COST ACCOUNTING

doi:10.1016/j.aos.2006.05.001 Reducing conXict in balanced scorecard evaluations Bernard Wong-On-Wing a,¤ , Lan Guo a, Wei Li b, Dan Yang c a Department of Accounting, Washington State University, P.O. Box 644729, Pullman, WA 99164-4729, USA b Department of Accounting, Kent State University, Kent, OH 44242-0001, USA c Southwestern University of Finance and Economics, 55 Guanghuacun, Chengdu 610074, Sichuan, China Abstract Recent studies [Ittner, C., & Larcker, D. (2003). Coming up short on nonWnancial performance measurement. Har- vard Business Review(November) 88–95; Ittner, C., Larcker, D., & Randall, T. (2003b). Performance implications of strategic performance measurement in Wnancial services Wrms. Accounting, Organizations and Society, 28, 715–741] pro- vide evidence of companies’ tendency to overlook the validity of the causal links between driver and outcome measures of the balanced scorecard (BSC), and to ignore the underlying strategically-linked causal business models. It is posited that this propensity leads to conXict between top management and divisional managers because of the failure of the former to evaluate and consider strategy eVectiveness in performance evaluation. The present study hypothesizes that individuals in the top-manager role do not take into account strategy eVectiveness unless they are explicitly required to do so. In contrast, individuals in the store-manager role automatically consider the quality of strategy without being prompted to do so. A study using 63 evening MBA students provides support for the hypotheses. The results have impli- cations for the study of evaluation biases in BSC as well as in other performance measurement systems, and for devising means to mitigate them.

Introduction Performance evaluation is an essential function in any organization. Consequently, it is important to understand how performance measurement sys- tems inXuence such appraisals. In a seminal study on the role of accounting data in performance evaluation, Hopwood (1972) highlighted problemswith traditional accounting measures of perfor- mance. In particular, he noted the lack of compre- hensiveness of the measures, the imprecision with which accounting systems measure performance, the limited focus on outcome measures, and the over-emphasis on short-term performance (pp.

157–158). Hopwood (1972) hypothesized and found that depending on the evaluation style, reli- ance on these accounting performance measures can result in dysfunctional consequences including disagreement and conXict between supervisors *Corresponding author. Fax: +1 509 335 4275.

E-mail address: [email protected] (B. Wong-On-Wing). 364B. Wong-On-Wing et al. / Accounting, Organizations and Society 32 (2007) 363–377 (raters) and subordinates (ratees). Subsequent studies (e.g., Hartmann, 2000; Otley & Pollanen, 2000) have similarly provided evidence suggesting that, depending on various contextual factors, neg- ative consequences (e.g., job-related tension and distrust in supervisor) may result from the reliance on conventional accounting measures when evalu- ating performance.

In contrast to traditional performance mea- surement systems, the balanced scorecard (BSC) introduced by Kaplan and Norton (1992) is expected to address many of the concerns raised by Hopwood (1972). The BSC is a multi-dimen- sional performance measurement system that includes Wnancial, outcome and short-term as well as non-Wnancial, driver and long-term measures. A key feature of the BSC is its emphasis on linking the performance measures with business unit strategy (Otley, 1999, pp. 374–375). Because of its comprehensive and strategy-linked measures, the BSC can be expected to reduce the likelihood of previously observed disagreement and conXict between raters and ratees by promoting a more holistic approach to performance evaluation. For example, when evaluating poor performance, the inclusion of strategy-linked outcome and driver measures in the BSC may direct raters to attend to strategy quality as an explanation, which they are unlikely to consider when using a system strictly based on outcome measures. Consequently, raters using the BSC may ascribe inferior performance less to ratees and more to strategy-related causes.

This would presumably lead to more favorable performance appraisal of ratees and reduce the likelihood of disagreement and conXict between raters and ratees.

Recent studies (Ittner & Larcker, 2003; Ittner, Larcker, & Randall, 2003b), however, suggest that the causal links between driver and outcome mea- sures are often overlooked. For example, Ittner et al.’s (2003b, p. 725) study of Wnancial services Wrms found that of those claiming to use a bal- anced scorecard, 76.9% place little or no reliance on their strategically linked causal business mod- els. In another Weld study of manufacturing and service companies, Ittner and Larcker (2003, p. 90) note more speciWcally that among those that cre- ate causal models, only 21% go on to validate thecausal links between driver and outcome mea- sures. They also observe “businesses often fail to establish such links partly out of laziness or thoughtlessness” (p. 89). Thus, the evidence from these studies indicates that although information about strategy eVectiveness is available in the BSC, it is not used as much as would have been expected presumably because of cognitive limita- tions.

The present study posits that the above-noted tendency to overlook the validity of the causal links between driver and outcome measures of the BSC is a potential source of conXict between top management and divisional managers. SpeciWcally, it is proposed that discrepancies in performance ratings between raters and ratees may result from top management’s failure to consider the quality of its chosen strategy when evaluating divisional managers’ performance. Based on research (Feld- man, 1994; Ilgen, Barnes-Farrell, & McKellin, 1993; Landy & Farr, 1980) that views performance appraisal primarily from a cognitive process per- spective, the present study further posits that top management’s failure to consider strategy eVec- tiveness is due to its selective attention bias. Thus, it is hypothesized that increasing top manage- ment’s awareness of the impact of strategy eVec- tiveness on performance, may reduce conXict resulting from diVerences in evaluation between top management and divisional managers.

The purpose of this research is to test the fore- going propositions. To establish the existence of a bias, the present research Wrst assesses the extent to which, when using the BSC, divisional perfor- mance evaluation diVers between individuals who adopt the perspective of top management (the rater) and those who assume the role of divisional managers (the ratees). The study then examines the eVectiveness of a mechanism for reducing top man- agement’s bias, and thus, the disagreement in per- formance appraisal between the two perspectives.

The speciWc procedure involves increasing the rater’s awareness of the quality of top manage- ment’s strategy in inXuencing divisional perfor- mance. This is achieved by requiring an explicit assessment of the role of strategy quality in deter- mining divisional performance prior to perfor- mance evaluation. B. Wong-On-Wing et al. / Accounting, Organizations and Society 32 (2007) 363–377365 Examining potential conXict and disagreement that may result from top management’s bias in performance evaluation within the context of the BSC is signiWcant for at least three reasons. First, the study is important given that, as noted earlier, features speciWc to the BSC are expected to lessen the likelihood of bias and thus, conXict and disagreement in performance evaluation. How- ever, contrary to expectations, recent Weld studies (e.g., Ittner, Larcker, & Meyer, 2003a; Malina & Selto, 2001) document disagreement and conXict between top management and divisional mana- gers when using the BSC to evaluate the latter’s performance. In order to obtain the intended ben- eWts of the BSC, it is important to understand the nature of biases that may result in conXict, and investigate ways in which the biases can be miti- gated.

