The modification of occupational safety behavior

The modification of occupational safety behavior

Journal of Occupational Accidents, 9 (1987) 177-197 Elsevier Science Publishers B.V., Amsterdam - Printed 177 in The Netherlands The Modification of...

2MB Sizes 4 Downloads 14 Views

Journal of Occupational Accidents, 9 (1987) 177-197 Elsevier Science Publishers B.V., Amsterdam - Printed

177 in The Netherlands

The Modification of Occupational Safety Behavior BETH SULZER-AZAROFF University of Massachusetts at Amherst, Amherst, MA (U.S.A.) (Received 9 February

1987; accepted

13 April 1987)

ABSTRACT Sulzer-Azaroff, B., 1987. The modification tional Accidents, 9: 177-197.

of occupational

safety behavior.

Journal of Occupa-

Because human performance is implicated in many of the accidents and injuries that occur in the workplace, behavior modification has much to offer. Using examination and analyses of contingencies of reinforcement and punishment to diagnose current problem situations and plan change, effective behavioral practice also emphasizes validity of safety targets and support from within the organization, Baselines, consisting of repeated measures of targeted performances, serve as a standard against which treatment effects can be judged, and can be incorporated within an overall experimental design. Effectively arranged positive reinforcement, supported by clear antecedents (e.g., goals, instructions and so on) that cue probability that the response will be reinforced, is the preferred method of modification. In cases of unsafe behaviors especially resistant to change, one can combine reinforcement of preferred alternatives with punishment of the offensive act. Effective feedback blends reinforcement, stimulus control and punishment in an optimal mix. Complex skills, or resistant habits can be modified by breaking the safe performance into small parts and heavily reinforcing improvements in the previously incorrect components or by using chaining or shaping. Especially amenable to objective evaluation of its outcomes, due to its objectivity, rigor of measurement and experimental conventions, behavior modification has begun to demonstrate its value as a powerful tool for promoting safety and health in the workplace.

THE MODIFICATION

OF OCCUPATIONAL

SAFETY BEHAVIOR

Safety in the workplace depends on a number of factors, including, among others, how hazardous the task or the physical or social environment might be, setting factors, such as hunger, fatigue, stress, distraction, and upon the ongoing behavior of the individual. That the inherent danger of the job is not solely responsible is apparent, as experts like steeplejacks, radiologists, or demolition specialists can carry out their job requirements safely even under relatively dangerous conditions. If one examines the data on occupational accidents, for instance those contained in Accident Facts (1984)) it is apparent that accidents continue to happen even in minimally hazardous environments. Nor does training guarantee

0376-6349/87/$03.50

0 1987 Elsevier Science Publishers

B.V.

178

safe performance. Workers often behave unsafely, even when they are fully apprised of the risks involved and capable of taking preventive action; and injuries sometimes result. Consequently, although careful job analyses and the efforts of ergonomists, safety engineers, officers, and trainers do contribute toward reducing risk of injury, unsafe human performance continues to be a major challenge. Workers may be employed in a setting as free of danger as possible and have already demonstrated their mastery of the skills involved in performing safely on the job. Nevertheless, sometimes they take risks. Why should this happen? The science of human behavior may offer an explanation. As Skinner (1938) and other behavior analysts have demonstrated repeatedly, behavicr is a function of contingencies (i.e. relations between responses and stimuli). When unsafe behavior occurs, we can assume that past and present contingencies have played a role. In the material to follow, just how contingencies operate to affect behavior will be examined conceptually and illustrated. Next, a system for assessing performance in the work context to identify targets for change will be presented. Then a number of experimental applications of behavioral principles to modify safety behavior and prevent relapse will be described, and finally, evaluation issues and methods will be addressed. A BEHAVIORAL

PERSPECTIVE

OF JOB SAFETY

As a general rule, people behave in ways that have been reinforced optimally in the past. They avoid acting in ways that effectively have been punished or placed on extinction (i.e., not reinforced). (See the discussion of optimal conditions of reinforcement and punishment below and Azrin and Holz, 1966 and Morse and Kelleher, 1977.) It is important to note that reinforcement, punishment and extinction are defined not by the intent of others but by their effect on behavior. Reinforcement is presumed to have occurred only if the rate of the behavior upon which it was contingent has increased or maintained; punishment and extinction, only if the behavior has decreased. Information about variables that influence human behavior is far more complex than this, and we continue to learn more on the subject daily. Knowing about those factors helps us to understand why people behave the way they do and how to alter those patterns. (For the sake of simplicity, however, in this paper, the discussion is limited to a few basic concepts.) Let us consider an example. One of Eric’s responsibilities as a cabinet maker is to keep his tools sharp. While polishing a carved table leg, he notices that a groove is not cut deeply enough. Finding that his preferred tool has become dull, he goes to the sharpening wheel, where his safet,y glasses are within reach. “This will only

