1 Muel Kaptein is professor of business ethics and integrity management at Rotterdam School ... investigates the behavioral effects of hard and soft controls.
THE SOFT SIDE OF HARD CONTROLS: A CONTROL CODING THEORY1 Muel Kaptein and Hein-Jan Vink
Summary Controls can have positive or increasingly negative effects on the behavior of employees. In this article we develop a theory to explain these effects. Central to this is the way employees interpret the management intentions in applying control measures. Since control measures reduce employee autonomy, there is a risk that employees either put too much faith in the measures or grow to mistrust them. Three mechanisms are presented for excessive trust and mistrust.
1
Muel Kaptein is professor of business ethics and integrity management at Rotterdam School
of Management, Erasmus University, in the Netherlands. He is also a partner of KPMG where he works as an auditor and consultant for organizations in the field of culture and behavior. Hein-Jan Vink, RA, is an audit manager at the Netherlands Court of Audit. His PhD research investigates the behavioral effects of hard and soft controls. He has written this article in a private capacity.
1
Electronic copy available at: http://ssrn.com/abstract=2378437
1. Introduction Organizations often call on hard, formal control measures to influence employee behavior so as to attain organizational goals. Examples of such formal hard controls are fixed procedures, registration systems, codes of conduct, and job descriptions. But does having more numerous, further-reaching hard controls (greater control density) result in more desirable behavior? Various studies show a positive correlation between specific hard controls and factors such as work satisfaction, innovation, and productivity (Cravens, Lassk, Low, Marshall, and Moncrief, 2004; Damanpour, 1991; Demming, 1986; Jaworski and Young, 1993). Others, however, find a negative correlation between specific hard controls and factors such as work satisfaction and employee motivation (Arches, 1991; Rousseau, 1978a, 1978b). These studies do not necessarily contradict one another, as illustrated in research by Katz-Navon et al (2005). The study investigated the relationship between the magnitude of treatment protocols (a hard control) and treatment errors (a form of undesirable behavior) in 46 Israeli hospitals. They found both positive and negative effects of hard controls. At first it appeared that the more detailed the treatment protocols, and therefore the more extensive, the greater the effectiveness. However, as treatment protocols were extended they reached a point of diminishing returns, until eventually there was no further benefit and extension of the protocols had negative effects. The magnitude of hard controls exhibits a curvilinear relationship. If this is true of the magnitude of measures, it may also apply to the number of hard controls. Differences in study results on the effectiveness of hard controls may therefore lie in the limited variation of control density used and the purely linear analyses carried out. The question for this article is why this curvilinear relationship emerges. The answer is useful in determining optimal control density and avoiding imposing too few or too many controls. We would also like to focus on the soft side of hard control measures, or the informal side of formal controls. We do this firstly by introducing a Control Coding theory (CCT). Since control measures represent an attempt to influence the behavior of the target group, it is important to determine how the target group interprets the measures. This interpretation primarily focuses on what the target group believes the measures are intended to achieve. In relation to an organization this means that employees decode what they believe the originators, often the management, have coded. Just as in Aristotle’s virtue ethics a virtue is the middle between two vices, a good control measure should be the midpoint between the vices of mistrust (leading to aversive behavior) and excessive trust (leading to passivity): that midpoint is trust (leading to desirable behavior). For each function of a control measure we will investigate how this can lead to mistrust or excessive trust. The underlying mechanisms help us explain the curvilinear relationship between control density and behavior. The article is structured as follows. In section 2 we define the functions of hard controls. In section 3 we then introduce CCT. On this basis in section 4 we investigate how employees interpret hard controls and subsequent behavior. In section 5 we discuss how increas-
2
Electronic copy available at: http://ssrn.com/abstract=2378437
ing control density influences these interpretations, potentially undermining the function of control measures. In the closing section we examine the practical implications. 2. Definition and functions of hard controls In the 1970s there were substantial changes to thinking on the function of controls, which had previously been seen as a cybernetic process, with the organization acting as a thermostat for achieving organizational aims effectively and efficiently. Since then, as sociologists and business economists have started to publish on the subject of internal control alongside engineers, there has been more attention for the effect of behavior: controls within organizations aim to influence employee behavior towards achieving organizational aims (Anthony, 1965; Lawler and Rhode, 1976; Merchant 1982). We define controls in line with this as the measures for influencing the behavior of managers and employees. This is about both avoiding undesirable behavior, such as errors, accidents, incidents, and transgressions, and about advancing desirable behavior, such as lawfulness, effectiveness, and efficiency. Controls can be subdivided into hard and soft controls (Karapetrovic & Willborn, 2001; Roth, 2003). Hard controls are formal (such as policy, systems, and manuals), whereas soft controls are informal (such as culture, climate, and awareness). Hard and soft controls do not only affect manager and employee behavior separately; the interaction between them can also have an effect (Kaptein, 2010). In this article we focus on hard controls and explore their inherent ‘soft’ side. In order to determine the effectiveness of hard controls, we therefore have to investigate what these controls