Second, the present study in the speciWc context of the BSC is important because top management’s biases may also inXuence the eVectiveness of the BSC as a strategic management system. Kaplan and Norton (1996, 2001) speciWcally note how the BSC provides companies with the capacity for strategic learning, enabling them to modify strate- gies when necessary. In evaluating performance, the exclusive focus on divisional managers’ ability and eVort to execute the company’s strategy, and the failure to consider the eVectiveness of the strat- egy, may impede strategic learning. SpeciWcally, because of the failure to see potential problems with the strategy, it may not be adjusted when it is beneWcial and justiWed to do so.

Third, Keeping and Levy (2000) and Cardy and Dobbins (1994) argue that one criterion to use in evaluating performance appraisal systems is the reaction of ratees (i.e., divisional managers, in the present context). Although the present study is conducted in the context of the BSC, the results may have implications for reducing conXict and tension between raters and ratees in any perfor- mance measurement system.

The remainder of this paper is organized as fol- lows. The next section provides the theoretical background, which leads to the development of the hypotheses. The research method is then described followed by the results, and a discussion of the implications as well as the limitations of the study.Theoretical background The balanced scorecard The BSC provides management with a compre- hensive framework that translates a company’s strategy into a coherent set of performance mea- sures (Kaplan & Norton, 1993). It supplements tra- ditional Wnancial measures with measures from three additional perspectives: those of customers, internal business processes, and learning and growth. The BSC has several potential beneWts.

First, compared with traditional measurement sys- tems that include only Wnancial measures, the BSC is designed to improve managers’ decision making by guiding their attention to a broader vision of the company’s operations (Kaplan & Norton, 1992). Second, as a holistic performance measure- ment system, the BSC provides causal links con- necting the multiple classes of non-Wnancial measures (“drivers of the performance”) and the Wnancial measures (“Wnal outcome”) (Campbell, Datar, Kulp, & Narayanan, 2002; Ittner et al., 2003b). As such, it clearly shows the links by which speciWc improvements in the drivers are expected to lead to desired outcomes according to the strat- egy. Third, the BSC can be used as a strategic man- agement system (Kaplan & Norton, 1996, 2001). In particular, the pattern of results on the causally linked driver and outcome measures may provide clues about the eVectiveness of the strategy. This then provides companies with the capacity for stra- tegic learning thereby enabling them to modify strategies when necessary.

The unique features of the BSC noted above, suggest that its use can be expected to reduce the likelihood of disagreement and conXict between rat- ers and ratees in performance evaluation. For exam- ple, the causal linkages between the driver and outcome measures are expected to lead raters to consider the eVectiveness of the strategy in apprais- ing performance. According to Kaplan and Norton (1996, p. 85), “If the unit’s employees and managers have delivered on the performance drivers (retrain- ing of employees, availability of information sys- tems, and new Wnancial products and services, for instance), then their failure to achieve the expected outcomes (higher sales to targeted customers, for 366B. Wong-On-Wing et al. / Accounting, Organizations and Society 32 (2007) 363–377 example) signals that the theory underlying the strategy may not be valid.” Thus, the pattern of results in a BSC should enable one to infer the eVec- tiveness of a given strategy. Consequently, a given poor performance would not presumably be ascribed, at least not entirely, to the ratee if weak linkages between the driver and outcome measures suggest an ineVective strategy. Such a possibility does not exist in performance measurement systems that only report outcome measures. Thus, relative to less comprehensive performance measurement sys- tems, the BSC would be expected to lead to lower rater bias in performance appraisal.

ConXict in BSC evaluations Contrary to the foregoing expectations, two recent Weld studies (Ittner et al., 2003a; Malina & Selto, 2001) provide evidence of bias and conXict in performance evaluation among Wrms that have adopted the BSC. Malina and Selto (2001) studied divisional managers of a large international manu- facturing company that adopted a BSC. They observed signiWcant conXict and tension between top and middle management regarding perfor- mance evaluation of the latter. The study also doc- umented the inaccuracy or subjectivity of the BSC measures, top–down instead of participative com- munication about the BSC, and the use of inappro- priate benchmarks for evaluation.

In a study of branch managers at a major Wnan- cial services Wrm that adopted a BSC-based reward system, Ittner et al. (2003a) likewise, documented complaints about favoritism in bonus awards and the uncertainty in the criteria used to determine rewards. In particular, the results of the study indi- cate that, “the subjectivity in the balanced score- card plan allowed area directors to incorporate factors other than the scorecard measures in per- formance evaluations, to change evaluation crite- ria from quarter to quarter, to ignore measures that were predictive of future Wnancial perfor- mance, and to weight measures that were not pre- dictive of desired results,” (p. 749). Qualitative analyses of data gathered from two internal sur- veys of branch managers provide further evidence of dissatisfaction with the BSC. For example, 48% of the respondents disagreed with the statementthat the scorecard process fairly assessed job per- formance. The authors conclude based on their qualitative analyses, that the BSC did not achieve its objectives.

The above Wndings provide evidence of disagree- ments in BSC evaluations that are similar to those that have been observed in performance appraisal in other contexts (see for example, Atwater, OstroV, Yammarino, & Fleenor, 1998; Atwater & Yamma- rino, 1997; Harris & Schaubroeck, 1988; Viswesva- ran, Ones, & Schmidt, 1996). More importantly, the Wndings of the Weld studies (Ittner et al., 2003a; Malina & Selto, 2001) suggest that the unique fea- tures of the BSC, such as its comprehensive and strategically linked measures, may not be suYcient to overcome performance evaluation bias that can result in disagreement and conXict between raters and ratees. Consistent with earlier studies (e.g., Lipe & Salterio, 2000, 2002), the present research explores how raters’ (i.e., top management’s) cogni- tive limitations may bias their performance evalua- tions, and consequently, result in conXict between them and ratees. Speci Wcally, the current study pos- its that one source of disagreement between top management and divisional managers, is top man- agement’s bias in BSC performance evaluation.