179

take a second,” he tells himself, “so I won’t bother with the glasses.” While he sharpens his tool a fragment flies off and lodges in his eye. After arranging for medical assistance, his boss and co-workers “How stupid of him. He knows better than that.”

comment

Despite no major lasting damage, everyone including Eric can recognize the foolishness of his act: that to save just a moment, he has suffered a serious injury and lost lots of time for himself and his company. Eric had known what precautions he was supposed to take and how serious the consequences of an injury might be, yet he went right ahead. Why? What does the science of human behavior have to tell us about such an episode? Let’s examine the circumstances affecting Eric’s performance. What were the consequences for Eric if he: used the dull tool? The groove would not be cut to his standards. That consequence certainly would be immediately punishing, so he would act to avoid it. (He has learned in the past to avoid using a dull tool because the consequences were consistently and immediately punishing.) put on his safety glasses before sharpening the tool? He would have had to make an effort and lose a moment or two in returning to his work. Here the immediate consequence, putting on the glasses, would be negative. Avoiding an injury, would be a positive consequence but one that would be delayed and unlikely to happen (not consistent), especially during brief times at the grinder. Also, in the past, he and his fellow workers had performed such brief sharpening tasks without using eye protection to no ill effect. In fact, workers who appeared overly cautious were ridiculed by their peers - another immediate, somewhat uncertain but very negative consequence. Avoiding the immediate negative consequences, the time and effort he would need to take, and the possibility of being ridiculed, outweighed the positive ones. did not put on his safety glasses? The reinforcement of returning to his carving sooner would be immediate, positive and certain, as would avoiding the likely negative ones. So he did not wear the glasses. Just as behavioral principles predict, he avoided the immediate, certain negative consequences, in favor of the immediate, certain, positive one, despite the delayed uncertain negative consequences. ( See Daniels and Rosen, 1982, for a discussion of this type of analysis.) Conducting an analysis of the antecedents and consequences of particular behaviors permits contingencies to be identified and, if possible, rearranged so the behavioral pattern can be modified. In Eric’s case, probably more immediate positive consequences for compliance, such as praise, recognition or even tangible rewards, along with more immediate negative consequences, such as disciplinary action when he were observed not using eye protection when indil

l

l

180

cated, would need to be planned. Finding ways to eliminate peer ridicule of use of safety precautions would also be important. The thesis of this paper is that performance on the job, as elsewhere, is governed in large part by the planned and unplanned contingencies of reinforcement and punishment, that have been or are in effect. To diminish unsafe behavior, those contingencies need to be identified, and rearranged in order to promote safe behavior instead. Figure 1 illustrates how different planned and unplanned antecendents and consequences might influence safe and unsafe practice on the job (Sulzer-Azaroff, 1982, p. 507). Items on the left side constitute the kinds of events that could occur prior to safe and unsafe acts, (abbreviated A for Antecedents). In the center, examples of safe or unsafe acts, (or performances or behauiors, B) are displayed, while in the right hand column are a variety of planned and unplanned positive and negative consequences ( C) . Only directly observable and measurable responses are included; inferred ones are avoided or redefined as operations. (For instance, an antecedent condition such as “having a poor attitude” might be restated as “saying ‘I don’t like to do it that way’ or frowning when requested to perform the job in a certain way.“) The figure illustrates the familiar ABC analysis used by behavior modifiers prior to identifying and functionally analyzing factors that may be influencing particular behaviors. ASSESSMENT

STRATEGIES

An ABC analysis is only as valuable as the information included. Identifying behaviors that relate to safety on the job is of primary importance. Conditions and acts that have been implicated in accidents or near-accidents (Carter and Menckel, 1985) need to be pinpointed if they are to become the subject of change strategies. Selecting effective modification techniques depends upon the discovery of antecendents and consequences that are lawfully influencing those acts. The challenge is to search for and identify valid and important performance targets and the antecedents and consequences that seem to be controlling them; then to rearrange or replace those contingencies in order to promote improvement. To accomplish that purpose, Sulzer-Azaroff and Fellner (1984 ) have proposed an assessment strategy that includes reviewing records (similar to the method endorsed by Heinrich et al., 1980)) conducting interviews, observing safety inspections and setting priorities. Figure 2 depicts the steps we have followed on several occasions, while identifying performance targets and planning change strategies. Reviewing

records

Despite the inadequacies of the accident recording and reporting process (Fox and Sulzer-Azaroff, in press), accident reports can provide valuable

181

4rranged preventive conditior laws; rules, instructions, lniormation, system. functional specification of roles,

1

~~“~~~~“t

Arranged positive consequences: e.g.. posrtwe personnel actions, positive feedback.