do to the behavior of the target group. In order to find out, we first need to establish the functions of controls, so that we can then identify the possible positive and negative effects on behavior for each function. Three functions of controls We believe control measures have at least the following three functions: informing, motivating, and equipping employees. These functions can be coupled with the conditions for desirable behavior in general. A distinction is made in the literature between knowledge, desire, and ability (for example see Cressey, 1953; Elffers, Van der Heijden, and Hezemans, 2003; and Kaptein, 2011). The first function of control measures is information. Control measures are set up to inform employees of the behavior expected of them. In order to act as desired, employees need to know what is desirable and what is undesirable. A strategic plan informs employees of the organization’s aims and the steps that are to be taken; a code of conduct sets out the rules that apply; an appraisal system informs employees of desirable and undesirable behavior; and features such as signs on company premises inform employees of what they should or should not do. The second function of control measures is motivation. Control measures are established to encourage and inspire employees both to exhibit desirable behavior and to avoid un-
3
Electronic copy available at: http://ssrn.com/abstract=2378437
desirable behavior. In order to guarantee that employees work in agreement with organizational goals, employees must be willing to meet these aims. A strategic plan can motivate employees in this direction (for example by expressing ambitious, popular goals); a code of conduct can motivate employees to comply (by pointing out its importance); an appraisal system can motivate employees to do what is rewarded (by rewarding good behavior and punishing undesirable behavior); and a sign can motivate employees to comply (for instance by pointing out danger or possible sanctions). The third function of control measures is to equip. Control measures are established to enable employees to behave as desired, as mentioned in Adler and Borys (1996). There is also a ‘coercive’ side, attempting to avoid undesirable behavior by impeding it or making it impossible. A good strategic plan equips employees with the means of realization; a good code of conduct enables employees to resist temptation and pressure (for instance by referring others to the code when they receive requests to act undesirably); a good appraisal system offers employees opportunities to fulfill the criteria (such as time, training, and information); and a sign (by the emergency exit, for example) enables them to use the exit in a disaster. 3.
Control coding theory
In order to understand how control measures influence employee behavior, it is important to understand employees’ views on and experience of the measures, as the way they interpret the controls affects their behavior. Here we take a sociopsychological and socioconstructivist approach (Vygotsky, 1978). How employees interpret controls depends on how the designers, the management, designed them and the models on which they are based. The management communicates with employees through the controls. The message is interpreted by the employees, following the process described in communication sciences (Shannon, 1948). Figure 1 gives a schematic overview of the coding and decoding process.
4
Figure 1: Coding and decoding of controls Environment
Noise
Management descision
Coding by management
Hard Control
Decoding by employee
Employee Behaviour
Noise
Environment
First comes the coding of controls. Controls are established according to the definition given above to influence employee behavior, and are set up by management or with their approval. The issues management raise (or omit) in controls, and the way they shape and introduce them, reveal their view of employees, anticipated risks, and how they believe they can influence them. In this respect every control has a reason (why it has been set up), a standard (desirable or undesirable behavior), and expectation (employee compliance). The management codes this, consciously or subconsciously, in the hard control. A hard control therefore consists of lots of messages that the management use to communicate with employees. The question is how employees interpret these, be it consciously or subconsciously. This is the decoding side. Employees interpret the control measure based on factors such as the reasons why it was established, standard behavior, and employee expectations. This interpretation then affects their behavior; they act according to how they view and experience the controls. A hard control here is a social construct because its interpretation occurs not purely through individual employees but also in mutual communication and against the background of the social aspects of the working environment. Coding and decoding are based on management and employee assumptions. These assumptions, however, are not necessarily the same. Incorrect or unintended interpretations are sources of noise or interference, causing good managerial intentions to be interpreted wrongly by employees, who then do not behave as intended. 4. CCT and hard controls Based on the three control functions, employees may view controls as informative, motivational, and equipping, thereby feeling informed, motivated, and equipped. In that case the controls are sufficient and promote desirable behavior. This is the inherent positive side of
5
