In the 1990’s, we entered what is called the third age of safety which is characterized by an increasing focus on safety management. The first-line supervisor is a part of the management but has several advantages to the top management in his possibilities for promoting safety, because of the frequent contact with the workers. There are several supervisory styles that are more or less in favor of promoting safety. First, the supervisor’s beliefs about the subordinates, his concern for them and his safety commitment are important. They determine how he interacts with the subordinates and gives an impression of whether he cares about their well-being and safety. Second, the way the supervisor makes decisions and what feedback is given to performance is important for the subordinates’ safety behavior as well as modeling the right behavior. However, even though the supervisor wants to promote safety, it is often limited by organizational constraints such as his own authority in safety decisions. Using the supervisor as a safety promoter presupposes both sufficient leader training and adjusting his authority and resources to the responsibility he has for safety.

Hale and Hovden (1996) describe the development of safety research as evolving from focus on human, then on interaction between human and technology, and finally on management. The period of the 1970s and the early 1980s was the age of human factors. The causes of accidents was attributed to individuals who didn’t perform their job the way they should. Research focused on accident proneness - the belief that some people are, because of their personality, more exposed to accidents than others (McAfee & Winn, 1989; Hoyos, 1992). However, one realized after some time that a focus on interaction between humans and technology was more fruitful. The matching of human and technology has been the preoccupation of the socio-technic school (Robinson, 1982). Finally, in the 1990s, we entered the third age of safety, with an increasing focus on safety management. The central question is: How can leaders in organizations manage the safety hazards employees face in their daily work ?

The leaders at different levels of organizations, are thus seen as responsible for promoting safety among employees. When accidents or injuries occur, one looks to the leaders to find the roots of the problems. Human behavior and socio-technic interaction are also common reasons for accidents and injuries, but the main focus is on safety management. In this paper, I will discuss the role of the leaders at the lowest level of an organization; the first-line supervisors. The first-line supervisor was already in 1931 characterized as important in accident prevention. "The supervisor or foreman is the key man in industrial accident prevention. His application of the art of supervision to the control of worker performance is the factor of the greatest influence in successful accident prevention" (Heinrich, 1931, p. 14). According to Weber (1992), Heinrich was by many considered the father of industrial safety. Other researchers have also emphasized the role of the first-line supervisor in safety (e.g. DeReamer, 1958; Konczal, 1979; Dunbar, 1975; Weber, 1992; Simard & Marchand, 1994).

There are several reasons for emphasizing the role of the first-line supervisor. First, for new safety programs and instructions to be applied, the management depends upon the first-line supervisor for accomplishment. The information from the top management will be filtered through the first-line supervisor when he presents new instructions and orders from the top management to the workers (Clarke, 1999). Second, the workers depends on the first-line supervisor for passing on their safety concerns to the top management. This means the first-line supervisor is a key man in safety both for the management and the workers. Third, the first-line supervisor plays an important role as a motivator for good safety performance. Because he is the person who has most frequent contact with the workers, he has the power to influence and motivate in a way other leaders in the organization do not have (DeReamer, 1958). For these reasons, among others, the first-line supervisor is of great importance for promotion of safety in industrial organizations.

But why is safety promotion important ? Most organizations experience injuries and accidents from time to time. Maybe one of the most common reasons for accidents or injuries, is violation of safety instructions. There are many reasons for not following the rules, such as feeling pressure to prioritize production before safety, little motivation to work safe, incomplete instructions, or that safety has not been stressed enough by superiors. This introduces a challenge for first-line supervisors in organizations where safety is an important aspect of daily work. But although the first-line supervisor is important for motivating safe behavior, there are great differences between first-line supervisors and the organizations they work for. The first-line supervisor can be placed along a dimension from promoting safety to discouraging safety, and the organizational conditions can be from ideal to hard for safety promotion. The difference between supervisors who promote safety and those who do not, is a matter of the supervisors’ beliefs, concerns and behavior. However, if his beliefs, concerns and behavior are positive to safety, his influence on the workers might be limited by organizational factors such as his span of control, or average number of workers per supervisor (Simard & Marchand, 1994).

The main questions to be discussed in this paper, are:

• What characterizes a supervisor who promotes safety ?
• How is the supervisor’s influence limited by organizational factors ?

The focus will mainly be on discussing the supervisor’s beliefs, concerns and behavior, but important organizational conditions, such as the possibility for influencing upward, the span of control and the top management’s commitment to safety will also be discussed throughout the paper. The supervisor’s beliefs and concerns are important because they shape the way he interacts and communicates with the subordinates. In the first part of the paper, I will describe perspectives on the supervisor’s beliefs and concerns. McGregor’s (1968) theory X and Y, the managerial grid (Blake & Mouton, 1968) and safety commitment are relevant to this. Managerial grid is both a theory of leader concerns and behavior, but the focus will be on concerns. In the second part of the paper, I will discuss some aspects of the supervisor’s behavior; how he makes safety decisions, how he models safety behavior and finally how he gives feedback to subordinates. When outlining feedback, operant learning principles will be examined, since these have been widely used in safety research. It is natural to discuss different questions in the first two parts of the paper and sum up in the discussion, where I will discuss the role of organizational factors and whether one leader style can be called ‘best’.

The setting is limited to first-line supervisors in industrial organizations, and also to personal safety such as use of personal protective equipment. The term supervisor will be used, referring to the first-line supervisor. Further, I will use "he" when referring to the supervisor, because it’s probably most common with male supervisors in industrial companies. Studies from safety research will be used when possible, and studies from general research on leadership style will be used where no safety studies are known. Mattila, Hyttinen, and Rantanen (1994) state that effective safety supervision has similar characteristics as general management principles, which is an argument for using studies from other research areas.

I will define promoting safety as influencing the workers so that they perform the work in a safe way, or have safe work practices. "Safe work practices are procedures or behavioral sequences that employees are expected to follow in order to reduce exposures to harmful agents or to minimize the probability of injury" (Cohen & Colligan, 1997, p.380).

At first, definitions of beliefs and concerns are needed. With beliefs, I mean how the supervisor looks upon his subordinates. With concerns, I mean how concerned the supervisor is with his employees’ well-being and safety. His beliefs and concerns will influence the workers both directly and indirectly. First of all, beliefs and concerns have a direct effect on subordinates, because they perceive him as more or less positive instilled to them and interested in their safety. This will probably affect their own concerns for safety and satisfaction with their job. Second, his beliefs and concerns will influence his behavior and his behavior will affect the subordinates, as I will discuss later. Different ways of thinking may thus have great consequences for the promotion of safety. Let’s look at three different aspects of the supervisor’s beliefs and concerns: Theory X and Y, the managerial grid, and safety commitment. McGregor’s (1968) theory X and Y describe two ways of thinking about subordinates, such as lazy or engaged. Blake and Mouton (1968) describe two dimensions of leadership style, concern for production and people, or the managerial grid. The safety commitment construct has been widely used in research on safety management and has been a good predictor of accident rates and risk behavior (e.g. Cohen, 1977; Smith, Cohen, Cohen, & Cleveland, 1978; Dedobbeler & Béland, 1991; Cohen & Cleveland, 1983; Simard & Marchand, 1994; Simard & Marchand, 1997; Hoffman and Morgeson, 1999).

Theory X and Y (McGregor, 1968) are two different ways leaders can look at subordinates. The original theory focuses on the leader in a general work setting, but for this purpose I will use the term supervisor and focus mainly on safety.

The traditional belief (theory X) of supervisors in organizations, according to McGregor (1968), is that workers are lazy and prefer to be led rather than have responsibility. The laziness is due to an inherent reluctance to work. Because of this, the workers have to be forced, controlled, instructed and threatened to be punished in order to perform sufficiently. This belief results in a supervisor always checking that people are at work and that they in fact are working (Petersen, 1976). Further, people are believed to work only because they need money. To threaten with punishment such as sacking, is therefore believed to be effective. The workers are also seen as having low ambitions and striving for security. The supervisor will therefore use harsh discipline, punishment, close supervision and threats to achieve the goals of the organization, such as the safety rules. He will probably also make safety decisions without asking the employees, assuming they prefer to be led rather than have responsibility. However, since the theory X supervisor has little respect for the subordinates’ worth, he might also have little respect for their safety (Denton, 1982). But if he is interested in enforcing the safety rules, he will probably do this by close supervision.