Bias in BSC evaluations Research by Lipe and Salterio (2000, 2002) sug- gests that the beneWts of the BSC may not be obtained because of raters’ cognitive biases and limitations. For example, Lipe and Salterio (2000) found that supervisors’ evaluation of divisional managers’ performance using BSC was based only on common measures across diVerent business units, and not on the measures that were unique to particular business units. To eliminate such com- mon-measures bias, recent studies have examined various de-biasing techniques, such as providing evaluators with strategy map (Banker, Chang, & Pizzini, 2004), disaggregating the BSC measures (Roberts, Albright, & Hibbets, 2004), and invoking process accountability and improving the per- ceived quality of BSC measures (Libby, Salterio, & Webb, 2004).

Following Lipe and Salterio (2000) who suggest that managers paid insuYcient attention to unique B. Wong-On-Wing et al. / Accounting, Organizations and Society 32 (2007) 363–377367 measures relative to common measures, the current study posits that raters’ (top management) bias may be attributable to their inadequate attention to the quality of the strategy compared to the eVort and ability of ratees (divisional managers), as a potential determinant of performance. As noted before, although the strength of the causal linkages among the BSC measures, as suggested by the pat- tern of results, should enable one to infer the eVec- tiveness of a given strategy (Kaplan & Norton, 1996), Weld studies (Ittner & Larcker, 2003; Ittner et al., 2003b), suggest that the causal links between driver and outcome measures are often over- looked. Ittner and Larcker (2003, p. 89) in parti- cular, note that businesses often fail to establish such causal links partly out of laziness or thought- lessness. The current study posits that one source of disagreement between top management and divi- sional managers, is top management’s failure to consider the potential inXuence of strategy quality on the latter’s performance. SpeciWcally, it is pro- posed that this failure is due to selective attention bias among top management.

It is important to distinguish between raters’ attention to the quality of the strategy and their awareness of strategically linked measures in the BSC. The study by Banker et al. (2004) found that in evaluating performance, raters attend to strate- gically linked measures when they are provided with a strategy map. SpeciWcally, in such a case, raters place more weight on strategically linked measures than on non-linked measures. The pres- ent study conjectures that if participants cannot diVerentiate between linked and non-linked mea- sures, they are unlikely to be able to assess the quality of the strategy and consider it in the evalu- ation of the store managers. However, although raters may be made aware of strategically linked measures, they will not necessarily assess the qual- ity of the strategy or consider it in appraising the performance of divisional managers.

Selective attention in performance appraisal Research (Feldman, 1994; Ilgen et al., 1993) that views performance appraisal primarily from a cog- nitive process perspective has identiWed several sources of bias and inaccuracies. One such sourceis the rater’s selective attention. The rater’s ability to selectively attend to information inXuences rat- ing accuracy (Cardy & Kehoe, 1984). The present study focuses speciWcally on problems of raters’ selective attention. The remainder of this section examines two types of selective attention biases that are particularly relevant to the present con- text. These provide the basis for (1) predicting the divergence in BSC performance evaluation between top management and divisional managers, and (2) proposing the mechanism for reducing top management’s bias, and ultimately, the diVerence in BSC performance appraisal between top man- agement and divisional managers. Below, consis- tent with the speciWc context of the present study, the relevant biases are discussed as they relate to the evaluation of poor divisional performance.

ConXict may be more likely to occur in such a situ- ation than when performance is superior.

The actor–observer bias. Jones and Nisbett (1971) proposed that actors tend to view their behavior as determined by situational factors whereas observers tend to attribute the same behavior to the actors’ dispositions (e.g., traits, character, etc.). In a review of the related literature, Watson (1982) examined research evidence of the actor–observer bias and found that it is largely due to the diVerential tendency of actors and observers to attribute causality to the environment, rather than a diVerential preference for dispositional inferences. In particular, actors and observers do not diVer signiWcantly in attributing behavior to actors’ dispositions. However, compared to actors, observers tend to attribute actors’ behavior less to situational factors.

Based on the above Wnding, it is postulated that when evaluating performance, top management (the observer) is less likely than divisional mana- gers (the actors) to perceive that performance of the latter is due to situational factors. In the pres- ent context, the quality (eVectiveness or ineVective- ness) of the strategy represents one situational determinant of divisional performance. The com- prehensive and strategically linked measures in a BSC are expected to facilitate the evaluation of the quality of the strategy (Kaplan & Norton, 1996, 2001). Such strategy evaluation would then pre- sumably be considered in making inferences about 368B. Wong-On-Wing et al. / Accounting, Organizations and Society 32 (2007) 363–377 divisional managers and assessing their perfor- mance. However, to the extent that the actor– observer bias exists, it is expected that when divisional performance is inferior, top manage- ment is less likely than divisional managers to ascribe it to a possibly ineVective strategy. Conse- quently, compared to divisional managers, top management will tend to rate the managers lower on performance.

The correspondence bias. One of the most robust Wndings in social psychology is the correspondence bias. The correspondence bias is the tendency to draw inferences about a person’s unique and endur- ing dispositions from behaviors that can be entirely explained by the situations in which they occur (Gilbert & Malone, 1995). As in the case of the actor–observer bias, the tendency is to ignore situa- tional factors that can clearly explain the observed behavior. In the present context, examining the comprehensive and the causally linked measures of the BSC presumably can highlight potential prob- lems with the strategy (Kaplan & Norton, 1996).

However, if top management is susceptible to the correspondence bias, it will ignore the role of its possibly ineVective strategy in contributing to divi- sional managers’ inferior performance. Instead, it will hold divisional managers entirely responsible for the poor performance. As a result, top manage- ment will rate divisional managers lower than if the bad performance had been, at least partly, ascribed to a poorly formulated strategy.

Reducing bias and conXict The preceding discussion describes biases, which explain why top management may assign lower performance ratings than divisional manag- ers who are evaluated using the BSC. To overcome this apparent predisposition, it is necessary to Wrst understand the reasons for the biases.

Both the actor–observer bias and the correspon- dence bias have been explained from the cognitive information processing perspective. For example, the evidence reviewed by Watson (1982) demon- strates the importance of the salience of situational factors in determining the extent of the actor– observer bias. SpeciWcally, an observer who is made more aware of situational factors by increas-ing their salience typically makes more situational attributions.

Similarly, Gilbert and Malone (1995) suggest that one possible reason for the correspondence bias is the lack of awareness of situational forces.

