .546 .795

.787 .660 .515 .569

.731 .598 .611 .770

.645 .642 .518 .763

.735 .617 .505 .681

.693 .609 .466 .750

.632 .449 .731 .578 .569 .635 .555

.541 .346 .688 .585 .577 .384 .447

.597 .384 .703 .544 .465 .532 .432

.408 .421 .655 .609 .679 .535 .419

.507 .356 .667 .560 .504 .482 .424

.561 .331 .729 .638 .611 .472 .488

.693

.671

.676

.731

.544

.745

.85 .78 .79

.79 .72 .70

.81 .77 .72

.86 .74 .82

.76 .72 .69

.81 .72 .74

PERSONAL NEED FOR STRUCTURE

all highly significant, indicating good 12-week test-retest reliability for the overall 11 -item scale (r = .76), for Factor 1 (r = .84), and for Factor 2 (r = .79). Summary. The PNS scale seems to assess the proposed construct in a face-valid and reliable manner, as indicated by our six samples. Moreover, it seems to capture two conceptually related, but somewhat independent, factors: the extent to which people prefer to structure their lives (Desire for Structure) and the manner in which people respond when confronted with unstructured, unpredictable situations (Response to Lack of Structure). Moreover, test-retest reliability over a 12-week period was high, suggesting that the scale is capturing a fairly stable construct. Study 2: PNS and Other Relevant Measures The results of our first investigation indicate that the PNS scale likely serves as a reliable operationalization of the desire for simple structure. The purpose of Study 2 was twofold: to assess the ability of other existing scales to reliably operationalize desire for simple structure as well as to begin assessing the convergent and discriminant validity of the PNS scale. Several other existing scales also appear to assess the desire for simple structure. For example, the Authoritarian syndrome has implied rigidity in both cognitive function and in the acceptance of social hierarchies (Adorno et al., 1950; Kirscht & Dillehay, 1967); Dogmatism (Rokeach, 1960) is concerned with the openness and closedness of belief systems; Intolerance of Ambiguity was theorized as a construct related to one's resistance to lack of conceptual clarity (Eysenck, 1954); Uncertainty Orientation addresses the ways in which people attempt to seek cognitive clarity (Sorrentino & Short, 1986); and Rigidity (Gough & Sanford, 1952) reflects the flexibility of thought and behavior. The similarities of these scales to PNS seem more shallow than initially apparent, however. Conceptually, these alternatives are characterized not only by features relevant to simple cognitive structuring but also by their interest in additional, unrelated constructs (e.g., Authoritarianism also purports to assess conventionalism, power and toughness, authoritarian aggression and submission, etc.). Moreover, a perusal of the items of these alternatives surprisingly suggests little operational overlap with the PNS scale. In addition, these other scales often use items of a sociopolitical nature, making it difficult to discern the underlying psychological construct or constructs being operationalized. Indeed, our own a priori content analysis revealed only a handful of items across these scales that seem to have any face validity as representations of the needfor-simple-structure construct (Rigidity was an exception; see later discussion). Psychometric analyses of these alternatives should help us assess their ability to reliably operationalize desire for simple structure, and the correlations of these scales with PNS will enable the assessment of PNS's discriminant validity. Second, the Need for Cognition scale (Cacioppo & Petty, 1982) has been proposed to represent chronic preferences of people to engage in effortful, elaborative thought. In fact, individuals high in need for cognition tend to be more likely to centrally process persuasive communications (Cacioppo, Petty,