McGregor (1968) was a representative for the school of human relations, proposing that employees have other needs and motives than security and payment. McGregor therefore presented theory Y as an alternative way of looking at subordinates. The assumptions of theory Y are quite different from those of theory X; Human beings do not despise work, but experience work as either satisfying or punishing, dependent on factors they control themselves. The use of close control and threats of punishment are not the only ways of making people work and accomplish the safety goals of the organization. If the workers are satisfied with the work and given opportunity to participate in goal-setting, they will perform well without supervision. If the workers identify with the safety tasks at hand, they will also be independent and control their own safety work rather than having to depend on external control. To identify with the safety goals and tasks at hand, the organization have to reward a successful execution of the tasks. The most valuable rewards are self-realization and regard, and not payment as in theory X. The theory Y supervisor will thus try to promote identification with safety tasks, and give the workers responsibility for acting safely. He will use positive reward such as praise and job promotion for acting safely, rather than strict control and punishment for acting unsafe. In contrast to theory X supervisors, who see the workers as someone who do not want responsibility and prefer to be led, the theory Y supervisors believe that most humans will try to achieve responsibility if the conditions are facilitating. The reason for not wanting responsibility is believed to be bad experience rather than inherent reluctance. As we can see, the theory Y assumptions lead to a very different supervisory style than those of theory X.

It may be artificial to categorize supervisors as believing in either theory X or Y. Most supervisors will probably be somewhere in between. Using the categorization might however help to describe which supervisors who promote or do not promote good safety practice. A few studies have outlined the differences. Eyssen, Hoffman, and Spengler (1980) found that foremen in districts with high injury rates believed in the need for strict discipline, close supervision, criticism of mistakes and enforcement of rules. It looks like these foremen believed in theory X, and treated their subordinates thereafter. But surprisingly, foremen who believed in the value of criticism were also found in districts with low injury rates. In a study by Zohar (1980a), the management in a large metal fabrication plant was production-oriented and authoritative, based on theory X assumptions. Despite this fact, the company was known for it’s "outstanding safety record due to top management involvement in all safety-related issues" (Zohar, 1980a, p. 80). Another study found shown that theory Y assumptions led to better worker performance than theory X assumptions (Duke, 1972). The evidence is not conclusive, though theory Y seems to be the best for safety promotion. More research is needed to find out what beliefs are best for promotion of safety. How the supervisor looks upon his subordinates will anyway give the workers an impression of how interested he is in their safety and well-being. A theory X supervisor who strictly controls safety performance, can be perceived as someone occupied with safety rules, but not very interested in the workers’ well-being. This can create reactance. Cohen and Colligan (1997, p.385) characterizes reactance as: "Individuals who feel that their behavior is being manipulated or unfairly coerced will react defiantly as a means of establishing their independence and restoring their sense of control." The workers’ might feel they are loosing control over the work situation when under close control, resulting in a counter reaction where the workers follow the safety rules only when the supervisor is present. Though theory X can lead to fewer accidents, it can create a negative atmosphere among the workers. Even if safety performance is high, it does not mean the employees are satisfied with the work. It would be better if both safety performance and satisfaction were high. But then again, some workers might tolerate a supervisor who uses punishment and close supervision, though in today’s society such supervisor behavior is probably not very accepted.

I have now taken a look at what beliefs the supervisors can have of his workers. These might be connected to the supervisor’s concern for production and people. A theory X supervisor can give an impression that he is not very concerned with the workers.

Blake and Mouton’s (1968) managerial grid has been a very popular theory when it comes to describing the traits of a good leader (Riggio, 2000). I will here concentrate on the leader’s concerns rather than behavior. Blake and Mouton describe leadership on two dimensions: concern for production and concern for people. Concern for people implies in a way concern for safety. Concern for workers is concern both for their physiological and psychological well-being. In industrial organizations, safety is mostly associated with physiological well-being. However, workers’ psychological well-being is also important for the workers’ motivation to work in a safe way. Denton (1982) proposes for instance that fulfilling the workers higher needs can motivate them to work safe, and this contributes to their psychological well-being.

The dimensions Blake and Mouton (1968) describe are especially relevant for safety. A high concern for production might result in a lower priority for safety since safe behavior usually takes some time and effort, and makes production slower. "A major concern in occupational safety is whether reinforcement for working rapidly, especially when reinforcement is based on productivity, jeopardizes safety performance" (Sulzer-Azaroff, 1982, p. 525). Studies have shown that workers’ perception of demands for production versus safety has been a good predictor of risk behavior (e.g. Rundmo, Hestad, & Ulleberg, 1998).

In the managerial grid, concern for production is placed on a scale from 1 (low concern) to 9 (high concern). The same scale is used on concern for people. Blake and Mouton (1968) describe five different leadership styles (see figure 1). Although they only have outlined five styles, 81 combinations of concerns are possible in the managerial grid. Blake and McCanse (1991, cited in Riggio, 2000) have named the five styles impoverished management, country club management, middle of the road management, authority compliance management and team management. Denton (1982) has applied them to safety issues, and I will in the following use his perspective. I will also use the term "supervisor", which here is more appropriate than "leader".

9	Country club management								Team management
8
7
6
5					Middle of the road management
4
3
2
1	Impoverished management								Authority compliance management
	1	2	3	4	5	6	7	8	9

1.1 Impoverished management: The 1.1 supervisor is not interested in high production rates, neither in the workers’ well-being. He believes the workers are innately lazy (Petersen, 1976), as in theory X. The purpose of the work is to just getting by. You can hardly call the 1.1. supervisor a leader at all, therefore the term impoverished management. He will have little influence over what the subordinates do or do not, and they will drift through the workday with no goals other than the ones they set for themselves. Impoverished management is approximately the same as Lippit and White (1943, cited in Brown, 1988) calls laissez-faire leadership style. They found that groups with laissez-faire leaders were the least productive groups. The 1.1. supervisor is probably preoccupied with other things than his leader responsibilities. One can assume that he therefore is a rather bad safety promoter.

• 1.9 Country club management: The 1.9 supervisor has little interest in production but a high concern for subordinates. He sees most accidents as caused by unsafe acts and prefer to talk his employees into working more safely. However, he will be so preoccupied with keeping good relationships with subordinates, that he might turn the back when safety regulations are broken (Denton, 1982). Safety problem records may not change because the supervisor is not willing to discipline his subordinates. The attitude is more like "let us have good relationships here at work" rather than "let us keep up the good work". Just concern for people is apparently not enough for good safety promotion.

• 5.5 Middle of the road management: The 5.5 supervisor has some interest for people and some for production. However, his concerns are not strong enough for acting upon. He is most affected by the company’s policy requirements and to do as the rules dictate. He wants to maintain status quo. He does not give any suggestions for improving safety, but rather goes along with what others suggest. A major problem may be that the 5.5. supervisor only will influence subordinates in a positive direction if the top management tells him to.
• 9.1 Authority compliance management: The 9.9 supervisor has a high concern for production but low for people. He will let production goals be prioritized before safety, which is seen as a hindrance to production efforts. He thinks the best is to press his subordinates to work for higher production (authority compliance), and this implies often in an unsafe way. Workers who perceive a strong pressure to perform, will focus on getting the job done and less on safety procedures. This may result in using shortcuts to meet the demands of production (Hofmann & Stetzer, 1996). Roethlisberger and Dickson (1940, cited in Eyssen et al., 1979) found that accidents occurred more frequently among workers who perceived their managers to lack concern for them.

• 9.9 Team management: The 9.9 supervisor is highly concerned both with production and people. He is the team-player, who seeks to maintain all company and employee performance at a high level. "He expects the best from people and gets it because he knows they will be happier as they do achieve goals on their job" (Petersen, 1976, p. 82). This belief resembles theory Y. Further, the 9.9 supervisors considers the employees’ safety as his responsibility and does not resist safety suggestions from the subordinates. "Because they see that their supervisor regards safety equally important as production, they can also expect that he will react positively when they work safely" (Andriessen, 1978, p.374).

The 9.9 leader is seen as the one best leader ship style in the managerial grid. Denton (1982) also states that 9.9 is the ideal leadership style for the supervisor who is to promote safety. One might think that high concern for people would be enough to promote safety, but apparently this can create problems. Concern for production and people should thus be integrated. "Safety should be regarded as an integral part of the production system closely related to the overall degree of control management has over production processes" (Zohar, 1980b, p. 101). Gardner (1969) writes that promotion of safety needs an active training and involvement from the supervisor: "He must supervise the workers in safety in much the same way as he supervises them in efficiency and production; that is, he must train them in safety, along with other requirements of the job, and follow up to see that their job performance continues to meet these requirements" (Gardner, 1969, p.6). Heinrich (1931) also states that if the supervisor already knows how to enforce production rules in a good manner, he just has to do the same to enforce safety rules.

