Goal-Based Learning and the Future of Performance Management · and understandings that foster intelligent behavior (Senge 1990). Characteristics of a learning culture include high

Goal-Based Learning and the Future of Performance Management 203

Donald P. MoynihanTexas A&M University

Goal-Based Learning and the Future ofPerformance Management

All levels of government have begun to pursue results-based reforms, which assume that manag-ers will use performance information to make better decisions. However, reforms have neglectedthe insights of a large and relevant literature on organizational learning. This article revisits thisliterature, treating results-based reform as an organizational learning mechanism and a deliber-ate structural effort to induce learning. From an organizational learning perspective, most re-sults-based reforms target narrow process improvement (single-loop learning) rather than a broadunderstanding of policy choices and effectiveness (double-loop learning), even though the latteris more critical for long-term organizational success. Case evidence from state governments illus-trates single- and double-loop learning and the importance of two frequently neglected aspectsof organizational learning: learning forums—routines where performance information is delib-erately examined—and the role of organizational culture in enabling or limiting learning.

Recent efforts to make government more results-orientedhave been well documented (Melkers and Willoughby1998; Snell and Grooters 2001). A common element ofsuch reforms—whether they are known as performance-based budgeting, performance management, or managingfor results (MFR)—is the regular collection and dissemi-nation of performance data. Similar reforms have struggledin the past: The underlying assumptions of rational deci-sion systems proved unrealistic, and the habitual and pass-ing nature of reforms increasingly bred cynicism amongpublic employees (Downs and Larkey 1986; Wildavsky1975). Some observers suggest the current wave of reformfaces a similar fate. Reforms have occurred without theadditional, necessary changes to organizational structureand environment (Schick 2001). Meeting MFR reportingrequirements can become a control mechanism for demand-ing compliance from agencies rather than a tool for im-provement (Franklin 2000). At the federal level, the Gov-ernment Accountability Office (formerly the GeneralAccounting Office) reported that as the Government Per-formance and Results Act was implemented between 1997and 2000, the percentage of managers using performanceinformation declined significantly (GAO 2001, 29–38).

What can help the current wave of MFR reforms to avoidfailure? The organizational learning literature can help toidentify and remedy the problems of managing for results.1

The key assumption that underpins MFR is essentially alearning theory: Decision makers will learn from perfor-mance information, and, in turn, they will make better-informed decisions and improve government performance.However, MFR has failed to incorporate many of the les-sons about how learning occurs in organizations. This ar-ticle revisits this literature, applying it to case studies ofMFR efforts in the state governments of Virginia and Ver-mont to develop useful insights into performance manage-ment. First, this literature distinguishes two types of learn-ing. Most MFR efforts are targeted toward narrow processimprovement (single-loop learning) rather than broad un-derstanding of policy choices and effectiveness (double-loop learning), even though the latter is critical for long-term organizational success. Second, this literature distin-guishes cultural and structural approaches to learning. Thelearning literature would classify MFR reforms as a struc-tural approach that is likely to struggle until it incorpo-rates cultural aspects into learning forums.

Donald P. Moynihan is an assistant professor at the George Bush School ofGovernment and Public Service at Texas A&M University. He received hisPhD from the Maxwell School of Citizenship and Public Affairs at SyracuseUniversity. His research analyzes the process of selecting and implementingpublic management reforms, especially in the area of performance manage-ment. His work has appeared in number of edited volumes and journals,including Public Administration Review and J-PART. E-mail: [email protected].

204 Public Administration Review • March/April 2005, Vol. 65, No. 2

Linking Organizational Learning Theoryto Managing for Results

Argyris and Schön present two types of learning thatmay occur in organizations. Single-loop learning is “in-strumental learning that leads to improvement in the per-formance of organizational tasks,” that “changes strategiesof actions or assumptions underlying strategies in waysthat leave the values of a theory unchanged” (1996, 20). Inthe context of governance, single-loop learning is appro-priate for routine, repetitive operations, where public-sec-tor goals are clear and widely accepted. In terms of man-aging for results, it implies specifying goals to the pointwhere they are measurable; tracking the achievement ofgoals; and judging these results in the context of a point ofcomparison, whether it be pre-set targets, previous perfor-mance, the performance of other organizations, or otherparts of the same organization. Such comparison promptsa dialogue that analyzes the factors and processes that un-derpin performance and how they might be changed. Inshort, single-loop learning allows organizations to do thesame things better.

Double-loop learning is “learning that results in a changein the values of theory-in-use, as well as in its strategiesand assumptions.… Strategies and assumptions maychange concurrently with, or as a consequence of, changein values” (Argyris and Schön 1996, 21). In the context ofgovernance, double-loop learning occurs when public ac-tors test and change the basic assumptions that underpintheir mission and key policies. It is more relevant for com-plex, nonprogrammable issues that are important to theorganization’s survival rather than short-term efficiency.Double-loop learning means questioning the goals of aprogram and asking whether the program is worth pursu-ing—or worth pursuing in the public sector. In the contextof managing for results, it implies a willingness to revisitthe basic organizational mission, goals, and strategies on aregular basis.

The organizational learning literature emphasizes theimportance of organizational culture to learning (Schein1992). Learning is based on shared experiences, norms,and understandings that foster intelligent behavior (Senge1990). Characteristics of a learning culture include highemployee empowerment, participation, and discretion(Argyris and Schön 1996; Fiol and Lyles 1985).2 Lipshitz,Popper, and Oz (1996) criticize the abstract nature of thiscultural approach to learning. They argue that learning canbe better studied and promoted by undertaking a structuralapproach, which they call organizational learning mecha-nisms: “institutionalized structural and procedural arrange-ments that allow organizations to systematically collect,analyze, store, disseminate, and use information that is rel-evant to the effectiveness of the organization” (293). While

a cultural approach emphasizes creating shared and func-tional norms among workers, the structural approach de-notes a reliance on formal rules and procedures to enablelearning.

The structural approach to learning is attractive to re-formers because structure and procedure are aspects of anorganization that can be changed through formal mandates.A structural approach is also consistent with rational-ana-lytic theories of learning that emphasize collecting, stor-ing, and distributing data (Mahler 1997). It is clear thatmanaging for results, as it is currently being implemented,is closer to the structural rather than the cultural approachto learning. The structure comes from legislative and ad-ministrative mandates that create formal rules and proce-dures to generate, collect, and disseminate data (Radin1998; Snell and Grooters 2001). Strategic plans are cre-ated to specify goals, and data are measured and collectedto describe performance. Program and agency data are usu-ally generated and reported in a process that closely re-sembles—and is often part of—annual budget reportingprocedures to central agencies. For good measure, the dataare likely to be verified by an independent auditor. All suchperformance information is typically published ingovernmentwide and agency documents.

The types of mandates described above largely ignorethe role that organizational culture plays in enabling theuse of performance data. The procedural nature of suchmandates is reinforced by recommendations from profes-sional bodies for greater performance reporting—the Gov-ernment Accounting Standards Board, the National Acad-emy of Public Administration, and the National Councilof State Legislatures have all issued such calls. And per-formance-reporting mandates make sense to elected offi-cials: They are clear and specific reforms that elected offi-cials can adopt to demonstrate they care about results-basedgovernment. In contrast, cultural reform is slow, difficult,hard to observe, and largely shaped at the agency level(Wilson 1989).

A focus on structure is not the only characteristic sharedby organizational learning mechanisms and managing forresults. Both approaches share the assumption that infor-mation forms the basis for improved decision making. Bothrely on systems theory for a logical account of how infor-mation is converted into decisions (Senge 1990; Moynihanand Ingraham 2003). Both assume that routines of infor-mation collection and dissemination are followed by rou-tines of information use. Studies of MFR efforts in stategovernment question whether such routines of performanceinformation use occur. The Government PerformanceProject surveyed MFR practices of all state governmentsduring 1998 and 2000 and found that almost all states hadadvanced strategic planning and performance-measurementprocedures that had produced a lot of performance infor-


mation. However, evidence of actual use of this informa-tion was anecdotal rather than systemic (Ingraham andMoynihan 2001). The weakness of most state MFR sys-tems, therefore, lies between the points of disseminationof the data (which is done well) and use (the ultimate pur-pose, which is done poorly). These findings appear to bereplicated at the federal level (GAO 2001).

