LP762-L2 - Week 5

Organizational Behavior and Human Decision Processes 113 (2010) 73–84

Contents lists available at ScienceDirect

Organizational Behavior and Human Decision Processes

journal homepage: www.elsevier .com/ locate/obhdp

What do you think of followers? Examining the content, structure, and consequences of implicit followership theories

Thomas Sy Department of Psychology, University of California, 900 University Ave., Riverside, CA 92521, United States

a r t i c l e i n f o a b s t r a c t

Article history: Received 21 October 2009 Accepted 28 June 2010 Available online 23 July 2010

Accepted by Daan van Knippenberg

Keywords: Implicit theory Followership Prototypes Leadership Categorization theory

0749-5978/$ - see front matter � 2010 Elsevier Inc. A doi:10.1016/j.obhdp.2010.06.001

E-mail address: [email protected]

Implicit followership theories (IFTs) are defined as individuals’ personal assumptions about the traits and behaviors that characterize followers. Goals of this research were to: (1) Identify the content and struc- ture of IFTs, (2) examine the relationship between IFTs and extant implicit theories in the leadership lit- erature, and (3) establish a preliminary nomological network of leaders’ implicit followership theories by examining its consequences for leader–follower interpersonal outcomes. This study included 1362 par- ticipants across five separate studies and seven samples. Results provide evidence for content, conver- gent, discriminant, criterion, and incremental validity, as well as internal and temporal consistency of the IFTs instrument. IFTs are represented by a first-order structure (Industry, Enthusiasm, Good Citizen, Conformity, Insubordination, and Incompetence), and a second-order structure (Followership Prototype and Antiprototype). Leaders’ IFTs predicted interpersonal outcomes: liking, relationship quality, trust, and job satisfaction. Future research and practical implications are discussed for this understudied branch of leadership research.

� 2010 Elsevier Inc. All rights reserved.

Introduction

What are the perceptions that individuals have about follow- ers? A fundamental task of all organisms is to classify and catego- rize stimuli in their environment (Rosch, 1978). In particular, research from the social cognition literature indicates that individ- uals have a natural propensity to classify others (Fiske & Taylor, 1991). In organizational settings, researchers have posited that individuals are naturally inclined to classify people as leaders and followers (Engle & Lord, 1997; Lord, Foti, & Phillips, 1982; Lord & Maher, 1993). While a rich body of research on implicit leader- ship theories (ILTs) has been established in the past 30 years, there is noticeably little research on the corresponding notion of implicit followership theories (IFTs) despite evidence in support of IFTs. For example, McGregor (1960) posited that Theory X leaders have dif- ferent assumptions about the personal attributes of followers than do Theory Y leaders. Likewise, Eden (1990) proposed that leaders’ performance expectations are influenced by their implicit theories about workers. In addition, research has demonstrated that trans- formational and transactional leaders have different follower-sche- mas (Goodwin, Wofford, & Boyd, 2000; Wofford & Goodwin, 1994). Likewise, followers possess different follower-schemas that are influenced by context (Carsten, Uhl-Bien, West, Patera, & McGre- gor, 2010). Thus, leadership theories and empirical evidence indi-

ll rights reserved.

cate that individuals have preconceived notions or implicit theories about followers.

Individuals’ cognitions shape their judgment of and response to others (Fiske & Taylor, 1991). Indeed, research has established that individuals utilize ILTs to judge and respond to leaders (Lord & Ma- her, 1993). Like ILTs, individuals’ IFTs may shape how they judge and respond to followers, which has implications for outcomes in leader–follower contexts. Research on IFTs addresses a major gap in the leadership literature on how leaders and followers ‘‘perceive, decide, behave, and take action” (Avolio, Walumbwa, & Weber, 2009). Furthermore, insights generated from research integrating both ILTs and IFTs may provide a more complete understanding of leadership (van Gils, van Quaquebeke, & van Knippenberg, 2010). As such, the goal of the present research is to examine im- plicit followership theories (IFTs). Specifically, three key research questions are addressed in the present research: (1) What is the content and structure of IFTs, (2) How are IFTs related to implicit theories in the extant leadership literature (i.e., implicit leadership theories and implicit performance theories), and (3) What are the consequences of IFTs for leader–follower interpersonal outcomes?

Theoretical foundation and construct definition of IFTs

Research indicates that individuals are cognitive misers who naturally categorize others to streamline cognition because cate- gories serve the function of simplifying the external world (cogni- tive economy), permitting abstract and symbolic thinking, and

74 T. Sy / Organizational Behavior and Human Decision Processes 113 (2010) 73–84

providing a shared system of labels that facilitate communication (Fiske & Taylor, 1991). This process of categorizing others is so fundamental that it operates automatically and spontaneously (within milliseconds) by the mere presence of a stimulus target (Boldenhausen & McCrae, 1998). Individuals categorize others on the basis of visually salient cues (e.g., race, gender, age, etc.) and social roles (Operario & Fiske, 2004) such as leader and fol- lower. Research on the categorization of leaders in the organiza- tional literature has been examined within the context of ILTs. Surprisingly, research on the corresponding notion of IFTs is noticeably absent despite its appearance in the literature nearly two decades ago (e.g., Eden, 1990; Lord & Maher, 1993).

IFTs are defined as individuals’ personal assumptions about the traits and behaviors that characterize followers. Consistent with the literature on implicit theories, IFTs represent individuals’ ‘‘lay” or ‘‘naïve” theories (vs. formal theories) (Rosenberg & Jones, 1972). IFTs likely form at an early age (Hunt, Boal, & Sorenson, 1990) via socialization, and further develop as individuals gain experiences with followers (Lord & Maher, 1993). Central components of IFTs are prototypes, defined as abstract composites of the most repre- sentative member or the most commonly shared attributes of a particular category (Lord et al., 1982; Rosch, 1978). Prototypes may be based on common taxonomy (central tendency prototypes, i.e., ‘‘how followers are”) or goal-derived (ideal prototypes, i.e., ‘‘how followers should-be”) (Barsalou, 1985; Schyns & Meindl, 2005). Goal-derived follower prototypes have received some atten- tion in the literature. For example, followers’ perceptions of effec- tive followers include the attributes of ‘‘integrity,” ‘‘dependable,” ‘‘communication skills,” etc. (Carsten et al., 2010). Goal-derived fol- lower prototypes also have been examined under the framework of implicit performance theories (IPTs), defined as performance expec- tations for employees, and include the parallel attributes of ‘‘hon- est,” ‘‘reliable,” ‘‘communicates effectively,” etc. (Engle & Lord, 1997; Wernimont, 1971). As such, IPTs conceptually overlap with goal-derived follower prototypes in that both reflect expectations for effective followers, thereby consisting of only positive attri- butes. In comparison, common taxonomic follower prototypes are broader based, consisting of both positive and negative attri- butes (Eden, 1990; Goodwin et al., 2000; McGregor, 1960; Pillai & Uhl-Bien, 2007; Wofford & Goodwin, 1994) that provide a more comprehensive representation of followers. Indeed, scholars have asserted that perceptions of followers may be more negative than positive (Pillai & Uhl-Bien, 2007). The focus of the current research is common taxonomic follower prototypes.

IFTs may consist of both universal dimensions that are pertinent across contexts and idiosyncratic dimensions that are context spe- cific (van Gils et al., 2010). Consistent with the trajectory of re- search on ILTs, it may be more relevant to examine first the universal dimensions of IFTs (Epitropaki & Martin, 2004), with sub- sequent stages focusing on context-specific dimensions (Ensari & Murphy, 2003; House et al., 2004; Johnson, Murphy, Zewdie, & Rei- chard, 2008). Furthermore, while context may influence individu- als’ implicit theories (Foti, Knee, & Backert, 2008; Hanges, Lord, & Dickson, 2000), contextual effects likely reflect differences in endorsement for the same set of core attributes rather than actual differences in attribute content (Epitropaki & Martin, 2004). In- deed, differences in ILTs as a function of gender (Ensari & Murphy, 2003; Johnson et al., 2008), culture (Gerstner & Day, 1994), and race and occupation (Sy et al., in press) primarily reflect differences in endorsement for the same set of core leadership attributes. Like- wise, contextual differences in perceptions of effective followers primarily reflect differences in endorsement for the same set of core follower attributes (e.g., differences in endorsement for obedi- ence as an effective follower attribute) (Carsten et al., 2010). Thus, identifying the core content and structure of IFTs in the current research affords cross-context comparisons, thereby facilitating

future research on contextual effects on IFTs (Epitropaki & Martin, 2004; House et al., 2004).

Theoretical relevance of IFTs

Research on IFTs answers the call to address two key challenges to advance the field of leadership (Avolio et al., 2009): (1) The need to be more holistic in the examination of leadership by integrating followers as a focal element, and (2) the emerging trend of exam- ining leaders’ cognitions, and how they process information, par- ticularly, leaders’ perception of followers. As such, IFTs may help advance our understanding of leadership.