controls. After all, by definition controls indicate that something is important and desirable (otherwise the management would not take the trouble to develop a control for a given issue), that desirable behavior can be promoted (this is inherent in the definition of a control as a measure to promote desirable behavior and therefore something that employees are willing to comply with), and that employees are capable of complying (otherwise there would be no point in formulating controls). This positive side is also an important element of CCT. However just as we can explain the positive effect of controls by their functions, we can also explain the negative effect. For this purpose we must first realize that controls reduce the autonomy of employees, even when the management’s intentions are purely positive. They are measures taken to influence behavior. Management measures prevent employees from deciding for themselves precisely how to behave, restricting their freedom to act or refrain from acting (cf. Cropanzano, 2001). Hard controls also reduce autonomy with respect to the three functions: they signal that employees do not know everything, and are unwilling or unable to handle everything. Even the formal rule that employees should be autonomous reduces autonomy, because employees then cannot determine for themselves whether they are autonomous. In that respect controls by definition express a problem. There must be a reason why the desired behavior does not occur automatically, as controls aim to influence behavior in the desired direction, so without controls there would be less chance of desirable behavior. The fact that controls restrict employee autonomy helps us distinguish and explain two negative effects that can emerge as a result of controls. The first effect is that the reduction of employee autonomy causes them to base their behavior too much on the controls. They put too much faith in controls, thereby becoming passive. The second effect is that the reduction of autonomy causes employees to see controls as an infringement of their autonomy and as an act of mistrust. This makes them mistrust the controls and their instigators, resulting in aversive behavior. In the next section we explain how each of the three functions can lead to these two negative behaviors. Table 2 gives the core points.
6
Table 2: Positive and negative interpretations and effects of hard controls Functions of controls:
A. Positive atti-
B. Passive atti-
C. Aversive atti-
tude:
tude:
tude:
Controls and
Employees trust
Employees mi-
management
controls and man-
strust controls and
show confi-
agement too much
management
Informed
Assumed
Pedantic
Motivated
Dependent
Offended
Equipped
Careless
Obstructed
dence in employees 1. Informational: expanding knowledge of desirable and undesirable behavior 2. Motivational: motivating engagement in desirable behavior and refraining from undesirable behavior 3. Equipping: expanding ability to engage in desirable behavior and avoid undesirable behavior Effect:
Positive beha-
Passive behavior:
Aversive behavior:
vior:
Doing the oppo-
Not engag-
er desirable be-
site of desirable
ing in unde-
havior
behavior
sirable be
Neglecting oth-
Engaging too
Engaging too
havior
much in desira-
little in desira-
Engaging in
ble behavior
ble behavior
desirable behavior Passive attitude Controls can inspire excessive trust on the part of employees. Based on these three control functions, this works through the following three mechanisms. Firstly due to the informational function a control can make employees think they know enough or are already omniscient, or that they no longer need to think because others (those who established the control) have already done so. This is inherent in controls because
7
informing assumes that what is communicated is true and that further information is unnecessary (Merchant, 1990). Otherwise controls would have been established for any other issues as well. Employees can assume too much due to controls, taking certain points for granted and seeing the measures as sufficient, and thereby becoming passive, in the sense of failing to think independently or gather information for themselves. People go through a protocol without thinking about other points of attention, assume they are safe as long as the lights on the dashboard are green, that as long as the alarm is not sounding they do not need to check for burglars, and that everything is allowed as long as no rules specifically prohibit it. Secondly, a control can lead to employees feeling less responsible as a result of their motivational function. If the control requires or prohibits particular actions, employees may feel less responsible for their own behavior. After all, they are doing as they are told. Research shows, for example, that transparency about rewards and bonuses makes people feel less responsible for bonus size (Cain, Loewenstein, and Moore, 2005). There is therefore a risk that employees will hide behind controls and shift responsibilities onto them. People become dependent, trusting controls too much and lacking the courage to take responsibility, for instance out of fear of doing the wrong thing. Thirdly, a control can lead to carelessness among employees due to the equipping function. Employees can come to trust systems and procedures too much, making them less alert to risks or even causing them to take more risks, just as drivers with winter tires drive faster when it snows and cause more accidents, People feel safe, secure, covered, and omnipotent, making them overconfident (Wilde, 2001). People may think, for example, that the automated systems are so watertight that they are insufficiently critical of possible systemic errors or abuse. The means and possibilities can also be taken for granted, making people resigned and only inclined to do what they have to. When such interpretations of controls arise, this increases the chance of passive behavior. Passive behavior means firstly that people focus more on what is prescribed and less on other desirable behavior. People meekly follow the control as if on autopilot, resulting in goal displacement (Merchant, 1982) or tunnel vision (Birnberg, 1983). People focus on expectations and therefore miss or neglect other relevant issues. For instance people may focus on measurable, quantifiable goals, neglecting qualitative goals that are not measured. They may focus on the short term, not the long term, because only the former is rewarded, and on one group of stakeholders because there are lots of controls for them, thereby neglecting others for whom there are no controls (cf. Gilliland and Landis, 1992; Shah, Friedman, and Kruglanski, 2002). Passive behavior can also lead to employees exceeding controls, engaging too much in desirable behavior. They can focus on what the control requires, thereby losing their sense of proportion. They do more of what is required because this tendency is encouraged by the control. There may be competition to achieve the control as well as possible, which may in practice be interpreted as exceeding it. Examples include delivering more information than required, asking for permission more often than needed, and refusing to accept any gifts at all when the limit is set at fifty dollars.
