taking charge at v^ork: extrarole efforts …...1999 morrison and phelps 405 it entail efforts to...

18
® Academy of Management Journal 1999. Vol. 42, No. 4. 403-419. TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS TO INITIATE WORKPLACE CHANGE ELIZABETH WOLFE MORRISON COREY C. PHELPS New York University In this study, we investigated a neglected form of extrarole behavior called taking charge and sought to understand factors that motivate employees to engage in this activity. Taking charge is discretionary hehavior intended to effect organizationally functional change. We ohtained hoth self-report and coworker data for 275 white- collar employees from different organizations. Taking charge, as reported hy cowork- ers, related to felt responsihility, self-efficacy, and perceptions of top management openness. These results expand current understanding of extrarole hehavior and suggest ways in which organizations can motivate employees to go heyond the bound- aries of their johs to hring ahout positive change. In everyday experience, it comes down to a conflict between those folks who dutifully work to manage established routines in order to ensure the success- ful functioning of their organization, and those who courageously challenge routines in order to do the same thing. (Hornstein, 1986: 8) During the past decade, there has been growing emphasis in the organizational behavior literature on extrarole behavior, or employee bebavior tbat goes beyond role expectations in a way tbat is or- ganizationally functional. Scholars bave argued tbat tbis'pbenomenon is critical for organizational effectiveness because managers cannot foresee all contingencies or fully anticipate tbe activities tbat tbey may desire or need employees to perform (Katz & Kabn, 1966; Organ, 1988). Scbolars bave also argued tbat extrarole bebaviors are not moti- vated by tbe same factors as traditional, role-pre- scribed activities (Organ, 1988, 1990). Given tbe recognized importance of extrarole bebavior, nu- merous studies bave focused on identifying factors tbat predict tbis pbenomenon (see Organ and Ryan [1995] for a review). We argue, bowever, tbat tbis researcb bas been limited by an overly narrow conceptualization of extrarole bebavior. Tbe most beavily researcbed form of sucb bebavior is organizational citizensbip bebavior (OCB). Organ defined OCB as "tbose or- ganizationally beneficial bebaviors and gestures tbat can neitber be enforced on tbe basis of formal role obligations nor elicited by contractual guaran- tee or recompense" (1990: 46). Altbougb tbis defi- We wish to thank two anonymous reviewers for their comments and suggestions on earlier versions of this article. nition is broad, researcb on OCB bas concentrated almost entirely on wbat Organ (1988: 6) referred to as "modest, some would even say trivial" bebaviors tbat sustain tbe status quo. Examples include belp- ing colleagues witb tbeir workloads, not taking ex- cessive breaks, attending functions tbat are not re- quired, and alerting otbers about work-related problems (Organ, 1988). Altbougb tbese extrarole activities are important, we argue tbat tbey are not sufficient for ensuring tbe continued viability of an organization and tbat organizations also need employees wbo are willing to cballenge tbe present state of operations to bring about constructive cbange. Hence, in tbis study, we focused on a form of extrarole bebavior tbat bas been largely neglected. We refer to tbis type of bebavior as taking charge. Taking cbarge entails voluntary and constructive efforts, by individual employees, to effect organizationally functional cbange witb respect to bow work is executed witbin tbe.contexts of tbeir jobs, work units, or organizations. It is similar to otber forms of extra- role bebavior in tbat it is discretionary (tbat is, not formally required). Unlike otber forms of extrarole bebavior tbat bave been assessed, bowever, taking cbarge is inberently cbange-oriented and aimed at improvement. It fits within tbe general class of extrarole bebaviors tbat Van Dyne, Cummirigs, and Parks (1995) referred to as "cballenging-promo- tive." As tbose autbors pointed out, tbis class of extrarole bebaviors bas received little researcb at- tention. Evidence from a variety of sources bigbligbts tbe potential value of taking cbarge. In tbeir work on socialization. Van Maanen and Scbein (1979) ar- gued tbat because organizations in tbe long run 403

Upload: others

Post on 14-Mar-2020

4 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS …...1999 Morrison and Phelps 405 it entail efforts to implement solutions. Two other constructs related to taking charge are task revision

® Academy of Management Journal1999. Vol. 42, No. 4. 403-419.

TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS TOINITIATE WORKPLACE CHANGE

ELIZABETH WOLFE MORRISONCOREY C. PHELPS

New York University

In this study, we investigated a neglected form of extrarole behavior called takingcharge and sought to understand factors that motivate employees to engage in thisactivity. Taking charge is discretionary hehavior intended to effect organizationallyfunctional change. We ohtained hoth self-report and coworker data for 275 white-collar employees from different organizations. Taking charge, as reported hy cowork-ers, related to felt responsihility, self-efficacy, and perceptions of top managementopenness. These results expand current understanding of extrarole hehavior andsuggest ways in which organizations can motivate employees to go heyond the bound-aries of their johs to hring ahout positive change.

In everyday experience, it comes down to a conflictbetween those folks who dutifully work to manageestablished routines in order to ensure the success-ful functioning of their organization, and those whocourageously challenge routines in order to do thesame thing. (Hornstein, 1986: 8)

During the past decade, there has been growingemphasis in the organizational behavior literatureon extrarole behavior, or employee bebavior tbatgoes beyond role expectations in a way tbat is or-ganizationally functional. Scholars bave arguedtbat tbis'pbenomenon is critical for organizationaleffectiveness because managers cannot foresee allcontingencies or fully anticipate tbe activities tbattbey may desire or need employees to perform(Katz & Kabn, 1966; Organ, 1988). Scbolars bavealso argued tbat extrarole bebaviors are not moti-vated by tbe same factors as traditional, role-pre-scribed activities (Organ, 1988, 1990). Given tberecognized importance of extrarole bebavior, nu-merous studies bave focused on identifying factorstbat predict tbis pbenomenon (see Organ and Ryan[1995] for a review).

We argue, bowever, tbat tbis researcb bas beenlimited by an overly narrow conceptualization ofextrarole bebavior. Tbe most beavily researcbedform of sucb bebavior is organizational citizensbipbebavior (OCB). Organ defined OCB as "tbose or-ganizationally beneficial bebaviors and gesturestbat can neitber be enforced on tbe basis of formalrole obligations nor elicited by contractual guaran-tee or recompense" (1990: 46). Altbougb tbis defi-

We wish to thank two anonymous reviewers for theircomments and suggestions on earlier versions of thisarticle.

nition is broad, researcb on OCB bas concentratedalmost entirely on wbat Organ (1988: 6) referred toas "modest, some would even say trivial" bebaviorstbat sustain tbe status quo. Examples include belp-ing colleagues witb tbeir workloads, not taking ex-cessive breaks, attending functions tbat are not re-quired, and alerting otbers about work-relatedproblems (Organ, 1988).

Altbougb tbese extrarole activities are important,we argue tbat tbey are not sufficient for ensuringtbe continued viability of an organization and tbatorganizations also need employees wbo are willingto cballenge tbe present state of operations to bringabout constructive cbange. Hence, in tbis study, wefocused on a form of extrarole bebavior tbat basbeen largely neglected. We refer to tbis type ofbebavior as taking charge. Taking cbarge entailsvoluntary and constructive efforts, by individualemployees, to effect organizationally functionalcbange witb respect to bow work is executedwitbin tbe.contexts of tbeir jobs, work units, ororganizations. It is similar to otber forms of extra-role bebavior in tbat it is discretionary (tbat is, notformally required). Unlike otber forms of extrarolebebavior tbat bave been assessed, bowever, takingcbarge is inberently cbange-oriented and aimed atimprovement. It fits within tbe general class ofextrarole bebaviors tbat Van Dyne, Cummirigs, andParks (1995) referred to as "cballenging-promo-tive." As tbose autbors pointed out, tbis class ofextrarole bebaviors bas received little researcb at-tention.

Evidence from a variety of sources bigbligbts tbepotential value of taking cbarge. In tbeir work onsocialization. Van Maanen and Scbein (1979) ar-gued tbat because organizations in tbe long run

403

Page 2: TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS …...1999 Morrison and Phelps 405 it entail efforts to implement solutions. Two other constructs related to taking charge are task revision

404 Academy of Management Journal August

require innovation and change, it is often valuablefor employees to reject and redefine aspects of theirwork roles. Similarly, Staw and Boettger (1990)highlighted the importance of employees' takingaction to correct faulty tasks or misdirected workroles. They argued that an employee who goes be-yond the call, of duty to accomplish an incorrectlyspecified role, or one who is extra conscientious infollowing procedures that are counterproductive,may ultimately be dysfunctional for an organiza-tion. If current role definitions, procedures, or pol-icies are inappropriate or inefficient, it may bemore adaptive for employees to channel some oftheir extrarole efforts toward changing rather thanmaintaining the status quo. Scholars in the area ofinnovation (Bunce & West, 1995; Scott & Bruce,1994), as well as strategy researchers (Burgelman,1994; Ghoshal & Bartlett, 1994), have also noted thepotential value of employee-initiated change forlong-term organizational adaptability and have be-come increasingly concerned with how organizationscan promote employee initiative (cf. Frohman, 1997).

Given the seeming importance of taking charge,the research question that we sought to address wasthe following: What factors motivate employees tostep outside the boundaries of their jobs to bringabout constructive change in their workplaces? Webegin this article by discussing how taking chargerelates to other forms of extrarole behavior. We thenpresent a model of the motivation to take charge anda set of hypotheses, based on that model, regardingspecific contextual and individual-level antecedents.