They note (p. 25), “To avoid the correspondence bias, an observer must realize that a situation is playing a causal role in an actor’s behavior.” They add, “one can implicate situational forces as causes only when one is aware that such forces exist in the Wrst place.” In a recent study, Choi and Nisbett (1998) found that increasing the salience of situa- tional constraints reduced correspondence bias among Korean participants. Thus, it appears that both the actor–observer bias and the correspon- dence bias can be reduced by making the observer more aware of the relevance of situational factors in inXuencing behavior. Research (Gilbert & Mal- one, 1995; Lieberman, Gaunt, Gilbert, & Trope, 2002; Reeder, 1993) suggests that once observers attend to situational factors, they may correct their initial inference. In the present context, if raters become aware of strategy quality (situational fac- tor) as a determinant of performance, they may revise their inferences about the store managers.

These inferences then presumably inXuence their performance evaluation.

Based on the foregoing, the present research proposes and tests the eVectiveness of a mechanism for reducing the divergence in BSC performance evaluation between top management and divi- sional managers. The current research examines the eVectiveness of a de-biasing mechanism for increasing raters’ awareness of the quality of strat- egy as a potential determinant of divisional perfor- mance. SpeciWcally, it involves a step that requires top management to explicitly assess the signiW - cance of its strategy quality in inXuencing divi- sional performance prior to the evaluation of the divisional managers. This procedure is expected to increase the salience of strategy ineVectiveness as a possible cause for observed poor divisional perfor- mance. Top management should then discount the responsibility of divisional managers for the infe- rior divisional performance, and thus, rate them more favorably. As a result, the performance eval- uations of top management and divisional mana- gers should be less divergent. B. Wong-On-Wing et al. / Accounting, Organizations and Society 32 (2007) 363–377369 Hypotheses Based on the discussion in the preceding section, two hypotheses are formulated. Because it is expected that when using the BSC, top manage- ment is less likely than divisional managers to con- sider a potentially ineVective strategy as a reason for poor divisional performance, the Wrst hypothe- sis predicts that when using the BSC, performance evaluations will diverge between top management and divisional managers. SpeciWcally, in the absence of any mechanism to reduce top manage- ment’s bias, and in the context of inferior divi- sional performance, top management will rate the performance of divisional managers lower than the latter. Thus, the Wrst hypothesis is:

H1: In the context of poor divisional performance, individuals who assume the role of top manage- ment will rate divisional managers signiWcantly lower than those who assume the role of a divi- sional manager.

The second hypothesis predicts that when using the BSC, the diVerence in performance evaluations between top management and divisional managers will be reduced as a result of increasing top man- agement’s awareness of the importance of its strat- egy in determining divisional performance. More speciWcally, it is expected that requiring an explicit assessment of the signiWcance of strategy eVective- ness in contributing to divisional performance will result in top management’s discounting of divi- sional managers as a cause of inferior divisional performance. Consequently, top management’s perceptions about divisional managers will be more favorable and it will rate them higher than when the ineVectiveness of the strategy is not considered as a possible cause of poor divisional performance. Because divisional managers pre- sumably ascribe the inferior performance to strat- egy without the de-biasing mechanism, the same eVect is not expected to be signiWcant among them.

Thus, the following is postulated:

H2: In the context of poor divisional performance, the diVerence in performance ratings between indi- viduals who assume the role of top management and those who assume the role of divisional man-agers will be signiWcantly smaller when they are Wrst required to assess the importance of strategy in determining divisional performance than when they are not required to do so.

Method Overview Participants read a case, which described a new strategic business unit (SBU) recently acquired by a clothing company. The SBU consisted of two stores (Store A and Store B) each managed by a newly hired store manager. The case also stated that top management adopted for the new SBU, a new sales growth strategy recommended by a well-known independent consultant. Participants were then asked to assume either the role of top management (top management condition) or a store manager (store manager condition). After examining the BSC, half of the participants (assessment condition) were required to assess the extent to which they believed the new SBU’s per- formance was due to the quality of the adopted sales growth strategy and due to the divisional managers’ ability and eVort in executing the strat- egy. The other half of the participants (no assess- ment condition) made no such assessment. All participants were subsequently instructed to pro- vide a preliminary performance evaluation of each store manager.

Participants A total of 68 evening MBA students at a Chi- nese university completed the survey during class time. Sixty-six percent were male, and the average age was 29 years. The participants had an average of 6.5 years of full-time work experience, and 67% held management positions.

Design This study employed a 2 (Role)£2 (Assess- ment)£2 (Store) design. The Wrst two factors were between-subjects factors and allowed the study’s hypotheses to be tested. Store was a within- 370B. Wong-On-Wing et al. / Accounting, Organizations and Society 32 (2007) 363–377 subjects factor. This manipulation had two pur- poses. First, it was designed to provide participants with BSC results that imply a potential problem with the strategy. This was achieved by presenting across the two stores, a consistent pattern of per- formance suggesting weak linkages between driver and outcome measures. This was necessary in order to examine whether as suggested by Kap- lan and Norton (1996), raters recognize the possi- bility of an ineVective strategy in the absence (H1) and in the presence (H2) of a de-biasing proce- dure.

Second, based on the Wndings of Banker et al.

(2004), another purpose of the store manipulation was to verify whether participants distinguished between the strategically linked and non-linked measures. As posited earlier, if participants can- not diVerentiate between linked and non-linked measures, they are unlikely to be able to assess the quality of the strategy and consider it in the evalu- ation of the store managers. In the present study, the BSC of the two stores were designed such that Store A performed better than Store B on the non-linked measures whereas Store B performed better than Store A on the strategically linked measures. If consistent with Banker et al.’s (2004) Wndings, participants are able to distinguish between linked and non-linked measures (by assigning a higher evaluation to the manager of Store B than the manager of Store A), it would suggest that any observed performance evaluation bias should not be attributed to participants’ inat- tention to the strategy-linked measures, but instead to other factors such as the failure to con- sider the quality of the strategy. In contrast, if par- ticipants do not diVerentiate between linked and non-linked measures (i.e., the manager of Store B is not evaluated higher than the manager of Store A), it is possible that subsequent performance evaluation bias may simply be due to participants’ inattention speciWcally to the strategy-linked mea- sures.

The three manipulated factors are described in the independent variables section. Each participant was randomly assigned to one of the four combi- nations of role and assessment. The same two stores and their BSC results were used in all four conditions.Case material The case used in the current study was adapted from Lipe and Salterio (2000) and Banker et al.

(2004). 1 It described a SBU recently acquired by a clothing company. Following Banker et al. (2004), the case also described the sales growth strategy (see Fig. 1) recommended by an independent con- sultant, and adopted by top management for implementation at the two stores of the new SBU.

Participants were told that both store managers were recently hired to execute the new strategy.