The gap between dissemination and use occurs partlybecause of an absence of routines in which data are exam-ined and interpreted—learning forums. Learning forumsare routines that encourage actors to closely examine in-formation, consider its significance, and decide how it willaffect future action. Such routines are unlikely to occur asan organic reaction to the simple provision of quantitativeinformation. Managers prefer to spend their time interact-ing with people and collecting oral data, not contemplat-ing quantitative data (Mintzberg 1975).

Leavitt and March (1990) point to routines as the criti-cal lever by which behavior can be adjusted. However,whether and how organizational actors decide to create andparticipate in routines depends on what they consider ap-propriate to their organization. This returns us to the issueof culture: Actors will learn if they have the information tolearn, but also if the organizational culture portrays rou-tines of data consideration as appropriate organizationalbehavior. The Government Accountability Office recentlytied the success of results-based management to creating“an evaluation culture through commitment to self exami-nation and learning” (GAO 2003, 3). Even if learning rou-tines can be established, the way that actors value, assess,and interpret information will be shaped by cultural biases(Mahler 1997). Structural approaches that ignore culturalaspects, as most MFR mandates and systems appear to do,are weakened.

There are other ways that culture and structure interact.The organizational learning literature sees high employeeempowerment, participation, and discretion as conduciveto learning. In contrast, punishment-oriented control sys-tems tend to discourage learning and lead to defensive re-actions (Argyris and Schön 1996). Centralized structuresreinforce past behaviors and make new learning more dif-ficult. The public sector has traditionally relied on central-ized controls on behavior, human and financial resources,and decision making, leading to goal displacement, trainedincapacity, and a decline in creativity and effectiveness(Klay 1994; Merton 1940; Osborne and Gaebler 1992).Even with the introduction of new rules and proceduresdesigned to facilitate learning, failure to remove the oldrules will thwart change (Mahler 1997). Hedberg (1981)points to the need to “unlearn” past behaviors before newlearning can take place. The ability to unlearn is most atrisk when old cognitive frameworks contradict new les-sons to be learned. Evidence suggests that, although state

governments have enthusiastically implemented strategicplanning and performance measurement, they have beenmuch slower to remove central controls on financial andhuman resources (Brudney, Hebert, and Wright 1999;Moynihan forthcoming). Forrester and Adams (1997) pointto the need for public management reforms to foster moreflexible and empowering organizations if they are to leadto organizational learning.

Managing for results has neglected another key aspectof the organizational learning field: the possibility ofdouble-loop learning. Single-loop learning appears to bethe only type of learning promoted in the MFR model thatis recommended to practitioners and realized in mandates:Bureaucrats will figure out ways to achieve organizationalgoals more efficiently rather than challenge these goals(Barnow 1992; Liner et al. 2001, 91–96, 99–135). Thepolicy-making aspect of strategic planning is rarely dis-cussed, and it can be conveniently overlooked if we as-sume that bureaucrats use the process only to clarify, notto challenge, preexisting goals, and that new goals will beadopted at the behest of elected officials and stakeholders.Occasionally there is an acknowledgement that data cancontribute to double-loop learning, but this usually occursin such narrow terms—should government be charged withproviding a service, or is it provided more cheaply by an-other sector?—that it does not appear to be a substantialdeparture from single-loop learning.

What explains this oversight? To a large degree it re-flects the constraints of normative assumptions about de-cision making in the public sector, that is, the hoary poli-tics–administration dichotomy. As a reform, managing forresults harkens back to the concept of the neutral adminis-trator, who seeks to implement the goals they are given byelected officials most efficiently. Managing for results cer-tainly does not promise to empower bureaucrats withpolicy-making authority or to question the nature of thegoals they pursue (Koteen 1989). While the dichotomy hassuffered many a battering, no reform has ever enhanced itspotential for adoption by explicitly rejecting its existence.The widespread success of MFR reforms is partly the re-sult of a willingness to reaffirm the dichotomy and to over-look how MFR may violate it.

The dichotomy is reinforced by the normative assump-tions of public choice, a theory that is associated with re-sults-based government (Boston 1991). Public choice seesmanaging for results as a means by which elected officialscan reassert control over administrators by setting goals inthe strategic planning stage and closely overseeing out-puts. If double-loop learning is to occur at all, it is the prov-ince of elected officials, not bureaucrats. However, the logicof the public-choice perspective questions whether any suchlearning can occur: Elected officials are motivated to winreelection, and therefore they will seek to change strategic


goals to match shifting public preferences, not because ofinformation about program performance.

Single-Loop Learning in VirginiaHere I analyze case studies to illustrate the problems

and possibility of learning in the context of managing forresults. I selected three state governments regarded as high,middle, and low performers in MFR—Virginia, Vermont,and Alabama, respectively.3 The case outcomes describedin this article are based on fieldwork and analysis under-taken in 2001, although in each state it was necessary toestablish a narrative of at least a decade to understand howthese outcomes occurred. In each state I examined the state-wide and agency procedures for MFR and the role of per-formance information in decision making. I controlled forfunction by focusing on corrections. Alabama’s MFR pro-cess was weakened by a lack of resources and support,and it provided little evidence that MFR had led to anytype of useful learning. Therefore, I discuss Alabama onlyto the extent that it provides inferences about the failure toinduce learning, and focus the analysis instead on Virginiaand Vermont. The characteristics and case outcomes forthese two states are summarized in table 1.

Virginia is a state that is widely regarded as a nationalleader in managing for results (GAO 1994). It has pioneeredthe use of a public Web site to maintain voluminousamounts of agency performance information. For all of thesophistication of the statewide performance informationsystem, interviewees conceded that for budget and otherpolicy decisions, performance information was rarely aninfluential factor—a finding that is true of Alabama andVermont also. Statewide actors lack the time, interest, in-centive, or functional expertise to examine the very spe-cific data emerging from agencies. If performance infor-mation provides relevant information on a politicallyimportant program, it may be considered, but largely tosupport preexisting policy positions. Not only is there toomuch information for statewide decision makers to use,there is not enough of it to substitute for function-specific

knowledge and experience. Learning that improves servicequality, efficiency, and effectiveness can occur only at theagency level, among those who are familiar with organi-zational processes and can understand the outputs a pro-cess will deliver (Bennett and Hill 2002).

Does managing for results promote single-loop learn-ing at the agency level? All three states had establishedperformance-information systems to provide the data nec-essary for such learning. All three corrections departmentshad strategic goals and matching performance targets; allreported actual performance on a regular basis. However,there is little evidence that they regularly evaluated thisknowledge in the search for improved alternative organi-zational processes, thereby crystallizing the lessons ofsingle-loop learning. Virginia made the greatest progress,but a common weakness in all three states was their failureto develop regular learning forums among target users ofperformance data.

The Department of Corrections in Virginia provides thestrongest example of learning forums that led to the evalu-ation and use of performance information. A series of 1996strategic planning meetings among senior managers wasused to draw attention to critical organizational issues.Performance information first drew attention to issues, butit also provided a metric by which to judge and comparealternative processes. Staff embarked on what they referredto as benchmarking excercises, creating teams that searchedfor ways to deal with these issues from other organiza-tions regarded as national leaders, or from their own em-ployees. The experience resulted in changed practices inthe critical areas, changes that the department’s managersregarded as successful in terms of cost and performance.