IFTs serve as ‘‘sensemaking” functions (Weick, 1995) from which individuals interpret, understand, and respond to follower behaviors (Poole, Gioia, & Gray, 1989). IFTs may influence individ- uals’ behaviors because they use IFTs as a benchmark to form impressions of followers (Lord & Maher, 1993). Individuals engage in a matching process of comparing their IFTs with a target person (e.g., a follower) and based on the degree of congruence, individu- als form an impression of followers that subsequently shapes their behaviors towards followers. Furthermore, inference-based pro- cesses (Lord & Maher, 1993) posit that IFTs can be inferred from followers’ performance. Inference-based processing involves mak- ing attributions of follower characteristics based on outcomes of salient events. For example, followers are perceived as industrious when business goals are successfully accomplished.

Individuals also may internalize and endorse certain IFTs over time, which may predispose them to judge and respond to followers in a certain fashion (Engle & Lord, 1997; Lord & Maher, 1993) be- cause implicit theories may operate spontaneously and automati- cally (Boldenhausen & McCrae, 1998). For example, theorists have argued that leaders’ management style is a function of their predisposed assumptions of followers (McGregor, 1960). Indeed, research demonstrates that leaders have different follower-sche- mas that predispose them to interpret events differently, which re- sults in correspondingly different behaviors toward followers (Goodwin et al., 2000). In sum, IFTs may influence outcomes in lea- der–follower contexts because IFTs (1) serve as benchmarks from which individuals judge and respond to followers, and (2) predis- pose individuals to judge and respond to followers in a certain fashion. The benchmark and predisposition propositions are con- sistent with theorists’ propositions that implicit theories affect judgments and behaviors via their impact on controlled and uncontrolled information processing (van Gils et al., 2010).

Because IFTs may be antecedents of leaders’ and followers’ af- fect, behaviors, and cognitions (Lord & Maher, 1993), IFTs may have particular relevance on leader–follower interpersonal outcomes (Graen & Uhl-Bien, 1995). Thus, leaders who have more positive IFTs may behave differently towards followers than leaders who have more negative IFTs (McGregor, 1960). Differences in leaders’ behaviors towards followers, as a function of their IFTs, should im- pact leader–follower interpersonal outcomes such as liking for leaders and followers, relationship quality, etc. (The impact of lead- ers’ IFTs on interpersonal outcome variables are examined in Study 5). It may be that the effect of leaders on followers is explained partly by IFTs. Indeed, existing evidence shows that performance differences between followers may largely result from leaders’ per- ceptions of and subsequent interactions with their followers (Goodwin et al., 2000; Wofford & Goodwin, 1994). Much can be gained by understanding how leaders utilize implicit theories to interpret their environment and subsequently guide leadership ac- tions (Avolio et al., 2009; Lord & Brown, 2004). Much can also be gained by understanding how leaders’ IFTs impact followers’ affect, behaviors, cognitions (i.e., subsequent information processing), and outcomes. Research has demonstrated that followers tend to

T. Sy / Organizational Behavior and Human Decision Processes 113 (2010) 73–84 75

fulfill the perceptions leaders have of them (Eden, 1992). In fact, the role of leaders in shaping followers is so central that scholars have recently reconceptualized leadership as the process by which leaders transform followers (Lord & Brown, 2004).

Furthermore, the utility of IFTs is that it may augment extant models of leadership. For example, whereas ILTs have explained how followers use their implicit theories of leaders to guide their actions toward leaders (e.g., determining who is a good leader and whether to follow their directives), IFTs may explain how lead- ers use their implicit theories of followers to guide their actions to- ward followers (e.g., determining whether to reward or punish) (Engle & Lord, 1997), as well as explain how followers use their im- plicit theories of followers to guide their own actions (Carsten et al., 2010). In a similar vein, IFTs may advance our understanding of Leader Member Exchange (LMX) theory by illuminating how congruence in leaders’ and followers’ ILTs and IFTs may account for relationship quality, i.e., leaders and followers both have ILTs and IFTs that function as interpretation frameworks from which relationship quality is judged (van Gils et al., 2010). Furthermore, IFTs may further our understanding of Pygmalion effects. A missing link in Pygmalion research is the lack of explanation for the ante- cedents of leaders’ performance expectation. Indeed, Eden (1990) has identified IFTs as potential antecedents of leaders’ performance expectations.

Overview of studies

The methods utilized by Offermann, Kennedy, and Wirtz (1994) and Epitropaki and Martin (2004) were consulted in the develop- ment and validation of the IFTs instrument because their proce- dures followed extensive and rigorous validation processes. Furthermore, this approach results in the development of a short measure of IFTs, which has practical value for organizational re- search by minimizing respondents’ workload, fatigue, and bias, while improving response rates (Epitropaki & Martin, 2004). This study also follows established scale development procedures (Hin- kin, 1998). Accordingly, the current research consists of five phases, each corresponding to Studies 1–5. The goal of Study 1 was to generate a comprehensive and representative pool of items for the IFTs instrument. The goals of Study 2 were to identify the underlying structure of IFTs, and examine internal consistencies of the identified factors. The goals of Study 3 were to cross validate the factor structure, and evaluate convergent and discriminant validity. The goal of Study 4 was to assess temporal consistency, with the secondary purpose of providing additional cross valida- tion evidence for the factor structure of IFTs. The goal of Study 5 was to examine the theoretical relevance and consequence of IFTs, and in doing so, provide evidence of criterion and incremental validity.

1 The list of items is available from the author upon request.

Study 1: Item generation

Study 1 utilized an inductive methodology. An inductive meth- od is appropriate when novel constructs lack substantial research to guide content development (Hinkin, 1995). Furthermore, the single adjectival approach (e.g., ‘‘productive”) was employed in- stead of the statement-based approach (e.g., ‘‘In general, followers are productive”) because accumulating evidence indicates that the former is methodologically superior (Gill & Hodgkinson, 2007).

Method

Participants Study 1 consisted of 149 workplace leaders from a variety of

industries (e.g., professional services, transportation, telecommu-

nication, retail, manufacturing, etc.). Participant demographics included: 73 (49%) males, average age of 35.85 years (SD = 11.54), 16.16 (SD = 10.33) average years of work experience, 32.9% Asian American, 23.5% Caucasian, 19.5% Hispanic, 6% African American, 7.4% ‘‘other,” and 10.7% who did not indicate their race.

Procedure Workplace leaders were recruited by a team of trained under-

graduates. Participants responded to one of four surveys that asked them to list up to 20 traits or characteristics of either a ‘‘follower,” ‘‘effective follower,” ‘‘ineffective follower,” or ‘‘subordinate.” These terms were included as stimulus cues in order to capture a wide domain of characteristics representing IFTs. Because there may be distinctions for perceived effectiveness of a target person, stim- ulus cues for effective and ineffective followers were included. In addition, the possibilities that distinctions exist as a function of the hierarchical structure of cognitive categories were considered. That is, cognitive categories may be organized hierarchically (e.g., superordinate, basic, etc.) with each level reflecting the degree of inclusiveness for the cognitive categories (for a review, see Lord et al., 1982). Consistent with past theorizing (Offermann et al., 1994), ‘‘follower” and ‘‘subordinate” reflect superordinate-level and basic-level terms, respectively. In addition, consistent with the literature on implicit theories as ‘‘lay” theories, no definitions of the stimulus cues were provided (Offermann et al., 1994).

Results

Once a set of items was generated, three faculty members from the leadership field reviewed them to establish preliminary face and content validity. The various follower cues (i.e., ‘‘follower,” ‘‘effective follower,” and ‘‘ineffective follower”) generated 1030 un- ique items. The typical items generated from the ‘‘subordinate” cue were well reflected in the pool of 1030 items, either in exact terms (e.g., shy), by synonym (e.g., timid), or by its exact opposite (e.g., outgoing). The frequency with which each item was nominated was then calculated. Items that were nominated only once or twice (e.g., skeptical) were deleted. Items that were obvious synonyms (e.g., smart and intelligent) were combined in order to reduce the number of items to a manageable size. This procedure yielded a pool of 161 items.1

Study 2: Exploratory factor analysis

The goal of Study 2 was to identify the underlying structure of IFTs utilizing exploratory factor analysis.

Method

Participants Data were collected from 428 workplace leaders representing a

wide range of industries (e.g., professional services, education, healthcare, telecommunication, retail, etc.). Participant demograph- ics included: 257 (60%) females, average age of 35.63 (SD = 12.02), average of 7 years of leadership experience (SD = 7.05), average work week of 42 h (SD = 13.11), 65% Caucasian, 13% Asian American, 9% Hispanic, 8% African American, and 5% ‘‘other.” Participants represented a wide range of leadership levels: Entry-level (39%), middle-level (23%), senior-level (16%), upper-middle-level (11%), and non-leader (i.e., had prior leadership experience but were not in leadership positions at the time of the study) (11%).