They were then instructed to assume either the role of a member of top management or one of the store managers of the new SBU. A description of the BSC measurement system recommended by the independent consultant was subsequently pre- sented. The performance results were then pre- sented in a BSC as shown in Table 1.

At this point, participants in the assessment condition were asked to indicate separately the extent to which they believed the new SBU’s per- formance was due to: (1) the quality of the adopted sales growth strategy, and (2) the divisional man- agers’ ability and eVort in executing the strategy.

The order of these two questions was alternated.

All participants were then asked to rate each of the two store managers. The last part of the instrument requested participants to answer a question to ver- ify the eVectiveness of the role manipulation, and to provide demographic information.

The instrument was originally written in English. After translating it into Chinese, it was back translated into English following Brislin (1970). There were no signiWcant problems in either the translation or back translation. 1The case used in the present study diVers from those used by Lipe and Salterio (2000) and Banker et al. (2004) as follows. Be- cause the focus of these studies was on judgmental eVects of common and unique measures, they compared two SBUs with diVerent strategies, measures and targets. The current study, however, compared two stores within one SBU. A main pur- pose of the store manipulation was to convey to participants the possibility of an ineVective SBU strategy through the con- sistent pattern of BSC results across both stores. Thus, in the present study, the two stores had the same strategy, measures and targets. B. Wong-On-Wing et al. / Accounting, Organizations and Society 32 (2007) 363–377371 Fig. 1. Sales growth strategy for ASL’s new SBU. ASL, Incorporated, owns and operates a chain of clothing stores. The company recently acquired two stores to form a new strategic business unit (SBU). The new SBU will specifically cater to fashion-conscious professional women with the goal of becoming a chain of stores in which women can shop for all of their wardrobe needs, from clothing to accessories to shoes, in one convenient location. To assist the company in planning its strategy for the new SBU, a well- known independent consultant was hired. Following the advice of the consultant, ASL chose the following sales growth strategy for its new SBU:

According to the sales growth strategy, ASL’s new stores will focus on the training of its sales associates, and improve employee satisfaction. Better-trained and more satisfied employees will enhance store appearance, and enable the creation of a "perfect in-store shopping experience." This will in turn, increase customer satisfaction and repeat sales, and result in sales growth at ASL's new SBU. Increase sales associates' training and employee satisfaction Enhance store appearance and customers' in-store experience Increase customer satisfaction and repeat sales Grow sales Table 1 Balanced scorecard for ASL’s new SBU a a The strate gicall y linked measures are italicized. However, in the actual instrument, the y were not. Measure Target for both storesStore A Store B Actual % better (or worse) than targetActual % better (or worse) than target Financial perspective 1. Sales margins 60.00% 58.0%¡3.33 54.0%¡10 2. Sales growth10% 5.4%¡46 6.0%¡40 3. Assets turnover 6 5.8¡3.33 5.8¡3.33 4. Return on expenses 40% 38.0%¡5 38.3%¡4.25 Customer perspective 1. Repeat sales %40.00% 35.4%¡11.5 37.8%¡5.5 2. Customer satisfaction rating85 75¡11.76 81¡4.71 3. Sales per square foot of retail space $30,000 $28,500¡5 $26,700¡11 4. Customer returns as a % of sales 3.50% 3.7%¡4.29 3.9%¡11.43 Internal process perspective 1. Store appearance rating85 87 2.35 95 11.76 2. Number of stock-outs <3 times 2.65 11.67 2.93 2.67 3. “Mystery shopper” rating on “perfect shopping experience”80 82 2.5 89 11.25 4. Time to process returns <4 min 3.55 11.25 3.92 2 Learning and growth perspective 1. Employee satisfaction80 82 2.5 90.5 13.13 2. Employee suggestions per year >3 times 3.4 13.33 3.1 3.3 3. Store computerization 60.00% 66.5% 10.83 61.0% 1.67 4. Hours of employee training>70 h 72 2.86 78 11.43 372B. Wong-On-Wing et al. / Accounting, Organizations and Society 32 (2007) 363–377 Independent variables The two between-subjects independent vari- ables were role and assessment. Role was manipu- lated by instructing the participant to imagine that he/she was a member of the top management or one of the store managers of the new SBU. The written instruction appeared in the case before the BSC and performance evaluation questions. Par- ticipants were also reminded of their role by one of the experimenters at the beginning of the sur- vey.

The assessment treatment consisted of requir- ing half of the participants to indicate on separate 11-point scales (0: not at all; 10: to a great extent), the extent to which they believed that the new SBU’s performance was due to the quality of its strategy, and due to the ability and eVort of the store managers. The assessment was made before the evaluation of the store managers. Participants in the no assessment condition did not perform this step.

Store was a within-subjects factor. To present a consistent pattern of performance across two stores, the results in the BSC showed that both stores performed better than target on the inter- nal business process and the learning and growth measures, and worse than target on Wnancial and customer measures. Moreover, to verify partici- pants’ ability to diVerentiate between strategi- cally linked and non-linked measures, the BSC of the two stores were designed such that Store A performed better than Store B on the non-linked measures whereas Store B performed better than Store A on the strategically linked measures (see Table 1). However, for each of the four catego- ries of the BSC, the sum of the variance (diVer- ence between actual results and target) as a percentage of the target was the same for the two stores.

Dependent measure Participants were asked to evaluate each of the two store managers on an 11-point scale (0:

extremely poor; 10: excellent). The performance evaluation ratings were used to test the hypothe- ses.Results Manipulation check The role manipulation was overall eVective.

Only two out of 34 participants who were assigned to the role of store manager indicated that they assumed the role of top management.

Three out of 34 participants who were assigned to the role of top management responded that they assumed the store manager role. The data analy- ses reported below exclude the Wve participants who failed the manipulation check question. The results are not signiWcantly diVerent when the responses of those Wve participants are included in the data analyses.

Participants were also asked to indicate on sep- arate 11-point scales, the perceived realism (0: not at all realistic; 10: extremely realistic) and the level of diYculty (0: not at all diYcult; 10: extremely diYcult) of the case. In general, they thought that the case was reasonably realistic (meanD6.63, sdD1.64), and moderately diYcult (meanD5.56, sdD1.92).

Hypothesis tests To test the hypotheses, a 2 (Role)£2 (Assess- ment)£2 (Store) repeated-measures analysis of variance (ANOVA) was performed on the perfor- mance evaluation ratings. As shown in Table 2, there was a signiWcant interaction between role and assessment (F(1, 59) D4.22, pD0.04). The test of each hypothesis is presented below.