MFR did promote single-loop learning in Virginia. How-ever, it is also clear there was a deliberate and largely suc-cessful effort to foster MFR not just through a series ofplanning and reporting procedures, but also through build-ing a results-oriented organizational culture. Departmentleadership decided to institutionalize MFR and other or-ganizational-improvement techniques through a revitalizedmanagement-training program. Managers reported the

Table 1 Case Characteristics and Outcomes

Case outcomes Virginia VermontSingle-loop learning

Double-loop learning

View of managing for results

Learning forums

Role of culture

Major challenge

Occurred in a number of instances and among avariety of organizational actorsDid not occur

Management tool to promote and demonstrateperformance improvementUsed throughout the organization, as in benchmarkingteams; promoted by managing for results requirements,but without formal routine or structure; focus on processchangeTied performance improvement to creating anemployee-centered, mission-based cultureFurther formalizing of performance-improvement efforts

Occurred to a limited degree, but without frequency

Occurred among agency leadership, which successfullypromoted dramatic policy changeLearning metaphor to understand and communicatepolicy choices and outcomesUsed infrequently by high-level actors for major strategicdecisions (strategic visioning) and annually for program-matic decisions

Open and experimental culture among leadership;challenged institutional culture of staffGaining buy-in from institutional staff on new policy goals


training helped to reshape the organizational culture byleading to a shared ethos of organizational improvementand a willingness to examine existing procedures to seekmore effective ways of doing things. Frequently, thechanges that resulted did not employ rigorous benchmark-ing of cost and performance data, but were based on thelogical advantages of a new procedure and the experienceand reputation of the source of the innovation. The 125most senior managers continued to meet once a year, shar-ing stories about organizational successes in achieving stra-tegic goals, and institutional staff report to monthly meet-ings where performance data is discussed. On an informalbasis, employees find time to share ideas, and they reportthat suggestions are frequently adopted as organizationalstandard operating procedures.

Such learning forums foster process change through thedissemination of innovations, taking advantage of theknowledge of the change agent and the audience. Suchknowledge is difficult to encode into a purely quantitativebenchmarking excercise, which might neglect implicit stan-dards that employees are constantly aware of, such as staffsafety. The change agent is likely to have experimentedwith different options before settling on the one that he orshe regards as innovative enough to promote. The recipi-ents also have a knowledge base that allows them to con-sider how an innovation will fit into their organizationalenvironment.

The ethos of performance improvement was also fos-tered by two other decisions made during the 1996 strate-gic planning meetings. The first was to create a sense ofempowerment among employees, at least within the stateconstraints the Department of Corrections had to obey.Lower-level employees would be increasingly involved infuture waves of strategic planning, and employees were toldto expect greater receptivity to their ideas than in the past.The second decision was to emphasize the centrality ofemployees to the organization’s mission. A focus on em-powerment fitted an organizational culture where the cen-tral managers tried to portray themselves as acting to sup-port field staff rather than conflicting with them. More thanhalf of the organizational vision statement that emerged fromthe 1996 meeting is given to the critical role of employees:“The employees of the Department are the cornerstone ofthe agency. They share a common purpose and a commit-ment to the highest level of professional standards and ex-cellence in public service. The Department, through itsunwavering commitment to its employees is a satisfyingand rewarding place to work and grow professionally.”

This is employee-centered MFR, an effort to simulta-neously build a culture that is characterized by a resultsfocus, a sense of organizational unity, and employee sup-port and empowerment. The department’s leadership be-lieved that in a people-intensive industry such as correc-

tions, improved performance depends on the active engage-ment of employees and the internalization of organizationalgoals—an outcome that is better generated by voluntaryrather than coerced actions, and therefore dependent onorganizational culture. The focus on culture was partly areaction to a similar strategic planning effort during the1980s, which resulted in “a nice document that I’m notsure anybody in particular paid attention to,” according toone manager.

The Department of Corrections sought to shape cultureby using symbols, concrete evidence of support and com-munication. Symbols include the vision statement, whichwas prominently displayed in every correctional facilityin the state, along with a mission statement and list of goals.Concrete benefits include an improved retirement system,improved safety through a revised classification system,and more days off by moving to 12-hour work shifts. Im-proved communication occurred primarily through train-ing and was the most frequently cited benefit arising fromMFR in Virginia. Better communication led to a greatersense of organizational unity, purpose, and especially cross-divisional cooperation between the traditionally separateinstitutional and community programs. There is nowgreater information sharing between institutions and com-munity staff about prisoners on probation or parole, andprocedures for preparing prisoners for release have beenimproved. In surveys, employees have largely shown sup-port for organizational changes. One staff member sum-marized, “He [former commissioner Ron Angelone] hasconvinced the corrections officers that they’re important.He has partnered with them from day one. Those folksknow that somebody’s looking out for them. Now I callthat a major policy of investing in the line people. Andthat’s been dramatic.”

Double-Loop Learning in VermontDespite its relative success, managing for results in the

Virginia Department of Corrections did not induce double-loop learning. The initial round of strategic planning in1996 did not challenge the existing organizational goalsand values. Instead, it reinforced the traditional values ofsafety and incarceration while emphasizing the key role ofemployees. In Vermont, by contrast, the primary benefit ofMFR was double-loop learning: questioning the basic out-comes of corrections and convincing the external environ-ment of the benefits of an alternative approach. How didthis come about? In November 1991, shortly after JohnGorczyk became commissioner, the Vermont Departmentof Corrections published a 20-year plan outlining many ofthe principles that would form the basis of policy docu-ments, presentations, budget requests, and other strategicplans over the following decade.


The principles resulted from a decade of searching for aguiding philosophy within the department. Gorczyk andother senior managers had been involved in this search,which was framed in the context of previous correctionpolicies in the state and the prevailing trend toward a morepunitive approach. During the late 1960s and 1970s, theVermont Department of Corrections focused on commu-nity treatment, with a minimal emphasis on incarcerationand a strong faith in rehabilitation. This approach cameunder criticism because it ignored the real dangers posedby violent offenders who repeat their behaviors. It also ig-nored the politics of public safety. During the 1980s and1990s, legislators in Vermont and across the nation reactedto public safety concerns through tougher sentencing, apunitive approach. Department leadership recognized thefailings of the purely rehabilitative approach, but correc-tions research and their own analysis led them to believethat increased incarceration had negative outcomes for themajority of the corrections population and simply cost toomuch to be a practical long-term policy.

The 1991 plan outlined a series of principles that incor-porated elements of both the rehabilitative and the puni-tive model, as well as some new ideas that became knownas “restorative justice.” Restorative justice provided theDepartment of Corrections with a coherent philosophy andits own language, values, and outcomes: victim-based ratherthan offender-based, restorative rather than just retributiveor rehabilitative.4 In searching for an alternative philoso-phy, the department conducted what it called market re-search, undertaking surveys and focus groups of the pub-lic and individual stakeholders in 1993. The results foundopposition to some aspects of the existing correctionspolicy, support for others, and guidance for new ways ofthinking about corrections. The market research found thatpeople in Vermont wanted a criminal justice system thatcould do the following:• Increase safety without inflicting needless harshness or

deprivation• Offer treatment when it is likely to work• Hold offenders accountable for their actions, making

them repair the damage they cause• Return value to the victim and community• Avoid spending money on incarcerating those who offer

little threat to society• Offer greater opportunity for public involvement in the

criminal justice process.While one part of the market research was based on find-

ing out what customers wanted, another part was to bettercategorize its inputs and outputs. The Department of Cor-rections began to track in great detail different aspects ofcorrections and the wider criminal justice system: crimerates, sentencing, and differentiating prisoners by the typeof crime and individual characteristics. With these find-

ings, the department identified a mandate to reshape itsown operations and, to the extent possible, the wider crimi-nal justice system. Since the early 1990s the departmenthas used this philosophy as the basis for a number of inno-vative criminal justice programs.