Table 1 Study 2: factor matrix, communalities, item means, and standard deviation.

Factors Variables Factor loading Communalities M SD

Industry (.86) Hardworking .87 .91 6.30 2.09 Productive .63 .72 6.14 2.08 Goes above and beyond .46 .50 5.37 2.36

Incompetence (.74) Uneducated .81 .63 4.37 2.33 Slow .57 .52 4.40 2.23 Inexperienced .56 .43 5.26 2.43

Conformity (.71) Easily influenced .82 .74 7.07 2.32 Follows trends .64 .42 7.15 2.26 Soft spoken .51 .42 6.01 2.42

Enthusiasm (.83) Excited .87 .75 5.29 2.12 Outgoing .81 .66 5.09 2.12 Happy .58 .55 5.71 2.04

Insubordination (.82) Arrogant �.83 .63 3.39 1.99 Rude �.76 .68 3.34 1.90 Bad temper �.68 .61 3.47 1.97

Good citizen (.81) Loyal .75 .64 7.07 2.14 Reliable .74 .62 6.73 2.12 Team player .73 .54 7.52 2.01

Note. Numbers in parentheses represent the scale reliabilities for each factor. N = 428. IFTs scale � 2008 Thomas Sy. Use of this scale for scholarly research purposes is granted without further permission to academic researchers at public or non-profit colleges or universities. All other rights reserved by Thomas Sy.

76 T. Sy / Organizational Behavior and Human Decision Processes 113 (2010) 73–84

Procedure Workplace leaders were recruited via a community networking

website to complete a survey on perceptions of followers. Data were collected from two diverse samples. Sample 1 consisted of 262 leaders who were compensated with a $10 gift certificate for an online retailer, and Sample 2 consisted of 166 leaders who were not compensated. The 161 items were combined in random order to form a single measure of follower perceptions. Using a 10 point scale, participants were asked to indicate how characteristic each of the items was for followers (1 = not at all characteristic; 10 = ex- tremely characteristic).2 As in Study 1, no explicit definition of the terms was provided. Responses from both samples were compared, and they were determined to be similar (t-tests indicated that re- sponses to the 161 IFTs items did not differ significantly between samples). As such, responses from both samples were combined for further analyses.

Results

The 428 participants reflected a good sample size for factor analysis (Tabachnick & Fidell, 2001). Exploratory factor analysis with oblique factor rotation was performed to determine the num- ber of factors. Multiple rounds of analyses were conducted, in which items with loadings below .30 were eliminated from subse- quent analyses. The number of factors to be extracted was guided by Cattell’s (1966) scree test and eigenvalues above 1. On this ba- sis, a six factor solution was chosen for three reasons: (1) It was the most conceptually interpretable, (2) it resulted in sound factor structures with strong internal consistency, and (3) it accounted for a substantial amount of total variance explained (60%). For par- simony, the top three items for each factor were retained, resulting

2 A potential issue with the scale anchor is that it could have been construed as prototypicality (as is intended) or the unintended meaning of atypicality (i.e., not at all relevant). However, the atypicality construal is less likely because (1) the content validity of the IFTs attributes were previously established with leadership scholars and industry executives, (2) the results provided evidence for the family resemblance structure for the attributes (Lord et al., 1982; Lord and Maher, 1993), (3) the identified attributes are corroborated by those of Carsten et al. (2010), and (4) none of the 1362 participants indicated that they were confused by the scale anchor. Furthermore, the scale anchor is identical to those utilized in validated ILTs measures (e.g., Epitropaki and Martin, 2004; Offermann et al., 1994).

in an 18-item measure of implicit followership theories. While there are no strict rules for the number of items per factor, three items are considered acceptable (Hinkin, 1998). Furthermore, three items per factor are consistent with those for ILTs, which range from two to six (Epitropaki & Martin, 2004). Table 1 presents the factor names, reliabilities, item variables, factor loadings, commu- nalities, means, and standard deviations. All items had factor load- ings above .40 and loaded on only one factor. The communalities ranged from .42 to .91. The reliabilities ranged from .71 to .86. Con- tent analysis of the items for each of the six factors yielded the dimensions of Industry, Enthusiasm, Good Citizen, Conformity, Insubordination, and Incompetence.3

Study 3: Confirmatory factor analysis

The goal of Study 3 was to cross validate the factor structure ob- tained in Study 2. A secondary goal was to examine convergent and discriminant validity. Particularly, the association of IFTs to the theoretically relevant constructs of IPTs and ILTs were of interest.

Method

Participants As in Study 2, data were collected from 397 workplace leaders

representing a wide range of industries (e.g., Professional services, leisure, retail, healthcare, telecommunication, etc.) Participant demographics included: 214 (54%) females, average age of 34.70 (SD = 11.28), average of 6 years of leadership experience (SD = 6.30), average work week of 39 h (SD = 12.65), 69% Caucasian, 14% Asian American, 8% Hispanic, 5% African American, and 4% ‘‘other.” Participants represented a wide range of leadership levels: Entry-level (33%), middle-level (27%), upper-middle-level (14%), senior-level (13%), and non-leader (i.e., had prior leadership

3 Content validation was conducted with 25 senior level executives utilizing Anderson and Gerbing’s (1991) measures of Proportion of substantive agreement, Psa

and Substantive-validity coefficient, Csv. Psa values for the IFTs items ranged from .76 to 1, with an average of .87. Csv values were all statistically significant (p < .01), with a ranged from .56 to 1, and an average of .77. Results provide support for the content validity of IFTs.

Table 3 Study 3: intercorrelations, reliabilities, means, and standard deviations of IFTs factors.

Factors 1 2 3 4 5 6

1. Industry (.88) 2. Enthusiasm .67*** (.87) 3. Good citizen .63*** .63*** (.78) 4. Conformity �.23*** �.17** .01 (.75) 5. Insubordination �.09 �.01 �.23*** .02 (.91) 6. Incompetence �.25*** �.22*** �.27*** .30*** .37*** (.85)

M 5.98 5.77 7.12 7.20 2.95 4.56 SD 2.29 2.13 1.86 1.99 1.84 2.11

Note. Numbers in parentheses represent reliabilities. N = 397. ** p < .01.

*** p < .001.

T. Sy / Organizational Behavior and Human Decision Processes 113 (2010) 73–84 77

experience but were not in leadership positions at the time of the study) (13%).

Materials IFTs were assessed with the scale derived from Study 2. IPTs

were assessed with the shortened version of Wernimont’s (1971) scale adapted by Engle and Lord (1997). Cronbach’s alpha for all 17 items was .97. ILTs were assessed with the shortened version of Offermann et al.’s (1994) scale adapted by Epitropaki and Martin (2004). Cronbach’s alpha for the six ILTs dimensions ranged from .88 to .94.

Procedure Surveys were administered to two diverse samples. Sample 1

consisted of 308 leaders recruited online (identical to Study 2 pro- cedures), and Sample 2 consisted of 89 leaders recruited from three locations of a coffee house franchise (the 89 leaders were patrons of the coffee house). Sample 1 leaders completed the surveys on- line and were compensated with a $5 gift certificate for an online retailer. Sample 2 leaders completed a paper version of the survey and were compensated with a $5 gift certificate for the coffee house franchise. Results for Sample 1 were no different from the combined samples. Thus, the results of the combined samples were reported.

Results and discussion

The combined sample of 397 participants was considered a good sample size (Tabachnick & Fidell, 2001). Maximum likelihood confirmatory factor analysis was used with AMOS 16.0 to examine the validity of the six factors of IFTs. In line with Kline’s (1998) rec- ommendation, four statistics for assessing model fit were reported: Normed chi-square measure (v2/df), comparative fit index (CFI), Tucker-Lewis index (TLI), and root–mean–square error of approxi- mation (RMSEA).

First-order six-factor model It was hypothesized a priori that (1) responses to the IFTs scale

would be explained by six factors, (2) each item would have a non- zero loading on its designated factor and zero loadings on all other factors, (3) the six factors would be correlated, and (4) measurement error terms would be uncorrelated. The hypothesized model of six factors was further tested with competing models: (1) A null model specifying that the IFTs items have no existing relationship, (2) a one factor model with all items loading onto a single factor because the IFTs scale may simply measure a general construct, and (3) a two-factor model with items loading on one prototypic and one antiprototypic factor that reflect two general dimensions of IFTs. Table 2 presents the results of these analyses. Results indicated that the six correlated factor model provided the most parsimonious fit to the data, v2(120) = 271.75, p < .001, (v2/df = 2.27, CFI = .96, TLI = .95, RMSEA = .06). Table 3 presents the intercorrelations, reli- abilities, means, and standard deviations of the IFTs factors.

Table 2 Overall fit indices for alternative factorial models of the IFTs scale.

Model v2 df v2/df

Null 4232.34 171 24.75 One factor 2295.63 135 17.01 Two correlated factors 1625.33 134 12.13 Six correlated factors 271.75 120 2.27 Second order CFA model 348.89 128 2.73

Note. Numbers in bold indicate acceptable fit indices. CFI = comparative fit index. TLI = Tuc factor analysis. *** p < .001. N = 397.