There are some problematic points about the managerial grid. According to Riggio (2000), the theory has been criticized for it’s universality. The 9.9 style is seen as the best in all situations. Riggio states that most researchers think the leader’s effectiveness depends on a fit between leader and situation. I will outline this some more in the discussion. Another problematic point, is that organizational factors might limit the influence of the supervisor’s concerns. The top management’s interest in safety or production affects the supervisor. The supervisor is most likely interested in doing a good job, which is the basis for getting promoted. Therefore, the supervisor is probably interested in following instructions from the top management. If the management is more interested in production, the supervisor might feel a press to emphasize production before safety, even though he is interested in safety. "…. the supervisor is vulnerable to and dependent upon the subordinate; inadequate work by the subordinate threatens the supervisor’s position with respect to his or her own boss." (Krantz, 1993, p. 6). Metzgar (1999) criticizes Heinrich’s (1931) emphasizing of the supervisor’s key role in accident prevention. He states that the supervisor is confronted with everlasting demands from the government’s regulations, what the boss wants and what he himself is willing to accept responsibility for. The supervisor’s boss is seen as more important than the supervisor, because of what he expects from the supervisor. Metzgard’s statement can be discussed, but he is probably right about supervisors being very influenced by their boss’ expectations in some situations. This clearly limits the supervisor’s possibilities for influencing subordinates in a positive direction, and might be an explanation of why some supervisors are seen as little committed to safety.

One of the most important attributes of the first-line supervisor, is his own commitment or concern for safety. In what extent does he think safety is an important issue ? Many studies have shown that management’s commitment to safety can predict both safety attitudes and safety behavior of the employees, and have a negative effect on accident rates (e.g. Cohen, 1977; Smith, Cohen, et al., 1978; Dedobbeler & Béland, 1991; Cohen & Cleveland, 1983; Simard & Marchand, 1994; Simard & Marchand, 1997; Hoffman and Morgeson, 1999). The workers’ commitment to safety is also important; "A strong commitment to safety benefits the organization by increasing safety compliance behavior (…), reducing the number of accidents and reducing the cost associated with accidents " (Hofmann & Morgeson, 1999, p. 288). If the supervisor is strongly committed to safety, his commitment can affect the workers’ commitment and in this way promote safety.

But what is safety commitment ? It is adopted from the construct organizational commitment, which means "the relative strength of an individuals identification with and involvement in an organization" (Mowday, Steers, & Porter, 1979, cited in Arnold, Cooper, & Robertson, 1998, p. 209). An interpretation could be that safety commitment is an identification with the organizations’ safety program, and involvement in the program. However, if the organization does not have values that promote safety, the supervisor can still be committed and occupied by safety. Nelson (1998, p. 41) states that safety commitment has been defined as "no injury is acceptable, all injuries are preventable and we will do whatever it takes to prevent injury". However, this definition is developed by companies in the construction industry and not in an academic context. It seems that the definition of safety commitment is missing in safety research. Nelson concludes the view that worker injury is unacceptable for humanitarian reasons, should be the definition of the construct. However, it might be more important to look at the implications of safety commitment rather than the construct’s meaning. Safety commitment has been used as an underlying factor for variables such as management’s safety attitudes, management’s involvement in safety program, in safety decisions and in safety activities, level of development of the safety program, the group leader’s concern for safety performance of his group, and how much the group leaders take responsibility for the organization’s safety record (Dedobbeler & Béland, 1991; Simard & Marchand, 1994; Simard & Marchand, 1997; Hofmann & Morgeson, 1999). Safety commitment is thus indicated by most activities and opinions that are associated with safety, and the safety activities of the supervisor should therefore be good indicators of the supervisor’s safety commitment.

However, for workers to be influenced by their supervisor’s safety commitment, they have to perceive his commitment correctly. The supervisor’s commitment can be perceived by how much he talks about safety and in what way. A supervisor with a strong commitment to safety, will most likely talk about safety quite often and in a positive way. If the supervisor fails to communicate his commitment, this might give the subordinates an impression that he is not interested in safety at all. As a result, they will not be very committed to safety either (Hofmann, Jacobs, & Landy, 1995). Konczal (1979) encourages supervisors to communicate their concern for safety, and states that effective communication can help formulate safety attitudes. But the problem is that even if the supervisor is committed to safety, the management might not be committed to safety, and this may affect the workers so that they work less safe (Andriessen, 1978).

As I have outlined, the supervisor’s safety commitment, his concern for people and production, and in what way he looks at subordinates are important for promotion of safety. I will now go on to examine the supervisor’s behavior, which also may have a profound effect on workers’ safety performance.

In addition to the effect of the supervisor’s beliefs and concerns, his behavior can also influence workers’ safety practices. Simard and Marchand (1994) found for example that supervisors’ effectiveness in accident prevention was dependent upon their behavior. I will discuss three aspects of the supervisor’s behavior; decision making, behavior modeling and feedback. Feedback is here limited to operant conditioning. According to Denton (1982), there are two approaches on how to improve safety performance by motivating change, the humanistic and the behavioristic. While the humanistic approach focuses on inner needs and motives, the behavioristic focuses on behavior. The use of participating or democratic decision making is a tool derived from the humanistic tradition, while behavior modeling and operant conditioning stems from the behavioristic tradition.

Decision making has been an important aspect of the leader’s behavior since Lippit and White (1943, cited in Brown, 1988) performed their studies with school boys. They found the way a group leader involves members in decisions, has a profound influence on the group’s productivity. In addition to decision making, the supervisor can also influence the workers by modeling appropriate behavior. Behavior modeling is a concept derived from social learning theory (Bandura, 1977). Finally, feedback is an essential part of the supervisor’s behavior. Many studies have shown that use of operant conditioning can improve safety performance (e.g. Rubinsky and Smith, 1973; Smith, Anger, & Uslan, 1978; Zohar, 1980a; McAfee & Winn, 1989; Guastello, 1993).

One of the leadership style dimensions that can be traced back to early research on group processes, is called democratic-autocratic leadership. Though autocratic and democratic leadership style are poles of a dimension, I will here concentrate on these two extremes to make the differences more clear. Lippit and White (1943, cited in Brown, 1988) examined leadership style of school boys, who was to lead performance of tasks in groups with other boys. They found that groups with autocratic leaders worked very hard, but when the leader left the room, they stopped to work. While the democratic groups didn’t work as hard as the autocratic groups, the work was little affected when the leader left the room. An autocratic supervisor makes all the decisions himself and then tells the workers what to do. A democratic supervisor lets the workers participate in decisions that are important to them.

Why is decision-making important for safety promotion ? Making decisions, according to Scandura, Graen, and Novak (1986, p. 579), "is one of the most important functions performed by managers in organizations." They also state that making decisions can enhance the effectiveness of the work unit. Thus, the way decisions are made can enhance safety performance as well, as safety performance is one of the tasks of the work unit. But why does the democratic and the autocratic supervisor create such different reactions from subordinates ?

When decisions that affect the workers’ situation are made without asking for their opinions, they may feel that they loose control over the situation, and reactance may be created. A supervisor who tries to promote safety by behaving autocratic, always telling the subordinates what to do, may thus result in subordinates following the safety rules while the supervisor is present and taking short-cuts, to restore a sense of control, when he leaves the room. This might give the supervisor an impression that autocratic behavior is effective. He might also have a theory X belief, that the workers do not want responsibility and prefer to be told what to do, resulting in a belief that autocratic decision making is the best. However, Logozzo (1989) found that theory X was associated with letting subordinates participate in decisions, and it is thus not clear whether theory X always leads to autocratic decision making. Anyway, autocratic decision making can create reactance. Eventually an accident might occur, and the supervisor does not understand why it happened, because he thought the employees were acting safe. Another reason for noncompliance with autocratic made decisions, might be that workers do not understand the reasons for doing what the supervisor says. A way to compensate for decisions that are taken autocratic, is to explain to the workers why it is important to follow instructions. For instance if an industrial plant has decided that all workers shall wear helmets when entering a special area, it’s easier to accept if explained how it can prevent injuries. To have an understanding that a behavior prevents injury is an essential part of the Health Belief Model (Rosenstock, 1974). Rosenstock states that people who do not believe a behavior prevents injury, will not perform the behavior. However, the knowledge of hazards, for example that exposure to noise causes permanent hearing loss, is not enough. The knowledge often lacks motivating qualities and personal relevance (Zohar, 1980a). Autocratic decision making might also lack motivating qualities. Several studies have found that autocratic leadership style increases the probability of accidents. Andriessen (1978) reviewed studies on leadership style and safety, and found that autocratic leadership style was associated with more accidents. Autocratic supervision was also associated with the occurrence of accidents in a study by European Community of Coal and Steel (1967, cited in Eyssen et al., 1980). However, Eyssen et al. did not find any association between autocratic management and high accident rates. The authors suggest this may be due to inadequate assessment of this aspect of the worker environment. Another explanation might be that in some companies, autocratic supervision is more accepted and therefore more effective.