8
Aversive attitude Controls can also lead to employee mistrust for the controls themselves and their instigators. Based on the three control functions, this works through the following three mechanisms. Firstly the informational function of the control can be seen as pedantic. Employees can take the control as evidence that management think they do not know how they should behave or what is expected of them. They see the control as evidence that management underestimate their knowledge and interpret the control as patronizing and paternalistic. When an organization introduced a 40-page dress code, employees took this as an undermining of their own sense of appropriate work clothing. Secondly, from the motivational function of the control, this can be seen as insulting. Instead of the control motivating people to comply, it is demotivating because employees take it as evidence that management do not trust them and do not think they are honest or well meaning (Knights and Collinson, 1987). The control therefore creates suspicion and fear. When personnel are searched on their way off the premises for possible stolen goods, it can create a feeling that the management sees every employee as a potential criminal. Thirdly, from the function of equipping, the control can be interpreted as obstructive, creating the impression among employees that they can no longer carry out their work properly but must be led by the hand. Controls then hinder desirable behavior and promote undesirable behavior. The controls can be seen as a straitjacket, bureaucratic, laborious, inefficient, or even pointless. Actions seem preprogrammed so employees feel they are on a leash and have insufficient freedom to deliver good work based on their own insight and expertise. When one or more of these interpretations of controls arise, this increases the chance of aversive behavior. Aversive behavior can mean first of all that employees engage in too little desirable behavior. They may do less in protest. They become less motivated to engage in desirable behavior, evading or circumventing the rules or doing the bare minimum. People may also work to rule, precisely following the control with the aim of showing management that without employees’ knowledge, motivation, and skill, the control is inadequate and undermining. Secondly aversive behavior can lead to employees engaging in undesirable behavior. Resistance theory (Brehm, 1966) states that people seek to reinstate their feeling of reduced autonomy by doing precisely what has been banned. Research shows that when people are persistently asked to stop doing something, they are more likely to do it (Grandpre, Alvaro, Burgoon, Miller, and Hall, 2003). Considerations such as ‘gaming’ and ‘beating the system’ play a role here (Jaworski and Young, 1992). The more insistent the request, the more people feel challenged and provoked to do the opposite. The feeling of being mistrusted hits back in mistrust of the instigators and therefore in a higher chance of opposition and manipulation.