Second-order two-factor model First-order factors that are correlated suggest the possibility of a

hierarchical model (Noar, 2003). Accordingly, on the basis of the ILTs literature demonstrating that implicit theories may be repre- sented by higher-order factor structures, it was hypothesized that the six first-order factors would be explained by a second-order factor structure. A second-order two-factor model is also consis- tent with research positing that individuals have two broad per- spectives of followers (McGregor, 1960). A second-order, two correlated factor confirmatory factor analysis was tested, whereby the three dimensions of Industry, Enthusiasm, and Good Citizen measured the higher latent construct of a Followership Prototype, and the three dimensions of Conformity, Incompetence, and Insub- ordination measured the higher latent construct of Followership Antiprototype. The prototype and antiprototype terms reflect dif- ferent notions in the literature. These terms have been utilized to reflect representativeness and distinctiveness of attributes. For example, prototypic (e.g., ‘‘coordinates groups” and ‘‘exercises influence”) and antiprototypic (e.g., ‘‘requests approval” and ‘‘agrees readily”) characteristics reflect leader and non-leader attri- butes, respectively (Lord, Foti, & De Vader, 1984). These terms also have been utilized to reflect the positive and negative valence of attributes. For example, prototypic (e.g., ‘‘Intelligence” and ‘‘Dedi- cation”) and antiprototypic (e.g., ‘‘Tyranny” and ‘‘Masculinity”) characteristics reflect positively and negatively valenced leader attributes, respectively (Epitropaki & Martin, 2004; Offermann et al., 1994). In the current research, Followership Prototype and Antiprototype reflect positively and negatively valenced dimen- sions of IFTs, respectively.

Results in Table 2 indicated that although the IFTs items were not best represented by two first-order factors (i.e., first-order pro- totypic and antiprototypic followership), the six first-order factors were well represented by two second-order factors. Results of the second-order confirmatory factor analysis model indicated a rea- sonable fit to the data, v2(128) = 348.89, p < .001, (v2/df = 2.73, CFI = .95, TLI = .93, RMSEA = .07). The significantly different chi- square values between the first-order six-factor model and the sec- ond-order two-factor model, Dv2(8) = 77.14, p < .001, suggested

Dv2 Ddf CFI TLI RMSEA

.25 1936.71*** 36 .47 .33 .20

670.30*** 1 .63 .53 .17 1353.58*** 14 .96 .95 .06

77.14*** 8 .95 .93 .07

ker-Lewis index. RMSEA = root–mean–square of approximation. CFA = confirmatory

Loyal

Reliable

Team player

Excited

Outgoing

Happy

Hardworking

Productive

Goes above and beyond

Easily influenced

Follows trends

Soft spoken

Uneducated

Slow

Inexperienced

Good Citizen

Enthusiasm

Industry

Conformity

Incompetence

Followership Antiprototype

Followership Prototype

.70

.79

.73

.85

.80

.83

.91

.85

.78

.86

.89

.42

.86

.72

.84

.98

.39

.35

.87

.88

.87

-.31

Arrogant

Rude

Bad tempered

Insubordination .81 .92 .89

Fig. 1. Second-order factor model of implicit followership theories. All factor loadings and the correlation between follower prototype and antiprototype are significant at p < .001. v2 (128, N = 397) = 348.89, v2/df = 2.73, comparative fit index = .95, root–mean–square error of approximation = .07.

78 T. Sy / Organizational Behavior and Human Decision Processes 113 (2010) 73–84

the first-order model best represented the data. However, all fit indices indicated that the second-order model was plausible, thus providing support for the existence of two higher factors of Follow- ership Prototype and Antiprototype. Thus, results supported both the two second-order factor model and the six first-order factor model. A graphic representation of this model is provided in Fig. 1.

Convergent and discriminant validity Two aspects of convergent and discriminant validity were

examined: (1) Convergent and discriminant validity among the six IFTs factors, and (2) convergent and discriminant validity of the six IFTs factors with theoretically relevant constructs (i.e., ILTs and IPTs). Regarding the former, convergent validity was examined by assessing whether each item had a statistically significant factor loading on its specified factor (Anderson & Gerbing, 1988). Conver- gent validity was supported as all factor loadings were significant with critical ratios ranging from 8.00 to 24.62 (p < .001), and stan- dardized loadings ranging from .42 to .92. Likewise, results of the IFTs latent factor correlations in Table 3 provided initial support for discriminant validity as none of the correlations exceeded the recommended cut-off point of .85 (Kline, 1998). Furthermore, dis- criminant validity was examined with a test suggested by Ander- son and Gerbing (1988). The test assessed whether the confidence intervals (two standard errors) around the correlation estimates between two factors included 1.0. None of the confi- dence intervals for the correlation estimates between two factors included 1.0, providing support for the discriminant validity of the six subscales.4

Convergent and discriminant validity were examined next by assessing the association of IFTs with theoretically relevant con-

4 A detailed presentation of all discriminant validity analyses is available upon request.

structs (i.e., ILTs and IPTs). It was expected that (1) IFTs would be associated with ILTs and IPTs (convergent validity) and (2) their association would not be high given that they are not measures of the same construct (discriminant validity). As reported in Ta- ble 4, the moderate associations between IFTs with ILTs and IPTs provided evidence for both convergent and discriminant validity (Campbell & Fiske, 1959; Hinkin, 1998). Although the results for Followership Antiprototype were less than perfect, further validity evidence was provided in Study 5. As a whole, the pattern of results provided preliminary evidence of convergent and discriminant validity.

Study 4: Temporal consistency and replication of factor structure

The primary goal of Study 4 was to examine temporal consis- tency. In addition, providing further cross validation evidence for the factor structure of IFTs beyond the initial confirmatory factor analysis is highly desirable (Hinkin, 1998). Data were collected from a sample of students. Students are likely to have developed IFTs because implicit theories are developed at an early age (Hunt et al., 1990), and students are likely to have had various experi- ences in leader–follower contexts. Thus, it would be informative to cross validate the factor structure of IFTs with a different popu- lation of participants.

Method

Participants and procedures Test-retest reliability was assessed with a sample of 228 under-

graduates. Participant demographics included: 135 (59%) females, average age of 21.81 years (SD = 3.53), 33% Asian American, 25% Hispanic, 19% Caucasian, 7% African American, 9% ‘‘other,” and 7%

Table 4 Study 3: correlations between IFTs, IPTs, and ILTs.

IFTs Subscales IPT Sens Intel Dedi Dyna Tyra Masc L-Proto L-Anti

1. Industry .14* .20*** .14* .11 .12* .16** .16** .17** .18**

2. Enthusiasm .14* .23*** .19** .12* .21*** .25*** .27*** .23*** .29***

3. Good Citizen .28*** .33*** .30*** .26*** .30*** .15** .17** .35*** .18**

4. Conformity .24*** .13* .21*** .17** .21*** .19** .09 .22*** .15*

5. Insubordination �.23*** �.22*** �.18** �.23*** �.26*** .27*** .23*** �.24*** .28***

6. Incompetence �.05 �.05 .01 �.04 �.03 .24*** .20** �.01 .24***

7. F-Proto .20** .28*** .23*** .18** .23*** .22*** .23*** .28*** .25***

8. F-Anti-Proto �.01 �.06 .02 �.04 �.03 .34*** .25*** �.01 .32***

Note. F-Proto = Followership Prototype. F-Anti-Proto = Followership Antiprototype. IPT = Implicit performance theories. Sens = Sensitivity. Intel = Intelligence. Dedi = Dedi- cation. Dyna = Dynamic. Tyra = Tyranny. Masc = Masculinity. L-Proto = Leadership Prototype. L-Anti = Leadership Antiprototype. N = 397.

* p < .05. ** p < .01.

*** p < .001.

T. Sy / Organizational Behavior and Human Decision Processes 113 (2010) 73–84 79

who did not indicate their race. The time lag between administra- tions (i.e., Time 1 and Time 2) of the IFTs scale was 4 weeks.

Results

The 4-week test-retest reliability estimates for the IFTs dimen- sions of Conformity, Insubordination, Incompetence, Industry, Enthusiasm, and Good Citizen were .83, .85, .82, .87, .85, and .89, respectively. As such, test-retest reliabilities for the IFTs dimen- sions were acceptably stable over time. Furthermore, internal con- sistency estimates for all IFTs dimensions were acceptable at Time 1 (.79–.89) and Time 2 (.82–.90).