It is too simplistic to suggest that managing for resultswas the factor that enabled double-loop learning. Clearlyother factors were at work, most prominently the agnosticpredisposition of the department’s leadership toward thetraditional and prevailing corrections model, partly becauseof the social science training (rather than criminal justicetraining) of many senior managers.5 This culture of open-ness to new ideas among the departmental leadership ledto a willingness (and need, given limited resources) tosearch for alternative goals and policies. It is not too much,however, to suggest that performance information bothshaped the policy outcomes that emerged and allowed theDepartment of Corrections to convince others of the le-gitimacy of these outcomes. Commissioner Gorczyk ismore assertive than this: “It’s [MFR] enabled us to signifi-cantly change the nature of the delivery of correctionalservices in Vermont. Without strategic planning I don’tbelieve in any way we could have gotten there.”

Learning MetaphorsWhat the corrections department’s leadership needed

was a tool to help them structure their questions, their searchfor solutions, and, ultimately, to persuasively communi-cate these solutions. They employed a series of “meta-phors,” of which MFR was just one.6 These metaphors pro-vided a means by which the Department of Correctionsstaff and their stakeholders were offered a credible andalternative perspective to challenge the otherwise domi-nant “mental model” about the corrections function (Senge1990). The credibility of the metaphors comes from theirassociation with a respected body of knowledge or way ofthinking and operating. The insights derived from the useof metaphors are more likely to be accepted by an audi-ence that places a value on the metaphor. Such knowledgecannot be classified as simply ideological or radical—likelythe reception that the restorative justice philosophy wouldhave received if it had been presented in the absence of anarrative that included the use of metaphors.

The use of metaphors facilitated the questioning of ex-isting technologies and the purposes of corrections, and itallowed the Department of Corrections to classify thesetechnologies as ineffective and traditional corrections out-comes as ideological and counterproductive. For instance,a departmental analysis showed the punitive approach toincarceration, in aggregate, led to reduced public safety,as those who entered institutions have higher rates of re-cidivism than those placed into alternatives to incarcera-tion, even after controlling for the risk profile of prisoners


(VDOC 2000b). One manager contrasted this with the pre-vious rehabilitative approach, which was “based on their[previous Department of Corrections staff] belief, not sci-ence, that prisons were bad and that people could growand learn if they were instituted in the community and treat-ment was provided.” The current departmental leadershipportrayed itself as scientific and rational, particularly withits risk-management approach.

MFR was not the only metaphor used. Business andscience were also used in conjunction with MFR. The busi-ness metaphor was particularly appropriate for communi-cative purposes during a time when the state was search-ing for more businesslike ways of government. Thismetaphor described the criminal justice system as a mar-ketplace, where demand for services was growing withoutan equivalent increase in resources. By identifying andinterviewing the customers of the criminal justice system—the public and corrections stakeholders—the Departmentof Corrections argued this demand was artificially createdby legislators and did not reflect what customers actuallywant or were willing to pay for. The market research es-sentially told legislators, “we have talked to the people,and what they want is different from what you are givingthem.”

All of the metaphors share the idea of logical, rational,nonideological judgment of performance based on empiri-cal analysis. Strategic visioning—identifying the mission,vision, and principles of the organization—was used as aforum in which information could be codified into a set ofphilosophical statements. In these forums, participantsmake sense of the information derived from the knowl-edge metaphors and relate it to the organization’s opera-tions and goals. As a process, strategic visioning occurredon a fairly frequent basis during the 1980s and 1990s, in-volving senior staff and intensive debate. As time passed,consensus developed on high-level strategic goals. Thefocus of the debate became specific programmatic plan-ning and the reporting of goals and performance measures,which occurs annually as part of performance-based bud-geting requirements.

Double-Loop Learning and Policy ShapingHow did the process of double-loop learning lead to new

policies? Prompted by the new departmental vision, theDepartment of Corrections leadership was responsible fordeveloping a set of new policies that sought to divert non-violent offenders from incarceration and convincing thepolitical environment to adopt these policies. Departmen-tal analysis found that more punitive sentencing was theprimary factor behind the increased inflow of offendersinto the criminal justice system (VDOC 2000b). One op-tion for dealing with prison overcrowding is to find waysto put offenders back into the community more quickly, or

to divert them from prisons in the first place. The depart-ment pursued this option in a number of ways. Probationand furlough jumped dramatically from the early 1990s.In 1994, the Department of Corrections created other in-termediate, community-based sentencing options to incar-ceration. These options were more intrusive and expen-sive than probation, offering less due process formisbehavior, but less expensive than prison. Intermediatesentencing options have been used increasingly as an al-ternative to incarceration and as a way of controlling thesize of the prison population.7

As new and traditional sentencing alternatives to incar-ceration reduced the demand for prison beds, the Depart-ment of Corrections developed techniques and programsto differentiate the corrections population based on theseverity of the crime and potential to reoffend. The depart-ment dubbed this a “risk-management approach,” wherebythe most expensive service, incarceration, is reserved foroffenders that pose the most threat to society. The approachis based on the assumption that it is possible to classifyoffenders by their threat to society, where threat is calcu-lated by multiplying the severity of offense by the likeli-hood of reoffense. According to the threat level of eachoffender, there is a logical response for their treatment inthe criminal justice system. This risk-management ap-proach portrays itself as pragmatic and scientific rather thanideological, relying on analysis of the corrections popula-tion and desired outcomes. It is the result of “the prag-matic judgment of whether the offender can safely re-en-ter society and be a productive citizen without intensiveState intervention. The judgment combines both the sci-ence of risk assessment and the experience of the correc-tions professional” (VDOC 2000b). In addition to appear-ing pragmatic, the risk-management approach also has thevirtue of being cost-efficient, offering a means of safelyreducing the expenses of incarceration.8

The Department of Corrections created reparative boardsin 1994. The program began by targeting low-risk, low-severity offenders who had pled guilty and received courtsentencing for nonviolent crimes, such as driving underthe influence, petty theft, or disturbing the peace. Repara-tive boards are made up of community volunteers appointedby the Department of Corrections, and board membersencourage offenders to accept responsibility for theircrimes. Offenders first tell their story, and the boards fre-quently organize meetings between offenders and victims.Contracts are created in which the offender and boardmembers agree on appropriate reparations to the commu-nity and to the victims. Offenders who fulfill their con-tracts to the boards’ satisfaction are recommended for ter-mination of probation.

Sixty reparative boards operate in 35 Vermont commu-nities, involving hundreds of citizens. The recidivism rate


is lower for offenders in the program than for those onprobation for similar offenses, and a portion of those whogo before reparative boards would otherwise be incarcer-ated. In terms of the size of the overall corrections popula-tion, reparative boards represent a small but growing frac-tion. However, the Department of Corrections has devoteda high amount of energy in creating the boards, viewingthem as the most tangible representation of the restorativejustice philosophy. The department’s strategic plan states,“Restorative justice focuses on the restoration of the vic-tim, the repair of the community, and the reintegration ofthe offender as a productive citizen who has acknowledgedresponsibility for his offenses, made amends and repairedthe damage. The restoration is done by the community, inthe community, with the victim as partner” (VDOC 2000b).

The success of the original reparative boards enabledthe Department of Corrections to extend such policies. Awave of market research conducted in 1999–2000 foundpublic support for a subsequent expansion of the programto include all types of violent offenders, particularly sexoffenders, who had participated in treatment programs. Theresearch found the public wanted to be involved in thesedecisions, but also wanted the expertise of traditional gov-ernmental actors as a support, particularly law enforcementofficials and community-based social services. To meet thisdemand, the department established community restorativejustice centers in major towns and restorative programs insmaller communities, and it also offered training to volun-teers. The centers provide a venue for reparative boards,as well as conflict resolution and neighborhood forums.Neighbors and officials can come together more easily,identify their particular capacity and concerns, and buildpartnership solutions. For example, a center in Burlingtoncreated a program for volunteers to help victims of prop-erty crime and vandalism.