*

engineered equ!pment, arranged physical and social enwronment. and so on Safe practNxs: all correct con ponents in proper sequence, appropriate coodnions, no incorrect components.

iI

1

Unplanned

preventwe

conditions;

history of remforcement and punishment of mdlwdual. soc.lal mflueoces, sk\lls. attributes. and others.

I *

Unplanned positwe consequences’ e.g., attentton, praw, positwe self-statements, slgnal for long range goal.

Unplanned negatwe consequences: e.g , critd statements by others, negatwe self-statements, loss of earntngs,

Arranged negative consequences: e.g., negative personnel actions. L

1 I Jnplanned

I

detrimental

c:onditmns:

fatigue, illness, physical disabilities, competing events, history of reioforcement and pu~i~ment, hazardous physical environment, soual environment, and others.

-z

Unsafe practices: componentr in. correct, absent, or in wrong sequence; conditions

Unplanned negative cowequences. e.g.. accidents, injury. prop. erty damage. negative statements-self and others.

tnappropriate.

Unplanned

posttive consequences:

Fig. 1. Examples of planned and unplanned events that might influence performance on the job. (From Sulzer-Azaroff, 1982. Reprinted with permission of John Wiley & Sons.)

information. By looking back over at approximately a two year period and using a checklist to ensure completeness of the survey, we try to obtain a fairly representative sample of episodes. Major changes, such as key personnel shifts or production modifications are noted as they may alter the relevance of the information. (For instance, in one plant with which we were affiliated, the

Organizational

Behavior Management.)

Fig. 2. Identifying performance targets for behavioral programming of occupational safety. (From Sulzer-Azaroff and Fellner, 1984. Reprinted with permission from the Journal of

Go back

r

psrronnrl sorting

safety conduct

I

brhovaorol

defmltlons

operotionol

I

183

position of safety officer was eliminated, producing sweeping changes in the safety program. ) Slow and peak production times, changes in supervisors, union contracts and other circumstances that conceivably might influence safety at the site are also identified. By reviewing reports of accidents or near-misses, we have been able to determine trouble spots, such as frequently slippery areas, blocked walkways and other conditions remediable by human intervention. From the review of records, one can derive the exact nature of the job being performed when the incident occurred, a description of the safe way of conducting it, along with the frequency, severity and attributed cause of any injuries, and potential corrective measures. For example, a series of studies currently being conducted by Alavosius and Sulzer-Azaroff (1985,1986,1987) involves prevention of back injuries by direct service providers, in an institution for mentally retarded people. Records had indicated that back injuries were especially frequent and severe in that setting, as positioning and transferring physically handicapped adult clients was a major staff responsibility, and often the job was not performed with necessary precision. Interviews

Additional important information can be obtained by discussing safety strengths and problems with employees at all strata of an organization: hourly workers, union representatives, members of safety committees and teams, managers, safety and human resource personnel and others. Their descriptions of unreported events can lead to the identification of potentially hazardous acts or conditions. For instance, workers see their peers behaving in the absence of their supervisors and sometimes they reported some pretty dangerous episodes. For instance, by guaranteeing our interviewees anonymity, through structured interviews we have learned about fork-lift truck operators who appeared purposely to speed around corners to frighten their fellow workers, employees who took short cuts by climbing over moving conveyor belts or who cleaned machinery while it continued to operate. Understanding the necessary operations involved in a job task is important when preparing observational instruments and training programs. Experienced workers can provide such information during an interview. They can also cooperate in the specification of corrective actions (Carter and Menckel, 1985 ) , and may provide clues about what sorts of interventions are not apt to be very successful. For instance, interviewees have identified capable and willing candidates to help with a particular safety intervention. Observing safety inspections

Directly observing in-house safety inspections increases one’s familiarity with the organization’s priorities and standard auditing procedures. Observing