Confirmatory factor analyses provided further support for the six first-order factor structure at Time 1, v2(120) = 261.81, p < .001, (v2/df = 2.23, CFI = .94, TLI = .91, RMSEA = .07) and Time 2, v2(120) = 262.29, p < .001, (v2/df = 2.19, CFI = .94, TLI = .92, RMSEA = .07). Furthermore, confirmatory factor analyses provided support for the two second-order factor structure at Time 1, v2(128) = 322.47, p < .001, (v2/df = 2.52, CFI = .92, TLI = .89, RMSEA = .08) and Time 2, v2(128) = 317.39, p < .001, (v2/df = 2.48, CFI = .92, TLI = .90, RMSEA = .08).

Study 5: Consequences of IFTs

The goal of Study 5 was to establish a preliminary nomological network of IFTs by examining the theoretical relevance and conse- quences of IFTs for leader–follower outcomes. As a preliminary investigation, Study 5 focused specifically on leaders’ IFTs. More- over, while results provided support for both first- and second-or- der factor models, Study 5 focused on the second-order factors of Followership Prototype and Antiprototype because (1) the notion of bandwidth-fidelity (Cronbach & Gleser, 1965) suggests the nat- ure of the criterion should dictate the choice of predictors in order to enhance validity (Hogan & Roberts, 1996; Ones & Viswesvaran, 1996), and the second-order model would be more appropriate in the current study context as the criterion variables (e.g., satisfac- tion, relationship quality, etc.) and research context (i.e., a wide range of companies, industries, locations, etc.) were broad in scope, and (2) theories suggest consequences of leaders’ IFTs could be ex- plained best by two broad perspectives of followers (e.g., Eden, 1990; McGregor, 1960).

A second goal of Study 5 was to provide additional validity esti- mates. Criterion validity is an important facet of construct validity and represents the extent to which a construct is related to other variables on the basis of theory (Cronbach & Meehl, 1955; Hinkin, 1998). In addition to criterion validity, it is important to demon- strate the incremental validity of IFTs over related implicit theory constructs, i.e., ILTs and IPTs. The variables of (1) liking, (2) rela- tionship quality, (3) trust, and (4) job satisfaction were of interest

because research has established their relevance and linkage to im- plicit theories (Engle & Lord, 1997; Giessner & van Knippenberg, 2008). Furthermore, examining the relationships between these variables and IFTs extend our understanding of the nomological network of IFTs, in comparison with other implicit theories (i.e., ILTs and IPTs). Criterion validity was examined by relating leaders’ IFTs to followers’ outcomes (Sample 1): (1) Liking for leaders, (2) relationship quality with leaders, (3) trust in leaders, and (4) job satisfaction. Also, incremental validity was examined by relating leaders’ IFTs to followers’ outcomes while controlling for IPTs. Like- wise, criterion validity was examined by relating leaders’ IFTs to leaders’ outcomes (Sample 2): (1) Liking for followers, and (2) rela- tionship quality with followers. Furthermore, incremental validity was examined by relating leaders’ IFTs to leaders’ outcomes while controlling for IPTs and ILTs. As discussed, leaders’ IFTs may shape how leaders judge and respond to followers, which influence lea- der–follower interactions that impacts leader–follower outcomes.

Liking and relationship quality

Liking and relationship quality are strongly linked such that lik- ing can be regarded as one facet of relationship quality (Wayne, Shore, & Liden, 1997). Leaders’ IFTs may impact liking and relation- ship quality because leaders’ IFTs may serve as sensemaking func- tions (Weick, 1995) that act as antecedents of leaders’ affect, behaviors, and cognitions toward followers. Thus, leaders’ IFTs may shape the pattern of interactions between leaders and follow- ers because many aspects of dyadic relationships are based on the automatic use of implicit theories in perceiving and interpreting the behaviors of one’s dyadic partner (Engle & Lord, 1997). Because evaluations often correspond with perceivers’ implicit theories (Johnson et al., 2008; Lord & Maher, 1993), leaders’ prototypic and antiprototypic perceptions of followers may influence leaders to evaluate and treat followers positively and negatively, respec- tively (Goodwin et al., 2000; Wofford & Goodwin, 1994), which in turn may have a corresponding positive and negative impact on liking and relationship quality. For example, leaders who per- ceive that followers are loyal may evaluate followers’ behaviors positively (e.g., followers are working late because they are team players who want to see our team succeed), which may positively influence leader–follower interactions (e.g., empowering followers and affording them wide discretion in managing their work and time) that impact interpersonal outcomes (i.e., liking and relation- ship quality). Conversely, leaders who perceive that followers are incompetent may evaluate followers’ behaviors negatively (e.g., followers are working late because they are slow), which may neg- atively influence leader–follower interactions (e.g., micro-manag- ing followers) that impact interpersonal outcomes. Accordingly, it was hypothesized:

80 T. Sy / Organizational Behavior and Human Decision Processes 113 (2010) 73–84

Hypothesis 1a: Leaders’ Followership Prototype will be positively related to followers’ liking for and relationship quality with leaders. Hypothesis 1b: Leaders’ Followership Antiprototype will be nega- tively related to followers’ liking for and relationship quality with leaders. Hypothesis 2a: Leaders’ Followership Prototype will be positively related to leaders’ liking for and relationship quality with followers. Hypothesis 2b: Leaders’ Followership Antiprototype will be nega- tively related to leaders’ liking for and relationship quality with followers.

Trust

Leadership and trust are inherently intertwined (Dirks, 2000). Leaders’ perceptions of followers may affect the level of trust be- tween leaders and followers. As posited above, leaders’ prototypic and antiprototypic perceptions of followers may influence leaders to evaluate followers positively and negatively, respectively. In turn, leaders’ positive and negative evaluations of followers may affect leaders’ trust in followers positively and negatively, respec- tively. Consequently, leaders will exhibit more or less trusting behaviors, and in accordance with Social Exchange Theory (Blau, 1964), followers will reciprocate in kind to leaders’ display of trust. Thus, it was hypothesized:

Hypothesis 3a: Leaders’ Followership Prototype will be positively related to followers’ trust in leaders. Hypothesis 3b: Leaders’ Followership Antiprototype will be nega- tively related to followers’ trust in leaders.

Job satisfaction

Leaders account for a significant portion of followers’ job satis- faction (Locke, 1976). In fact, leaders can buffer or be a source of stress that influences followers’ job satisfaction, depending on leaders’ treatment of followers and the nature of leader–follower interactions (Bono, Foldes, Vinson, & Muros, 2007) that may be a function of leaders’ IFTs (Goodwin et al., 2000; Wofford & Good- win, 1994). In line with previous rationale, leaders’ prototypic and antiprototypic perceptions of followers may influence leaders’ positive and negative treatment of followers, respectively. Leaders’ positive treatment of followers satisfies their interpersonal needs, and meeting these needs increases followers’ job satisfaction (Mayer, Bardes, & Piccolo, 2008). Conversely, followers who receive negative treatment from leaders may not have their interpersonal needs satisfied, which may negatively impact followers’ job satis- faction. Accordingly, it was hypothesized:

Hypothesis 4a: Leaders’ Followership Prototype will be positively related to followers’ job satisfaction. Hypothesis 4b: Leaders’ Followership Antiprototype will be nega- tively related to followers’ job satisfaction.

Method

Sample 1 participants Data were collected from 80 dyadic pairs (160 participants) of

workplace leaders and their followers from various industries lo- cated in a large metropolitan area in the Western United States (e.g., Education, professional services, healthcare, food service, re- tail, etc.). Ethnicity of participants was diverse: Caucasian (32%), Asian American (27%), Hispanic (21%), African American (14%), and other (6%). Leaders and followers worked an average of 46 (SD = 12.83) and 32 (SD = 15.58) hours per week, respectively. Regarding followers, 42 (53%) were female and their age ranged from 18 to 70, averaging 29.34 years (SD = 11.03). Regarding lead-

ers, 44 (56%) were male and their age ranged from 20 to 66, aver- aging 38.81 years (SD = 11.10). Leaders had an average of 11 years of leadership experience (SD = 7.89) and represented a wide range of leadership levels: Senior-level (38%), upper-middle-level (21%), middle-level (18%), and entry-level (23%).

Sample 1 procedures A team of trained undergraduates received approximately 1 h of

training on study protocols, ethical guidelines, and were provided with detailed written protocol instructions, including standardized recruitment scripts, to recruited working adults from their existing network of contacts. Leaders completed an online survey consist- ing of the IFTs and IPTs scales. Followers completed an online sur- vey reporting their liking for leader, relationship quality with leader, trust in leader, and job satisfaction.

Sample 2 participants and procedures Participants and procedures were the same as those in Study 3

for the sample of 308 leaders who completed an online survey con- sisting of measures of IFTs, IPTs, liking for followers, and quality of relationship with followers.

Materials for Samples 1 and 2 IFTs were assessed with the same scale described in Study 3. As

control variables, IPTs (Engle & Lord, 1997) and ILTs (Epitropaki & Martin, 2004) were assessed with the same scales described in Study 3.

Liking for leader was measured with a four item scale (e.g., ‘‘I like my work group supervisor very much”) (Wayne et al., 1997). Followers rated these items using a 7-point Likert-type scale (1 = strong disagree; 7 = strongly agree).