TAKING CHARGE IN RELATION TO OTHEREXTRAROLE BEHAVIORS

As noted, taking charge stands in marked con-trast to organizational citizenship behavior, thebest known and most heavily researched form ofextrarole behavior. Taking charge can also be com-pared with such change-oriented behaviors as prin-cipled organizational dissent and whistleblowing.Graham defined principled dissent as "the effort byindividuals in the workplace to protest and/or tochange the organizational status quo because oftheir conscientious objection to current policy orpractice" (1986: 2), and Miceli and Near (1992)defined whistleblowing as the act of reporting ille-gal, immoral, or illegitimate practices to persons ororganizations that might be able to effect a remedy.There are three primary differences between theseactivities and taking charge. First, principled dis-sent and whistleblowing are typically motivated bysuper organizational interests and are often under-taken irrespective of whether the change will beorganizationally functional. Taking charge, in con-

trast, is motivated by desire for organizatioijal im-provement and is not necessarily rooted in princi-ple or in a belief that current practices are wrong orbad. Second, the primary goal of whistleblowing orprincipled dissent is to expose, criticize, or elimi-nate something negative. Taking charge, however,is aimed at implementing something positive.Third, unlike principled dissent and whistleblow-ing, taking charge occurs solely through internaland organizationally sanctioned tactics.

The taking charge construct can also be distin-guished from voice as conceptualized in the workof Withey and Cooper (1989) and Rusbult, Farrell,Rogers, and Mainous (1988). Voice has been de-fined as "any attempt at all to change, rather thanto escape from, an objectionable state of affairs" ^(Hirschman, 1970: 30). Although voice is change-oriented, its objective is not to bring about organi-zational improvement, but to eliminate personaldissatisfaction (Withey & Cooper, 1989). Voice isalso broader and more multifaceted than takingcharge. It includes grievance filing, union partici-pation, use of suggestion boxes, sharing concernswith others, complaining to supervisors, appealingto a higher authority, and external protest (Farrell,1983). Although this list includes activities thatwould be considered taking charge, it also includesactivities that involve much less effort (for in-stance, complaining), as well as activities intendedto harm the employing organization (for instance,external protest). Because voice is defined sobroadly, measures have tended to exhibit low in-ternal consistency, and researchers have had diffi-culty predicting voice (Withey & Cooper, 1989). Inlight of these difficulties, we saw it as valuable tofocus specifically on activities that (1) entailed ac-tive efforts to initiate and implement change (thatis, not just voicing concerns) and (2) were intendedto be constructive for the organization.

Taking charge also has features in common withissue selling (Ashford, Rothbard, Piderit, & Dutton,1998; Dutton & Ashford, 1993). Issue selling entailscalling an organization's attention to key trends,developments, and events that have implicationsfor its performance. Like taking charge, issue sell-ing may enhance an organization's effectiveness bybringing about adaptive change (Ashford et al.,1998). A key difference between these two con-structs, however, is that issue selling focuses onstrategic issues (Dutton & Ashford, 1993), whereastaking charge focuses on the internal means foraccomplishing organizational goals, such as workmethods, policies, and procedures. Fvirther, whereasissue selling calls attention to problems or opportu-nities, it does not offer suggestions about how toaddress those problems or opportunities, nor does

Page 3: TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS …...1999 Morrison and Phelps 405 it entail efforts to implement solutions. Two other constructs related to taking charge are task revision

1999 Morrison and Phelps 405

it entail efforts to implement solutions. Two otherconstructs related to taking charge are task revisionand role innovation. The former is defined as "tak-ing action to correct a faulty task or misdirectedwork role" (Staw & Boettger, 1990: 537), and thelatter refers to changing or improving a job role orthe procedures for performing that role (VanMaanen & Schein, 1979). An important difference,however, is that taking charge can be directed be-yond the scope of an individual's job role.

Recently, Frese and his colleagues (Frese, Kring,Soose, & Zempel, 1996) proposed a construct thatthey termed personal initiative. This construct alsoshares features with taking charge. Personal initia-tive refers to "a behavioral syndrome resulting inan individual's taking an active and self-startingapproach to work and going beyond what is for-mally required in a given job" (Frese et al., 1996:38). Some of the examples of initiative that Freseand colleagues identified could be classified as ex-amples of taking charge (for instance, attempts toreorganize a work structure). However, other exam-ples of personal initiative reflect more traditionalforms of extrarole behavior (such as offering sug-gestions or trying to prevent problems). It is alsoimportant to note that Frese and colleagues concep-tualized personal initiative as a relatively stablebehavioral tendency. Yet we believe that takingcharge will be variable depending on the situation.

A MODEL OF TAKING CHARGE

Despite the recognized importance of extraroleefforts to bring about change in the workplace, rel-atively little is known about what motivates suchbehavior. Because taking charge significantly dif-fers from more traditional forms of extrarole behav-ior, such as OCB, models that have been advancedto explain those behaviors are inappropriate forexplaining taking charge. Thus, to develop ourmodel, we drew from works on discretionary activ-ities that, like taking charge, challenge the statusquo. Specifically, we looked to the literatures onvoice (Rusbult et al., 1988; Withey & Cooper, 1989),issue selling (Ashford et al., 1998; Dutton & Ash-ford, 1993), task revision (Staw & Boettger, 1990),innovation (Bunce & West, 1995; Scott & Bruce,1994), personal initiative (Frese et al., 1996), prin-cipled dissent (Graham, 1996), and whistleblowing(Miceli & Near, 1992). Our review of these litera-tures provided a foundation for our model andindicated two important starting points. The firstwas that effortful, discretionary behavior such astaking charge reflects a calculated, deliberate deci-sion process. The second was that both contextual

and individual-level factors affect this decisionprocess and, through it, taking charge.

Ashford and colleagues' (1998) model of issueselling and Withey and Cooper's (1989) model ofvoice were particularly helpful for understandingthe decision process behind taking charge. Both ofthese works provided coherent theoretical frame-works rather than merely sets of variables, and bothbuilt upon established theories of human behavior(e.g., Ajzen, 1991; Vroom, 1964). From these mod-els, we can conclude that two key judgments un-derlie the decision to take charge. First, employeesassess the probability that taking charge will besuccessful. This argument is consistent with evi-dence that willingness to engage in issue sellingrelates to beliefs about likely success (Ashford etal., 1998) and with evidence that "expected effi-cacy" affects the decision to engage in voice(Withey & Cooper, 1989) and whistleblowing(Miceli & Near, 1992). It is also consistent-with theprominent role of expectancy beliefs within moregeneral models of motivated behavior (Ajzen, 1991;Vroom, 1964).

Models of issue selling and of voice also indicatethat an assessment of anticipated consequenceswill play an important role in the decision to takecharge. Ashford and colleagues (1998) focused spe-cifically on impression management risks and ar-gued that employees will be less likely to engage inissue selling if they fear that doing so will harmtheir images. Missing from this framework, how-ever, is an explicit discussion of the role of antici-pated benefits. Whereas taking charge clearly haspotential risks—a damaged reputation if the initia-tive fails or disapproval if it is seen as inappropri-ate or threatening, for example—it also has poten-tially positive consequences. Subjective expectedutility frameworks such as expectancy theory(Vroom, 1964) suggest that employees will weighanticipated costs against anticipated benefits whendeciding whether to engage in taking charge. Inother words, it is not anticipated risks alone thatenter into the decision process, but rather, a jointconsideration of relative costs and benefits. Thisargument is consistent writh research on voice.Withey and Cooper (1989) argued that, in decidingwhether to engage in voice, employees implicitlyweigh possible payoffs against likely costs. Rusbultand coauthors (1988) argued that actors consider-ing voice assess consequences both for themselvesand for others with whom they are interdependent.

To conclude, we argue that the decision to takecharge will be affected by two judgments: an assess-ment of likely success and an assessment of likelyconsequences. It is important to note, however, thatwe did not empirically assess the two proposed

Page 4: TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS …...1999 Morrison and Phelps 405 it entail efforts to implement solutions. Two other constructs related to taking charge are task revision

406 Academy of Management Journal August

judgmerits in this study. These judgments varyfrom one taking charge opportunity to another, andour objective was to understand the type of personand context that might favor this activity in general.Hence, we focused on identifying specific contex-tual and individual-level factors that would predictan employee's overall tendency to take charge, andwe used the mediating judgments as theoreticaljustifications for relating these variables to our phe-nomenon of interest.

There was no single theory that we could use toidentify a set of relevant contextual and individual-level variables. We therefore drew from the litera-tures on voice, innovation, issue selling, personalinitiative, and principled dissent. Although thoseliteratures have addressed a wide range of indepen-dent variables, in the interest of parsimony we fo-cused on variables suggested by more than one ofthose literatures. Our main criterion in selectingvariables was consistency with our underlyingmodel. Specifically, we focused on variables thatcould be directly related, at a theoretical level, tothe two judgments underlying the decision to takecharge. Our conceptualization of taking charge asbased on a deliberate decision process also steeredus away from the attitudinal and dispositional vari-ables that have been emphasized in the OCB liter-ature, as very different motivational processes havebeen posited for those variables (Organ, 1990). Thecontextual variables that we identified as importantdeterminants of taking charge were top manage-ment openness and supportive group norms. Theindividual-level variables that we focused uponwere self-efficacy, felt responsibility for change,and expert power.