Managing for Results in Shaping theEnvironment

The Vermont case illustrates an organization that hasbeen highly successful in controlling and shaping its ex-ternal environment. The Department of Corrections enjoyednot only increased financial support as the offender popu-lation increased, it also reshaped sentencing and releasepolicies in ways that were at odds with the stated prefer-ences of elected officials. In arguing for restorative jus-tice, the department developed a standard narrative thatpoints to the problems of overcrowding and excess demandon corrections, the causes of this problem, and the benefitsof the innovative, pragmatic, and cost-effective optionsinspired by a restorative justice approach. In making itscase, the Department of Corrections relied heavily on as-pects of MFR—public surveys, strategic goal setting, popu-lation trends, performance data and targets, and charts—

to make these policy preferences appear logical, cost con-scious, pragmatic, and nonideological. One senior man-ager admitted that “data has allowed us to establish sig-nificant credibility with the legislature, and traditionallyhave more influence on the overall criminal justice systemand its practices than we would have absent that informa-tion.” The narrative is calculated to impress that the de-partment has thought long and hard about corrections, hascarefully and rationally investigated every aspect of thisfunction—relying on facts rather than ideology—and hasdeveloped solutions with demonstrably positive outcomes.

The Department of Corrections had a high degree ofsuccess in convincing the three branches of the state gov-ernment and the public that a punitive approach to correc-tions is wrong, that incarceration is largely wasteful andcounterproductive for most offenders, and that rehabilita-tion is sometimes possible and should be pursued. Thesearguments ran counter to the prevailing criminal justicepolicy trends in corrections throughout the country and inVermont (Economist 2002). Proponents of these argumentshave had limited success elsewhere and have been seen assoft on crime. Corrections officials in Vermont were largelyable to avoid this characterization by emphasizing publicsafety and arguing for increased incarceration for violentoffenders.

The Department of Corrections also argued for com-munity involvement to reduce the distance between citi-zens and the justice system, a move that helped it to buildwider and deeper public support for its mission through-out the 1990s. In Vermont, as elsewhere, corrections wasan unpopular function. The only groups that cared about itdeeply—offenders and their families—were not politicallyinfluential. Surveys of the public and particular focusgroups provided a basis for building a constituency by find-ing out how the department could improve the public per-ception of corrections. The emphasis on public safety, re-parative boards, and the wider restorative justice movementproved popular with Vermont residents.9 In addition, thedepartment has built strong support from stakeholders withhigh political legitimacy, such as victims groups and a con-stituency that the department actually created, the mem-bers of the reparative boards.

As the Department of Corrections developed new prin-ciples and goals, it created performance measures to re-flect these changes. Most of the traditional performancemeasures for corrections—recidivism, escapes, disciplin-ary infractions—represent negative outcomes, that is, out-comes for which improved performance means reducingthe incidence of the measure. Negative outcomes, by theirvery nature, do not gain public or elected official attentionunless they are moving in the wrong direction. The publicdoes not cheer when zero escapes occur, but it does be-come upset if many prisoners escape (Barker 1998). De-


partment officials believed that positive measures wouldhelp to convince the public and the legislature of the im-portance and benefits of corrections (for instance,postincarceration employment, hours or dollar value re-turned to the community).

The support of the public and stakeholders has been criti-cal in winning over elected officials. In 2000, the state leg-islature added the principles of restorative justice to itscriminal justice policy: “It is the policy of this State thatprinciples of Restorative Justice be included in shapinghow the criminal justice system responds to personscharged with or convicted of criminal offenses. The policygoal is a community response to a person’s wrongdoing atits earliest onset and a type and intensity of sanction tai-lored to each instance of wrongdoing” (28 V.S.A. Sec 2a).This followed a four-year period during which a speciallegislative committee had been created on the future of thecorrections system. The Department of Corrections targetedthe committee, providing it with the standard narrativeabout corrections and reasons why the restorative philoso-phy worked.

Structure and Culture in Managingfor ResultsIncorporating Learning Forums

This article has argued that designers of MFR systemsneed to take routines that consider and discuss data (learn-ing forums) as seriously as they do the routines that col-lect and disseminate data. Without such forums, MFR isan insufficiently specified structural approach to learning.As an organizational learning mechanism, MFR has beenwidely if imperfectly adopted. Almost all states have simi-lar procedures in place that guarantee the collection andreporting of performance data, but they have struggled toestablish new routines to ensure data is carefully evalu-ated. Even in Virginia, the primary learning forums are abyproduct of strategic planning procedures and training,not the conscious effort of organizational design. Strate-gic planning is an opportunity to contemplate the goalsand performance of the past, but it has disadvantages as alearning forum. First, the primary purpose of strategic plan-ning is goal setting, not process change. Strategic plan-ning further limits single-loop learning by frequently ex-cluding lower-level managers whose input can informprocess change. Virginia partially overcame these prob-lems by building process-review elements into its depart-mental strategic planning and by encouraging lower-levelstrategic planning at the regional and institutional levels.Does strategic planning work as a double-loop learningforum? It should, but instances such as Vermont are rare.Even in setting goals, most organizations limit their po-

tential to learn by implicitly or explicitly making any di-rect challenge to the existing organizational goals tabooand inappropriate.

The standard assumption that performance informationwill automatically become a factor in existing decisionprocesses continues to shape MFR system design. Somedecision routines are amenable to being substantially reor-dered to facilitate the use of data, such as contracting withproviders of public services (Andrews and Moynihan 2002;Heinrich 1999). Budget and other policy decision routinesappear to be largely impervious to the use of performancedata in anything other than a marginal or advocacy role(Bennett and Hill 2002; Joyce and Tompkins 2002). In suchcases alternative learning forums should be developed.

The organizational learning literature and case studiesoffer insights on various aspects of learning forums in termsof the nature of the forum, who is involved, how partici-pants relate to one another and the role of performanceinformation (see table 2).

Table 2 Elements of Learning Forums

• Routine event• Facilitation and ground rules to structure dialogue• Nonconfrontational approach to avoid defensive reactions• Collegiality and equality among participants• Diverse set of organizational actors responsible for producing the

outcomes under review• Dialogue centered, with dialogue focused on organizational goals• Basic assumptions are identified, examined, and suspended (especially

for double-loop learning)• Quantitative knowledge that identifies successes and failures, including

goals, targets, outcomes, and points of comparison• Experiential knowledge of process and work conditions that explain

successes, failures, and the possibility of innovation

Leavitt and March (1990) emphasize the need for suchlearning as a routine matter. The key characteristic of suchforums is to create dialogue, which is considered a precur-sor to learning. Senge points out that dialogue allows par-ticipants to examine their own thinking and create com-mon meaning. Dialogue gives managers an opportunity topractice, experimenting with decisions styles in a way thatis not feasible in practice. He suggests the necessary as-pects of successful dialogue include the suspension of as-sumptions, facilitation that explains and enforces theground rules for dialogue, the active involvement of mem-bers, collegiality among participants, and a willingness formembers to raise the most pressing organizational issuesfor dialogue (Senge 1990, 238–49, 259–60). Argyris andSchön (1996) discuss the dangers of confrontational usesof data, leading to defensive reactions rather than to learn-ing; therefore, an open and nonconfrontational approachamong colleagues is preferable to a top-down analysis offailure. Lipshitz, Popper, and Oz (1996) point out the ben-efit of an equal footing for all members of a learning fo-


rum. Collegiality not only defuses defensive reactions, italso encourages the sharing of information about successes.