Relationship quality was measured with the Dual-Perspective LMX7 scale that was developed specifically to assess both leader and follower perspectives (Paglis & Green, 2002). Each subscale consisted of 7 items and sample items included, ‘‘My supervisor understands my problems and needs” (follower perspective), and ‘‘I think I understand my subordinate’s problems and needs” (lea- der perspective). Leaders and followers rated these items using a 5-point Likert-type scale (1 = strong disagree; 5 = strongly agree).

Trust was measured with a seven item scale (e.g., ‘‘I believe my supervisor has high integrity”) (Gabarro & Athos, 1976). Followers rated these items using a 5-point Likert-type scale (1 = strong dis- agree; 5 = strongly agree).

Job satisfaction was measured with a four item scale (e.g., ‘‘All in all, I am very satisfied with my current job”) (Quinn & Shepard, 1974). Followers rated these items using a 7-point Likert-type scale (1 = strong disagree; 7 = strongly agree).

Results

Table 5 presents the means, standard deviations, reliabilities, and correlations. Leaders’ Followership Prototype was positively related to all follower outcomes (i.e., liking for leaders, relationship quality with leaders, trust in leaders, and job satisfaction). Like- wise, leaders’ Followership Antiprototype was negatively related to all follower outcomes. Similarly, leaders’ Followership Prototype was positively related to all leader outcomes (i.e., liking for follow- ers, and relationship quality with followers). Likewise, leaders’ Followership Antiprototype was negatively related to leaders’ re- port of relationship quality with followers, but not with leaders’ liking for followers. The results suggested that while Followership Prototype consistently predicted both leader and follower out- comes, Followership Antiprototype was less consistent (i.e., did not predict leader liking for follower). Overall, the results provided support for the criterion validity of IFTs and all hypotheses, except hypothesis 2b.

Table 7 Study 5: partial correlations for IPTs and ILTs controlling for Leaders’ IFTs.

L.Liking L.LMX

IPTs .13* .21**

ILTs – Leadership Prototype .10 .08 ILTs – Leadership Antiprototype .05 .05

Table 5 Study 5: correlations, means, standard deviations, and reliabilities for leaders’ IFTs, IPTs, ILTs and criterion variables.

Sample 1 Sample 2

Variables 1 2 3 4 5 6 7 5 6 7 8 9 10 11

1. F.Liking (.94) 2. F.LMX .66*** (.87) 3. F.Trust .50*** .50*** (.82) 4. F.Satisf .53*** .57*** .61*** (.87) 5. L.Proto .42*** .40*** .23* .29* (.90) (.91) 6. L.Antiproto �.36** �.49*** �.32** �.33** �.34** (.80) �.27*** (.78) 7. L.IPTs .20 .05 .26* .14 .24* �.04 (.96) .20** �.01 (.97) 8. L.Liking .39*** �.10 .26*** (.87) 9. L.LMX .41*** �.18** .37*** .64*** (.88) 10. L.ILTs-P .28*** �.01 .68*** .28*** .35*** (.94) 11. L.ILTs-A .25*** .32*** �.03 .12* .08 .04 (.94)

M 5.66 4.08 4.15 5.39 6.41 4.40 8.15 6.31 4.91 8.69 5.14 3.70 8.43 4.56 SD 1.33 .70 .69 1.39 1.67 1.27 1.37 1.81 1.38 1.39 1.24 .75 1.33 2.36

Note. ‘‘L” indicates leaders’ ratings. ‘‘F” indicates followers’ ratings. Liking = Liking for dyadic partner. LMX = Relationship quality. Satisf = Job satisfaction. Proto = Followership Prototype. Antiproto = Followership Anti-prototype. IPTs = Implicit Performance Theories. ILTs-P = Implicit Leadership Theories – Leadership Prototype. ILTs-A = Implicit Leadership Theories – Leadership Anti-prototype. Sample 1 N = 160. Sample 2 N = 308.

* p < .05. ** p < .01.

*** p < .001.

Table 6 Study 5: partial correlations for leaders’ IFTs and criterion variables.

Criterion Controlling for IPTs Controlling for IPTs and ILTs Sample 1 Sample 2

L.Proto L.Antiproto L.Proto L.Antiproto

1. F.Liking .37*** �.36**

2. F.LMX .37*** �.51***

3. F.Trust .17 �.32***

4. F.Satisf .29** �.35**

5. L.Liking .29*** �.14*

6. L.LMX .31*** �.22***

Note. ‘‘L” indicates leaders’ ratings. ‘‘F” indicates followers’ ratings. Liking = liking for dyadic partner. LMX = relationship quality. Satisf = job satisfaction. Proto = Fol- lowership Prototype. Antiproto = Followership Anti-prototype. IPTs = implicit per- formance theories. ILTs = implicit leadership theories. Sample 1 N = 160. Sample 2 N = 308.

* p < .05. ** p < .01.

*** p < .001.

T. Sy / Organizational Behavior and Human Decision Processes 113 (2010) 73–84 81

To examine the incremental validity of IFTs, partial correlations were conducted to examine the association between leaders’ Fol- lowership Prototype and Antiprototype with follower and leader outcomes, while controlling for related implicit theory constructs (i.e., IPTs, and ILTs). Table 6 presents the partial correlation results. Consistent with the pattern of results for criterion validity, Follow- ership Prototype and Antiprototype accounted for additional vari- ance, above and beyond IPTs and ILTs, for all leader and follower outcomes, with the exception of followers’ trust in leaders. Overall, after controlling for IPTs and ILTs, the results provided support for the incremental validity of IFTs and all hypotheses, with the excep- tion of hypothesis 3a. An analysis also was conducted on Sample 2 examining the relationships between IPTs and ILTs with the out- come variables while controlling for leaders’ IFTs. Results in Table 7 indicated that just as leaders’ IFTs accounted for additional

variance above and beyond IPTs, IPTs also accounted for additional variance beyond that of leaders’ IFTs. Thus, while the two con- structs may be related, they are distinct in accounting for unique variances in the outcome variables (i.e., Liking and LMX). In addi- tion, the results provided evidence indicating that leaders’ IFTs are broader constructs that encompasses IPTs, given that leaders’ IFTs accounted for more variance than IPTs.

General discussion

Results suggest that IFTs are most accurately represented by a first-order six-factor structure that includes Industry, Enthusiasm, Good Citizen, Conformity, Insubordination, and Incompetence. IFTs also are accurately represented by a second-order two-factor structure: Followership Prototype (Industry, Enthusiasm, and Good Citizen) and Followership Antiprototype (Conformity, Insubordina- tion, and Incompetence). In addition to identifying the content and structure of IFTs, an assessment instrument was developed and validated to facilitate research. Results support the psychometric properties of the IFTs instrument, providing preliminary evidence for content, convergent, discriminant, criterion, and incremental validity, as well as internal and temporal consistency. In turn, these provide evidence for construct validity (Hinkin, 1995). Further- more, the theoretical relevance and consequences of IFTs were demonstrated. Specifically, leaders’ IFTs have consequences for the interpersonal aspects of leader and follower outcomes. As dis- cussed below, IFTs may also have consequences beyond interper- sonal outcomes, i.e., performance outcomes.

Future research

IFTs may further augment our understanding of well-substanti- ated theories in the organizational science literature. Future re- search could extend the examination of IFTs beyond leaders to followers. While Study 5 examined leaders’ IFTs, followers also have IFTs (Carsten et al., 2010; Engle & Lord, 1997; van Gils et al., 2010). Indeed, Study 4 demonstrated that the structure and content of IFTs for undergraduate students (befitting the label of followers) were similar to leaders. It may be fruitful to examine the congru- ence between leaders’ and followers’ IFTs and ILTs on leader–fol- lower outcomes (Engle & Lord, 1997; van Gils et al., 2010), thus

82 T. Sy / Organizational Behavior and Human Decision Processes 113 (2010) 73–84

answering calls for leadership research to be more integrative (Avolio et al., 2009).

In addition to interpersonal outcomes, IFTs may be particularly applicable to theories positing the role of leaders in shaping fol- lower performance outcomes. Specifically, Eden’s Pygmalion the- ory (e.g., 1990, 1992) has established that leaders’ performance expectations may significantly impact follower performance out- comes. However, Pygmalion theory is incomplete (McNatt, 2000) because it does not explain the antecedents of leaders’ perfor- mance expectations. In classic Pygmalion studies, expectations are artificially manipulated by the researchers (i.e., researchers cre- ate the expectations for participants). One potential mechanism may be leaders’ perceptions of followers (Eden, 1990). That is, lead- ers’ IFTs may be antecedents of leaders’ performance expectations for followers that lead to Pygmalion effects whereby followers ful- fill leaders’ positive perceptions of followers (i.e., Followership Pro- totype), or Golem effects whereby followers fulfill leaders’ negative perceptions of followers (i.e., Followership Antiprototype) (Eden, 1990, 1992). In sum, leaders’ IFTs may determine leaders’ perfor- mance expectations for followers, which then shape followers’ per- formance outcomes. Furthermore, consistent with inference-based processing (Lord & Maher, 1993), this process may be cyclical such that the corresponding follower performance outcomes may serve as feedback that reinforces leaders’ IFTs (Goodwin et al., 2000; Wofford & Goodwin, 1994).