117

& Morris, 1983). Because the desire to simply structure one's world need not imply effortless processing—there may indeed be cases in which one goes to great cognitive lengths to create or maintain a simple set of structures or beliefs—we expected only a small negative relationship between PNS and Need for Cognition. Indeed, we argue here that Need for Cognition may best represent preferences for the amount of cognitive activity, whereas PNS may better represent preferences for the desired outcome (simple structure) of cognitive activity. Third, we felt that identifying the relationship between PNS and the constructs of the Big Five personality dimensions (e.g., see Goldberg, 1981, 1990; John, 1990; McCrae & Costa, 1985, 1987) would be useful both for purposes of assessing convergent and discriminant validity and for assessing proposed distinctions between the two PNS subfactors. Specifically, we anticipated that PNS should be particularly related to the Openness and Conscientiousness dimensions, with Factor 1 (Desire for Structure)—given its proactive nature—associating to a greater extent with Conscientiousness and Factor 2 (Response to Lack of Structure)—because of its social-affective nature—associating to some extent with Neuroticism and Extraversion. Fourth, because the Response to Lack of Structure subfactor items seem to possess an anxiety component, we thought it appropriate to include several measures of trait anxiety and affective distress as a means of assessing hypothesized distinctions between it and Factor 1 (Desire for Structure). Fifth, we expected that PNS should have only a weak negative relationship with intelligence, such that those who prefer a simpler structure might be lower in intelligence; a weak association here would imply that PNS captures a motivation for simple structure, as opposed to merely reflecting a limited cognitive capacity. Sixth, to assess the breadth of the PNS operationalization, we also included scales measuring behavioral routinization and self-monitoring (Snyder, 1974,1979), reasoning that an overall syndrome of desiring simple structure should be reflected both in the creation of behavioral routines and in a greater reluctance to rely on the behavior of others as a guide for one's own. Finally, we also included a measure of social desirability. Method Four independent groups of Arizona State University undergraduates completed scale batteries in large testing sessions in partial fulfillment of introductory psychology subject pool requirements. Depending on the session, subjects were administered some subset of the following scales, in randomized order: the PNS scale, the MarloweCrowne Social Desirability Scale (MCSDS; Crowne & Marlowe, 1964), the Balanced F Scale (Byrne, 1974), the Dogmatism scale, Form E (Rokeach, 1960), the Rigidity scale (Gough & Sanford, 1952), the Intolerance of Ambiguity Scale (Eysenck, 1954), the Uncertainty Orientation inventory (Sorrentino & Hewitt, 1984; Sorrentino & Short, 1986), the Need for Cognition scale (Cacioppo & Petty, 1982), the Big Five Inventory of the Omni Personality Survey (John, Donahue, & Kentle, 1991). the Social Anxiety subscale of the Self-Consciousness Scale (Fenigstein, Scheier, & Buss, 1975), the Taylor Manifest Anxiety Scale (Taylor, 1953), the Routinization scale (Reich & Zautra, 1991), the Self-Monitoring scale (Snyder, 1974), and the Wonderlic Personnel Test (Form A; Wonderlic, 1983)—an easily administered test of gen-