Performance data highlight the relative success or fail-ure of a unit or process (relative to the past, a target, an-other part of the organization, or a different organization),but only dialogue can help to identify and disseminate thereasons why success occurs. Interviewees from Virginiaand Vermont noted that a sense of professional pride is atleast as strong a motivator as high-level demands for per-formance. Mintzberg (1994) argues that forum participantsshould be the decision makers whom reformers hope willuse performance information, not central planners whoproduce formal policy evaluations that are removed frommanagerial realities. For single-loop learning this meanslower-level employees who oversee organizational pro-cesses, and for double-loop learning it means more senior-level employees who understand the entire organizationand its environment. Kaplan and Norton (1996) empha-size a team-based approach that links dialogue to criticalorganizational goals. Learning is enabled by “a team prob-lem-solving process that analyzes and learns from the per-formance data and then adapts the strategy to emergingconditions and issues” (252). Such analyses should be opento both quantitative and qualitative evidence, includingcorrelations between process changes and intended out-comes, in-depth management gaming/scenario analysis,anecdotal evidence, interim review of the progress of pro-cess changes, and peer review of performance. Both theVirginia and Vermont cases illustrate how managersbrought different types of information to the table: some-times quantitative data, but sometimes experiential infor-mation or new policy ideas. Dialogue among peers sub-jected all types of information to questioning and review.Kaplan and Norton also argue that diversity of expertiseimproves the potential for team learning, and they advo-cate teams that are both cross-functional and mix seniorand operational managers, as Virginia did. They also warnthat routine learning forums based on performance datamaintain a constant risk of devolving into dialogues thatare exclusively about operations rather than strategy.

The Inescapable Role of CultureWhat these cases make abundantly clear is that the suc-

cess of structural learning procedures is inextricably boundup with the culture of the organization. The Department ofCorrections in Alabama sought to implement MFR with-out changing any other aspect of the organization. As aresult, MFR is simply considered a reporting mechanism,the sum of its reporting mandates and not a means to learn.The performance-improvement ethic in Virginia found itsway to lower-level supervisors in institutions because thosesupervisors considered it a shared norm of expected be-havior—they followed the logic of appropriateness estab-

lished in their culture. In Vermont, senior managementsought metaphors that acted as alternative frameworks forconsidering their function—a cultural approach to learn-ing, but one in the context of a structured MFR process.The departmental leadership group was able to engage indouble-loop learning because of a culture of openness andexperimentation. However, the department struggled to gainthe degree of buy-in among institutional staff that Virginiaenjoyed. To a large degree this was because the VermontDepartment of Corrections undertook a more ambitioustask, seeking to challenge the existing goals of the organi-zation and emphasizing nonincarceration programs. Un-derstandably, this was met with some resistance by prisonofficers, whose daily existence is shaped by the demandsof keeping prisoners safely incarcerated (Shtull 1999).10

In contrast, Virginia did not seek to change its organiza-tional goals, but to increase effectiveness toward goals thatstaff were familiar with and supported.

Learning, Authority, and the Tradition ofManagement Controls

Closely related to a learning culture are a series of struc-tural constraints that limit managerial authority, hence thepotential for learning-based change. Senge notes that “Frus-trations appear to occur in settings where teams seeking todevelop their learning capabilities lack the power to act inthe domains about which they are learning” (1990, xvii).The potential for this type of frustration is particularly pro-nounced for public managers who are subject to the deci-sions of elected officials and other external actors.11 Pub-lic bureaucracies became associated with rules and culturesthat have traditionally restricted the scope of managerialaction, fostering learning in terms of rule observance orthe entrepreneurial avoidance of rules. A shift to goal-basedlearning is difficult, given the continued existence of theold culture and rules alongside results-based reforms. Evenin Virginia, where the Department of Corrections attemptedto create a sense of empowerment, there remained state-imposed legal restrictions on uses of financial and humanresources. The case evidence suggests a circular relation-ship between learning and the authority to initiate changeand the frequent mismatch of these qualities for the pur-poses of performance improvement. Performance enhance-ment is based on learning, but the willingness to first en-gage in learning is influenced by whether managers believelearning can be put to good use. The manager who be-lieves learning from performance information is unlikelyto be used because of controls limiting authority is lesslikely to engage in the process of creating forums for learn-ing in the first place.

The mismatch between learning and authority is par-ticularly pronounced for single-loop learning and processimprovement. Those who have the time, interest, and ex-


pertise to examine information and to make well-informedjudgments are likely to be lower-level managers, who lackthe authority to make the appropriate changes. Senior man-agers or elected officials with high authority are likely tolack the interest, motivation, and operational expertise toconsider and make informed judgments about specific pro-cesses. As a result, performance information is likely toremain unused, potential learning opportunities untaken,and ineffective managerial processes unchanged. One re-sponse, pursued in the three cases, is to provide more andmore information to elected officials and senior managersin the hope they will take an interest in management is-sues. Ultimately, however, this response does not changethe realities of limited time, interest, and expertise that re-strict higher-level officials from engaging in such decisions.

The Effect of Performance-Reporting MandatesThe factors that permit learning from performance

data—time, interest, expertise, and contextual knowl-edge—suggest that MFR should be an agency-centeredorganizational learning mechanism. However, MFR hasbeen adopted as a series of state-level requirements thatapply equally to all agencies (Snell and Grooters 2001).The heavy reporting burdens and the one-size-fits-all na-ture of these requirements may induce a sense of compli-ance and frustration among agency staff (Franklin 2000).Alabama is typical of many states, where performance in-formation is required as part of the annual budget report-ing process. The Department of Corrections reacted bytreating MFR as a reporting mandate rather than as an op-portunity to learn. The already overworked departmentalbudgeters were given responsibility for meeting this re-quirement. Unsurprisingly, the budget staff saw MFR as atool to argue for increased resources, not as a tool to changemanagement practices. Once the department completed itsreporting requirements, managers were not involved in —or from their point of view, bothered by—MFR until thenext reporting cycle.

Such a dismal picture was not repeated in Vermont andVirginia, which also have reporting mandates. In part thiswas because both states enjoyed greater organizational re-sources that could be applied to supporting MFR with train-ing programs and specialized staff support. It was also be-cause the corrections leadership saw MFR mandates as anopportunity to further the organizational agenda they werepursuing. The Department of Corrections in Vermont hadbeen effectively undertaking MFR long before all stateagencies were required to do so, first using MFR as a toolto question existing goals, and then to promote the restor-ative justice philosophy. MFR in the Virginia Departmentof Corrections had been abandoned in the 1980s until a1996 requirement to undertake it once more. The leader-ship saw this as an opportunity to examine the critical is-

sues facing the organization and a mechanism to imple-ment change.

Statewide MFR mandates, therefore, may be a mixedblessing. They create a real reporting burden for agen-cies, and statewide officials appear unlikely to actuallyuse the information presented to them. But they can alsoaid managers who are already intent on pursuing change.As Weick (2001) notes, organizational interpretation isshaped by senior-level managers and then transferred tothe rest of the organization. In Vermont and Virginia, MFRprovided a tool to engage in interpretation and dissemi-nate the results.

The Political Context of Double-Loop LearningDouble-loop learning, whether it occurs among elected

officials, political appointees, or bureaucrats, is the divi-nation of dramatic policy alternatives. It is a political act.It requires dealing not just with matters of uncertainty,which additional information can resolve, but also reduc-ing ambiguity, which demands both interpreting informa-tion and employing value judgments (Feldman 1989).Transforming learning into a reality requires political skills.The Vermont case demonstrates how MFR not only en-abled double-loop learning, but also shaped the externalenvironment as part of a pragmatic rationale for signifi-cant policy change. The Department of Corrections lead-ership was politically skilful, creating a convincing narra-tive about how traditional correctional policies were failingand showing the need for alternatives. They cultivated sup-port from stakeholders and the public and expanded thetraditional corrections constituency.

The Department of Corrections leadership was also for-tunate. Their narrative may have been persuasive, but theywere lucky to find elected officials willing to listen andwilling to move away from the punitive approach. Theyalso benefited from the long-term continuity of a group ofreform-minded individuals among senior managers, withCommissioner Gorcyzk leading the organization from 1991to 2001. Continuity allowed the development and promo-tion of a distinct vision. Such continuity is rare, and it iseven more rarely combined with a willingness to engagein dramatic change. Weick (2001) notes that it is difficultto overcome agreed-upon organizational goals and under-standings. After 20 years of consistently shifting toward anew model of corrections, the Department of Correctionsstill struggles to gain the full support of many line em-ployees. Without such continuity, the prospects would havebeen substantially dimmer.