In addition to Pygmalion effects, future research should exam- ine the relationship between leaders’ IFTs and followers’ self-con- cept, an emerging area of research (Avolio et al., 2009; Lord & Brown, 2004; van Knippenberg & Hogg, 2003). Scholars have pos- ited that leadership is the process by which leaders transform the way followers envision themselves (Lord & Brown, 2004). Con- sistent with research positing that leaders’ treatment of followers is determined by their assumptions of followers (McGregor, 1960), it could be that Theory X leaders who endorse more antipro- totypic perceptions of followers may treat followers differently from Theory Y leaders who endorse more prototypic perceptions of followers. In turn, their leadership style and treatment of follow- ers impact followers’ conceptualization of themselves (Lord & Brown, 2004).

IFTs also may be antecedents for individuals’ affective experi- ences. For example, leaders’ IFTs may account for leaders’ affect and the transference of affect from leaders to followers (Bono et al., 2007; Sy, Cote, & Saavedra, 2005). Leaders’ IFTs may predis- pose leaders to certain affective experiences. That is, leaders’ posi- tive or negative perception of followers (i.e., leaders’ IFTs) may elicit corresponding affective experiences in leaders (e.g., percep- tions of insubordination in followers may elicit anger). Further- more, leaders are key sources of uplifts and hassles for followers (Dasborough, 2006), which may be a result of leaders’ behaviors and interactions with followers that are shaped by leaders’ IFTs. Leaders who endorse more prototypic perceptions of followers may be more likely to engender positive group affective tone, and leaders who endorse more antiprototypic perceptions of fol- lowers may be more likely to engender negative group affective tone (Sy et al., 2005). Thus, future research could examine the interaction of IFTs with leadership and emotions.

Study limitations

This study has several limitations. First, the majority of the par- ticipants in the different samples were primarily Caucasian and fe- male. Thus, it is possible that the results may be skewed to the perspective of these participants. However, the major minority groups were represented in the study samples and the racial com- position reflects those of the US population. Likewise, the propor- tion of female to male participants in the samples is consistent

with studies in the ILTs field (e.g., Epitropaki & Martin, 2004; John- son et al., 2008; Offermann et al., 1994) and the broader social sci- ence literature.

Second, this study included individuals from a variety of differ- ent companies, industries, and job functions. As such, it was not possible to control for extraneous factors that may confound the study results. However, the diversity of respondents, particularly the multiple samples utilizing multiple methodologies, may also increase the generalizability of the results. Nevertheless, the gener- alizability of the results may be limited to Western perspectives of IFTs as implicit theories may vary as a function of context (Foti et al., 2008; Hanges et al., 2000).

Third, the type of follower was not specified. Theorists argue that implicit theories may be examined at three hierarchical levels (for a review, see Lord et al., 1982). This study primarily focused on IFTs at the superordinate level (i.e., followers in general). This ap- proach is consistent with researchers’ recommendations (Lord & Maher, 1993) to examine perceptions beyond the basic level (i.e., specific types of followers, e.g., military, business, minority, reli- gious, etc.). However, given that the majority of the participants were from the business sector, it is also likely that they were refer- ring to business followers in their responses. On the other hand, unlike individuals’ romance with leaders, which may lead individ- uals to pay more attention to and make distinctions between dif- ferent leaders, it may be that followers are less likely to be differentiated because individuals do not afford the same level of attention and fascination with followers (Pillai & Uhl-Bien, 2007). Thus, followers are often viewed and treated as an undifferentiated mass or collective (Collins, 2006). As such, the results of this study may be generalized to a broader category of followers. That is, the conceptualization of IFTs may be most relevant at the superordi- nate level.

Practical implications

IFTs have implications for evaluations of followers. Personnel decisions are often based on individuals’ perceptions of followers, which are often systematically biased (Heslin & VandeWalle, 2008). Individuals who endorse more antiprototypic followership theories may be prone to providing punitive evaluations of follow- ers (e.g., severity errors). Conversely, individuals who endorse more prototypic followership theories may be prone to providing positive evaluations of followers (e.g., leniency errors).

Furthermore, because evaluations often correspond with per- ceivers’ implicit theories (Johnson et al., 2008; Lord & Maher, 1993), leaders may more easily recognize potential in followers that fit their implicit theories of Followership Prototype and may not recognize potential in equally capable followers who exhibit less congruence. This is particularly relevant in multicultural envi- ronments. For example, among other traits, Western leaders may recognize the potential of followers who show Enthusiasm, and la- bel and treat these individuals as ‘‘high potentials.” However, Wes- tern leaders may overlook the same potential in equally capable followers who may not exhibit Enthusiasm because their cultural values may inhibit expressions of emotions (e.g., Eastern cultures such as Japan and China). This bias may also occur for gender (Johnson et al., 2008). As such, an important practical implication is that leaders should develop awareness of their IFTs profile, and how these perceptions may bias their cognitions and behaviors to- ward followers.

In sum, IFTs are an understudied branch of leadership research that complements ILTs and IPTs. IFTs address the historical gap in leadership research that has focused primary on leaders, and an- swer recent calls for leadership research to be more integrative by incorporating followers as a focal element, as well as calls for much needed research on leaders’ cognitions (Avolio et al., 2009).

T. Sy / Organizational Behavior and Human Decision Processes 113 (2010) 73–84 83

Thus, an integrative examination of implicit theories of leaders and followers may provide more holistic understanding of leadership, as well as inform leadership practice because IFTs are application oriented. As demonstrated in Study 5 and consistent with the Pyg- malion literature, leaders’ IFTs may have consequences for leaders’ and followers’ cognitions, affect, behaviors, and outcomes. Consis- tent with this line of reasoning, it is possible that the performance of leaders, followers, and organizations can be elevated through IFTs.

Acknowledgments

I thank Daan van Knippenberg and his team for valuable edito- rial guidance. I extend my gratitude to Paul Adelman, Uma Kedhar- nath, Tina Kim, Sharon Rogala, Susanna Tram, and Paul Whiteley for assistance with data collection and support. I also thank Jin Nam Choi, Dov Eden, Howard Friedman, Stefanie Johnson, Bob Lord, Bob Rosenthal, and Susan Sy for helpful comments.

References

Anderson, J. C., & Gerbing, D. W. (1988). Structural equation modeling in practice: A review and recommended two-step approach. Psychological Bulletin, 103, 411–423.

Anderson, J. C., & Gerbing, D. W. (1991). Predicting the performance of measures in a confirmatory factor analysis with a pretest assessment of their substantive validities. Journal of Applied Psychology, 76, 732–740.

Avolio, B. J., Walumbwa, F. O., & Weber, T. J. (2009). Leadership: Current theories, research and future directions. Annual Review of Psychology, 60, 421–449.

Barsalou, L. W. (1985). Ideals, central tendency, and frequency of instantiation as determinants of graded structure in categories. Journal of Experimental Psychology: Learning, Memory, and Cognition, 11, 629–654.

Blau, P. (1964). Exchange and power in social life. New York: Wiley. Boldenhausen, G., & McCrae, C. N. (1998). Stereotype activation and inhibition. In R.

S. Wyer (Ed.), Stereotype activation and inhibition. Advances in social cognition (Vol. XI, pp. 1–52). Mahwah, NJ: Earlbaum.

Bono, J. E., Foldes, H. E., Vinson, G., & Muros, J. P. (2007). Workplace emotions: The role of supervision and leadership. Journal of Applied Psychology, 92, 1357–1367.

Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminant validation by the multitrait–multimethod matrix. Psychological Bulletin, 56, 81–105.

Carsten, M. K., Uhl-Bien, M., West, B. J., Patera, J. L., & McGregor, R. (2010). Exploring social constructions of followership: A qualitative study. Leadership Quarterly, 21, 543–562.

Cattell, J. L. (1966). The scree test for the number of factors. Multivariate Behavioral Research, 1, 245–276.

Collins, D. (2006). Rethinking followership: A post-structuralist analysis of follower identities. Leadership Quarterly, 17, 179–189.

Cronbach, L. J., & Gleser, G. C. (1965). Psychological tests and personnel decisions. Urbana, IL: University of Illinois Press.

Cronbach, L. J., & Meehl, P. C. (1955). Construct validity in psychological tests. Psychological Bulletin, 52, 281–302.

Dasborough, M. T. (2006). Cognitive asymmetry in employee emotional reactions to leadership behaviors. Leadership Quarterly, 17, 163–178.

Dirks, K. T. (2000). Trust in leadership and team performance: Evidence from NCAA basketball. Journal of Applied Psychology, 85, 1004–1012.