Contextual Factors

Top management openness is defined as the de-gree to which top management is believed to en-courage and support suggestions and change initi-atives from below. As Ashford and her coauthors(1998) argued in their work on issue selling, em-ployees attend to the actions of top management toobtain clues about how management is likely torespond to various types of risky initiatives. Theirwork suggests that if employees perceive that topmanagement will respond favorably (or at least notnegatively) to a risky change-oriented activity, theywill feel more confident of success and less con-cerned about political and image risks. Researchfrom other areas also suggests that top managementopenness will be an important variable affectingtaking charge. Scott and Bruce (1994) emphasizedthat employees will be more likely to engage inindividual innovation if they perceive that there is

a "climate for innovation," or in other words, thattheir organization supports new ideas and changeefforts. Further, Graham (1986) argued that princi-pled dissent will be more likely if an organizationalculture supports independence and innovative re-sponses from employees.

Hypothesis 1. Taking charge will be positivelyrelated to perceptions of top managementopenness.

Research on innovation (Bunce & West, 1995;Scott & Bruce, 1994), as well as work on issueselling (Ashford et al., 1998), suggests that workgroup norms that support and encourage changewill also motivate employees to take charge. Normsare shared standards of behavior that emergewithin a group. Because most individuals attachpositive value to meeting the behavioral expecta-tions of their group, they gain personal benefits bybehaving in a way that meets those expectationsand incur costs by violating them (Ajzen & Fish-bein, 1980). Hence, if there are work group normsthat support change, employees should be morelikely to take charge because they will regard doingso as a way to gain group approval (Scott & Bruce,1994). Further, they will regard taking charge asless costly (Ashford et al., 1998), because support-ive norms imply less risk of group disapproval orrejection. Supportive group norms also imply ahigher likelihood of success, because coworkerswill not stand in the way of the change effort andmay even actively facilitate it (Ashford et al., 1998).

Hypothesis 2. Taking charge will be positivelyrelated to work group norms that supportchange.

Individual-Level Factors

Research from several related literatures suggeststhat one of the most important individual-levelvariables that will predict taking charge is self-efficacy, or an employee's estimate of his or hercapacity to perform (Gist & Mitchell, 1992). Al-though self-efficacy is typically conceptualized as atask- or situation-specific construct, Sherer and col-leagues (Sherer, Maddux, Mercandante, Prentice-Dunn, Jacobs, & Rogers, 1982) argued that individ-uals also possess generalized self-efficacy beliefsthat apply across situations. Evidence that general-ized self-efficacy will relate to taking charge comesfrom recent work showing that self-efficacy in-creases personal initiative at work (Speier & Frese,1997). We argue that, relative to employees withlow self-efficacy, those with high self-efficacy willtend to attach a higher likelihood of success to

Page 5: TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS …...1999 Morrison and Phelps 405 it entail efforts to implement solutions. Two other constructs related to taking charge are task revision

1999 Morrison and Phelps 407

taking change and will thus be more likely to at-tempt this behavior. This argument is consistentwith Graham's (1986) conceptual model of princi-pled dissent, which suggests that employees withhigh self-confidence see principled dissent as amore feasible (that is, potentially effective) way tobring about change than employees with low self-confidence. It is also consistent with Withey andCooper's (1989) argument that individual differ-ences in beliefs about personal efficacy will affectthe decision about whether to engage in voice.There is also evidence that self-efficacy will affectthe perceived costs and benefits associated withtaking charge. In particular, Sitkin and Pablo (1992)demonstrated that individuals with high self-effi-cacy tend to underestimate the risks associatedwith any given course of action and tend to over-estimate their ability to overcome those risks.

Hypothesis 3. Taking charge will be positivelyrelated to self-efficacy.

A second individual-level variable hypothesizedto affect taking charge is felt responsibility, or anindividual's belief that he or she is personally ob-ligated to bring about constructive change. Felt re-sponsibility has been cited as an important variablein several works on discretionary employee behav-ior that entails risks (Frese et al., 1996; Graham,1986; Pearce & Gregersen, 1991; Withey & Cooper,1989). Most notably, Graham (1986) argued that thedecision to respond to an issue of principle isheavily dependent on perceived responsibility, andFrese and coauthors (1996) proposed that felt re-sponsibility relates to employee initiative. Al-though neither of these works is explicit about themediating process linking initiative taking to feltresponsibility, we would argue that this effect ismediated both by judgments about likely outcomes(costs and benefits) and by judgments about likelysuccess. To the extent that employees have a highsense of personal responsibility regarding change,they will attach positive valence to taking chargebecause it will provide a sense of personal satisfac-tion and accomplishment (Frese et al., 1996; Gra-ham, 1986). Indeed, Graham (1986) suggested thatemployees with high felt responsibility may attachnegative valence to not taking action when an op-portunity arises. Graham also proposed that indi-viduals with a high level of felt responsibility willregard challenging behavior as more feasible, or inother words, as more likely to succeed.

Hypothesis 4. Taking charge will be positivelyrelated to an employee's felt responsibility for .bringing about change.

Finally, our framework suggests that takingcharge will relate to an employee's expert power, orthe degree to which the employing organization isdependent on the employee for critical knowledgeor skills (French & Raven, 1959; Salancik & Pfeffer,1977). Because power implies greater discretionand credibility and less resistance from others, em-ployees with a high level of expert power shouldfeel more confident that they can bring aboutchange successfully. Similar arguments have beenmade in discussions of issue selling (Dutton & Ash-ford, 1993), role innovation (Ashford & Taylor,1990), task revision (Staw & Boettger, 1990), andpersonal initiative (Frese et al , 1996). Power is alsolikely to encourage taking charge by reducing theperceived costs associated with that activity. Rela-tive to an employee with little expert power, onewith a high level of expert power will be less likelyto suffer organizational or group sanctions if he orshe tries to initiate change. This argument is sup-ported by research on voice, which has shown thatindividuals with more job alternatives, and hencegreater power, are more likely to engage in voicebecause they are less concerned about retaliation(Rusbult et al., 1988; Withey & Cooper, 1989).

Hypothesis 5. Taking charge will be positivelyrelated to the amount of expert power that anemployee possesses.

METHODS

Sample and Procedures

The sample pool for this study consisted of 1,010employees enrolled in a part-time master's of busi-ness administration (M.B.A.) program at a largeurban university. In July 1996, we visited each of22 classes to introduce the study and encourageparticipation. Surveys were then sent by mail toemployees' homes, with a cover letter assuring con-fidentiality. To provide an incentive to participate,we told potential respondents that their nameswould be entered into a drawing w^hose winnerwould receive dinner for two at the restaurant ofhis or her choice (maximum value $200). The sur-veys were followed by reminder postcards after twoweeks and by replacement surveys after fourweeks. Four-hundred eighty-one completed sur-veys were returned (47.6%). A drawing was heldfor the prize, which was subsequently awarded.

The focal employee survey assessed the five in-dependent variables in the model. It also assessedage, gender, organizational and job tenure, hierar-chical level, and number of "direct reports," as

Page 6: TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS …...1999 Morrison and Phelps 405 it entail efforts to implement solutions. Two other constructs related to taking charge are task revision

408 Academy of Management Journal August

potential control variables. The average age of re-spondents was 28.7 years (s.d. = 3.8), the averageorganizational tenure was 3.8 years (s.d. = 2.7), theaverage job tenure was 2.8 years (s.d. = 1.4), and 39percent of the respondents were women. For hier-archical level, 38 percent indicated that they heldnonsupervisory positions, 27 percent indicatedthat they held lower-level management positions,and 23 percent indicated that they were in middlemanagement. Forty-nine percent had no direct re-ports, and only 12 percent had more than fourdirect reports.

The dependent variable, taking charge, was as-sessed via a different survey sent to coworkers. Thelast part of the focal eniployee survey asked respon-dents to provide the names, phone numbers, andaddresses of two coworkers with whom theyworked closely and who they thought would bewilling to provide some additional information.They were told that they could list their directsupervisors, peers, or subordinates—anyone whoworked with them and was familiar with theirwork. To minimize selection bias, they were nottold the nature of the questions that would beasked. We asked for the names of two coworkers inthe hope that we could obtain data from at least oneof them. Two-hundred thirty-six employees pro-vided names and addresses for two coworkers, and75 provided the name and address of one coworker.In total, 65 percent of the respondents (n = 311)provided the name and address of at least one co-worker.

The survey sent to each of the identified cowork-ers [n = 547) asked them to assess the relevant focalemployee's taking charge behavior. Assessing ourdependent variable via coworker surveys enabledus to avoid common source bias and also to mini-mize the social desirability bias that might signifi-cantly distort self-reports of taking charge. Eachcoworker was also asked to indicate age, gender,job and organizational tenure, hierarchical position relative to the focal employee, and how long he orshe had known the focal employee. The surveyswere followed by reminder postcards after twoweeks and then by phone calls to anyone who hadnot responded after four weeks. In total, 417 co-workers returned completed surveys (76%), pro-viding us with data on taking charge for 275 focalemployee respondents. This final sample of 275(which was used for hypothesis testing) repre-sented 88 percent of the respondents who had pro-vided at least one coworker name, 57 percent ofthose who had completed surveys, and 27 percentof those originally solicited.

Scale Development

Because taking charge is a new construct, weengaged in a multistage process to develop a mea-sure. First, we administered an open-ended surveyto 148 part-time M.B.A. students. The survey askedthem to think of individuals with whom they hadworked who had actively tried to bring about im-provement within their organization. They weretold that these change efforts could be aimed at anyaspect of the organization, including the person'sjob, how work was performed within their depart-ment, and organizational policies or procedures.Further, they were asked to focus on efforts thatwent beyond the person's formal role or, in otherwords, efforts that were not required or formallyexpected. Respondents were then asked to list spe-cific behaviors that reflected or exemplified theperson's change efforts. The number of behavioralstatements respondents made ranged from 1 to 5,with an average of 3.00. In total, respondents pro-vided 445 statements.