H1. The Wrst hypothesis predicts that in the absence of any mechanism to reduce top manage- ment’s bias, participants who assume the top man- agement role will rate performance signiWcantly lower than those who assume the divisional man- ager role. To test this hypothesis, a simple-eVects contrast of the roles within the no assessment con- dition was performed (Neter, Kutner, Nachtsheim, & Wasserman, 1996, pp. 860). The results show that participants who were assigned to the top management role (4.86, sdD1.58) gave signiW- cantly (tD1.75, pD0.04, one-tailed) lower evalua- tions to the store managers on average than those who assumed the store manager role (5.62, B. Wong-On-Wing et al. / Accounting, Organizations and Society 32 (2007) 363–377373 sdD1.56). Similar results were obtained using the evaluation of the individual stores. Participants who were assigned to the top management role gave lower ratings to store managers A (4.50, sdD1.63) and B (5.22, sdD1.49) than those who assumed the store manager role (AD5.06, sdD1.71; BD6.18, sdD1.19). Therefore, H1 is sup- ported.

H2. The second hypothesis predicts that the diVerence in performance ratings between individ- uals who assume the role of top management and those who assume the role of divisional managers will be signiWcantly smaller when they are Wrst required to assess the importance of the quality of the strategy in determining overall performance than when they are not required to do so. As shown in Table 2 and Fig. 2, there was a signiWcant interaction between role and assessment on divi-sional performance evaluation (F(1, 59)D4.22, pD0.04).

To examine the speciWc nature of the interaction and in particular, how assessment inXuenced the performance evaluation among participants in the top management and the divisional manager roles, simple-eVects contrasts of assessment within each role were performed. As shown in Fig. 3A, among participants who assumed the role of top manage- ment, the evaluation of store managers was signiWcantly (tD2.41, pD0.01, one-tailed) higher when they were required to make the assessment (5.9, sdD1.12) than when they were not (4.86, sdD1.15). In contrast, the eVect of Assessment was not signiWcant (tD¡0.50, pD0.62) among partici- pants who assumed the role of store managers (see Fig. 3B).

The results of the test of H1 established that in the no assessment condition, individuals who assumed the role of top management rate perfor- mance signiWcantly lower than those who assumed the role of a divisional manager. To determine whether this diVerence remained signiWcant in the assessment condition, the performance ratings for the two roles were compared for that condition.

The results showed that the diVerence in evalua- tion between roles was no longer signiWcant (tD1.16, pD0.25). The average performance rat- ings of the top management and store manager roles were 5.9 (sdD1.12) and 5.4 (sdD1.27) respec- tively. Together, the results of the preceding tests support H2. Table 2 Assessment£role£store ANOVA on performance evaluation a (nD63) aThe dependent variable is the performance evaluation score of the individual store manager. Source of variation dfSum of squares Mean squaresFp-value Between subjects Assessment 1 5.32 5.32 1.80 0.184 Role 1 0.52 0.52 0.18 0.675 Assessment£role 1 12.43 12.43 4.22 0.044 Error 59 173.92 2.95 Within subjects Store 1 20.57 20.57 13.87 <0.001 Assessment£store 1 0.37 0.37 0.25 0.617 Role£store 1 0.22 0.22 0.15 0.704 Assessment£role£store 1 0.43 0.43 0.29 0.594 Error 59 87.51 1.48 Total 125 301.80 Fig. 2. Role£assessment interaction eVect on store manager performance evaluation. 4.865.9 5.62 5.4 3 4 5 6 7 No Assessment Average store manager performance evaluation Top Management Store Manager Assessment 374B. Wong-On-Wing et al. / Accounting, Organizations and Society 32 (2007) 363–377 Store main eVect Recall that the performances of the two stores were comparable except that store B (A) per- formed better than store A (B) on the strategically linked (non-linked) measures. Table 2 shows a sig- niWcant main eVect of store (F(1, 59)D13.87, p< 0.01). Neither the interaction eVect between store and assessment (pD0.62) nor that between store and role (pD0.70) was signiWcant. Across both role and both assessment conditions, store manager B was given a higher rating (5.85, sdD1.41) than store manager A (5.03, sdD1.60).

Thus, consistent with the results of Banker et al.

(2004), participants placed more weight on the strategically linked measures than on the non- linked measures. This Wnding suggests that partici- pants in both role and both assessment conditions were able to distinguish between strategically linked and non-linked measures. This evidence is consistent with the current study’s conjecture thatthe performance evaluation bias observed under the no assessment condition is not attributable to the lack of awareness of the strategically linked measures, but instead due to the failure to attend to the quality of the strategy.

Discussion Summary of results Following research (e.g., Lipe & Salterio, 2000, 2002) that examines biases in BSC evaluations, the present study explored the role of rater’s selective attention in contributing to conXict between raters and ratees. SpeciWcally, based on research on biases and performance appraisal, the present study investigated (1) the extent to which perfor- mance evaluation using the BSC diverges between individuals who assume the role of top manage- ment (rater) and those who adopt the role of a store manager (ratee), and (2) the degree to which the diVerence in BSC performance evaluation is reduced as a result of requiring participants to assess the signiWcance of strategy eVectiveness in inXuencing performance, prior to rating divisional performance.

As hypothesized, individuals who assumed the role of top management rated divisional perfor- mance lower than those who assumed the role of store managers. This is consistent with the evidence of disagreement in BSC evaluations between top management and divisional managers, gathered from Weld studies (Ittner et al., 2003a; Malina & Selto, 2001). While this Wnding is not unlike those observed in other performance evaluation contexts (Atwater et al., 1998; Atwater & Yammarino, 1997; Harris & Schaubroeck, 1988; Viswesvaran et al., 1996), it is particularly important in the present sit- uation since the comprehensive and strategically linked measures of the BSC are expected to lead to lower rater bias than traditional performance mea- surement systems that include only Wnancial and outcome measures. Thus, the Wnding suggests that contrary to expectations, the unique features of the BSC do not signiWcantly reduce selective attention bias that presumably leads to conXict between top management and divisional managers. Fig. 3. (A) Top management evaluation of store manager per- formance. (B) Store manager evaluation of store manager per- formance. 4.55.53 5.226.27 3 4 5 6 7 No Assessment Store manager performance evaluation 5.065.07 6.18 5.73 3 4 5 6 7 Store manager performance evaluation Store Manager A Store Manager B No Assessment Assessment Store Manager A Store Manager B Assessment (A) (B) B. Wong-On-Wing et al. / Accounting, Organizations and Society 32 (2007) 363–377375 The predicted eVect of increasing top manage- ment’s awareness of the importance of strategy eVectiveness was also obtained. SpeciWcally, as pos- tulated, individuals who were assigned to the role of top management rated store managers signiW- cantly better when they were required to assess the role of strategy quality in determining divisional performance than when they were not. Apparently, without explicit prompting, participants assigned to the role of top management do not automati- cally consider the strategy. In contrast, among par- ticipants who assumed the role of store managers, there was no signiWcant diVerence in performance ratings between those prompted to assess the role of strategy eVectiveness and those who were not. It appears that participants in the store manager role consider the strategy even when they are not explicitly required to do so.