Letting workers participate in decisions that affect them, might be a better alternative than autocratic decision making. Participating decision making is here the same as democratic decision making. Several studies have shown that participation in decision making can improve safety performance of subordinates (e.g. Matilla et al., 1994; Lee, 1998; Thompson, Hilton, & Witt, 1998). Denton (1982), following Maslow’s (1970) hierarchy of needs, states that using participation as means of motivating safe behavior, applies to the subordinates’ higher needs, such as esteem and self-actualization. Most people have fulfilled their lower needs, such as physiological and love needs. Applying to the higher needs should therefore be effective for motivating to safe behavior. Letting workers participate in decisions that affect them also lets them keep control, and can create motivation rather than reactance. Thompson et al. found that supervisor fairness (how much the workers get to participate in decisions that affect them) predicted supervisor support for safety (how concerned the supervisor was about the workers’ safety), that again predicted the workers’ safety compliance. Lee (1998) reviewed studies on low accidents plants, and found that one of the characteristics of such plants is a democratic, cooperative, participative and humanistic management leadership style, as distinct from autocratic and adversarial. Mattila et al. (1994) also found that supervisors with a high safety index (high level of conditions and safety behaviors in their department), made more use of participatory style in discussions. However, the performance of subordinates can also be a reason for including the subordinates in decision making (Scandura et al., 1986). The use of surveys as empirical evidence for the positive effect of democratic decision making can thus be questioned. The relationship between performance and participation can go both ways. In the future, studies should be performed to reveal the causal relationship between the two variables.

Even if democratic decision making does have an affect on performance, it is not always possible to use it. The supervisor might not have the authority to make the safety decisions that affect the workers’ situation (Schoenwald & Kopp, 1986). The decisions concerning safety might be made by a safety executive, safety committee or the top management. If the subordinates perceive their supervisor as having little influence in safety decisions, their motivation may decrease. Simard and Marchand (1997) found that supervisor’s participation in safety decisions correlated with safety rule compliance. Simard and Marchand also state that safety compliance is higher when supervisors have some power over decisions that affect the workgroup’s safety. "To function effectively, a supervisor must have sufficient influence with his own superior to be able to affect the superior’s decisions when required" (Likert, 1961, p. 94). However, the supervisor’s authority in safety decisions depends on the organizational structure. The more levels of management between the supervisor and the safety decision makers, the less possibility the supervisor has for influencing safety decisions. Child and Partridge (1982, p. 39) state although: "…it is possible for supervisors to influence decisions for which they have no formal authority". Even if the supervisor does not have authority, he can still try to communicate the workers’ concerns to those who make the decisions. Child and Partridge examined supervisors in an engineering and a food plant in Birmingham. The plants had formal procedures for disciplining and safety, that limited the supervisors authority. However, many supervisors claimed their responsibilities for safety had increased. Child and Partridge found that supervisors had somewhat more influence than formal authority over safety decisions. As we can see, supervisors can in some situations influence safety decisions where they themselves do not have formal authority.

However, it is more common today to give the supervisors more authority than before. Cohen and Colligan (1997) states that worker participation has become more and more common as the result of two lines of development. The need to compete in the market has caused companies to downsize and restructure their organizations. This has given middle-management and workers at lower levels more autonomy. The second development is the acceptance of total quality management principles. This empowers workers to solve problems, help improve processes and ensure quality etc. It also means that the supervisor has more authority concerning safety today than he had before. But even though the supervisor does not have the power to make all the decisions concerning safety, there are always some decisions in which he can involve the workers. One decision that is quite important for improving safety performance, is setting goals. According to Fleming and Lardner (1999), involvement of workers in goal-setting, gives a positive effect. Several studies have succeeded in using goal-setting as means of improving safety performance (e.g. Chhokar & Wallin, 1984 and those cited in McAfee & Winn, 1989: Reber & Wallin, 1984; Reber, Wallin, & Chhokar, 1984).

In addition to supervisor’s authority, the characteristics of the work force can also be a problem. One is that the more people you involve, the more time-consuming the decision making will become. Thus, "if there is a pressure for an immediate decision, participative decision making may not be feasible" (Locke & Sweiger, 1979, p. 322). If the supervisor is in charge of many subordinates, it will be quite difficult to involve all the workers. And even though there are not so many subordinates, the supervisor probably has a lot of work to do, and it’s much easier for him to decide autocratically, because it is the most convenient thing to do. But then again, this depends on how he sees his workers (theory X or Y) and how concerned he is with safety. Another problem that arise when workers participate in safety-decisions, is that they might not have the information they need to participate. A suggestion from the supervisor of improving safety might be voted down because the subordinates do not understand how it can affect the probability of an injury or accident, as one can hypothesize from the Health Belief Model (Rosenstock, 1974). Locke and Sweiger reviewed literature on decision making and found that one of the most important contextual factors that determined usefulness of participation, was the subordinates’ expertise. It is therefore important that the supervisor explains why safety improvements are made and what they are supposed to accomplish.

Democratic supervisors seem to be the best safety promoters, but this depends on organizational factors such as the workers’ expertise, the supervisor’s authority in safety decisions, and his span of control. However, if the supervisor is a poor role model for safety performance, it might not matter how the decisions are made.

"Employees tend to watch and model those above them. Exclaiming ‘Do not do as I do, do as I say !’ will not get a supervisor very far." (Kirkpatrick and Lewis, 1995, p.17). A supervisor who speaks warmly of safety, but acts unsafe when performing operations is a rather bad safety promoter. The supervisor may be used as a role model for subordinates, whether he works safely or not. Role models are persons who influence the behavior of persons around them, by the power of their position, according to Kirkpatrick and Lewis. Wogalter, Allison, and McKenna (1989) found that the presence of a confederate who complied to a warning to wear safety equipment while performing a task, had a profound effect on the subjects’ compliance with the warning. Behavior modeling can thus be used to enhance safety performance. Modeling behavior is a central part of social learning theory (Bandura, 1977). Subordinates can observe the behavior of their supervisor and then imitate his behavior. However, in order to imitate, one has to pass through several stages. In the first stage, the individual attends to the model’s behavior. This happens for example when a worker observes a supervisor putting on safety equipment before performing an operation. Second, the information from observation must be recalled. If the worker attended to the situation long enough, he creates a mental image of the supervisor’s behavior. Third, the observer must be able to perform the observed behavior. The worker must know how to put on the safety equipment and adjust it, to perform the modeled behavior. Finally, the observer must be motivated to perform the behavior, such as using safety equipment. When the four conditions are met, the likelihood of imitation increases (Bandura, 1977; Racicot & Wogalter, 1995).

There are three different effects of behavior modeling; observational learning, the inhibitory or disinhibitory effect and the response facilitation effect (Bandura, 1969; 1977). First, a person that observes another person’s behavior may acquire a set of new behaviors that he didn’t have before. Observational learning is for instance when a new employee watches the supervisor using personal protective equipment, and afterwards uses the equipment himself. Second, observation of modeled behavior may strengthen or weaken inhibitory responses. This means when a person perceives the behavior of the model as having positive effects, imitation will increase (disinhibitory). But when the effects are perceived as punishing, imitation will decrease (inhibitory). If a worker observes the supervisor wearing a safety helmet, but the only effects he observes are complains of heat and itching, the effects are seen as punishing, and imitation will decrease. Third, the behavior of others serve as discriminative stimuli in facilitating previously learned behavior. In this situation, no new responses are learned. A worker who has been employed for a long time knows how to wear a safety helmet, but if he watches the supervisor who wears a safety helmet, this can increase the likelihood of him also using a helmet.

Since new employees are involved in more accidents than experienced workers (Levitt & Samelson, 1993), they need special attention and modeling for acting safely. If the newcomer has the supervisor as a role-model, he will know how to perform safely. But unfortunately, there is nothing automatic in observation and modeling. As I outlined, the observer has to pass through several stages for imitation to occur. However, before he passes through these stages, a model is chosen. Unfortunately, the model that is chosen can be someone else that the supervisor. Coworkers are also role models. An essential point is therefore attention. According to Weiss (1978), some types of models are more often chosen than others. The extent of imitation is influenced by three groups of factors; factors tied to the model (status, competence, nurturance, attractiveness, power, age, sex), factors tied to the observer (self-esteem, dependency, reinforcement history of imitation) and factors in the situation (task ambiguity, reward contingencies). All these interact to influence the amount of imitation (Bandura, 1969; Weiss, 1978). Attributes of the model such as success, competence and nurturance carry information of how appropriate imitation is (Weiss, 1978). Sulzer-Azaroff (1982) also states that similarity to the model determines amount of imitation. If the supervisor is seen as an appropriate model, imitation will increase; "Employees who see an admired supervisor performing in a certain manner will respond the way they think the supervisor would respond" (Denton, 1982, p. 76). But if the supervisor is not seen as an appropriate model, there will be no imitation. For example if the supervisor has worked in the department shorter than oneself, he might not be seen as an appropriate model because he lacks competence. The coworkers might also be seen as someone who are more similar to the observer than the supervisor. Simard and Marchand (1997) found that supervisors’ experience could predict compliance of safety rules. An explanation could be that the supervisor is not seen as an appropriate model. Choosing another model, such as a coworker, might then be more appropriate. However, this may depend on how competence is defined. If the supervisor has the highest education in his department, he might be seen as competent enough and an appropriate model.