We reduced the list of 445 statements to 180 byeliminating redundancies, statements that reflectedgeneral tendencies rather than behaviors, and state-ments that were too vague to be formulated in termsof observable behaviors. We then sorted the behav-ioral statements into groups based upon their sim-ilarity to one another. For each group we developeda statement that we believed captured the generalmeaning of the specific behaviors in the group.This process yielded a total of 19 general behav-ioral statements. We then independently assessedwhich statements best refiected the domain of theconstruct. Inconsistent assessments were resolvedduring discussion. In the end, we constructed a listof ten prototypical activities that best reflected theconstruct of taking charge.

We then presented the list of ten behaviors to 20M.B.A. students with the following instructions:"Please think about your last full-time job and theperson with whom you worked most closely. Indi-cate how frequently or infrequently that personengaged in each of the activities below." Responseswere on a five-point scale (1 = very infrequently,5 = very frequently). Respondents were then askeda series of questions about the clarity of the instruc-tions and were asked for suggestions about how thesurvey might be improved. Respondents indicatedthat they considered the instructions to be clearand that they had little difficulty thinking of spe-cific instances that corresponded to each listed ac-tivity. We made two changes based on their feed-back: We provided a "not applicable" option, incase some of the activities were not relevant in agiven context, and we changed the response scale

Page 7: TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS …...1999 Morrison and Phelps 405 it entail efforts to implement solutions. Two other constructs related to taking charge are task revision

1999 Morrison and Phelps 409

to an agree/disagree format, altering the wording ofthe items accordingly. The agree/disagree format isconsistent with other scales used to assess extrarolebehaviors (e.g., Konovsky & Pugh, 1994; Podsakoff,MacKenzie, Moorman, & Fetter, 1990).

The third stage of instrument development in-volved pretesting the revised measure using a sam-ple of 152 individuals who were similar to thosewho would be included in the actual study: part-time M.B.A. students with full-time jobs. The in-structions were the following: "We are studyingvoluntary change-oriented behaviors at work.Please think about a particular coworker withwhom you work closely. Then indicate the degreeto which each of the statements below character-izes that person's behavior."

There were three objectives of this pretest. First,we wanted to assess the internal consistency of theitems. We found the scale to be unidimensionaland highly reliable, with an alpha coefficient of .92.Second, we wanted to ensure that there was rea-sonable variance on the items. The average stan-dard deviation for the items was 1.12, which indi-cated that responses did sufficiently vary. Third,we wanted to test whether the ten activities couldbe considered extrarole for the types of employeeswho would be included in this study. To determinethis, we asked respondents to indicate with a checkany behaviors that went above and beyond whatwas expected as part of their coworker's job. Weintentionally worded the question this way (in-stead of asking them to indicate any behavior thatwas not extrarole) in order to provide as conserva-tive a test as possible. The responses left us reason-ably confident that the activities represented extra-role behaviors for our sample. Seventy-sevenpercent of the respondents checked a majority (sixor more) of the behaviors.

The final stage of instrument development wasdesigned to more rigorously assess the levels ofdiscriminant validity provided by the scales mea-suring (1) taking charge and in-role behavior and(2) taking charge and other forms of extrarole be-havior. We collected questionnaire data from a newsample of part-time M.B.A. students [n = 165) whowere instructed to rate a coworker. The question-naire assessed taking charge, in-role behavior, andOCB. We measured in-role behavior with the fourpositively worded items from Williams and Ander-son's (1991) scale (cf. Van Dyne & LePine, 1998)and measured the altruism and civic virtue dimen-sions of OCB using the scales developed by Podsa-koff et al. (1990). The latter two scales have beenused in several studies of OCB. Cronbach's alpha

was .94 for in-role behavior, .79 for civic virtue(four items), .91 for altruism (five items), and .95 fortaking charge. All items were measured using afive-point scale (1 = strongly disagree, 5 = stronglyagree).

We investigated the construct validity of the tak-ing charge measure via confirmatory factor analysis(CFA) using LISREL 8 (Joreskog & Sorbom, 1993).We specified a four-factor measurement model (in-role behavior, altruism, civic virtue, and takingcharge) with a 23 X 23 "polychoric" correlationmatrix. Convergent validity was evaluated in termsof whether each item had a statistically significantloading on its posited underlying construct factor.As shown in Table 1, the factor loadings were allhighly significant (p < .01) and corresponded to thehypothesized latent constructs. These results dem-onstrate clear convergent validity.

To establish discriminant validity, we tested aseries of hierarchically nested models: a one-factormodel (all items together), a two-factor model (in-role behavior versus extrarole behavior), and ourhypothesized four-factor model. We employed achi-square difference test to examine which modelbest fitted the data. The hypothesized four-factormodel provided a fit superior to the next-best-fit-ting model (A^^ = 564.88, p < .001, n = 165),providing evidence of good discriminant validity.Discriminant validity was also supported by thefact that correlations among the four latent con-structs were all significantly less than 1.0. Thesevalues ranged from .34 to .63, with a mean of .49.

To assess overall model fit, we considered severalgoodness-of-fit indexes: the incremental fit index (IFI;Bollen, 1989), recommended by Gerbing andAnderson (1993); the normormed fit index (NNFI;Bentler & Bonett, 1980); and both the comparative fitindex (CFI) and the root-mean-squared error of ap-proximation (RMSEA), recommended by Jaccard andWan (1996). The IFI, NNFI, and CFI indicate therelafive improvement in fit of a hypothesized modelover a null model. The general rule of thumb for theseindexes is that values of .90 or greater indicate goodfit (Bentler & Bonett, 1980; Bollen, 1989; Jaccard &Wan, 1996). For the RMSEA, Browne and Cudeck(1993) argued that values of less than .08 imply ade-quate fit. As reflected in these various indexes, ourhypothesized measuremerit model demonstratedgood fit with the data (IFI = .93, NNFI = .93, CFI =.93, RMSEA = .07). In sum, the results of the CFAprovide strong evidence that the taking chargemeasure is distinct from both in-role behavior andtwo of the most commonly assessed forms ofextrarole behavior.

Page 8: TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS …...1999 Morrison and Phelps 405 it entail efforts to implement solutions. Two other constructs related to taking charge are task revision

410 Academy of Management Journal August

TABLE 1Results of Factor Analysis of the Hypothesized Measurement Model with Taking Charge, In-Role

Behavior, Civic Virtue, and Altruism"

ItemIn-Role Civic Taking

Behavior Virtue Altruism Charge

This person fulfills the responsibilities specified in his/her job description.This person performs the tasks that are expected as part of the job.This person meets performance expectations.This person adequately completes responsibilities.This person attends meetings that are not mandatory, but that are considered important.This person attends functions that are not required, but that help the company image.This person keeps abreast of changes in the organization.This person reads and keeps up with organization announcements, memos, and so on.This person helps others who have been absent.This person helps orient new people even though it is not required.This person helps others who have heavy work loads.This person helps others who have work-related problems.This person is always ready to lend a helping hand to those around him/her.This person often tries to adopt improved procedures for doing his or her job.This person often tries to change how his or her job is executed in order to be more

effective.This person often tries to bring about improved procedures for the work unit or

department.This person often tries to institute new work methods that are more effective for the

company.This person often tries to change organizational rules or policies that are nonproductive .

or counterproductive.This person often makes constructive suggestions for improving how things operate

within the organization.This person often tries to correct a faulty procedure or practice.This person often tries to eliminate redundant or unnecessary procedures.This person often tries to implement solutions to pressing organizational problems.This person often tries to introduce new structures, technologies, or approaches to

improve efficiency.

.95

.92

.93

.90.63.57.89.81

.80

.82

.87

.84

.90.82.83

.89

.90

.73

.81

.83

.74

.87

.87

° Data are from the final scale validation sample [n = 165). All factor loadings are completely standardized lambda loadings and aresignificant at p < .01.

Assessment of Dependent Variable forHypothesis Testing

The taking charge measure appeared on the sur-veys sent to coworkers. The coworkers w ere pre-sented with the ten statements and asked to indi-cate (on a five-point agree/disagree scale) thedegree to which each statement characterized focalemployees' behavior. The coworkers were alsoasked to indicate the extent to which the activitiesexceeded what was expected as part of the focalemployees' jobs (1 = not at all, 3 = somewhat, 5 =completely) and the extent to which the focal em-ployees had discretion to engage in the variousactivities (1 = none at all, 3 = a moderate amount,5 = complete discretion).

The data from each coworker were used to com-pute a measure of taking charge for the correspond-ing focal employee. This measure was formed byaveraging the responses to the ten items (a = .93).For approximately half of the focal employees (n =142), we had two separate measures of taking

charge-^ one from each coworker. This was a muchgreater proportion than anticipated, because we didnot expect to have such a high response rate fromthe coworkers.

To determine whether it was appropriate to cre-ate an average rating for each coworker pair, wecomputed an intraclass correlations coefficient(ICC). Because our data consisted of ratings fromdifferent judges for each focal employee, we em-ployed ICC(1,1) (Shrout & Fleiss, 1979). This formof intraclass correlation provides a point estimateof the agreement of ratings made by two or morejudges (James, 1982). The ICC indicated a low, andnot quite significant, level of agreement (ICC = .11,p < .10).