As a result of the required assessment, the diVer- ence in performance evaluation between partici- pants in the top management and the store manager roles was no longer signiWcant. This pat- tern of results is consistent with the present study’s conjecture that the observed performance evalua- tion bias may be due to top management’s failure to attend to the quality of the strategy when appraising divisional managers’ performance.

Consistent with the Wndings of Banker et al.

(2004), the Store main eVect observed in the cur- rent study implies that participants were able to distinguish between strategically linked and non- linked measures, and assigned more weight to the former. Since performance bias is observed in the no assessment condition when participants are able to diVerentiate between strategically linked and non-linked measures, this evidence suggests that participants’ attention to strategically linked measures is not suYcient to reduce selective atten- tion bias that presumably leads to conXict between top management and divisional managers.

Implications Together, the above results are important in attempting to understand possible reasons for con- Xict and tension similar to those observed in Weld studies (Ittner et al., 2003a; Malina & Selto, 2001) between evaluators and those being evaluatedusing the BSC. SpeciWcally, the Wndings of the pres- ent study suggest that one source of the disagree- ment in performance ratings appears to be top management’s (the rater’s) insuYcient attention to the quality (eVectiveness or ineVectiveness) of its strategy in evaluating divisional manager perfor- mance. Such bias may reduce the likelihood of achieving the beneWts of the BSC. For example, divisional managers (ratees) may ignore the BSC if they perceive it to be completely subjective, unfair or irrelevant to their performance evaluation.

The results also indicate that in the context of poor divisional performance, it may be possible to reduce top management’s bias by requiring an assessment of the inXuence of strategy quality on divisional performance. Moreover, the apparent eVectiveness of the assessment procedure suggests that top management’s bias observed in the cur- rent study may be cognitive and not motivational or self-serving. These Wndings have potential impli- cations to the extent that they suggest a means to mitigate the observed bias and possibly, the con- Xict and tension between top management and divisional managers. This is necessary if the bene- Wts of the BSC are to be achieved.

Reducing top management’s bias may in addi- tion increase the eVectiveness of the BSC as a stra- tegic management system. In evaluating divisional performance, emphasizing the eVectiveness of the strategy may facilitate the identiWcation of its limi- tations, and consequently, highlight the need for its revision. This is consistent with the strategic man- agement system function of the BSC (Kaplan & Norton, 1996, 2001).

It is also important to note the implications of the present Wndings for other performance measure- ment systems. First, if as observed in the present study, the comprehensiveness of the BSC is not suY- cient enough to reduce selective attention bias in performance evaluation, it can be expected that con- Xict would be even more likely in other performance measurement systems that do not include strategy- linked outcome and driver measures. In other words, performance raters are less likely to evaluate strategy if they receive performance reports that are less clearly strategy-based. Second, reminding raters of other situational causes of ratees’ perfor- mance may have similar de-biasing eVects in other 376B. Wong-On-Wing et al. / Accounting, Organizations and Society 32 (2007) 363–377 performance evaluation systems. Nonetheless, the speciWc de-biasing procedure (assessment of the role of strategy quality) used in the present study for reducing potential conXict may not be as eVective when using other performance measurement sys- tems that are strictly based on outcome measures as when using the BSC. This is because those perfor- mance measurement systems may not enable the quality of the strategy to be inferred.

The foregoing implications of the current study must nevertheless be qualiWed in light of the exper- imental research method employed. Because the present study was conducted in a controlled envi- ronment, it omitted many contextual factors that would be present in the natural setting. Thus, the results of the current research do not imply that top management’s selective attention bias can only be reduced by the proposed assessment mecha- nism. Several factors that are present in the natural business environment may similarly reduce the likelihood of such a bias. For example, communi- cations through participative decision making between top management and divisional managers may highlight the former’s tendency to be suscepti- ble to selective attention in performance evalua- tion. Moreover, according to Balser and Stern (1999, p. 1033), evaluative and corrective processes not only take place within formal appraisal sys- tems but may happen informally as well. Hence, formal and informal information exchanges between raters and ratees may reduce the extent of top management’s selective attention bias as eVec- tively as the mechanism proposed in the current study.

Limitations and future studies The foregoing results should be interpreted in light of the study’s limitations. First, this study’s focus was limited to sources of bias due to top management’s (the rater’s) selective attention in evaluating poor performance. Sources of bias other than those identiWed by Lipe and Salterio (2000, 2002) and in this study, are likely to exist in BSC evaluations. Although the results of the pres- ent study suggest that the observed bias among top management may be cognitive, motivational or self-serving biases may still be present in the natu-ral environment. For example, Moers (2005) exam- ined archival data of a Wrm and found evidence suggesting that when using subjective and multiple performance measures, superiors may be moti- vated to compress and give more lenient perfor- mance ratings. The present study was also limited to rater bias in BSC evaluation of poor divisional performance only. Both rater and ratee biases may be present when using the BSC to evaluate both good and bad performance. For example, ratees may be motivated to take credit for good perfor- mance, whereas raters may tend to attribute it to their choice of an eVective strategy. This is likely to result in conXict between raters and ratees. The nature of other rater and ratee biases, as well as ways to mitigate them can be examined in future studies.

Second, participants were not asked to assume the role of a speciWc store manager. Instead, they were asked to assume the role of one of the store managers. Thus, the “role” manipulation was not as strong as it would have been if participants had been assigned to the role of a speciWc store man- ager. It is possible that the diVerence in perfor- mance evaluation between roles under the no assessment condition may have been even more signiWcant had participants been asked to assume the role of a speciWc store manager.

Third, the current study does not provide direct evidence of the process by which the assessment procedure reduces performance evaluation bias.

SpeciWcally, it cannot be demonstrated that, as posited in the present study, the eVect of the de- biasing mechanism on performance evaluations is mediated by inferences about management since the latter was not measured. Future research can examine whether inferences about ratees mediate the eVect of strategy assessment on performance evaluation. References Atwater, L. E., OstroV, C., Yammarino, F. Y., & Fleenor, J. W.