ConclusionThere are two main ways to interpret this article; both

are valid and not completely inconsistent with one another.


However, one is pessimistic and one is optimistic. The pes-simistic interpretation argues that the promise of results-based government is not on course to be kept. Such re-forms have been interpreted and implemented withoutconsideration of factors that will enable learning—andtherefore, more efficient and effective government—tooccur. As an organizational learning mechanism, these re-forms have failed to structure learning forums that allowfor the consideration of performance data among its targetusers. The highly structural bias of these reforms also ig-nores the role of culture and the logic of appropriatenessin learning. Goal-based learning is only likely to occurwhen the design of results-based reforms take these fac-tors into account.

The optimistic interpretation celebrates the cases foridentifying the possibility that goal-based learning can

occur. Results-based reforms may need to be reconsidered,but if done correctly, they can create value. From this per-spective, the real quandary for organizational actors is notwhether to pursue organizational learning, but what typeof learning to pursue. March (1999) argues there is a trade-off between the exploration of new possibilities (akin todouble-loop learning) and the exploitation of old certain-ties (akin to single-loop learning). Both types of learningcompete for resources and attention. Ultimately, the long-term capacity of the organization demands a balance be-tween the two approaches: “Adaptiveness requires bothexploitation and exploration. A system that specializes inexploitation will discover itself becoming better and bet-ter at an increasingly obsolescent technology. A systemthat specializes in exploration will never realize the ad-vantages of its discovery” (March 1999, 185).

Notes

11. Argyris (1999) points out that the learning literature tendsto be divided into two subsets: the learning organizationand organizational learning literatures. The learning orga-nization approach emphasizes the adaptability of organi-zations, has a positive action bias, and is associated withconsultants and practitioners. Organizational learning, bycontrast, tends to be more academic and theoretical, focus-ing on the barriers to and difficulties of learning. It pointsout that learning can occur only through individual mem-bers of the organization, and therefore it depends on thelimits and weaknesses of human cognition and the poten-tial for coordination. Learning organization writers are pre-scriptive, assuming the ability of organizations to performheroic feats of adaptability. The organizational learningapproach treats “observed impediments as unalterable factsof organizational life” (Argyris 1999, 14). This article at-tempts to straddle the two literatures, looking at deliberateefforts to foster organizational learning and pointing outits failings and possibilities

12. The focus on culture and the central role of the employeereflects the strong influence of the human relations schoolon the organizational learning literature.

13. To increase the reliability of the selection process, I usedtwo distinct but overlapping approaches. First I used 1998and 2000 Government Performance Project grades in thearea of managing for results. The high performer, Virginia,received an A– in both years, Vermont received a B– and aB in 1998 and 2000, respectively, and Alabama received anF in 1998, improving to a D+ in 2000. Second, I took close-ended survey questions and content analysis form the Gov-ernment Performance Project to create a quantitative scalethat measured the degree of implementation of strategic plan-ning and performance-measurement systems for all 50 stategovernments. Because MFR mandates typically require bothagency-level and state-level reporting, the scale measured

the availability of both statewide performance informationand agency-specific information (sampling the functions ofcorrections, education, and transportation for all states) instate and agency budgets, performance reports, and strate-gic planning documents. States were divided into three cat-egories: high, medium, and low levels of MFR implemen-tation. To be eligible for selection, a state had to satisfy thecategory both in terms of statewide and agency performanceinformation.

14. The concept of restorative justice did not originate in Ver-mont. Variations of the restorative justice idea were occur-ring in many places during the 1980s (Lemley 2001). Theseideas appealed to managers in Vermont, but their own learn-ing process also led them to emphasize other elements ofthe justice system that were not associated with the restor-ative justice movement, including a focus on punishing vio-lent offenders and the use of alternative sentencing.

15. Simon (1991) observes that organizations themselves do notlearn, but learning occurs within organizations through theirindividual members. Therefore, new organizational learn-ing can occur only if existing members acquire new knowl-edge or if the organization hires new members with a vari-ety of knowledge. The latter mechanism was clearly at workin the case of Vermont and in the educational backgroundof its senior managers.

16. The term “metaphor” was used by a senior manager whowas closely involved in the search for knowledge. Describ-ing the business approach to corrections, he stopped to em-phasize that he viewed business as a metaphor for operat-ing, not as a literal model that the public sector should adopt.It is worth noting that Lipshitz, Popper, and Oz (1996) alsouse the term “metaphor” to describe the cultural approachto learning.


17. Between 1994 and 1999 the population under correctionscontrol expanded from 7,511 to 12,386. During this timethe percentage incarcerated stayed relatively steady, drop-ping from 12.3 percent to 11.6 percent. The percentage ofthose on probation also dropped from 77 percent to 74 per-cent of the total population. The difference was largely madeup through the introduction of intermediate sanctions, ac-counting for 4.8 percent of the corrections population in1999 (VDOC 2000a).

18. In its 2000 strategic plan, the department claims the devel-opment of intermediate sanctions has prevented the creationof an additional 1,000 bedspaces, which would have cost anadditional $21.5 million per year (Vermont DOC 2000b).

19. Between 1994 and 1999 surveys administered to the publicfound the department’s approval rating jumped from 37 per-cent to 44 percent.

10. The actors involved in the double-loop learning in Vermontare senior department staff, not institution or field staff. Thishelps to further explain the disjunction between the depart-ment philosophy and the culture of institutions. Having notundergone the same process of learning, institutional staffare less accepting of the philosophy developed by seniormanagers. Greater emphasis on communication may over-come this issue to some extent, but it cannot replicate theprocess of learning for the institutional staff.

11. The role of the judicial branch and the American Correc-tional Association as particularly influential environmental

actors in the area of corrections requires some brief discus-sion. As with any other external stakeholders, they can dic-tate their priorities, limiting the opportunities for organiza-tional change and innovation. Courts have becomeincreasingly active in asserting the rights of prisoners andin intervening to direct the provision of correctional ser-vices (DiIulio 1987). The American Correctional Associa-tion defines a set of accreditation standards that almost allcorrections systems seek to achieve. Given the attention theaccreditation process takes and the legitimacy it provides, itmight reasonably be argued that managers are less likely toinnovate as a result. However, the courts and accreditationcan also promote change, and they have done so for MFRreforms. One reason for adopting MFR is to demonstrate tocourts that a department of corrections has a plan for thefuture and does not require judicial oversight. This was cer-tainly the case in Alabama, where the department was se-lected as a pilot for performance management partly be-cause it had frequent interactions with the courts. Thewillingness of a professional organization such as the Ameri-can Correctional Association to adopt an innovation is themost reliable way to promote that innovation among itsmembers (DiMaggio and Powell 1983). The American Cor-rectional Association has adopted the rhetoric of MFR, andit is converting its 21 manuals of accreditation standardsinto performance-based standards that include outcomemeasures.

References

Andrews, Matthew, and Donald P. Moynihan. 2002. Why Re-forms Don’t Always Have to Work to Succeed: A Tale of TwoManaged Competition Initiatives. Public Performance andManagement Review 25(3): 282–97.

Argyris, Chris. 1999. On Organizational Learning. 2nd ed.Malden, MA: Blackwell.

Argyris, Chris, and Donald Schön. 1996. Organizational Learn-ing: A Theory of Action Perspective. 2nd ed. Reading, MA:Addison-Wesley.

Barker, Anthony. 1998. Political Responsibility for U.K. PrisonSecurity—Ministers Escape Again. Public Administration75(1): 1–23.

Barnow, Burt S. 1992. The Effects of Performance Standards onState and Local Programs. In Evaluating Welfare and TrainingPrograms, edited by Charles F. Manski and Irwin Garfinkel,277–309. Cambridge, MA: Harvard University Press.