184

the procedure may reveal that stated goals may or may not be exactly the ones upon which safety inspectors concentrate. For instance, the stated priority may be on proper conduct of job tasks, while watching shows that audits emphasize housekeeping and placement of guards. Also, from a future perspective, to increase the probability that change will sustain, it is important to know about standard operating procedure. When possible, new change strategies should fit into the normal routine, rather than intruding artificially in that setting. Setting priorities By the time the search for performance targets is complete, an unmanageable pool of potential items probably will have been amassed. These need to be distilled down to a set that reasonably can be targeted. Selection criteria might include the validity of the item (i.e., its relation to frequent and/or severe accidents), its potential for being observed often and the requirement that it be mediated by human performance (otherwise the problem is addressed most appropriately by engineering or ergonomics.) When choosing from too large an array of potential safety targets we have also eliminated most of those items that were consistently found to be acceptable on the many occasions when they were audited. (We keep a few so the program can begin with a certain amount of success. ) When an entire organization is to identify a common pool of targets, informed and interested personnel can be asked to sort each item in terms of its importance. Alternatively, as different work areas may each contain their own unique safety challenges, targets can be tailored to individual work units. For instance, in a recently completed study in a paper mill, one unit’s target behaviors included lifting with legs, and adjusting the conveyor belt via external controls while another focused on unobstructed fire exits, lockouts and eye protection. ( Sulzer-Azaroff et al., 1987). This assessment process yields a set of performance targets, that once refined and, if necessary clarified, readily lend themselves to observation. The items and their definitions can be placed directly on a recording form. (Later, during the intervention phases, those very same forms or a simplified version, can be used to provide feedback to the workers involved.) At this point, it is time to begin formally observing the targeted safety performances and conditions. OBSERVATIONAL

RECORDING

Objective, accurate observational recording is essential to an effective behavior modification program. Who is to conduct the recording and how are objectivity and accuracy to be maintained?

185

Selecting observers

Up until now, considerations for research and/or practice have blended. Yet, the question of who is to conduct the observations needs to be decided; and that depends on the purpose of the program. If research is of prime import, then observers should be people who are less likely to contribute any bias. That means keeping them uninformed about the details and expectations of the study and uninvolved in the inner workings of the organization. Bright, conscientious college students provide an excellent pool from which to select observers. People from within the organization will also have to be involved as observers if the program is to function effectively during and following the development of the behavioral program. Who these people will be, depends on their roles within the organization and upon the sort of prevention program that probably will receive strong durable support. Looking at the current operation helps here. In several of our studies, (e.g., Sulzer-Azaroff, 1978; Sulzer-Azaroff and de Santamaria, 1980)) members of the safety committee who already were accustomed to conducting safety audits (largely of equipment and hazardous conditions) were trained to conduct periodic safety audits of selected targets. Other personnel may serve as well, provided that they too receive training and supervision sufficient to keep their observations objective and accurate. In a recently completed study (SulzerAzaroff et al., 1987) for example, workers themselves served as observational recorders for their own work groups. Their recording form was simplified from the one used by the research team. Training and calibrating observers

.Whether simply as a basis for intervention or for purposes of research, accurate observational measurement is essential. Providing observers with guided, reinforced practice, under simulated and actual conditions helps them to acquire the skills for accurately observing. To train his observers accurately to record patient-transfer performance by staff, Alavosius ( Alavosius and Sulzer-Azaroff, 1985) used video taped demonstrations first and then accompanied observers to the actual site in which the transfers were conducted. In both settings, observers continued to practice until indices of agreement (number of agreements for each task component occurring in the proper order, divided by total number of task components observed) sustained at above 0.85. “Observer drift” was managed by conducting similar comparisons intermittently throughout the study. Any drift in accuracy can be corrected via a retraining session similar to the initial series.

186

I I I

“‘”

I4 I

rlcpt.2

I I

Dept.4

:c, I I

1cpt. 3

I

lept.6

-b

I Kl

2'0

$0 SESSlOMS

40

w

50

If+-

=&

187

Conducting baseline assessment Regardless of the purpose of the program, obtaining a reasonably representative picture of ongoing performance is essential. This is accomplished by recording target acts and conditions repeatedly until a stable series of data points has been recorded. (One criterion for stability is “no new highs nor lows for three observations in a row”.) The information will show where the strengths and troublespots actually are and serve as a standard against which any subsequent change can be measured. Figure 3, from the Sulzer-Azaroff and de Santamaria’s study (1980)) displays a series of baselines of varying lengths. These were useful as a way of determining how serious hazardous conditions were, whether or not, when implemented, the feedback strategy was effective, and had one additional advantage: By applying the feedback intervention at different times, in “multiple baseline” fashion, (Baer et al., 1968; Komaki, 1982) the design demonstrated the lawful relation between the method and performance change. Providing an opportunity to catch and rectify unanticipated problems with the observational recording system is another potential advantage of a baseline period of reasonable length. (Naturally, baseline must be reestablished following such refinements, for consistency of comparisons between conditions.) MODIFYING