9
5. Curvilinear relationship between control density and behavior In the previous sections we explored how hard, formal control measures can have positive and negative effects on employee behavior. The question now is how this helps explain the curvilinear relationship between controls and behavior. In order to understand this properly we must first make it clear what we mean by increasing controls, referring to control density. Control density is the extent to which controls are applied to influence employee behavior. Control density is the product of a number of separate controls and the specificity and detail of these controls. The key point is the product of these two elements. In this respect it does not matter whether an organization has one code of conduct with ten items or ten codes of conduct with one item. In our view the curvilinear relationship between control density and behavior consists of the result of the three effects described above. Employee behavior is explained by the extent to which controls have a positive, passive, and aversive effect. In the case of increasing control density, the positive effect decreases and the negative effect increases: the passive effect follows an S-curve and the aversive effect an exponential curve. Figure 3 gives an example of how, other things being equal, these separate relationships can lead, like the overall effect, to undesirable behavior. These lines are mirrored in desirable behavior. Figure 3 Positive, negative, and total effects of increasing control density U N D E S I R A B L E B E H A V I O R
4 A
C
B
D 3
2
1 CONTROL DENSITY
The positive effect of increasing control density exhibits diminishing returns (line 1). An important reason for this is a saturation effect with increasing control density. The more controls, the less a new control contributes a desirable effect, and the more specific the control, the less an extra detail will add. The passive effect takes the form of an S-curve (line 2). When control density is low, the same will apply to passivity. When control density increases, so will passivity. The more the organization seeks to influence employees, at least according to this effect, the more employees will allow themselves to be influenced and led. However, this effect is limited. People
10
cannot be led beyond the scope of their activities. For this reason beyond a certain level of control density, further increases will not result in greater passivity or submissiveness because people are already entirely passive and obedient. We believe the aversive effect works exponentially (line 3). As expected, this negative effect occurs later than the previous two effects because people only become irritated beyond a certain level of control density, but when that happens every increase in control density will further fan the flames, thereby causing the undesirable behavior to increase faster. The total effect of these three lines is given in line 4. The path of this control curve involves four phases. Phase A is the positive phase, in which control density leads to less undesirable behavior. In phase B returns on increasing control density are substantially diminished by the increasing effect of passivity. This means that increasing control density leads to more undesirable behavior. Phase C brings stabilization. Increasing control density does not reduce the positive effect, nor does it increase the passive effect. If control density continues to increase, aversive attitudes come into play, so that phase 4 is a phase of escalation. We therefore come to the following four propositions: Proposition 1: As control density increases, increases in the positive effect become smaller. Proposition 2: As control density increases, the passive effect increases logarithmically. Proposition 3: As control density increases, the aversive effect increases exponentially. Proposition 4: The curvilinear relationship between control density and undesirable behavior is the result of a positive, passive, and aversive effect.
6. Conclusion and practical implications In this article we have examined the factors that explain the curvilinear relationship between hard controls and behavior. The core of our explanation is that formal control measures are coded by managers and decoded by employees consciously and subconsciously. The proposed Control Coding Theory (CCT) helps in elucidating this process and the noise that can emerge in the data. CCT states that controls have a composite effect on behavior. Controls have an inherently positive side, as well as a negative side that can result in both passive and aversive attitudes among employees. On the basis of these three aspects, along with the three functions of controls, nine mechanisms can be distinguished in total. These nine mechanisms help explain the influence and effectiveness of controls on behavior.
11
Practical implications This article has various practical implications. Firstly we have attempted to show that those who design, introduce, and make decisions on hard controls should realize that every control they impose conveys an inherent normative message, whether they are aware of that or not. The target group then makes their own interpretation of this message. For effective controls it is therefore important that the designers, executors, and decision makers realize how the target group interprets them, the effect this has on their behavior, and the extent to which that agrees with what the designers and executors intended. In testing, this soft side of hard controls should be taken into account. In order to determine the effect of hard controls, it is essential to find out how the target group interprets them. Without knowing this, it is impossible to comprehend their effect. In determining this effect we can investigate the extent to which each of the nine mechanisms distinguished in this article arises and its total effect. For every control we can then locate its phase in the control curve described. Another important implication is that for the design, introduction, and testing of controls it is not only a matter of treating the controls in isolation but also as a group. In this article we have introduced the term control density for this purpose. The potential and actual interpretations of controls by the target group should be considered when determining the optimal control density, because hard controls have a soft side.