Because the ICC was so modest, we examined theraw data, making two observations. First, respon-dents tended to use the high end of the five-pointscale, causing a restriction of range that most likelyattenuated the correlation. Second, whereas inmost cases there was reasonable agreement be-

Page 9: TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS …...1999 Morrison and Phelps 405 it entail efforts to implement solutions. Two other constructs related to taking charge are task revision

1999 Morrison and Phelps 411

tween the two coworkers, in a few notable cases thetwo coworkers provided highly discrepant ratings.In those cases, it was impossible to determinewhich coworker (if either) was the more accurate,but it seemed clear that averaging the two re-sponses was unlikely to provide a valid measure ofthe focal employee's behavior.

Civen these conclusions, we decided that themost appropriate course of action was to eliminatethe cases in which the two coworkers providedhighly divergent assessments. We were conserva-tive in doing so, eliminating ten cases where thedifference between the two coworker measures ex-ceeded 2.00 on the five-point scale. The ICC be-tween the two measures of taking charge for theremaining pairs of coworkers was .36 (p < .001).This value compares favorably to other reportedICC values. In an extensive review of organizationalclimate studies, James (1982) reported that therange of agreement as measured by ICC(1,1) was .00to .50, with a median of approximately .12. With anICC of .36, we decided that it was appropriate toaverage the two ratings when we had data from twocoworkers [n = 133) and to use the single ratingwhen we had data from a single coworker [n =132). Thus, the final sample size for hypothesistesting was 265.

Assessment of Independent Variables

The surveys sent to the focal employees con-tained measures of the five independent variables:top management openness, group norms, self-effi-cacy, felt responsibility, and power. All variableswere measured on five-point scales (1 = stronglydisagree, 5 = strongly agree).

Top management openness was assessed with asix-item scale used by Ashford and colleagues(1998), which they adapted from House and Rizzo's(1972) top management receptiveness measure.House and Rizzo (1972) demonstrated their scale tobe reliable, and Ashford and colleagues also pro-vided strong support for the reliability of the scale(a = .92). In addition, Ashford and colleagues pro-vided evidence of the scale's convergent validityand demonstrated that it could be empirically dis-criminated from other perceptual constructs mea-sured via self-reports. A sample item from thisscale is "Good ideas get serious consideration frommanagement above me." Coefficient alpha for thescale was .90.

We measured group norms using a 12-item scaleadapted from Scott and Bruce's (1994) support forinnovation measure. Scott and Bruce demonstratedthis scale to be both unidimensional and highlyreliable (a = .92), and they provided evidence of

convergent, discriminant, and nomological valid-ity. Whereas the original scale asked individuals toindicate the degree to which their organizationssupported innovation, we adapted the scale by ask-ing respondents to use their immediate workgroups (the individuals with whom they worked ona regular basis) as their referents in responding. Weprovided a "not applicable" option for respondentswho did not consider themselves to be part of awork group. A sample item is "My work group canbe described as flexible and continually adapting tochange." The internal reliability coefficient (a) forthe 12-item scale was .91. As described in the re-sults section, however, because of a strong reverse-coding effect, we eliminated the 7 negativelyworded items. Coefficient alpha for the remainingitems was .84.

To assess self-efficacy, we used the general self-efficacy scale developed by Sherer and colleagues(1982) and used by several other researchers (e.g.,Schaubroeck & Merrit, 1997; Tharenou, Latimer, &Conroy, 1994). This scale originally consisted of 17items (a = .86), but we deleted an item that Shereret al. found to be problematic. Those authors dem-onstrated their scale to be adequately unidimen-sional and provided evidence of both nomologicaland criterion validity. A sample item from the scaleis "When I make plans, I am certain I can makethem work." The 16 items had a coefficient alpha of.85.

Felt responsibility was measured using five itemsdesigned specifically for this study (a = .80). Theseitems were as follows: "I feel a personal sense ofresponsibility to bring about change at work," "It'sup to me to bring about improvement in my work-place," "I feel obligated to try to introduce newprocedures where appropriate," "Correcting prob-lems is not really my responsibility," and "I feellittle obligation to challenge or change the statusquo." Although we did not engage in any system-atic construct validation for this scale, the fact thatit was significantly correlated with hierarchicallevel (r = .24, p < .01) provides at least someevidence of validity. That is, one would expect thatindividuals at higher organizational levels wouldfeel more responsible for change than those atlower organizational levels.

We measured expert power using four items de-veloped and used by Morrison and Robinson(1997). The items all related to an employee's pos-session of critical and unique skills or expertise. Asample item from this measure is "I possess uniqueskills that increase my value to the organization."The internal reliability coefficient (a) for the scalewas .79 in both this study and in Morrison andRobinson.

Page 10: TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS …...1999 Morrison and Phelps 405 it entail efforts to implement solutions. Two other constructs related to taking charge are task revision

412 Academy of Management Journal August

Confirmatory Factor Analysis of IndependentVariables

Because all 5 independent variables were as-sessed via self-reports on a single questionnaire, wefelt that it was important to assess discriminantvalidity before testing the hypotheses. This wasdone via a first-order confirmatory factor analysis(CFA) using the data from the entire sample of focalemployee respondents, who numbered 462 after"listwise" deletion of cases with missing data. Asthere were 44 items, it would have been extremelydifficult to fit a measurement model (Bentler &Chou, 1987; Harris & Schaubroeck, 1991; Joreskog& Sorbom, 1989). Thus, we followed Bentler andChou's and Harris and Schaubroeck's recommen-dations that only 20 or so observed variables beanalyzed.

To reduce the number of observed variables, wefollowed procedures delineated by Bagozzi andHeatherton (1994). For the scales that employed 5or fewer items (responsibility and expert power),we used the individual scale items as observedindicators of the latent constructs. For the remain-ing scales, we reduced the number of observedindicators by creating subscales from pairs or trip-lets of original scale items. We reduced the numberof indicators for top management openness from 6to 3 by randomly pairing items and reduced thenumber of indicators for group norms from 12 to 4in a similar manner.

We employed a slightly different approach withthe self-efficacy items. Although Sherer and col-leagues (1982) argued that their scale represented asingle construct, there is evidence that it may havethree dimensions, representing the strength, mag-nitude, and generality of efficacy (Woodruff &Cashman, 1993). Bagozzi and Heatherton (1994)provided a framework for modeling such multidi-mensional constructs. In their partial aggregationmodel, a single higher-order latent construct is rep-resented by multiple lower-order dimensions. Eachlower-order dimension is represented by a compos-ite of observed indicators. Bagozzi and Heathertonargued that because there is a one-to-one corre-spondence between each composite and the asso-ciated lower-order factor, the composites can betreated, in a CFA, as multiple indicators of a singlehigher-order factor. Drawing on their work, we re-duced the number of observed indicators for self-efficacy by developing three composite measuresconsisting of the items that correspond to the threeproposed dimensions of general self-efficacy (Wood-ruff & Cashman, 1993).

After reducing the number of observed indicatorsfrom 44 to 20, we performed a CFA using LISREL 8.

We specified our five-factor measurement model(top management openness, group norms, self-effi-cacy, felt responsibility, and expert power) using a20 X 20 polychoric correlation matrix. Fit statisticsfor the hypothesized measurement model indicateda relatively poor fit with the data. An examinationof factor loadings, standardized residuals, andmodification indexes indicated that the misspecifi-cation resided with the group norms scale. Theevidence indicated that the scale did not representa unidimensional construct, violating a crucial as-sumption of measurement theory (Hattie, 1985).

In order to modify the measurement model, wefollowed the cross-validation approach suggestedby Cudeck and Browne (1983), whereby the sampleis divided into two random subsamples. Modifica-tions are based on results from the first subsampleand are then cross-validated with the second sub-sample. Using the first subsample (n = 231), weconducted an exploratory factor analysis (EFA) ofthe group norms items. The results indicated twofactors. Seven items loaded on the first factor, withloadings ranging from .56 to .84, and the remainingfive items loaded on the second factor, with load-ings ranging from .64 to .87.

Further investigation indicated that the first fac-tor corresponded to all negatively worded items,and the second factor corresponded to all posi-tively worded items. This phenomenon is knownas a reverse-coding effect and is not uncommon(Herche & Engelland, 1996; Magazine, Williams, &Williams, 1996). As Herche and England pointedout, this methodological artifact poses a significantthreat to a scale's unidimensionality and thus, va-lidity. Although numerous theoretical rationaleshave been offered to explain this phenomenon (seeHerche and England [1996] for a review), there hasbeen little work on how to best solve the problem.Following the arguments of Nunnally (1978) andSchriesheim, Eisenbach, and Hill (1991), we de-cided to eliminate the negatively worded items. Wethen performed an EFA with the five remainingitems together with the items for the other fourindependent variables. Five clear factors emerged,corresponding to the five constructs. Primary load-ings ranged from .56 to .93, and cross-loadings didnot exceed .20.

Next, we performed a CFA using the second sub-sample [n = 231). Fit statistics for the hypothesizedmeasurement model indicated a good fit (IFI = .91,NNFI = .90, CFI = .91, RMSEA = .08). FollowingCudeck and Browne (1983), we then tested themodified measurement model in the full sample.Fit statistics indicated a good fit (IFI = .92, NNFI =.91, CFI = .92, RMSEA = .07). In addition, thecorrelations among the latent constructs were all

Page 11: TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS …...1999 Morrison and Phelps 405 it entail efforts to implement solutions. Two other constructs related to taking charge are task revision

1999 Morrison and Phelps 413

significantly less than 1.0. They ranged from .16 to.55, with a mean of .29. From these results, wecould conclude that there was sufficient discrimi-nant validity among the five independent variables.