BEHAVIOR

When beset with any of a variety or organizational problems, including those in the area of occupational safety, a prevalent tendency is to suggest one of two methods: provide more training or punish misdemeanors. Yet, research has shown that often the first produces either ineffective or ephemeral change (e.g. Komaki et al., 1980) and the second might cause more problems than it solves (Azrin and Holz, 1966; Davis and M&tire, 1969). Behavior modification selects its procedures on the basis of experimental findings, choosing those that have been shown to effect the desired changes with similar populations or conditions. Knowing how an organization and its members function in relation to their physical and social surroundings and the patterns of safe and unsafe behaviors, provides information valuable for designing programs of intervention. The ABC analysis suggests which contingencies may be supporting the behaviors of concern, both wanted and unwanted, and which need to be put in place to promote change.

Fig. 3. An example of a multiple baseline design. Data depict frequencies of hazards recorded prior to (baseline), during, and following the feedback/suggestions intervention. Letters a and b depict personnel actions in the plant. (From Sulzer-Azaroff and de Santamaria, 1980. Reprinted with permission from the Journal of Applied Behavior Analysis.)

188

Returning to the case of Eric, we can see how natural contingencies encouraged risk-taking. Achieving desired change would require that other consequences be added to overcome the intrinsic reinforcement of getting the job done at minimal cost, in terms of effort and time. Which specific consequences would be best altered and how? Fortunately, behavior analysis has discovered a broad array of potential procedures from which to select, but each has its advantages and shortcomings. Space does not permit a thorough presentation of those many strategies, but a number of texts on the subject can provide valuable suggestions (e-g, Kazdin, 1975; Sulzer-Azaroff and Mayer, 1977; and others). Here the heaviest focus is on the methods that have particular relevance for occupational safety: reinforcement, especially combined with judicious use of antecedents; punishment, when absolutely necessary; feedback; shaping; and chaining.

By definition, reinforcement works. When a consequence of a behavior has the effect of increasing or maintaining a behavior’s rate, it is said to be reinforcing. Regardless of how rewarding a consequence appears to an observer, it is only a reinforcer if the rate of the behavior it follows increases or sustains. So reinforcers must be selected because they influence behavior, not because they look appealing. Fortunately, many consequences have been found to function effectively as reinforcers for many people. So trying them out would be a good place to start. Among others, these include recognition, praise, (Hopkins et.al., 1986) privileges, material or monetary rewards (which can be embedded within compensation or performance appraisal systems ( Ritschl et al., 1976)) and preferred activities or assignments. (See Sulzer-Azaroff, 1982 for examples in the area of occupational health and safety. ) In general, it is wise to depend most heavily on reinforcers that are not especially intrusive in a particular setting, as they may be discarded or lose their value after a while. Unless used with care, (individualized, presented irregularly and sometimes unexpectedly) prizes and awards soon begin to lose their appeal, while having one’s efforts specifically and sincerely acknowledged or praised by a respected peer or supervisor will tend to keep serving a reinforcing function. The costs involved in the latter are so comparatively minimal that continuing to deliver those consequences pose no major challenge. Along with carefully selecting reinforcing consequences, they need to be delivered consistently and with minimal delay, especially during the early phases of a behavior change program. Later, inconsistency and delay actually may prove beneficial to maintenance (e.g., Baer et al., 1984; Guevremont et al, 1986). As we shall see in the discussion of feedback below, communicating the conditions under which reinforcement is probable is also important for producing rapid results. If Eric’s boss informed him of his plan to pay a bonus for