Another factor that determines imitation, is the supervisor’s power. Adler (1983) found that similarity between supervisors and subordinates were related to the supervisor’s power, assuming that similarity was caused by subordinates’ imitation. People with high self-esteem are also less likely to imitate a model than people with low self-esteem, because people with low self-esteem are less confident in ambiguous situations. Weiss (1978) performed a study where he examined the degree of similar work values between subordinates and their supervisors, assuming that subordinates develop the same work values as their supervisors by imitation. He hypothesized that the degree of value similarity was related to the supervisors success, competence and consideration for subordinates. He also hypothesized that similarity would be higher among subordinates who had lower self-esteem. As social learning theory proposes, sex was correlated with value similarity of the subordinate, but the correlation was low. Weiss also found a positive correlation between supervisor-subordinate value similarity and the supervisors’ consideration. However, for subordinates with low self-esteem, the value similarity was related to the supervisors success and competence in a positive direction (measured as frequency of good/bad decisions) but in a negative direction for people with high self-esteem. This can be explained with the fact that competence and success are not necessarily mediated by the model. The observer must therefore compare the role information from the model with information generated internally. With consideration, the outcomes are mediated by the model, and is therefore more relevant. Adler (1983) did however not find any correlation between self-esteem and similarity of supervisor and subordinates. Self-esteem is thus a matter of discussion in relation to modeling.

In order for the supervisor to promote safety, he thus has to be a good role model, he must show the workers that he cares for them (high concern for people), and he must be seen as successful. If the supervisor is not looked upon as an appropriate model, the workers might turn to other models such as coworkers. If the coworkers does not follow safety rules, neither will the observer. It is however more likely that the supervisor is chosen to be a model, because he has more power and status. Vandenput (1970) made a study on social influences on safety habits. He found that the supervisors had more influence on safety behavior than the coworkers had. The supervisors and the coworkers punished and rewarded the same behaviors, but the supervisors’ pressure covered a wider range of actions. This means that the supervisor often has higher possibility for influence, because he has greater power than coworkers. Social learning theory also predicts that the amount of imitation is related to the model’s power.

In addition to the problem of other models, how much time the supervisor spends together with his workers can be a limit for his influence. What opportunities does the supervisor actually have to be a model for all his workers ? The obvious limit is his span of control. When the number of subordinates increases, the possibility for modeling the right behavior for all the workers decreases, and the likelihood of the workers imitating other models increases.

As I have outlined, modeling safe behavior can be a powerful tool for influencing subordinates behavior. However, modeling is limited by the supervisors span of control and if he is seen as an appropriate model. To be an appropriate model, he must have success, competence and concern for his workers. Once the desired safety behaviors are attained, they can be maintained by giving appropriate feedback to subordinates.

In this paper, feedback is limited to reinforcement and punishment of behavior. Reinforcement and punishment are central concepts in behaviorism, here limited to operant learning theory (Skinner, 1953). As proposed by behaviorism, behavior is a function of it’s consequences (McAfeee & Winn, 1989). This means that a person acts according to what he expects will follow a behavior. A person will perform a behavior when a positive outcome is expected, and avoid behavior associated with a negative outcome. The goal of operant conditioning is to change behavior by changing the outcome of a behavior. The changes in outcome alter expectations, and expectations change behavior. The principles of operant conditioning have been applied to safety research in what has been called behavior modifications programs. Several studies have found that behavior modification has a profound effect on safety behavior (e.g. Smith, Anger, & Uslan, 1978; Zohar, 1980a; McAfee & Winn, 1989). In a review of accident prevention programs, behavior modification was also the approach with the strongest effect on accident rates (Guastello, 1993). Let’s take a look at the central concepts of operant conditioning; positive and negative reinforcement, punishment and extinction (Kirkpatrick & Lewis, 1995). While the use of positive reinforcement and extinction can be associated with theory Y, punishment and negative reinforcement can be associated with theory X.

A positive reinforcement, also called reward, is when an attractive consequence is presented as a result of a specific behavior. Examples of positive reinforcers which can be used to enhance safety are payment, praise, improvement in working conditions and increased responsibility. In a study of Smith, Anger, and Uslan (1978), supervisors managed to lower the accident rate with 7.48 % in a shipyard, by giving praise for wearing safety equipment. Zohar (1980a) describes a study where workers’ use of earplugs increased from 35 to 90 % during one week by the use of token economy. The workers received a token on each audit tour where they were observed using earplugs. The tokens gave the right to buy a variety of consumer products. Each product had a price in terms of number of tokens required for purchase. McAfee and Winn (1989) reviewed 24 studies that used behavior modification with positive reinforcement. The major finding was that incentives/feedback reduced accidents in all studies, at least in the short term.

Negative reinforcement occurs when a disfavored condition is removed or avoided as a result of an individual’s behavior. An example is a worker who puts on his safety helmet because he knows the supervisor will yell at him if he does not wear a helmet. McKelvey, Engen, and Peck (1973, cited in Peters, 1991) found that negative reinforcement significantly reduced the number of unsafe behaviors in a laboratory experiment.

Another form of conditioning is punishment. This is not the same as negative reinforcement. Punishment is when an undesired behavior is followed by an unpleasant consequence. If the supervisor gives the workers who do not use safety equipment extra work, this is a form of punishment. Rubinsky and Smith (1973) found that simulating accidents in the laboratory decreased the number of unsafe behaviors and accidents. This study was conducted with introductory psychology students, and has it’s limitations, but accident simulation might be a good way of training for safe behavior. The occurrence of an accident functioned as a punisher of unsafe behavior.

Extinction occurs when one removes the reinforcement of an undesired behavior. To ignore unsafe acts and praise safe acts should lead to an extinction of the unsafe acts, unless the unsafe acts are naturally reinforced, as I will discuss later.

The supervisor might enter a negative circle of punishment and negative reinforcement. When the non-use of safety helmet is followed by punishment, the supervisor is negatively reinforced by the fact that punishment leads to a reward – the supervisor achieves the goal of punishment. This may influence the supervisor to use punishment, but can create reactance among the workers and a rather unpleasant work environment (Kirkpatrick & Lewis, 1995). The supervisor will believe that the workers always wear safety helmets until an accident occurs. It is therefore very important for supervisors to realize how punishment affects the workers’ behavior, and learn to use other ways of dealing with non-compliance.

Positive reinforcement and extinction is usually more efficient than punishment and negative reward, because of the negative side effects of the two last ones, such as reactance and dissatisfaction with the social work environment. Positive reinforcement, as opposed to negative reinforcement and punishment, enhances the relationship between the supervisor and the employees by creating a positive and friendly atmosphere. This increases job satisfaction in addition to avoiding the negative side effects of punishment. Positive reinforcement is also more direct than punishment in effect. "It increases the probability of safe behavior rather than reducing the probability of unsafe behavior." (Mcafee & Winn, 1989, p.8). Simard and Marchand (1994) also state that punishment can produce undesirable behavior, because the workers feel they have been treated unfairly. Katz, Maccoby, Gurin, and Floor (1950, cited in Likert, 1961) found that high-producing supervisors tended to either ignore mistakes (extinction), knowing that their men had learned from experience, or used the situations to show them how to do the job correctly. The low-producing foremen used punishment and criticism when the workers made mistakes. Though these studies used production rates as a dependent variable, it might be generalized to safety as safety is a part of the daily work in the organization too. Mattila et al. (1994) found that supervisors with high safety index, gave more often feedback to the employees. The feedback was also more positive or neutral of nature than for the supervisors with low safety index. Supervisors with lower safety index also gave more often feedback about events with negative consequences. Both the frequency and the quality of the feedback is thus important. However, positive reinforcement can give negative side effects too, if they are only used to attain desired behavior. Cohen and Colligan (1997) states that reactance can be generated by use of positive incentments and reinforcements. If the workers feel the management is using the reward system to avoid responsibility in safety effort, noncompliance and hostility can occur.