(1998). Self-other agreement: does it really matter? Personnel Psychology, 51(3), 577–598.

Atwater, L. E., & Yammarino, F. Y. (1997). Self-other rating agreement: a review and model. Research in Personnel and Human Resources Management, 15, 121–174. B. Wong-On-Wing et al. / Accounting, Organizations and Society 32 (2007) 363–377377 Balser, D. B., & Stern, R. N. (1999). Resistance and cooperation:

a response to conXict over job performance. Human Rela- tions, 52(8), 1029–1053.

Banker, R. D., Chang, H., & Pizzini, M. (2004). The balanced scorecard: judgmental eVects of performance measures linked to strategy. The Accounting Review, 79(1), 1–23.

Brislin, R. W. (1970). Back-translation for cross-cultural research. Journal of Cross-Cultural Psychology, 1, 185–216.

Campbell, D., Datar, S., Kulp, S., & Narayanan, V. (2002).

Using the balanced scorecard as a control system for moni- toring and revising corporate strategy. Working Paper. Har- vard NOM Research.

Cardy, R. L., & Dobbins, G. H. (1994). Performance appraisal:

alternative perspectives. Cincinnati, OH: South-Western Publishing.

Cardy, R. L., & Kehoe, J. F. (1984). Rater selective ability and appraisal eVectiveness: the eVect of a cognitive style on the accuracy of diVerentiation among ratees. Journal of Applied Psychology, 69, 589–594.

Choi, I., & Nisbett, R. E. (1998). Situational salience and cul- tural diVerences in the correspondence bias and actor– observer bias. Personality and Social Psychology Bulletin, 24(9), 949–960.

Feldman, J. M. (1994). On the synergy between theory and application: social cognition and performance appraisal.

Handbook of Social Cognition, 2, 339–397.

Gilbert, D. T., & Malone, P. S. (1995). The correspondence bias.

Psychological Bulletin, 117(1), 21–38.

Harris, M. M., & Schaubroeck, J. (1988). A meta-analysis of self-supervisor, self-peer, and peer-supervisor ratings. Per- sonnel Psychology, 41, 43–62.

Hartmann, F. G. H. (2000). The appropriateness of RAPM:

toward the further development of theory. Accounting, Organizations and Society, 25(4/5), 451–482.

Hopwood, A. G. (1972). An empirical study of the role of accounting data in performance evaluation. Journal of Accounting Research, 10, 156–182.

Ilgen, D., Barnes-Farrell, J., & McKellin, D. (1993). Perfor- mance appraisal process research in the 1980s: what has it contributed to appraisal in use? Organizational Behavior and Human Decision Processes, 54, 321–368.

Ittner, C., & Larcker, D. (2003). Coming up short on nonWnan- cial performance measurement. Harvard Business Review(November), 88–95.

Ittner, C., Larcker, D., & Meyer, M. (2003a). Subjectivity and the weighting of performance measures: evidence from a balanced scorecard. The Accounting Review, 78(3), 725–758.

Ittner, C., Larcker, D., & Randall, T. (2003b). Performance implications of strategic performance measurement in Wnan- cial services Wrms. Accounting, Organizations and Society, 28, 715–741.

Jones, E. E., & Nisbett, R. E. (1971). The actor and the observer:

divergent perceptions of the causes of behavior. Morristown, NJ: General Leaning Press.

Kaplan, R. S., & Norton, D. P. (1992). The balanced score- card—measures that drive performance. Harvard Business Review(January/February), 71–79.Kaplan, R. S., & Norton, D. P. (1993). Putting the balanced scorecard to work. Harvard Business Review(September/ October), 134–142.

Kaplan, R. S., & Norton, D. P. (1996). Using the balanced score- card as a strategic management system. Harvard Business Review(January/February), 75–85.

Kaplan, R. S., & Norton, D. P. (2001). The strategy focused organization: how balanced scorecard companies thrive in the new business environment. Boston, MA: Harvard Business School Press.

Keeping, L. M., & Levy, P. E. (2000). Performance appraisal reactions: measurement, modeling, and method bias. Journal of Applied Psychology, 85, 708–723.

Landy, F. J., & Farr, J. (1980). Performance rating. Psycholo- gical Bulletin, 87, 72–107.

Libby, T., Salterio, S. E., & Webb, A. (2004). The balanced scorecard: the eVects of assurance and process accountabil- ity on managerial judgment. The Accounting Review, 79(4), 1075–1094.

Lieberman, M. D., Gaunt, R., Gilbert, D. T., & Trope, Y. (2002).

ReXexion and reXection: a social cognitive neuroscience approach to attributional inference. Advances in Experimen- tal Social Psychology, 34, 199–249.

Lipe, M. G., & Salterio, S. (2000). The balanced scorecard: judg- mental eVects of common and unique performance mea- sures. The Accounting Review, 75(3), 283–298.

Lipe, M. G., & Salterio, S. (2002). A note on the judgmental eVects of the balanced scorecard’s information organization.

Accounting, Organizations, and Society, 27, 531–540.

Malina, M., & Selto, F. (2001). Communicating and controlling strategy: an empirical study of the eVectiveness of the bal- anced scorecard. Journal of Management Accounting Research, 13, 47–90.

Moers, F. (2005). Discretion and bias in performance evalua- tion: the impact of diversity and subjectivity. Accounting, Organizations and Society, 30, 67–80.

Neter, J., Kutner, M. H., Nachtsheim, C. J., & Wasserman, W.

(1996). Applied linear statistical models. Boston: WCB/ McGraw-Hill.

Otley, D. T. (1999). Performance management: a framework for management control systems research. Management Accounting Research, 10, 363–382.

Otley, D. T., & Pollanen, R. (2000). Budgetary criteria in perfor- mance evaluation: a critical appraisal using new evidence.

Accounting, Organizations and Society, 25, 483–496.

Reeder, G. D. (1993). Trait-behavior relations and dispositional inference. Personality and Social Psychology Bulletin, 19(5), 586–593.

Roberts, M., Albright, T., & Hibbets, A. (2004). Debiasing bal- anced scorecard evaluations. Behavioral Research in Accounting, 16, 75–88.

Viswesvaran, C., Ones, D. S., & Schmidt, F. L. (1996). Compara- tive analysis of the reliability of job performance ratings.

Journal of Applied Psychology, 81(5), 557–574.

Watson, D. (1982). The actor and the observer: how are their perceptions of causality divergent? Psychological Bulletin, 92(3), 682–700.