189

using safety equipment regularly, Eric probably would begin to use the equipment more rapidly than if he remained unaware of the arrangement. Arranging

antecedents

The unplanned antecedents of targeted behaviors frequently are difficult or impossible to control. No one can alter an individual’s history of reinforcement, and modifying such conditions as illness or problems outside of work are at best difficult. Recognizing those limitations, safety programs often have turned to heavily emphasizing behavioral antecedents directly within the workplace. Signs, verbal admonitions, threats, promises, goals, and training are among the more frequently utilized antecedents. Yet, research has shown that emphasizing antecedents in isolation has succeeded but irregularly (e.g., Komaki et al., 1980; Laner and Sell, 1960). Why is this the case? Behavior analysis provides a compelling hypothesis: The antecedents that work signal predictable consequences reasonably accurately; those that don’t function very effectively probably fail to signal predictable consequences. Extensive basic and applied research on stimulus control has demonstrated the importance that responses occurring following (or in the presence of) antecedent stimuli be correlated with reinforcing or other controlling consequences, if those antecedents rapidly are to begin to evoke the behavior (See Mackintosh, 1977, for a discussion of the factors that promote stimulus control) . In other words, if A eventually is to evoke B, C should follow B when B occurs right after or in the presence of A. For instance, a sign such as “use safety glasses when sharpening tools” is more likely to be effective when, rather than being ignored, adherence to the instructions is reinforced (and/or noncompliance punished). The situation is similar in goal-setting, a popular technique used to encourage workers to meet production quotas and other goals, such as those encountered in safety programs (e.g, Fellner and Sulzer-Azaroff, 1985). Goals are antecedent statements of a particular level to be reached by a given time, such as specific reductions in costs or lost-time injuries within a given period of time. Research has shown that pairing goals and consequences works better than presenting goals by themselves. For instance, Kim and Hamner (1976) found that a combination of goal setting, feedback, and praise was superior to goal setting alone. (Fellner and Sulzer-Azaroff, 1984a, discuss goal setting from a behavior analytic perspective. ) Instruction to promote acquisition of skills, another kind of antecedent arrangement, seems to function similarly. Provide a set of detailed instructions, or even model a difficult response and often little change takes place. But regularly follow each correctly executed response with a reinforcing consequence, and the skill begins to be acquired. This is what we have been finding

190

in our work on teaching health care workers how properly to lift and transfer their patients ( Alavosius and Sulzer-Azaroff, 1987). Planned consequences need not follow every single antecedent-behavior combination, but should occur as often as possible when beginning to try to bring a response under control of antecedent stimuli such as rules, instructions, or goals. Before long, (extrapolating from Baer et al., 1984)) reinforcement can begin to be delivered at different times; not just when the response corresponds to or complies with the antecedent, but occasionally right after the antecedent or when the response happens even in the absence of the antecedent. To promote Eric’s use of eye protection, he could set a goal of using safety glasses each time he sharpens his tool. His foreman might point to the reminder sign and ask Eric what he plans to do when he sharpens his tools. Then, if Eric replies “Wear safety glasses” and he actually does next time, his supervisor could then praise him or provide some other effective reinforcer. Were this to continue consistently, the foreman could then occasionally reinforce Eric’s stated intentions, as long as he periodically reinforced the correspondence between Eric’s stated intention and actual follow through. Reducing unsafe behavior The difficulties involved with arranging to apply negative consequences to reduce unsafe behavior were alluded to earlier. Negative consequences, such as reprimands, fines, penalties and adverse personnel actions, can rapidly reduce unwanted behavior. Send home the vehicle driver who speeds around corners to scare his fellow workers and cause him to lose a few hours pay, and he may discontinue the game in the future. (Or, at least, he may desist when he knows his boss is on grounds.) But, along with the punishment, other reactions are predictable: He probably will become angry, increasing his potential for committing aggression or vandalism, and he may avoid the person who delivered the punishment. He may also temporarily experience some unhealthy physiological reactions, such as increased heart rate or changes in the blood supply to parts of his body. One more problem with using punishment is that when it is the only procedure applied, the person whose behavior has been punished may have no preferred alternative action available to practice. An example is a worker whose job requires replacing raw materials as they are used up. His boss passes by and reprimands him for lifting too heavy a load. The worker may continue to lift the same way in the future because the material must be replaced if he is to continue performing his job and no one else is available to help him. Nor does he have access to hoists or other mechanical aids. In other words, no alternative behaviors are available for him to practice. Yet, there are times when negative treatment is warranted. Reprimanding a