Although positive reinforcement of safe behavior might be the most optimal, unsafe behavior is often positive reinforced in what is called natural reinforcement. Working in an unsafe way may have rewarding features like enabling the work to be done faster and more comfortable (Cohen & Colligan, 1997; Fleming & Lardner, 1999). If the costs of performing a safe act exceed the benefits, these may interfere with the workers’ willingness to follow the safety practices (Cohen & Colligan, 1997). Reason, Parker, and Lawton (1998) states the problem is that the goals of the organization and the workers are often not the same. One of the primary motives of humans is to save time and effort. An operation that costs extra time and effort, is not psychologically rewarding. Rewarding behavior is the behavior that achieves the goal of the individual. The workers learn quickly that violations of rules carry no penalties, but are rewarded by saved time and effort. Further, the separation in time and space of the violation and the negative consequence (when an accident finally occurs) might reinforce the belief that violations are innocuous. Robinson (1982) states that desired behavior should be reinforced in an organization, but that this principle often is violated. The problem is that safety is communicated in rules, regulations, reprimands etc, but that short-term production is paid for in direct incentive systems. The supervisors are promoted for production (short-term), and berated for safety-failures. He states further that workers find shortcut solutions that increase production but also increase hazard levels. Management then encourages the clever techniques that promote production until an accident occurs, and then "drag out the dusty rule book" (Robinson, p. 127). Over time, unsafe routines can become habitual so that everybody does it the same way. This may eventually cause an accident or injury (Hofmann & Stetzer, 1996). According to Sulzer-Azaroff (1982), unsafe practices exist because the natural punishers for unsafe acts (accidents) are weak, delayed and infrequent. Unsafe acts are reinforced immediately by saving time and effort. "Immediate, regular, specific feedback is more effective than feedback which is delayed, infrequent and vague" (Fleming & Lardner, 1999, p.3). This means that the supervisor needs to identify natural reinforcers of unsafe behavior and reinforce safe behavior in stead. Since the natural reinforcers of safe behavior (prevention of accidents) are not very effective, they have to be supplemented by contrived reinforcers, such as praise (Sulzer-Azaroff, 1982). Cohen and Cleveland (1983) studied plants with good safety records, and found they had a system of immediate feedback on safety performance. When a worker did a good job, he was immediately recognized by his supervisor. If he violated safety rules, he was immediately corrected. Cohen and Cleveland also found that use of monetary incentives was played down. Instead, information materials was used for eliciting workers’ safety awareness. The use of behavior modification can thus be criticized for not giving the workers information about why they should use protective equipment etc.

Other problems with behavior modification in addition to natural reinforcers, is the supervisor’s span of control and power. If the supervisor is in charge of 100 subordinates, it’s quite obvious he does not have the time to observe all their behaviors and reinforce them consistently. The supervisor does not necessarily have the power to punish or reward all the behaviors either. The workers might be motivated by something the supervisor cannot give them. If they are motivated by payment, the supervisor might not have the authority to change their salary. To reinforce behavior also requires consistent feedback, something that demands a lot of effort. The supervisor does not necessarily have the time available to accomplish the conditioning that is necessary. A behavior modification program where the management is inconsistent is less likely to succeed according to Fleming and Lardner (1999).

The supervisor’s feedback is a way of influencing the subordinates’ safety performance, but the supervisor is limited again by his span of control, time and power. The most effective feedback seems to be positive reinforcement. I will now go on to discuss these organizational limitations, which also can influence other ways of safety promotion.

As I have outlined throughout this paper, several organizational factors limit the supervisor’s influence on the workers’ safety performance. His power, his span of control and the management’s commitment to safety are all important constraints of the supervisor’s influence. The supervisor’s role as a key man in accident prevention and safety promotion can thus be questioned. As Metzgar (1999) noted, the supervisor has a lot of pressure from several sources, and his own view might not be the one that is promoted.

The supervisor’s power determines his possibilities for using democratic decisions. If he does have the authority he needs, democratic decision making can be a powerful tool for motivating to safe behavior. If he does not have authority to make safety decisions or influence safety decisions, he cannot use democratic decision making to motivate the employees either. However, he can involve them in setting goals for safety performance in the department, and in this way motivate to safe behavior. In addition to limiting decision making, the supervisor’s power also limits his possibilities for reinforcing safe behavior. To use behavior modification for safety promotion, the supervisor first needs to identify what reinforcers that are useful. If the employees are motivated by payment, the supervisor might not have the power to change salaries. However, praise is something the supervisor always have the power to use, and studies have also found that praise can have a profound effect on safe behavior (e.g. Smith, Anger, & Uslan, 1978). Although praise and goal setting can be used despite little power, the use is unfortunately limited by the supervisor’s span of control.

If the supervisor is in charge of many subordinates, it’s difficult and time-consuming to involve all the workers in setting goals for safety improvement. With a large span of control, it is also difficult to give sufficient feedback to all the workers and promote safety by giving praise. Another problematic effect of a large span, is smaller possibilities for modeling safe behavior. As we have seen, the supervisor can influence the subordinates by being a role model. However, being in charge of many subordinates limits the possibilities for influencing by modeling safe behavior for all the subordinates. As a result, other models are chosen, which might influence employees to work unsafe rather than safe. If the supervisor does have a small span of control, behavior modeling can be used to influence the workers. That is, if the supervisor is a good model. Good modeling will probably result from a strong commitment to safety.

Several studies have shown how management’s commitment to safety can influence safety behavior (e.g. Cohen, 1977; Smith, Cohen et al., 1978; Dedobbeler & Béland, 1991; Cohen & Cleveland, 1983; Simard & Marchand, 1994; Simard & Marchand, 1997; Hoffman & Morgeson, 1999). The supervisor’s commitment can therefore have a knock-on effect on workers. However, the supervisor’s commitment is often not enough. Even if he is strongly committed to safety, his commitment might not lead to better safety performance. The top management’s attitude to safety is important in this matter. If the management is little committed to safety and more preoccupied by high productivity, the supervisor might be forced to prioritize production before safety. His own commitment to safety might then be of less importance, because he will model unsafe behavior rather than safe behavior.

As we can see, several organizational factors limit the supervisor’s influence. His role as a safety promoter is greatly dependent upon the organization he works for. If the supervisor has little authority, a large span of control and a management who is little concerned with safety, the supervisor has probably little opportunity for influencing the workers’ safety behavior. In those organizations, the supervisor is not the most important person for safety promotion either. However, in organizations where the supervisor has some power, a reasonable span of control and a management who is concerned for safety, his role in safety is more important. In these situations, one must consider which supervisory style best promotes safety.

As I have discussed, theory X, 9.9 leadership style, democratic leadership style, modeling right safety behavior, and positive reinforcement or extinction seem to be the best ways of promoting safety in the organization. However, are supervisors with these beliefs and behaviors always the ones who promotes safety the most ? Dubin, Homans, Mann, and Miller (1965, cited in Thurley and Wirdenius, 1973, p.54) states that "there is no ‘one best’ method of supervision." Likert (1961, p.89) also states that "Industry is spending substantial sums on training supervisors and managers in human-relations skills, and yet the results are often disappointing (…) One of the difficulties appears to be a widespread but erroneous assumption that there are specifically ‘right’ and ‘wrong’ ways to supervise. This leads to the belief in the efficacy of certain prescribed ‘rules of supervision’ ". Pelz (1951; 1952, cited in Likert, 1961) studied hundreds of supervisors and several thousand non-supervisory employees. He found only a slight relationship between the behavior of the supervisor and the employees’ attitudes and morale. The reason might be that characteristics of the workforce are important determinants for what supervisory style is best. The leadership style approaches (theory X/Y, management grid and democratic-autocratic leader) I have presented can be called trait approaches. They seek to find the one best way of supervising. However, there are another group of theories that suggest the interaction between situation and leader determines whether the leader (supervisor) is effective or not (Riggio, 2000).

Fiedler (1967, cited in Thurley and Wirdenius, 1973), states that when the relationship between the leader and the subordinates is poor, focusing on doing the task is more appropriate than focusing on building good relationships. Focus on task is approximately the same as concern for production. If the relationship is moderate, a focus on building good relations is appropriate, while when the relationship is very good, a focus on task is more appropriate Fiedler summarizes his research on leadership of fifteen years. He found that both task-oriented leaders and relationship-oriented leaders (similar to concern for production and people) can perform effectively, dependent on conditions. The relevance of these studies is questionable though. The supervisors’ success are measured by production, while we here are discussing safety matters. A supervisor that succeed in production does not necessarily succeed in safety promotion. But as Riggio (2000) stated, the management grid has been criticized for proposing 9.9 as the best style in all situations. If we should follow Fiedler, the other leadership styles could also be effective, depending on the situation. More research should therefore be executed to examine the effect of the different supervisory styles in Blake and Mouton’s (1968) management grid on safety performance in different organizations.