Test for Sampling Bias

Of the 481 focal employees who responded to thesurvey, 206 lacked coworker data and were thusexcluded from the study. This exclusion raisedconcerns about whether there were systematic dif-ferences between those respondents with coworkerdata and those without. To test for such differences,we performed a set of ten ^tests comparing the twogroups on the five control variables: age, gender,hierarchical level, organizational tenure, and jobtenure. Only two of the differences were signifi-cant. The individuals with coworker data wereslightly younger (mean difference = .82 years, t =2.30, p < .05) and had slightly less time with theircompany (mean difference = .64 years, t = 2.52,p < .05). These data suggested little basis for con-cluding that there were meaningful differences be-tween the two groups.

RESULTS

Descriptive Statistics and Correlations

Means, standard deviations, and correlations arein Table 2. The mean for taking charge was 3.84(s.d. = 0.72), indicating that taking charge was rel-atively common. There were significant correla-tions between taking charge and each of the inde-pendent variables.

Although the pretest provided evidence that tak-ing charge was distinct from in-role behavior, we

also assessed the extent to which the coworkerinformants regarded it as extrarole. In response tothe item "To what extent do the above activitiesexceed what is actually expected for this person?"65 percent responded with a 4 or a 5 on the five-point scale (5 = completely). The mean for thisitem was 3.9 (s.d. = 0.99). This value indicated tous that the activities were seen as more extrarolethan in-role. Coworkers also responded that thefocal employees had some discretion about engag-ing in the ten activities if they wished to do so. Themean for the item "How much discretion does theperson have to engage in each of the activities?" was3.4 (s.d. = 0.82), where 3 corresponded to "a mod-erate amount" and 5 reflected "complete discre-tion."

Tests of Hypotheses

To test the hypotheses, we regressed the takingcharge measure on the measures representing eachof the five independent variables: top managementopenness, group norms, self-efficacy, felt responsi-bility, and expert power. We entered job tenure andhierarchical level as control variables, because theywere correlated both with taking charge and withsome of the independent variables. We did notcontrol for age, gender, or organizational tenure.These variables were not significantly correlatedwith taking charge or with any of the independentvariables under investigation. Hence, there was nobasis for including them as control variables.

As shown in Table 3, the independent variablesexplained a significant amount of variance in tak-

TABLE 2Means, Standard Deviations, Scale Reliabilities, and Correlations"

Variable

1. Taking charge2. Top management openness3. Group norms4. General self-efficacy5. Felt responsibility6. Expert power7. Organization level8. Job tenure9. Organizational tenure

10. Age11. Gender''

Mean

3.843.223.614.123.703.391.922.843.81

28.710.39

s.d.

0.720.920.780.460.780.840.911.362.723.800.49

1

(.93).29*.22*.31*.25*.24*.13*

-.08-.03-.02-.07

2

' (.90).49**.14*

' .40**' .22**

.27**-.09-.05-.03-.05

3

(.84).09.32**.29**.18**

-.12*.03.07.04

4

(.85).20**.23**.04

-.08-.11

.02-.01

5

(.80).32**.28**.04.02.03

-.03

6

(.79).23**.07.06.06

-.08

7

.10

.39**

.21**-.03

8

.17*

.21**-.06

9

.45**-.01

10

- .15*

" The final sample size was 265 because 10 cases were eliminated. Cronhach's alphas appear on the diagonal for multiple-item measures.Coding: 1 = woman.* p < .05

** p < .01

Page 12: TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS …...1999 Morrison and Phelps 405 it entail efforts to implement solutions. Two other constructs related to taking charge are task revision

414 Academy of Management Journal August

TABLE 3Results of Regression Analysis for Taking

Charge^

Variable

Top management opennessGroup normsGeneral self-efficacyFelt responsibilityExpert powerJob tenure*"Organizational level''

Adjusted R^

F

P

.15

.02

.20

.28

.05-.06

.02

.27

.23

17.50***

t

2.87**0.364.43***5.62***1.02

-1.300.32

° Tbe final sample size was 265 because 10 cases were elim-inated,

'' Control variable.*p < .05

**p < .01***p < .001

ing charge [R^ = .27, adjusted R'^ = .23, F = 17.50,p < .001). In support of Hypothesis 1, top manage-ment openness was significantly related to takingcharge {)3 = .15, t = 2.87, p < .01). Hypothesis 2, onthe other hand, did not receive support. Groupnorms were unrelated to taking charge ()3 = .02, t =0.36, n.s.). Hypotheses 3 and 4 were both sup-ported. Taking charge was related to both self-effi-cacy (/3 = .20, t = 4.43, p < .001) and felt respon-sibility (/3 = .28, t = 5.62, p < .001). However,taking charge was unrelated to expert power(/3 = .05, t = 1.02, n.s.), providing no support forHypothesis 5.

DISCUSSION AND CONCLUSIONS

Consistent w ith our predictions, the results ofthis study indicated that employees were morelikely to take charge when they perceived top man-agement as open to employee suggestions and toemployee-initiated change. This finding highlightsthe importance of a context that conveys to employ-ees that this form of extrarole behavior will not bemet with resistance or entail high political risks.When employees perceive that top managementsupports constructive efforts to bring about im-provement, they may be more confident that takingcharge will be effective and less concerned aboutpotential costs.

Also consistent with our predictions, employeeswere more likely to take charge to the extent thatthey had a high level of self-efficacy and an inter-nalized sense of responsibility for bringing aboutchange in their workplaces. These results demon-

strate that the decision about whether or not to takecharge is affected not only by context, but also byindividual characteristics. Hence, even within thesame organization, some individuals may be morelikely to take charge than others—specifically,those with high self-efficacy and felt responsibility.

We had predicted that taking charge would alsobe encouraged by supportive group norms. How-ever, norms did not explain any unique variancein the dependent variable; there was a significantzero-order correlation between norms and takingcharge, but the fact that norms did not have aneffect when other variables were controlled impliesthat the correlation was most likely spurious. Onepossible explanation for the nonsignificant effect isthat we did not control for group cohesiveness.Hackman (1983) noted that group norms only affectindividual behavior where group cohesiveness ishigh, and respondents in this study may not havebeen members of cohesive work groups. The lack ofsupport for our prediction may also reflect the factthat we did not assess perceived need for change. Itis possible that within some work groups that arehighly supportive of change, enough change is ini-tiated by others that a given employee will feel lessneed to initiate change him- or herself. Becausethese ideas are merely speculative, we encourageresearchers to explore in more depth the relation-ship between work group characteristics and takingcharge.

The other unsupported prediction was that theamount of power possessed by an employee wouldexplain unique variance in taking charge. We fo-cused specifically on expert power (French &Raven, 1962; Salancik & Pfeffer, 1977). It is possi-ble, however, that this is not the most relevant formof power with respect to taking charge. Alterna-tively, the lack of support for this hypothesis mayreflect methodological shortcomings. The scale thatwe used to assess expert power has not been usedin previously published work, and there is rela-tively limited evidence of the nieasure's constructvalidity.

Implications for Research and for Practice

The results of this study make an important con-tribution to the literature on extrarole behavior.Despite a growing body of work in this area, exist-ing research has provided a limited view of extra-role behavior by neglecting activities aimed atchanging the status quo (cf. Van Dyne et al., 1995).This study takes a first step toward addressing thatgap, by providing insight into more challenging,risky, and effortful forms of discretionary employeebehavior. It thereby broadens current conceptual-

Page 13: TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS …...1999 Morrison and Phelps 405 it entail efforts to implement solutions. Two other constructs related to taking charge are task revision

1999 Morrison and Phelps 415

izations of extrarole behaviors within organiza-tions, going beyond the more mundane cooperativeand helping behaviors that have been the focus ofexisting research. Our results indicate that takingcharge is conceptually distinct from these moretraditional forms of extrarole behavior, and theysuggest that it is motivated by factors that have notpreviously been studied in the context of thesemore traditional forms of extrarole behavior.

This study also contributes to research on extra-role behavior by developing and validating a mea-sure that can be used in subsequent studies. Weengaged in a rigorous and multistep process to de-velop a measure of taking charge. This process fo-cused on developing a representative set of activi-ties with high face validity. Steps were also takenduring the scale development process to demon-strate discriminant validity and, in particular, thatthe activities fell outside of what is generally ex-pected or required for the types of employees in-cluded in this study. An additional methodologicalstrength of this study was the use of coworker re-ports and the fact that for many of the respondents,we had data from two different coworkers. Manypast studies of extrarole behavior have relied uponself-reports, which tend to exhibit inflated correla-tions with predictor variables because of commonmethod bias (Organ & Ryan, 1995). Further, of thosestudies that have assessed extrarole behavior fromsupervisors' or peers' reports, most have reliedupon a single report per focal employee, whichmakes it impossible to determine whether thatsource is providing a reliable assessment of thefocal employee's behavior. Although the level ofagreement between coworkers in this study was notas high as we would have liked, it provides at leastreasonable evidence of interrater reliability.

In addition to their implications for research onextrarole behavior, the results of this study are alsorelevant to research on other forms of employee-initiated change. Although several scholars havehighlighted the importance of employee-drivenchange for ensuring organizational effectiveness(Ashford et al., 1998; Frese et al., 1996; Frohman,1997; Hornstein, 1986; Kanter, 1983; Scott & Bruce,1994; Staw & Boettger, 1990; Van Dyne et al., 1995),relatively little is known about motivating factors(Scott & Bruce, 1994). This study addressed thatgap in the literature by suggesting ways in whichorganizations can encourage employees to bringabout change in their workplaces. At a practicallevel, our results suggest that top management canencourage more employee-directed change by con-veying that they are open to recommendations frombelow and by behaving in a way that signals thisopenness. The results also suggest that in situations

where a high level of employee-initiated changeand innovation are desirable, organizations shouldtry to select employees with high levels of self-efficacy and felt responsibility, or alternatively, tryto develop those attributes among their employees.