118

STEVEN L. NEUBERG AND JASON T. NEWSOM

desire for simple structure (albeit with weak reliability; a = .64), although the scale in its original form seems quite unreliable and nonunitary. Other findings relevant to convergent and discriminant validity. First, as expected, the correlations between PNS and Need for Cognition tended to be negative and weak, suggesting Results that PNS and Need for Cognition may be relatively orthogonal Do alternative operationalizations reliably capture the desireconstructs, as hypothesized. Note that this was also the case when we broke the Need for Cognition Scale into its three profor simple structure? First, confirmatory factor analyses and posed subcomponents: cognitive persistence, cognitive confiaverage interitem correlations indicate that the Authoritariandence, and cognitive complexity (Tanaka, Panter, & Winborne, ism, Dogmatism, Intolerance of Ambiguity, and Uncertainty 1988). Orientation measures all seem to assess multiple constructs, as Second, the relationships between PNS and the Big Five facsingle-factor models provided very weak fits to the data and tors were also consistent with expectations. Both PNS factors average interitem correlations were quite low (see Table 3). were negatively correlated with Openness, Desire for Structure Thus, although these scales may possess a desire-for-simple(but not Response to Lack of Structure) was positively correstructure component, they likely contain other constructs as lated with Conscientiousness, Response to Lack of Structure well, limiting their usefulness as a straightforward operation(but not Desire for Structure) was positively correlated with alization of the desire for simple structure (or of any other single Neuroticism and negatively correlated with Extraversion, and construct, for that matter). Of course, given their diffuse, multiPNS was uncorrelated with Agreeableness. These data not only ple-construct conceptual underpinnings, this is not surprising. provide important information on the convergent and discrimiMoreover, despite the levels of statistical significance (given nant validity of the PNS scale, but they also provide support for our large subject sizes), the magnitude of the correlations of the proposed distinction made between the scale's two subfacthese scales with PNS ranges from only low to moderate (see tors. Table 4, for correlations of PNS with all scales). Indeed, it is Third, as expected, correlations of PNS with measures of illustrative that only one item from each of the Authoritarianaffective distress—the Social Anxiety subscale of the Selfism, Dogmatism, and Intolerance of Ambiguity scales correConsciousness Scale and the Taylor Manifest Anxiety Scale— lated above .30 with the PNS scale or either of its subfactors, and reveal that such distress is better reflected in PNS Factor 2. none of these three items did so consistently across scale adminResponse to Lack of Structure. C. Showers (personal communiistrations.1 Thus, although desire for simple structure may be cation, 1990) reported similar findings using Spielberger et al.'s relevant to these conceptualizations, PNS is clearly not redunState-Trait Anxiety Inventory (Spielberger, Gorsuch, & Ludant with these alternative operationalizations. shene, 1970). These data thus provide additional evidence supIn contrast, there are reasons to believe that the Rigidity porting the proposed conceptual distinction between the two scale may fare somewhat better as a dispositional measure of PNS factors. desire for simple structure, although it too, in its original form, Fourth, PNS was only minimally negatively correlated with clearly consists of multiple factors and is unreliable (see Table our measure of intelligence. This suggests that high-PNS indi3). An a priori perusal of its items, however, suggested that six viduals are not merely those who are constrained to simplify of its items may indeed capture the desire-for-simple-structure their world because of a general lack of intellectual capabilities construct—four of which appear compatible with the PNS Debut rather may be those who prefer to do so. Of course, this sire for Structure (Factor 1; e.g., "I find it easy to stick to a accords quite well with our common observations of highly certain schedule, once I have started it") and two of which apintelligent individuals who nonetheless seem to engage disposipear compatible with the PNS Response to Lack of Structure tionally in structural simplification strategies. (Factor 2; e.g., "I do not enjoy having to adapt myself to new and unusual situations"). The remaining 16 items appear less directly relevant to the desire-for-simple-structure construct (e.g., 1 "I never miss going to church"). The average of the lowest item-total correlations for the overall PNS scale across the six administrations discussed earlier was .35, with Consistent with this, a two-factor solution based on our conthe worst of the PNS items (Item 2 on the Group 6 administration) ceptual analysis (i.e., the six desire-for-simple-structure items vs. the remaining 16 items) significantly improved the model fit correlating with the overall PNS scale at .31; the average of the lowest item-total correlations for PNS Factor 1 across administrations was .45. (see Table 3), although this fit too remains far from desirable. with the worst of the PNS items (Item 6 on the Group 6 administration) More specifically, as Table 4 indicates, the moderate correlacorrelating with the factor at .41; the average of the lowest item-total tions of Rigidity with PNS seem to stem primarily from the correlations for PNS Factor 2 across administrations was .32, with the 6-item subscale. Whereas the average correlations between worst of the PNS items (Item 2 on the Group 2 administration) correthese six Rigidity items and the overall PNS scale were relalating with the factor at .26. We thus felt that .30 would seem a reasonable, and highly conservative, cutoff point to assess the ability of any tively high (fall 1988 = .32; summer 1991 = .41; fall 1991 = .40), item from alternative scales to inform the desire-for-simple-structure the average correlations between PNS and the remaining set of construct. To the extent that few items from other scales reach this 16 items were predictably less so (fall 1988 = . 14; summer 1991 criterion, we can be additionally confident that they are not assessing = .14; fall 1991 = .13). It seems to be the case, then, that one desire for simple structure per se. subcomponent (27%) of the Rigidity scale began to capture the eral intelligence that correlates highly with both the original and revised versions of the Wechsler Adult Intelligence Scale (Dodrill, 1981, 1983; Dodrill & Warner, 1988; Hawkins, Faraone, Pepple, Seidman, Tsuang, 1990).

119

PERSONAL NEED FOR STRUCTURE

Table 3 Goodness-of-Fit Indices From One-Factor Confirmatory Analyses of Authoritarianism, Dogmatism, Intolerance of Ambiguity, Uncertainty Orientation, and Rigidity Scale Authoritarianism Fall 1988 (77= 191) Summer 1991 (n = 360) Dogmatism Fall 1988 (n= 191) Intolerance of Ambiguity" Fall 1988 (n= 191) Summer 1991 (n = 360) Uncertainty Orientation Summer 1991 (« = 80)" ^Uncertainty component* Authoritarianism component Rigidity Fall 1988 (n= 191) Summer 1991 (« = 360) Fall 1991 (« = 851) Two-factor solution11 Six-item subscale Remaining 16 items