191

worker who was endangering herself or her co-workers, would certainly be justifiable, particularly if paired with other provisions for effective punishment. No other procedure is known to function more rapidly. Like reinforcement, punishment is defined by its effect on behavior; except in this case, the consequence reduces the behavior. If the behavior doesn’t diminish, punishment hasn’t happened. Given its potential disadvantages, it is important to know how to use punishment as optimally as possible: It must be powerful for the individual under the circumstances in which it is applied and be dispensed rapidly and consistently every time it is called for. Another contributor to its success is communicating the conditions under which punishment will be meted out and avoiding delivering punishment only in the presence of particular people, times of day or other readily discriminated conditions. Otherwise the behavior may remain under control only under those particular circumstances. Most importantly, as implied earlier, the individual should be capable of expressing and receiving reinforcement for engaging in an acceptable alternative behavior. For the person lifting too heavy a load, that could mean having a mechanical hoist or co-worker available to assist in replacing the necessary materials. The reinforcement would be delivered regularly at first, and occasionally thereafter, whenever the proper method for handling materials was seen to be followed. (See Azrin and Holz, 1966 for the experimental evidence underlying these guidelines for effective punishment.) Using feedback A hospital worker carefully transfers a patient from his bed to a wheelchair, following a prescribed procedure to avoid injuring his back. An observer watches the transfer being executed and scores a form on which each of the components of a safe transfer are contained, checking those aspects performed safely and in the proper order; noting those needing improvement. He then hands the form to the worker and discusses the quality of the transfer, complimenting correctly performed components, suggesting or demonstrating methods for rectifying problematic ones. This example, from Alavosius and Sulzer-Azaroff (1986)) illustrates the use of feedback. Notice that feedback has the potential to function in a variety of ways: as a reinforcer, when it conveys success, as apunisher, when it conveys failure, and as an antecedent, (i.e. discriminative stimulus) when it prompts or cues the conditions under which responses will be reinforced (and/or punished). Perhaps it is because it can operate in all these ways at once that feedback has been found to be an especially powerful modification technique. Feedback has demonstrated its effectiveness despite many variations in its application: presented right away or following a delay, in oral, written, or graphic form; provided by different individuals: external agents, supervisors, peers, workers

192

themselves; conveying simple or complex information to varied recipients; displayed in public or communicated privately; paired with other consequences, such as performance appraisals, praise, reprimands, bonuses, prizes and so on. (See Fairbank and Prue, 1982,for an extensive discussion of feedback and Table 1 for a sample list of recent studies in which feedback was used to promote occupational health and safety.)

TABLE 1 A sample list of use of feedback in occupational health and safety First author

Date

Behavioral target

Form

Public/ private

Source

Other contingencies

Alavosius

1986

Safe lifting

checklist

private

researcher

Fellner

1984b 1985

graphic

public

researcher

Hopkins

1986

oral

private

trainer

praise, suggestions

Komaki

1978

graphs

public

1980

graphs

public

trainer/ researcher supervisor health & safety inspector

discrimination training, praise

Rhoton

various safety behaviors & conditions several to reduce exposure to styrene Bakery safety (various) reducing ventilation violations

praise, suggestions none

SulzerAzaroff

1978

copies of observational records

private

safety auditor/ peers

SulzerAzaroff

1980

copies of observational records

public

researcher

SulzerAzaroff

1987

various safety behaviors & conditions various safety behaviors & conditions various safety behaviors & conditions

direct comments, praise, correction, shut-down of production suggestions for improvements support-notices from director none

copies of observational records simplified observational records audio- grams

public

researcher

none

public

peers

private

audiologist/ experimenter

Zohar

1980

use of ear protection

information

193

Teaching complex skills and changing established habits The foregoing material has dealt with target behaviors that the individual is capable of expressing, or in technical jargon, behaviors that are established as part of the repertoire of the individual. The modification procedures were designed either to increase or reduce the rates of those behaviors. Yet, sometimes reinforcement of the target behavior is not possible because the behavior is never properly expressed. In that case one must ask whether stronger contingencies need to be identified. When people appear extraordinarily motivated yet still fail to display the behavior, they may have failed to have learned some of the components of the performance; or they may have learned them improperly or expressd them in the wrong sequence. Recall the earlier example of transferring a patient from a bed to a wheelchair. Let us say that a worker conducts the transfer successfully with one or two exceptions: she places herself too far away from the client and bends her back instead of her knees. The target, conducting a safe transfer, is very resistant to change, because the worker has been lifting patients that way for many years. Even though she suffers periodic back pain and knows something is wrong, she can’t seem to correct the problem. Behavioral psychology provides a potential solution. Just as in his early work, Skinner (1938) demonstrated how to produce complex performances in laboratory animals by reinforcing successive approximations to a target behavior (shaping) or reinforcing increasingly more complex sequences of responses (chaining), we can take the same approach. For complicated behaviors with many sub-components, we can identify and describe each of those sub-components, reinforce each when it is expressed correctly, then reinforce combinations and eventually reinforce the composite set of responses when executed correctly as a unit. Helping the hospital worker to break the habit of not moving in close enough and bending improperly can be accomplished by taking the complex behavior and breaking it into its sub-parts. Each can be listed on a check list such as that displayed in Fig. 4 from Alavosius and Sulzer-Azaroff, 1985. Immediately following the transfer, the completed checklist is handed to the worker and she is shown which components she performed well and which she needs to improve. Recently we have been exploring whether or not massing the feedback sessions (i.e., providing feedback for essentially every transfer over a period of several days or until performed correctly five times in a row for two days in a row) might help to reverse those kinds of habits more rapidly than providing feedback only weekly. Sometimes just providing feedback is insufficient, and other aids must be used to help shape the behavior or its component responses. Demonstrations, technically modeling, can help. One particularly clever method is to use self modeling (Dowrick and Dove, 1980). A video recording is taken of the worker