This study also provides insight (albeit specula-tive) into the emergence of informal leaders withinorganizations. Kotter argued that a fundamentalfunction of leadership is "constructive or adaptivechange" (1990: 5). As such, individuals who takecharge can be viewed as demonstrating a form ofleadership, and our results suggest conditions un-der which they will be more likely to do so. Thisargument is significant since the term leadershipoften implies a particular hierarchical level androle—one in which the incumbent is not just ex-pected to initiate change, but also has legitimate au-thority and the power to do so. In contrast, individu-als who take charge are not formally expected toinitiate change and may undertake change initiativeswithout the benefit of formal authority. Individualswho are effective at taking charge are those who canexercise infiuence without relying upon mere posi-tion, a form of leadership that is becoming increas-ingly important as organizations become less hierar-chical and more reliant on horizontal networks,cross-functional teams, and informal influence.

Limitations and Suggestions for Future Research

The contributions of this study must be consid-ered in light of its limitations. One limitation is thatnot all of the individuals surveyed responded to thesurvey, and of those who did respond, not all pro-vided coworker names. It is possible that individ-uals who do not exhibit much taking charge behav-ior were less likely to complete the survey or lesslikely to provide coworker names. It is also possiblethat coworkers were less likely to respond if theemployee whom they were rating did not exhibitmuch taking charge behavior. Either of these dy-namics may have created selection bias. The highmean for taking charge suggests that we may haveindeed ended up with a sample of people dispro-portionately inclined to take charge, although thereis no way to determine this with certainty.

Another factor that may have contributed to thehigh mean for taking charge is that some of the focalemployees were in managerial positions, wheretaking charge is apt to be more common and ac-cepted. Our data provide some evidence of a linkbetween managerial position and taking charge.There was a small yet positive correlation betweentaking charge and our measure of organizationallevel (r = .13, p < .05). Given our focus on profes-sional/managerial employees, we recommend that

Page 14: TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS …...1999 Morrison and Phelps 405 it entail efforts to implement solutions. Two other constructs related to taking charge are task revision

416 Academy of Management Journal August

future studies test whether our model will general-ize to other populations, particularly populationswhere taking charge is likely to be more unusual(for instance, clerical or blue-collar employees).

Another limitation of this study is that in somecases where there were two coworker informants,those individuals provided divergent evaluationsof the focal employee's behavior. Although we tookdetailed steps to develop the measure of takingcharge, these divergent evaluations raise somequestions about construct validity and about thescale's reliability. We recommend that in futurestudies in this area, researchers obtain ratings frommultiple sources so that issues of interrater agree-ment can be more clearly addressed.

The discriminant validity of our measure of tak-ing charge should also be further assessed. Al-though taking charge is part of a large domain ofextrarole behaviors, we view it as conceptually dif-ferent from extrarole behaviors that do not chal-lenge the status quo. The results of our confirma-tory factor analysis support this viewpoint: takingcharge was shown to be related to, yet distinctfrom, the altruism and civic virtue forms of OCB.However, the fact that it was related to altruism andcivic virtue raises the question of whether the an-tecedents of taking charge that we identified areunique to taking charge. In other words, is therediscrimination among the nomological networks ofthe various forms of extrarole behavior?

Although w e did not collect data to address thisissue, we do have some indirect evidence that thenomological network for taking charge is distinctfrom the nomological network for OCB, which hasbeen the main focus within the extrarole behaviorliterature. There have been numerous studies ofantecedents to OCB, yet none has identified thevariables that we assessed in this study (Organ &Ryan, 1995). That suggests to us that these variablesare unlikely to be major determinants of nonchal-lenging forms of extrarole behavior. The anteced-ents to taking charge that we studied have beendiscussed in the context of other change-orientedbehaviors, because we intentionally drew upon re-search on related constructs in developing ourmodel. The presence of these antecedents in theliterature on related constructs raises concernsabout whether the set of antecedents that we stud-ied is equally predictive of other forms of challeng-ing behavior. Examining this issue was outside ofthe scope and intent of this study. However, werecommend future research assessment of the de-gree to which the nomological networks of variousextrarole behavior constructs are convergent or dis-criminable.

An additional limitation of this study is that.

because data were not collected over time, the re-sults fail to reflect the dynamic nature of the deci-sion about whether to take charge. We suspect thatthere are feedback loops in this process and that atany given point in time, this decision is heavilyinfluenced by the success of past efforts to initiatechange. Further, the independent variables affect-ing the decision about whether to take charge maychange as individuals attempt to take charge andare either successful or unsuccessful. These changesmay in turn affect an individual's level of persis-tence. For example, an employee who decides totake charge partially because of his or her evalua-tion that top management is supportive may decideto discontinue the effort if he or she learns that topmanagement views the change initiative the em-ployee is putting forth unfavorably. By obtainingmeasures of our variables at a single point in time,we obviously missed some of these temporal dy-namics. We encourage longitudinal research so thatthis dynamic process can be more fully understood.

Although this study provides a useful first steptoward understanding the construct of takingcharge, it is important to recognize that it assessedonly a few of the many variables that may encour-age this activity. Future research should focus onidentifying a broader set of predictors, especiallythose that are counterintuitive or novel. This initialstudy also leaves unanswered some importantquestions about the taking charge construct. Onesuch question relates to the organizational implica-tions of this activity. We have argued that, likeother forms of extrarole behavior, taking charge isorganizationally functional. Yet this argument is abit simplistic. In most cases, it is impossible topredict with complete certainty whether the out-comes associated with a given course of action willbe positive or negative. Thus, an action that isintended to bring about functional change may insome cases have a dysfunctional effect. Further-more, because organizations are characterized bymultiple stakeholders and multiple goals, an out-come regarded as positive from the perspective ofone stakeholder or goal may be viewed as negativefrom the perspective of another.

It must also be recognized that taking chargeentails behavior that deviates from prescribed rolesand, consequently, that it may be viewed as threat-ening by peers or supervisors. Thus, an employeewho is trying to bring about improvement mayactually incite disharmony and tensions that willdetract from performance. It is also possible thatexcessive amounts of taking charge will be dys-functional. Unfortunately, it is often hard to deter-mine the dividing line be.tween change-directedbehavior that contributes something valuable and

Page 15: TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS …...1999 Morrison and Phelps 405 it entail efforts to implement solutions. Two other constructs related to taking charge are task revision

1999 Morrison and Phelps 417

that which goes too far and thus results in thedestruction of a reasonably well-functioning sys-tem. An important task for future research is toidentify conditions under which taking charge islikely to harm rather than help an organization.Identifying these conditions will provide an evenricher understanding of this phenomenon.

REFERENCES

Ajzen, I. 1991. The theory of planned behavior. Organi-zational Behavior and Human Decision Processes,50: 179-211.

Ajzen, I., & Fishbein, M. 1980. Understanding attitudesand predicting social behavior: Englewood Cliffs,NJ: Prentice-Hall.

Ashford, S. J., Rothbard, N. P., Piderit, S. K., & Dutton,J. E. 1998. Out on a limb: The role of context andimpression management in issue selling. Adminis-trative Science Quarterly, 43: 23-57.

Asbford, S. J., & Taylor, M. S. 1990. Adaptation to worktransitions: An integrative approach. In G. R. Ferris &K. M. Rowland (Eds.), Research in personnel andhuman resource management, vol. 8: 1-39. Green-wicb, CT: JAI Press.

Bagozzi, R., & Heatberton, T. 1994. A general approacb torepresenting multifaceted personality constructs:Application to self-esteem. Structural EquationModeling, 1: 35-67.

Bentler, P. M., & Bonett, D. G. 1960. Significance testsand goodness of fit in analysis of covariance struc-tures. Psychological Bulletin, 88: 588-606.

Bentler, P. M., & Gbou, G. 1987. Practical issues in struc-tural modeling. Sociological Methods and Re-search, 16: 78-117.

Bollen, K. A. 1989. Structural equations with latentvariables. New York: Wiley.

Browne, M. W., & Gudeck, R. 1993. Alternative ways ofassessing model fit. In K. Bollen & J. S. Long (Eds.),Testing structural equation models: 136-162. New-bury Park, GA: Sage.

Bunce, D., & West, M. 1995. Self perceptions and percep-tions of group climate as predictors of individualinnovation at work. Applied Psychology: An Inter-national Review, 44: 199-215.

Burgelman, R. 1994. Fading memories: A process tbeoryof strategic business exit in dynamic business envi-ronments. Administrative Science Quarterly, 39: 24-56.

Gudeck, R., & Browne, M. W. 1983. Gross-validationof covariance structures. Multivariate BehavioralResearch, 18: 147-167.

Dutton, J. E., & Asbford, S. J. 1993. Selling issues to topmanagement. Academy of Management Review,18: 397-428.

Farrell, D. 1983. Exit, voice, loyalty and neglect as re-sponses to dissatisfaction: A multidimensional scal-ing study. Academy of Management Journal, 26:596-607.

Frencb, J. R. P., & Raven, B. 1959. Tbe bases of socialpower. In D. Gartwrigbt (Ed.), Studies in social pow-er: 150-167. Ann Arbor, MI: Institute for SocialResearch.

Frese, M., Kring, W., Soose, A., & Zempel, J. 1996. Per-sonal initiative at work: Differences between Eastand West Germany. Academy of Management Jour-nal, 39: 37-63.