Bennett, Charlie, and Herb Hill. 2002. Performance Manage-ment Systems: The Importance of Defining their Purpose.http://www.dpb.state.va.us/VAResults/Material/PolicyService.pdf.

Boston, Jonathan. 1991. The Theoretical Underpinnings of Pub-lic Sector Restructuring in New Zealand. In Reshaping theStates, edited by Jonathan Boston, John Martin, June Pallot,and Pat Walsh, 1–26. Auckland, New Zealand: Oxford Uni-versity Press.

Brudney, Jeffrey L., F. Ted Hebert, and Deil S. Wright. 1999.Reinventing Government in the American States: Measuringand Explaining Administrative Reform. Public Administra-tion Review 59(1): 19–30.

DiIulio, John, Jr. 1987. Governing Prisons. New York: Free Press.DiMaggio, Paul J., and Woody W. Powell. 1983. The Iron Cage

Revisited: Institutional Isomorphism and Collective Ratio-nality in Organizational Fields. American Sociological Re-view 48(4): 147–60.

Downs, George W., and Patrick D. Larkey. 1986. The Search forGovernment Efficiency. New York: Random House.

The Economist. 2002. Too Many Convicts. August 10, 13.Feldman, Martha S. 1989. Order without Design: Information

Production and Policy-Making. Stanford, CA: Stanford Uni-versity Press.

Fiol, C. Marlene, and Marjorie A. Lyles. 1985. OrganizationalLearning. Academy of Management Review 10(4): 803–13.

Forrester, John P., and Guy B. Adams. 1997. Budgetary Reformthrough Organizational Learning: Toward an OrganizationalTheory of Budgeting. Administration and Society 28(4): 466–88.

Franklin, Aimee. 2000. An Examination of Bureaucratic Reac-tion to Institutional Controls. Public Performance and Man-agement Review 24(1): 8–21.


Hedberg, Bo. 1981. How Organizations Learn and Unlearn. InThe Handbook of Organizational Design, edited by Paul C.Nystrom and William H. Starbuck, 8–27. London: OxfordUniversity Press.

Heinrich, Caroyln J. 1999. Do Government Bureaucrats MakeEffective Use of Performance Management Information?Journal of Public Administration and Research Theory 9(3):363–93.

Ingraham, Patricia W., and Donald P. Moynihan. 2001. BeyondMeasurement: Managing for Results in State Government.In Quicker, Better, Cheaper? Managing Performance inAmerican Government, edited by Dall Forsythe, 309–35. Al-bany, NY: Rockefeller Institute Press.

Joyce, Philip G., and Susan S. Tompkins. 2002. Using Perfor-mance Information for Budgeting: Clarifying the Frameworkand Investigating Recent State Experience. In Meeting theChallenges of Performance-Oriented Government, edited byKathryn Newcomer, Ed Jennings, Cheryl Broom, and AllenLomax, 61–96. Washington, DC: American Society for Pub-lic Administration.

Kaplan, Robert S., and David P. Norton. 1996. The BalancedScorecard: Translating Strategy into Action. Boston, MA:Harvard Business School Press.

Klay, William Earle. 1994. Management through Budgetary In-centives. Public Productivity and Management Review 41(2):59–71.

Koteen, Jack. 1989. Strategic Management in Public and Non-profit Organizations: Thinking and Acting Strategically onPublic Concerns. New York: Praeger.

Leavitt, Barbara, and James March. 1990. Chester I. Barnardand the Intelligence of Learning. In Organization Theory:From Chester Barnard to the Present and Beyond, edited byOliver E. Williamson, 11–37. New York: Oxford UniversityPress.

Lemley, Eileen C. 2001. Designing Restorative Justice Policy:An Analytical Perspective. Criminal Justice Policy Review12(1): 43–65.

Liner, Blaine, Harry Hatry, Elisa Vinson, Ryan Allen, PatDusenbery, Scott Byrant, and Ron Snell. 2001. Making Re-sults-Based State Government Work. Washington, DC: Ur-ban Institute. http://www.urban.org/UploadedPDF/results-based-stategovt.pdf.

Lipshitz, Raanan, Micha Popper, and Sasson Oz. 1996. Build-ing Learning Organizations: The Design and Implementationof Organizational Learning Mechanisms. Journal of AppliedBehavioral Science 32(3): 292–305.

Mahler, Julianne. 1997. Influences of Organizational Culture inLearning in Public Agencies. Journal of Public Administra-tion Research and Theory 7(4): 519–40.

March, James. G. 1999. The Pursuit of Organizational Intelli-gence. Malden, MA: Blackwell.

Melkers, Julia E., and Katherine Willoughby. 1998. The State ofthe States: Performance-Based Budgeting Requirements in47 out of 50. Public Administration Review 58(1): 66–73.

Merton, Robert. 1940. Bureaucratic Structure and Personality.Social Forces 18(4): 560–68.

Mintzberg, Henry. 1975. The Nature of Managerial Work. NewYork: HarperCollins.

———. 1994. The Fall and Rise of Strategic Planning. HarvardBusiness Review, February, 107–14.

Moynihan, Donald P. Forthcoming. Managing for Results in StateGovernment: Evaluating a Decade of Reform. Public Admin-istration Review.

Moynihan, Donald P., and Patricia W. Ingraham. 2003. Look forthe Silver Lining: When Performance Based AccountabilitySystems Work. Journal of Public Administration Researchand Theory 13(4): 469–90.

Osborne, David E., and Ted Gaebler. 1992. Reinventing Gov-ernment: How the Entrepreneurial Government is Transform-ing the Public Sector. New York: Penguin.

Radin, Beryl A. 1998. The Government Performance and Re-sults Act (GPRA): Hydra-Headed Monster or Flexible Man-agement Tool? Public Administration Review 58(4): 307–17.

Schein, Edgar. 1992. Organizational Culture and Leadership.2nd ed. San Francisco: Jossey-Bass.

Schick, Allen. 2001. Getting Performance Measures to MeasureUp. In Quicker, Better, Cheaper? Managing Performance inAmerican Government, edited by Dall Forsythe, 39–60. Al-bany, NY: Rockefeller Institute Press.

Senge, Peter. 1990. The Fifth Discipline. New York: Doubleday.Shtull, Penny. 1999. Assessment of the Vermont Department of

Corrections Vision, Mission, and Values, and Principles 1998Revision. Waterbury: Vermont Department of Corrections.

Simon, Herbert. 1991. Bounded Rationality and OrganizationalLearning. Organization Science 2(1): 125–34.

Snell, Ron, and Jennifer Grooters. 2001. Governing-for-Results:Legislation in the States. A Report to the Urban Institute. http://www.urban.org/UploadedPDF/results-based-stategovt.pdf.

U.S. Government Accountability Office (GAO). 1994. Manag-ing for Results: State Experiences Provide Insights for Fed-eral Management Reforms. GAO/GGD-95-22. Washington,DC: GAO.

———. 2001. Managing for Results: Federal Managers Viewson Key Management Issues Vary Widely Across Agencies.GAO-01-152. Washington, DC: GAO.

———. 2003. Program Evaluation: An Evaluation Culture andCollaborative Partnerships Help Build Agency Capacity.GAO-03-454. Washington, DC: GAO.

Vermont Department of Corrections (VDOC). 2000a. Facts andFigures FY 1999. www.doc.state.vt.us.

———. 2000b. Corrections in Vermont: A Five-Year Plan—Making Vermont Safe for the 21st Century.www.doc.state.vt.us.

Weick, Karl. 2001. Making Sense of the Organization. Oxford,UK: Blackwell Publishing.

Wildavsky, Aaron. 1975. Budgeting: A Comparative Theory ofBudgeting Processes. Boston: Little, Brown.

Wilson, James Q. 1989. Bureaucracy: What Government Agen-cies Do and Why They Do It. New York: Basic Books.

Documents

Goal-Based Learning and the Future of Performance Management · and understandings that foster intelligent behavior (Senge 1990). Characteristics of a learning culture include high