12
Bibliography Adler, P. S. and B. Borys (1996), Two types of bureaucracy: Enabling and coercive, Administrative Science Quarterly, vol. 41, pp. 61-89. Anthony, R. N. (1965), Planning and Control Systems: Framework for analysis, Boston: Graduate School of Business Administration, Harvard University. Arches, J. L. (1991), Social structure, burnout, and job satisfaction, Social Work, vol. 36, pp. 202-206. Birnberg J. G., L. Turopolec, and S. M. Young (1983), The organizational context of accounting. Accounting, Organizations and Society, vol. 8, pp. 111-129. Brehm, J. W. (1966), A Theory of Psychological Reactance. New York: Academic Press. Cain, D. M., G. Loewenstein, and D. A. Moore (2005), The dirt on coming clean: Perverse effects of disclosing conflicts of interest, Journal of Legal Studies, vol. 34, pp. 1-25. Cravens, D. W., F. G. Lassk, G. S. Low, G. W. Marshall, and W. C. Moncrief III (2004), Formal and informal management control combinations in sales organizations: The impact on salesperson consequences, Journal of Business Research, vol. 57, pp. 241-248. Cressey, D. R. (1953), Other People’s Money, Glencoe: Free Press. Cropanzano, R. and Byrne, Z. (2001), When it’s time to stop writing policies: An inquiry into procedural injustice, Human Resource Management, vol. 11, pp 31-54. Damanpour, F. (1991), Organizational innovation: A meta-analysis of effects of determinants and moderators, Academy of Management Journal, vol. 34, pp. 555-590. Deming, W. E. (1986), Out of the Crisis, Cambridge: MIT Press. Elffers, H., P. van der Heijden, and M. Hezemans (2003), Explaining regulatory noncompliance: A survey study of rule transgression for two Dutch instrumental laws, applying the randomized response method. Journal of Quantitative Criminology, vol. 19, pp. 409-439. Fayol, H. (1916), Administration Industrielle et Générale, Paris: Dunod. Gilliland, W. and R. S. Landis (1992), Quality and quantity goals in a complex decision task: Strategies and outcomes, Journal of Applied Psychology, vol. 77, pp. 672-681. Grandpre, J. and E.M. Alvaro, M. Burgoon, C.H. Miller, and J.R. Hall (2003), Adolescent reactance and anti-smoking campaigns: A theoretical approach, Health Communication, vol. 15, pp. 349-366. Jackson, S. E. and R. S. Schuler (1985), A meta-analysis and conceptual critique of research on role ambiguity and role conflict in work settings, Organizational Behavior and Human Decision Processes, vol. 36, pp. 16-78. Jaworski, B. J. and S. M. Young (1992), Dysfunctional behavior and management control: An empirical study of marketing managers. Accounting, Organizations and Society, vol. 17, pp. 17-35.
13
Kaptein, M. (2010), The ethics of organizations: A longitudinal study of the U.S. working population. Journal of Business Ethics, vol. 92: pp. 601-618. Kaptein, M. (2011), Understanding unethical behaviour by unravelling ethical culture, Human relations, vol. 64, pp. 843-869. Karapetrovic, S. and Willborn, W. (2001), Audit and self-assessment in quality management: comparison and compatibility, Managerial Auditing Journal, vol. 16: pp.366–377. Katz-Navon, T., E. Naveh, and Z. Stern (2005), Safety climate in healthcare organizations: A multidimensional approach, Academy of Management Journal, vol. 48, pp. 1075-1089. Knights, D. and D. Collinson (1987), Disciplining the shopfloor: A comparison of the disciplinary effects of managerial psychology and financial accounting, Accounting Organizations and Society, vol. 12, pp. 457-477. Lawler, E. E. and J. G. Rhode (1976), Information and Control in Organizations, Pacific Palisades: Goodyear. Merchant, K. A. (1982), The control function of management, Sloan Management Review, vol. 23, no. 4, pp. 43-55. Merchant, K. A. (1990), The effects of financial controls on data manipulation and management myopia, Accounting, organizations and society, vol. 15, pp. 297-313. Roth, J. (2003), How do internal auditors add value?, Internal Auditor, vol. 60, pp.33–37. Rousseau, D. M. (1978a), Characteristics of departments, positions, and individuals: Contexts for attitudes and behavior, Administrative Science Quarterly, vol. 23, pp. 521-540. Rousseau, D. M. (1978b), Measures of technology as predictors of employee attitude, Journal of Applied Psychology, vol. 63, pp. 213-218. Shah, J. Y., R. Friedman, and A. W. Kruglanski (2002), Forgetting all else: On the antecedents and consequences of goal shielding, Journal of Personality and Social Psychology, vol. 83, pp. 1261-1280 Shannon, C. E. (1948), A mathematical theory of communication. Bell System Technical Journal, vol. 27, pp. 379–423 and pp. 623–656. Simons, R. (1995), Levers of Control: How managers use innovative control systems to drive strategic renewal, Boston: Harvard Business School Press. Smith, P. (1995), On the unintended consequences of publishing performance data in the public sector, International Journal of Public Administration, vol. 18, pp. 277-310. Vygotsky, L. S. (1978), Mind and Society: The development of higher psychological processes, Cambridge: Harvard University Press. Wilde, G. J. S. (2001), Target Risk 2: A new psychology of safety and health, Toronto: PDE Publications.
14