Frohman, A. L. 1997. Igniting organizational change frombelow: Tbe power of personal initiative. Organiza-tional Dynamics, 25(3): 39-53.

Gerbing, D. W., & Anderson, J. G. 1993. Monte Garloevaluations of goodness of fit indices for structuralequation models. In K. Bollen & J. S. Long (Eds.),Testing structural equation models: 40-65. New-bury Park, GA: Sage.

Ghosbal, S., & Bartlett, G. A. 1994. Linking organizationalcontext and managerial action: Tbe dimensions ofquality of management. Strategic ManagementJournal, 15: 91-112.

Gist, M. E., & Mitchell, T. R. 1992. Self-efficacy: A theo-retical analysis of its determinants and malleability.Academy of Management Review, 17: 183-211.

Grabam, J. W. 1986. Principled organizational dissent: Atheoretical essay. In L. L. Gummings & B. M. Staw(Eds.), Research in organizational behavior, vol 8:1-52. Greenwich, GT: JAI.

Hackman, J. R. 1983. Group influences on individuals. InM. D. Dunnette (Ed.), Handbook of industrial andorganizational psychology: 1455-1624. New York:Wiley.

Harris, M. M., & Scbaubroeck, J. 1991. Gonfirmatorymodeling in organizational bebavior/buman re-source management: Issues and applications. Journalof Management, 16: 337-360.

Hattie, J. 1985. Methodology review: Assessing unidi-mensionality of tests and items. Applied Psycholog-ical Measurement, 9: 139-164.

Herche, J., & Engelland, B. 1996. Reversed-polarity itemsand scale unidimensionality. Journal of the Acad-emy of Marketing Science, 24: 366-374.

Hirscbman, A. O. 1970. Exit, voice, and loyalty: Re-sponses to decline in firms, organizations, andstates. Gambridge, MA: Harvard University Press.

Hornstein, H. A. 1986. Managerial courage. New York:Wiley.

House, R. J., & Rizzo, J. R. 1972. Toward tbe measurementof organizational practices. Journal of Applied Psy-chology, 56: 388-396.

Jaccard, J., & Wan, G. K. 1996. LISREL approaches to

Page 16: TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS …...1999 Morrison and Phelps 405 it entail efforts to implement solutions. Two other constructs related to taking charge are task revision

418 Academy of Management Journal August

interaction effects in multiple regression. In M. S.Lewis-Beck (Ed.), Quantitative applications in thesocial sciences, vol. 114. Tbousand Oaks, GA: Sage.

James, L. R. 1982. Aggregation bias in estimates of per-ceptual agreement. Journal of Applied Psychology,67: 219-229.

Joreskog, K. G., & Sorbom, D. 1989. LISREL 7: A guide tothe program and applications (2nd ed.). Gbicago:SPSS.

Joreskog, K. G., & Sorbom, D. 1993. LISREL 8: Structuralequation modeling with the SIMPLIS commandlanguage. Hillsdale, NJ: Scientific Software.

Kanter, E. M. 1983. The change masters. New York:Simon & Schuster.

Katz, D., & Kabn, R. L. 1966. The social psychology oforganizations. New York: Wiley.

Konovsky, M. A., & Pugb, S. D. 1994. Gitizenship bebav-ior and social exchange. Academy of ManagementJournal, 37: 656-669.

Kotter, J. P. 1990. A force for change: How leadershipdiffers from management. New York: Free Press.

Magazine, S. L., Williams, L. J., & Williams, M. L. 1996. Aconfirmatory factor analysis examination of reversecoding effects in Meyer and Allen's affective andcontinuance commitment scales. Educational andPsychological Measurement, 56: 241-250.

Miceii, M. P., & Near, J. P. 1992. Blowing the whistle.New York: Lexington Books.

Morrison, E. W., & Robinson, S. 1997. The developmentof psychological breach and violation: A longitu-dinal study. Paper presented at the annual meetingof tbe Academy of Management, Boston.

Nunnally, J. G. 1978. Psychometric theory (2nd ed.).New York: McGraw-Hill.

Organ, D. W. 1988. Organizational citizenship behav-ior: The good soldier syndrome. Lexington, MA:Lexington Books.

Organ, D. W. 1990. The motivational basis of organiza-tional citizenship bebavior. In L. L. Gummings &B. M. Staw (Eds.), Research in organizational be-havior, vol 12: 43-72. Greenwicb, GT: JAI.

Organ, D. W., & Ryan, K. 1995. A meta-analytic review ofattitudinal and dispositional predictors of organiza-tional citizensbip bebavior. Personnel Psychology,48: 775-802.

Pearce, J. L., & Gregersen, H. B. 1991. Task interdepen-dence and extrarole bebavior: A test of tbe mediatingeffects of felt responsibility. Journal of Applied Psy-chology, 76: 838-844.

Podsakoff, P. M., MacKenzie, S. B., Moorman, R. H., &Fetter, R. 1990. Transformational leader bebaviorsand tbeir effects on followers' trust in leader, satis-faction, and organizational citizensbip bebaviors.Leadership Quarterly, 1: 107-142.

Rusbult, G. E., Farrell, D., Rogers, G., & Mainous, A. G.1988. Impact of exchange variables on exit, voice,loyalty, and neglect: An integrative model of re-sponses to declining job satisfaction. Academy ofManagement Journal, 31: 599-627.

Salancik, G. R., & Pfeffer, J. 1977. Wbo gets power—andbow tbey bold onto it: A strategic-contingencymodel of power. Organizational Dynamics, 5(3):3-21.

Scbaubroeck, J., & Merritt, D. E. 1997. Divergent effects ofjob control on coping witb work stressors: Tbe keyrole of self-efficacy. Academy of Management Jour-nal, 40: 738-754.

Scbriesbeim, G. A., Eisenbacb, R. J., & Hill, K. D. 1991.Tbe effect of negation and polar opposite item rever-sals on questionnaire reliability and validity: An ex-perimental investigation. Educational and Psycho-logical Measurement, 51: 67-78.

Scott, S. G., & Bruce, R. A. 1994. Determinants of inno-vative bebavior: A patb model of innovation in tbeworkplace. Academy of Management Journal, 37:580-607.

Sberer, M., Maddux, J. E., Mercandante, B., Prentice-Dunn, S., Jacobs, B., & Rogers, R. W. 1982. Tbe self-efficacy scale: Gonstruction and validation. Psycho-logical Reports, 51: 663-671.

Sbrout, P. E., & Fleiss, J. L. 1979. Intraclass correlations:Uses in assessing rater reliability. PsychologicalBulletin, 86: 420-428.

Sitkin, S. B., & Pablo, A. L. 1992. Reconceptualizing tbedeterminants of risk bebavior. Academy of Manage-ment Review, 17: 9-38.

Speier, G., & Frese, M. 1997. Generalized self-efficacy asa mediator and moderator between control and com-plexity at work and personal initiative: A longitudi-nal field study in East Germany. Human Perfor-mance, 10: 171-192.

Staw, B. M., & Boettger, R. D. 1990. Task revision: Aneglected form of work performance. Academy ofManagement Journal, 33: 534-559.

Tbarenou, P., Latimer, S., & Gonroy, D. 1994. How do youmake it to tbe top? An examination of influences onwomen's and men's managerial advancement. Acad-emy of Management Journal, 37: 899-931.

Van Dyne, L. & LePine, J. 1998. Helping and voice extra-role bebavior: Evidence of construct and predictiveValidity. Academy of Management Journal, 41:108-119.

Van Dyne, L., Gummings, L. L., & Parks, J. M. 1995.Extra-role bebaviors: In pursuit of construct and def-initional clarity. In L. L. Gummings & B. M. Staw(Eds.), Research in organizational behavior, vol.17: 215-285. Greenwicb, GT: JAI Press.

Van Maanen, J., & Scbein, E. H. 1979. Toward a tbeoryof organizational socialization. In B. M. Staw (Ed.),

Page 17: TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS …...1999 Morrison and Phelps 405 it entail efforts to implement solutions. Two other constructs related to taking charge are task revision

1999 Morrison and Phelps 419

Research in organizational behavior, vol. 1: 209-264. Greenwicb, GT: JAI Press.

Vroom, V. H. 1964. Work and motivation. New York:Wiley.

Williams, L. J., & Anderson, S. E. 1991. Job satisfactionand organizational commitment as predictors of or-ganizational citizensbip and in-role bebavior. Jour-nal of Management, 17: 601-617.

Witbey, M. J., & Gooper, W. H. 1989. Predicting exit,voice, loyalty, and neglect. Administrative ScienceQuarterly, 34: 521-539.

Woodruff, S. L., & Gasbman, J. F. 1993. Task, domain,and general efficacy: A reexamination of tbe self-efficacy scale. Psychological Reports, 72: 423-432.

Elizabeth Wolfe Morrison is an associate professor at tbeLeonard N. Stern Scbool of Business, New York Univer-sity. Sbe received ber Pb.D. in organizational bebaviorfrom Northwestern University. Her researcb interests in-clude employee proactivity, extrarole bebavior, em-ployee voice, newcomer socialization and adjustment,and psycbological contracts.

Corey C. Phelps is a doctoral candidate in tbe Depart-ment of Management and Organizational Bebavior at tbeLeonard N. Stern Scbool of Business, New York Univer-sity. His current researcb interests include tbe micro- andmacrodynamics of organizational and tecbnological in-novation, interfirm collaboration, and tbe interface be-tween technology and firm strategy.

Page 18: TAKING CHARGE AT V^ORK: EXTRAROLE EFFORTS …...1999 Morrison and Phelps 405 it entail efforts to implement solutions. Two other constructs related to taking charge are task revision