x2

df

BBNFI

TLI

CFI

a

MIIR

433.01 628.34

209 209

.51 .55

.62 .60

.66 .64

.75 .75

.13 .12

1306.71

740

.32

.48

.51

.82

.11

132.93 145.80

77 77

.36

.43 .67

.52 .72

.46 .55

.06

.58

538.23 399.16

298 — 209

.24 — .29

.31 — .36

.37 — .42

— .13 .71

.04 .11

369.20 445.80 946.15 940.34 44.49 506.22

209 209 209 208 9 104

.48 .60 .65 .65 .91 .63

.63 .71 .67 .67 .88 .62

.67 .74 .70 .70 .93 .67

.73 .75 .76 .76 .64 .66

.11 .12 .13 .13 .23 .11



.08 —

Note. Goodness-of-fit indices are for one-factor confirmatory analyses unless otherwise indicated. BBNFI = Bentler-Bonett Normed Fit Index; TLI = Tucker-Lewis Index; CFI = Comparative Fit Index; MIIR = mean interitem correlation. * Because of the dichotomous scoring format, we used the EQS maximum likelihood robust method of estimation here; note, however, that other relevant methods yield similar fits. b We tested Sorrentino, Roney, and Hanna's (1989) proposed two-factor model, using their assumption that Uncertainty Orientation's two components (^Uncertainty and Authoritarianism) are uncorrelated. Because of the different response formats for the two components (i.e., sentence completion and Likert-type scale, respectively), we did not generate overall alphas or mean interitem correlations. c Because of poor interitem correlations within the ^Uncertainty component, and several nonsignificant loadings, the EQS program was unable to generate an interpretable solution, thus dashes indicate no score was available.d x2(l , iV = 851) = 5.71, /? < .05 = chi-square difference for one- and two-factor models.

Fifth, as expected, high-PNS individuals tend toward behavioral routinization, as indicated by the positive correlation between PNS and Routinization. In a similar vein, high-PNS individuals also weakly tended toward being low self-monitors, suggesting that these individuals can be somewhat characterized by their use of own structures as a determinant of behavior. These findings suggest the breadth of the PNS conceptualization, moving beyond cognitive styles to behavioral preferences and self-reliance in behavior selection. Finally, as anticipated, PNS and its two subscales were uncorrelated with the MCSDS. Summary. It seems that PNS possesses reasonably good convergent and discriminant validity. First, although related to some extent, PNS is clearly distinct from Authoritarianism, Dogmatism, Intolerance of Ambiguity, and Uncertainty Orientation; it is highly unlikely that PNS serves merely as a proxy for any of these scales. Instead, given the multicomponent character of these alternatives, it strikes us that PNS may be reliably operationalizing one of these components. Also, as predicted, PNS was largely uncorrelated with Need for Cognition. Moreover, not only did PNS correlate with the expected factors of the Big Five, but also the pattern of correlations across the two PNS factors suggest that these factors are somewhat distinct in the manner posited. Correlations among

the two PNS factors and social anxiety measures likewise support the distinction between these two factors. In addition, its relationship with behavioral routinization and self-monitoring suggests that PNS's realm includes both the cognitive and the behavioral. Finally, PNS is only slightly related to intelligence and seems uninfluenced by social desirability concerns. The one alternative scale that PNS correlates with notably is Rigidity, although this association is primarily due to the six Rigidity items that are highly related conceptually to the desire for simple structure; its remaining conceptually unrelated items show a much weaker relationship. It thus appears that only one part of Rigidity (representing merely a quarter of the entire scale) begins to assess desire for simple structure per se. On the whole, then, the data reported to this point suggest that M. Thompson et al.'s (1989,1992) PNS scale is unique in its ability to reliably assess individual differences in desire for simple structure. Indeed, although the findings reported from the studies to follow—as well as from recent work done by other researchers—are all easily generated by the desire for simple structure construct, they cannot all be accounted for by any of the alternative conceptualizations investigated here. We return to this point later, as we strengthen our case for the discriminant validity of the PNS scale. We now present data from three additional studies—studies

120

STEVEN L. NEUBERG AND JASON T. NEWSOM

Table 4 Correlations of Personal Need for Structure W 'ith Other Scales Personal Need for Structure

Scale Authoritarianism Fall 1988 (n = 191) Summer 1991 (n = 360) Dogmatism Fall 1988 (/; = 191) intolerance of Ambiguity Fall 1988 (n = 191) Summer 1991 (n = 360) Uncertainty Orientation Summer'1991 (n = 80) ^Uncertainty component Authoritarianism component Need for Cognition Overall Fall 1988 (n =191) Summer 1991 (n = 360f Persistence Fall 1988