According to Likert (1961) democratic decision making is not always the best either. Likert cites Tannenbaum’s (1954) study, who found that subordinates who had a dependent personality reacted adversely to increase in participation in decisions about their work. Although this study is quite old, it might still be applicable today. Some workers do not want responsibility, and giving them a lot of responsibility will not motivate them to perform better either. Likert also cites other studies that have shown the response to participation in decisions is dependent upon the personality of the subordinates. He states that responses are dependent on the subordinates’ past experience and the traditions of the working situation. Locke and Sweiger (1979) also found that the success of participation in decisions was dependent on the subordinates’ expertise. They state that other problems associated with participative decision making are intragroup conflicts, conformity and group-think. The last two can lead to poor decision quality. As we can see, it is not clear cut that democratic decision making is the best in all situations. The supervisor needs to assess his workers’ expertise and personality to determine whether a more autocratic decision-making style is more appropriate.

McGregor (1968) also proposes one best way of leadership as the management grid and the democratic-autocratic decision tradition. He states that theory Y is a better way of looking at subordinates. Although this might be true, it is not clear-cut whether theory Y always leads to the best safety performance. Both Eyssen et al. (1980) and Zohar (1980a) found that theory X leaders could lead to high safety performance. The effects of the supervisor’s beliefs are probably again dependent on characteristics of the work force, such as the personality trait dependency. Huse and Price (1970) found for instance that immature workers were more motivated to work when the organizational climate was characterized by theory X, while mature workers were more motivated when the climate was characterized by theory Y. Age of the work force might also be important as older employees are probably used to a more directive and controlling style.

Modeling safe behavior can be a good way of promoting safety, but as we have seen, the effects of modeling are also dependent on characteristics of the observer (the workers) and the model (supervisor). As Weiss (1978) noted, workers with high self-esteem was less influenced by modeled behavior than those of low self-esteem. Modeling safe behavior was also most appropriate if the supervisor had some power and competence. The supervisor therefore needs to assess his workers and his own position to find out whether behavior modeling will have an effect and is a good way of promoting safety.

As I have mentioned, the use of behavior modification also requires an assessment of each individual worker. Different people are motivated by different incentments. Some studies found that punishment and negative rewards could have an effect on safety performance (McKelvey, Engen, & Peck, 1973, cited in Simard & Marchand, 1994; Rubinsky & Smith, 1973). Though several researchers (e.g. Mcafee & Winn, 1989; Mattila et al., 1994) think praise and extinction are better ways of giving feedback, this may also depend on characteristics of the work force. The supervisor therefore needs to assess his workers to figure out what motivates them the most.

Examining all these approaches to supervision, the main conclusion is that to determine the optimal supervisory style, the supervisor has to assess his subordinates. "Sensitivity to the values and the expectations of others is an important dimension of effective supervision." (Likert, 1961, p. 95). What motivates them and what supervisory behavior are

they used to ? The best supervisor is the one who can change his leadership style after characteristics of the work force. However, safety commitment is important in all situations. If the supervisor has a high commitment to safety, this is a good starting point for trying to find ways of promoting safety.

Throughout this paper, I have given a lot of suggestions for how the supervisor can promote safety in his work place. However, it is presupposed that they are practicable. As we have seen, the supervisor may gore against organizational constraints. A supervisor is often expected to have a high level of safety in his department, but there is often a mismatch between these expectations and the authority the supervisor is given from the management. In order to be a good safety promoter, the supervisor must be given the authority and the resources he needs to live up to the expectations in addition to being trained for good supervision by assessing the workers and choosing the right supervisory style. However, in spite of little authority and resources, there is always something the supervisor can do to promote safety, and his limits should not be used as an excuse for not doing anything when he perceives problems associated with safety.

A lot of industrial organizations have problems with accidents occurring because employees have violated safety rules. The first-line supervisor is the leader working nearest the employees, and can therefore influence them in a positive way. However, there are different supervisors, who care more or less about safety and the subordinates. A supervisor who sees the subordinates as independent and ambitious, is interested in their well-being, and who models and rewards safe behavior, seems to be the best for safety promotion. However, this depends on characteristics of the work force. The best supervisor is probably the one who has the ability to assess his subordinates and adjust his supervisory style after their characteristics. But although the supervisor has a strong commitment to safety, he is limited by organizational constraints such as span of control, the management’s safety commitment and his own power. Organizations who want to utilize the supervisors as a safety promoter should therefore give them both the leader training, the authority and resources they need for promoting safety.

Adler, S. (1983) Subordinate imitation of supervisor behavior: The role of supervisor power and subordinate self-esteem. Social Behavior and Personality, 11, 5-10.

Andriessen, J.H.T.H. (1978) Safe behavior and safety motivation. Journal of Occupational Accidents ,1, 363-376.

Arnold, J., Cooper, C.L. & Robertson, I.T. (1998) Attitudes at work. In Work psychology. Understanding human behaviour in the work place (pp.190-219). London: Financial Times.

Bandura, A. (1969) Modeling and vicarious processes. In Principles of behavior modification (pp.118-216). New York: Holt, Rinehart and Winston.

Bandura, A. (1977) Origins of behavior. In Social learning theory (pp.16-55). Englewood Cliffs, N.J: Prentice Hall.

Blake, R.R. & McCanse, A.A. (1991) Leadership dilemmas-grid solutions. Houston, TX: Gulf.

Blake, R.R. & Mouton, J.S. (1968) Ledelsens problematikk. I Lederstil : Utvikling av arbeidsmiljø og organisasjon (pp.18-28). Oslo: Bedriftsøkonomen.

Brown, R. (1988) Structural aspects of groups. In Group processes. Dynamics within and between groups (pp.51-89). Malden, MA: Blackwell Publishers.

Chhokar, J.S. & Wallin, J.A. (1984) Improving safety through applied behavioral anlysis. Journal of Safety Research ,15, 141-151.

Child, J. & Partridge, B. (1982) One and many supervisory roles. In Lost managers : Supervisors in industry and society (pp.35-62). Cambridge: Cambridge University Press.

Clarke, S. (1999) Perceptions of organizational safety: Implications for the development of safety culture. Journal of Organizational Behavior ,20, 185-198.

Cohen, A. (1977) Factors in successful occupational safety programs. Journal of Safety Research ,9, 168-178.

Cohen, A. & Colligan, M.J. (1997) Accepting occupational safety and health regimens. In D.S. Gochman (Ed.), Handbook of health behavior research (Vol. 2, pp.379-394). New York: Plenum Press.

Cohen, H.H. & Cleveland, R.J. (1983) Safety program practices in record-holding plants. Professional Safety, 9, 26-33.

Dedobbeleer, N. & Béland, F. (1991) A safety climate measure for construction sites. Journal of Safety Research ,22, 97-103.

Denton, D.K. (1982) Motivating, Safety leadership. In Safety management: Improving performance (pp.59-81, 104-119). New York: McGraw-Hill.

DeReamer, R. (1958) Safety - a management practice. In Modern safety practices (pp.3-14). New York: Wiley.

Dubin, R., Homans, C.G., Mann, F.C. & Miller, D.C. (1965) Leadership and productivity, facts of industrial life. San Francisko: Chandler Publishing Company.

Duke, E.E. (1972) A case study of management technique. Personnel Journal ,51, 32-35. (Abstract from PsychINFO, accession number 1972-25986-001)

Dunbar, R.L.M. (1975) Manager's influence on subordinates' thinking about safety. Academy of Management Journal ,18, 364-369.

European Community for Coal and Steel. (1967) Human factors and safety. Luxemburg.

Eyssen, G.M., Hoffman, J.E. & Spengler, R. (1980) Managers attitudes and the occurence of accidents in a telephone company. Journal of Occupational Accidents ,2, 291-304.

Fiedler, F.E. (1967) A theory of leadership effectiveness. New York: McGraw-Hill.

Fleming, M., & Lardner, R. (1999) Behaviour modification to improve safety. Literature review. Edinsburgh, Scotland: The Keil Centre.

Gardner, J.E. (1969) The supervisor as a key figure in safety, Causes of unsafe acts. In Safety training for the supervisor (pp.3-11, 31-46). Reading, MA: Addison-Wesley Publishing Company.

Guastello, S.J. (1993) Do we really know how well our occupational accident prevention programs work. Safety Science ,16, 445-463.

Hale, A.R. & Hovden, J. (1996, February) Management and culture : The third age of safety. A review of approaches to organisational aspects of safety, health and environment. Paper to Occupational Injury Conference, Sydney, Australia.

Heinrich, H.W. (1931) Basic philosophy of accident prevention. In Industrial accident prevention: A scientific approach (pp.13-65). New York: McGraw-Hill.

Hofmann, D.A., Jacobs, R. & Landy, F. (1995) High-reliability process industries - individual, micro, and macro organizational influences on safety performance. Journal of Safety Research ,26, 131-149.

Hofmann, D.A. & Morgeson, F.P. (1999) Safety-related behavior as a social exchange: The role of perceived organizational support and leader-member exchange. Journal of Applied Psychology ,84, 286-296.