Eden, D. (1990). Pygmalion in management. Lexington, MA: Lexington Books. Eden, D. (1992). Leadership and expectations: Pygmalion effects and other self-

fulfilling prophecies in organizations. Leadership Quarterly, 3, 271–305. Engle, E. M., & Lord, R. G. (1997). Implicit theories, self schemas, and leader–

member exchange. Academy of Management Journal, 40, 988–1010. Ensari, N., & Murphy, S. E. (2003). Cross cultural variations in leadership perceptions

and attributions of charisma to the leader. Organizational Behavior and Human Decision Processes, 92, 52–66.

Epitropaki, O., & Martin, R. (2004). Implicit leadership theories in applied settings: Factor structure, generalizability, and stability over time. Journal of Applied Psychology, 89, 239–310.

Fiske, S. T., & Taylor, S. E. (1991). Social cognition (2nd ed.). New York: McGraw-Hill. Foti, R. J., Knee, R. E., & Backert, R. G. (2008). Multi-level implications of framing

leadership perceptions as a dynamic process. Leadership Quarterly, 19, 178–194. Gabarro, J. J., & Athos, J. (1976). Interpersonal relations and communications.

Englewood Cliffs, NJ: Prentice-Hall. Gerstner, C. R., & Day, D. V. (1994). Cross-cultural comparison of leadership

prototypes. Leadership Quarterly, 5, 121–134. Giessner, S. R., & van Knippenberg, D. (2008). ‘‘License to fail”: Goal definition,

leader group prototypicality, and perceptions of leadership effectiveness after leader failure. Organizational Behavior and Human Decision Processes, 105, 14–35.

Gill, C. M., & Hodgkinson, G. P. (2007). Development and validation of the five-factor model questionnaire (FFMQ): An adjectival-based personality inventory for use in occupational settings. Personnel Psychology, 60, 731–766.

Goodwin, V. L., Wofford, J. C., & Boyd, N. G. (2000). A laboratory experiment testing the antecedents of leader cognitions. Journal of Organizational Behavior, 21, 769–788.

Graen, G. B., & Uhl-Bien, M. (1995). Relationship-based approach to leadership: Development of leader–member exchange (LMX) theory of leadership over 25 years: Applying a multi-level multi-domain perspective. Leadership Quarterly, 6, 219–247.

Hanges, P. J., Lord, R. G., & Dickson, M. W. (2000). An information processing perspective on leadership and culture: A case for connectionist architecture. Applied Psychology: An International Review, 49, 133–161.

Heslin, P. A., & VandeWalle, D. V. (2008). Managers’ implicit assumptions about personnel. Current Directions in Psychological Science, 17, 219–223.

Hinkin, T. R. (1995). A review of scale development practices in the study of organizations. Journal of Management, 21, 967–988.

Hinkin, T. R. (1998). A brief tutorial on the development of measures for use in survey questionnaires. Organizational Research Methods, 1, 104–121.

Hogan, J., & Roberts, B. W. (1996). Issues and non-issues in the fidelity-bandwidth trade-off. Journal of Organizational Behavior, 17, 627–637.

House, R. J., Hanges, P. J., Javidan, M., Dorfman, P., Gupta, V., & Associates, G. L. O. B. E. (2004). Cultures, leadership, and organizations: The GLOBE study of 62 societies. Thousand Oaks, CA: Sage.

Hunt, J. G., Boal, K. B., & Sorenson, R. L. (1990). Top management leadership: Inside the black box. Leadership Quarterly, 1, 41–65.

Johnson, S. K., Murphy, S. E., Zewdie, S., & Reichard, R. J. (2008). The strong, sensitive type: Effects of gender stereotypes and leadership prototypes on the evaluation of male and female leaders. Organizational Behavior and Human Decision Processes, 106, 39–60.

Kline, P. (1998). The new psychometrics: Science, psychology, and measurement. Florence, KY: Taylor & Francis.

Locke, E. A. (1976). The nature and causes of job satisfaction. In M. D. Dunnette (Ed.), Handbook of industrial and organizational psychology (pp. 1297–1349). Chicago, IL: Rand McNally.

Lord, R. G., & Brown, D. J. (2004). Leadership processes and follower identity. New Jersey: Lawrence Earlbaum Associates.

Lord, R. G., Foti, R. J., & De Vader, C. L. (1984). A test of leadership categorization theory: Internal structure, information processing, and leadership perceptions. Organizational Behavior and Human Performance, 34, 343–378.

Lord, R. G., Foti, R. J., & Phillips, J. S. (1982). A theory of leadership categorization. In J. G. Hunt, U. Sekaran, & C. Schriesheim (Eds.), Leadership: Beyond establishment views (pp. 104–121). Carbondale, IL: Southern Illinois University Press.

Lord, R. G., & Maher, K. J. (1993). Leadership and information processing: Linking perceptions and performance. New York: Routledge.

Mayer, D. M., Bardes, M., & Piccolo, R. (2008). Do servant-leaders help satisfy follower needs? An organizational justice perspective. European Journal of Work and Organizational Psychology, 17, 187–190.

McGregor, D. (1960). The human side of enterprise. New York: McGraw-Hill. McNatt, D. B. (2000). Ancient Pygmalion joins contemporary management: A meta-

analysis of the result. Journal of Applied Psychology, 85, 314–322. Noar, S. M. (2003). The role of structural equation modeling in scale development.

Structural Equation Modeling, 10, 622–647. Offermann, L. R., Kennedy, J. K., Jr., & Wirtz, P. W. (1994). Implicit leadership

theories: Content, structure and generalizability. Leadership Quarterly, 5, 43–58.

Ones, D. S., & Viswesvaran, C. (1996). Bandwidth–fidelity dilemma in personality measurement for personnel selection. Journal of Organizational Behavior, 17, 609–626.

Operario, D., & Fiske, S. T. (2004). Stereotypes: Content, structures, processes, and context. In M. B. Brewer & M. Hewstone (Eds.), Social cognition (pp. 120–141). Malden, MA: Blackwell Publishing.

Paglis, L. L., & Green, S. G. (2002). Both sides now: Supervisor and subordinate perspectives on relationship quality. Journal of Applied Social Psychology, 32, 250–276.

Pillai, R., & Uhl-Bien, M. (2007). The romance of leadership and the social construction of followership. In B. Shamir, R. Pillai, M. C. Bligh, & M. Uhl-Bien (Eds.), Follower-centered perspectives on leadership (pp. 187–210). Charlotte, NC: Information Age Publishing.

Poole, P. P., Gioia, D. A., & Gray, B. (1989). Influence modes, schema change, and organization transformation. The Journal of Applied Behavioral Science, 25, 271–289.

Quinn, R. P., & Shepard, L. J. (1974). The 1972–1973 quality of employment survey. Ann Arbor, Michigan: Survey Research Center.

Rosch, E. (1978). Principles of categorization. In E. Rosch & B. B. Lloyd (Eds.), Cognition and categorization. Hillsdale, NJ: Erlbaum.

Rosenberg, S., & Jones, R. (1972). A method for investigating and representing a person’s implicit theory of personality: Theodore Dreiser’s view of people. Journal of Personality and Social Psychology, 22, 372–386.

Schyns, B., & Meindl, J. R. (2005). An overview of implicit leadership theories and their application in organization practice. In B. Schyns & J. R. Meindl (Eds.), Implicit leadership theories (pp. 15–36). Greenwich, CT: Information Age Publishing.

Sy, T., Cote, S., & Saavedra, R. (2005). The contagious leader: Impact of the leader’s mood on the mood of group members, group affective tone, and group processes. Journal of Applied Psychology, 90, 295–305.

Sy, T., Shore, L., Strauss, J., Shore, T., Tram, S., Whiteley, P., et al. (in press). Leadership perceptions as a function of race-occupation fit: The case of Asian Americans. Journal of Applied Psychology.

84 T. Sy / Organizational Behavior and Human Decision Processes 113 (2010) 73–84

Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics (4th ed.). Boston: Allyn & Bacon.

van Knippenberg, D., & Hogg, M. (Eds.). (2003). Leadership and power: Identity processes in groups and organizations. London: Sage.

van Gils, S., van Quaquebeke, N., & van Knippenberg, D. (2010). The X-factor: On the relevance of implicit leadership and followership theories for leader–member exchange (LMX) agreement. European Journal of Work and Organizational, 19, 333–363.

Wayne, S., Shore, L., & Liden, R. (1997). Perceived organizational support and leader–member exchange: A social exchange perspective. Academy of Management Journal, 40, 82–111.

Weick, K. E. (1995). Sensemaking in organizations. Thousand Oaks, CA: Sage. Wernimont, P. F. (1971). What supervisors and subordinates expect of each other.

Personnel Journal, 50, 204–208. Wofford, J. C., & Goodwin, V. L. (1994). A cognitive interpretation of transactional

and transformational leadership theories. The Leadership Quarterly, 5, 161–186.