Hofmann, D.A. & Stetzer, A. (1996) A cross-level investigation of factors influencing unsafe behaviors and accidents. Personnel Psychology ,49, 307-339.

Hoyos, C.G. (1992) A change in perspective: Safety psychology replaces the traditional field of accident research. The German Journal of Psychology ,16, 1-23.

Hoyos, C.G. (1995) Occupational-safety - progress in understanding the basic aspects of safe and unsafe behavior. Applied Psychology ,44, 233-250.

Huse, E.F. & Price, P.S. (1970) Relationship between maturity and motivation in varied work groups. Proceedings of the Annual Convention of the American Psychological Association ,5, 587-588. (Abstract from PsychINFO, accession number 1970-19640-001)

Katz, D., Maccoby, N., Gurin, G. & Floor, L.G. (1950) Productivity, supervision and morale among railroad workers. Ann Arbor, MI: Institute for Social Research.

Kirkpatrick, T.O. & Lewis, C.D. (1995) What is supervision ?, Meeting the performance challenge, Principles of organizing, The supervisor as leader, Motivating employees. In Effective supervision. Preparing for the 21st century (pp.4-25, 26-45, 116-137, 138-157, 222-241). Orlando, FL: The Dryden Press.

Konczal, E. (1979) The supervisors' role in accident prevention. Supervisory Management ,24, 31-34.

Krantz, J. (1993) The managerial couple: Superior-subordinate relationships as a unit of analysis. In L. Hirschhorn, & C.K. Barnett (Eds.), The Psychodynamics of organizations (pp.3-17). Philadelphia: Temple University Press.

Lee, T. (1998) Assessment of safety culture at a nuclear reprocessing plant. Work and Stress ,12, 217-237.

Levitt, R.E. & Samelson, N.M. (1993) Managing your organization for zero accidents. In Construction safety management (2nd ed., pp.49-69). New York: John Wiley & Sons.

Likert, R. (1961) Leadership and organizational performance, Effective supervision: An adaptive and relative process. In New patterns of management (pp.5-25, 89-96). New York: McGraw-Hill.

Lippit, R. & White, R. (1943) The "social climate" of children's groups. In R.G. Barker, J. Kounin, & H. Wirght (Eds.), Child behaviour and development. New York: McGraw Hill.

Locke, E.A. & Scweiger, D.M. (1979) Participation in decision making: One more look. In B.M. Staw (Ed.), Research in organizational behavior : An annual series of analytical essays and critical reviews (Vol. 1, pp.265-339). Greenwich, CT: JAI Press.

Logozzo, R.N. (1989) The relationship of manager's theory X and theory Y assumptions to managerial participative behavior, employee commitment, and employee absenteism. (Doctoral dissertation, University of Massachusetts, 1989). Dissertation Abstracts international, 50, 1367A.

Maslow, A.H. (1970) Motivation and personality (2nd ed.). New York: Harper & Row.

Mattila, M., Hyttinen, M. & Rantanen, E. (1994) Effective supervisory behavior and safety at the building site. International Journal of Industrial Ergonomics ,13, 85-93.

McAfee, R.B. & Winn, A.R. (1989) The use of incentives feedback to enhance work place safety - a critique of the literature. Journal of Safety Research ,20, 7-19.

McGregor, D. (1968) Teori X: Det tradisjonelle syn på ledelse og kontroll, Teori Y: Tilpasning mellom individuelle og organisatoriske mål. I Mennesket og bedriften (pp.33-40, 41-49). Oslo: Gyldendal.

McKelvey, R., Engen, T. & Peck, M. (1973) Performance effeiciency and injury avoidance as a function of positive and negative incentives. Journal of Safety Research ,5, 90-96.

Metzgar, C.R. (1999) The boss who is not there: The key man dethroned. Professional Safety ,44, 40-42.

Mowday, R., Steers, R. & Porter, L. (1979) The measurement of organizational commitment. Journal of Vocational Behavior ,14, 224-247.

Nelson, E.J. (1998) Safety commitment redefined. Professional Safety ,43, 41-43.

Pelz, D.C. (1951) The influence of the supervisor within his department as a conditioner of the way supervisory practices affect employee attitudes. Unpublished doctoral dissertation. University of Michigan.

Pelz, D.C. (1952) Influence: A key to effective leadership in the first-line supervisor. Personnel, 29, 209-217.

Peters, R.H. (1991) Strategies for encouraging self-protective employee behavior. Journal of Safety Research ,22, 53-70.

Petersen, D. (1976) What turns them on. In Safety supervision (pp.77-84). New York: Amacom.

Racicot, B.M. & Wogalter, M.S. (1995) Effects of a video warning sign and social modeling on behavioral compliance. Accident Analysis and Prevention ,27, 57-64.

Reason, J., Parker, D. & Lawton, R. (1998) Organizational controls and safety: The varieties of rule-related behaviour. Journal of Occupational and Organizational Psychology ,71, 289-304.

Reber, R.A. & Wallin, J.A. (1984) The effects of training, goal setting, and knowledge of results on safe behavior: A component analysis. Academy of Management Journal ,27, 544-560.

Reber, R.A., Wallin, J.A. & Chhokar, J. (1984) Reducing industrial accidents: A behavioral experiment. Employee Relations ,23, 119-124.

Riggio, R.E. (2000) Leadership. In Introduction to industrial/organizational psychology (3rd ed., pp.339-371). Upper Saddle River, NJ: Prentice Hall.

Robinson, G.H. (1982) Accidents and sociotechnical systems: Principles for design. Accident Analysis and Prevention ,14, 121-130.

Roehtlisberger, F.J. & Dickson, W.J. (1940) Management and the worker. Harvard, MA: Harvard University Press.

Rosenstock, I. (1974) Historical origins of the health belief model. Health Education Monographs ,2, 1-8.

Rubinsky, S. & Smith, N. (1973) Safety training by accident simulation. Journal of Applied Psychology ,57, 68-73.

Rundmo, T., Hestad, H. & Ulleberg, P. (1998) Organisational factors, safety attitudes and workload among offshore oil personnel. Safety Science ,29, 75-87.

Scandura, T.A., Graen, G.B. & Novak, M.A. (1986) When managers decide not to act autocratically: An investigation of leader-member exchange and decision influence. Journal of Applied Psychology ,71, 579-584.

Schoenwald, I.S. & Kopp, R.R. (1986) Life style and the roles of the supervisor in an organization: An integration. Individual Psychology ,42, 350-359.

Simard, M. & Marchand, A. (1994) The behavior of first-line supervisors in accident prevention and effectiveness in occupational safety. Safety Science ,17, 169-185.

Simard, M. & Marchand, A. (1997) Workgroups' propensity to comply with safety rules: The influence of micro-macro organisational factors. Ergonomics ,40, 172-188.

Skinner, B.F. (1953) Operant behavior. In Science and human behavior (pp.59-90). New York: Macmillan.

Smith, M.J., Anger, W.K. & Uslan, S.S. (1978) Behavioral modification applied to occupational safety. Journal of Safety Research ,10, 87-88.

Smith, M.J., Cohen, H.H., Cohen, A. & Cleveland, R.J. (1978) Characteristics of successful safety programs. Journal of Safety Research ,10, 5-15.

Sulzer-Azaroff, B. (1982) Behavioral approaches to occupational health and safety. In L.W. Frederiksen (Ed.), Handbook of organizational behavior management (pp.505-538). New York: Wiley.

Tannenbaum, A.S. (1954) The relationship between personality and group structure. Unpublished doctoral dissertation, Syracuse University, NY.

Thompson, R.C., Hilton, T.F. & Witt, L.A. (1998) Where the safety rubber meets the shop floor: A confirmatory model of management influence on workplace safety. Journal of Safety Research ,29, 15-24.

Thurley, K.E. & Wirdenius, H. (1973) Analysis II: Theory. In Supervision: A reappraisal (pp.43-57). London: Heinemann.

Vandenput, M.A. (1970) Unsafe habits and social pressures. Psychologica Belgica ,10, 99-107.

Weber, J.O. (1992, ) The front-line supervisors' role in safety. Professional Safety, 34-49.

Weiss, H.M. (1978) Social learning of work values in organizations. Journal of Applied Psychology ,63, 711-718.

Wogalter, M.S., Allison, S.T. & McKenna, N.A. (1989) The effects of cost and social influence on warning compliance. Human Factors ,31, 133-140.

Zohar, D. (1980a) Promoting the use of personal protective equipment by behavior modification techniques. Journal of Safety Research ,12, 78-85.

Zohar, D. (1980b) Safety climate in industrial organizations: Theoretical and applied implications. Journal of Applied Psychology ,65, 96-102.