194 ONE PERSON

TRANSFER

(TOTAL

LIFT)

Unit/Ward

Employee Dste/Time

Locrtion

OQNIf

Purpose ot Lift

Sketch CHECK

TASK

Position:

COMPONENT

1. .Positiont wheelchair near goal (transfer across shortest distence). Ninetydegree anglr is best. 2. Explains to client whet they ere to do (words or gestures). 3. tccks wheelchrir brakes. 4. Removes adaptive devices: b. erm rests Iif possiblel. c. seatbelt( other adaptive equipment. Positrons for lift by: a. Standing at side of chair, at client’s hip angle. b. with feet apert (width of hips, at least). Bending posture, knees bent, AND ‘. spine straight, may be slight bend forward at waist. Slides client forward on seat. to permit adequate room for 7.

Supports client for IIf?: a. One arm beneath client’s arms and shoulders to support head, neck, and upper torso. b. Other arm beneath cl~ent’r Ihighs to support pelvis. 8. Hugs cltent (reduces distance betMen cl~cnt and staff). 9. Lgfts straight up by untrndtng knees (back remams stratght and erect). SMOOTH MOVEMENT 0. Pivots (turns on bails of feet, or shon steps, without twirrlng torso) and aligns client wth new surface. 11. Bends knees, lowers client to new surface Back stratght. 12. Securely positions client on new surface, then releases. 13. Fastens seatbelts. where appropriate.

Describe unsafe components,

if sny:

How long did tt take to complete

this observation?

Checklist for task analysis of total lift transfer by one employee. (From Alavosius and Sulzer-Azaroff, 1985. Reprinted with permission of Applied Ergonomics. )

Fig. 4.

carrying out the performance, and the tape is examined to find instances in which the response has been properly executed. That is then shown to the individual, who tries to repeat the performance just as it was displayed. When the behavior involves an especially difficult motor skill component, sometimes graduated guidance can be used. (See Foxx and Azrin, 1973, for a detailed description of this procedure. ) Just as the term implies, the trainer

195

actually guides the person’s movements physically. The “graduated” aspect means that slowly, as the trainee begins to assume more control over executing the correct motion, the pressure exerted by the trainer is lessened and the placement of the guidance moved away from the locus of the response, until the performance consistently is being executed correctly and the assistance has faded away. Graduated guidance could prove very helpful in preventing conditions such as tendonitis, especially in cases in which it is caused by improper hand or wrist movements, as in reaching or packing. Research on this type of habit reversal remains to be conducted. EVALUATION

Evaluating the processes and outcomes of behavior modification strategies that attempt to deal with occupational health and safety is facilitated by the methods the field has evolved. Objective specification of performance, rather than of inferred internal processes, and careful recording techniques, minimize bias. Repeated measurement permits more representative assessment of ongoing behavior or its products than do the single pre- and post-assessent points, so frequently encountered in social science experimentation. Baselines, rather than arbitrarily determined criteria, serve as standards against which subsequent progress can be evaluated, providing for realistic assessment of improvement. Baselines of adequate length also permit aspects of the observational method to be eliminated as possible sources of change, as those conditions are in place during the baseline as well as during the treatment. Any inflation or deflation in rate should be reflected equivalently under both conditions. Yet the issue of whether measured improvement might actually be a function of time, expectations or a placebo effect needs to be tackled another way: By putting experimental controls in place. As described earlier, this can be done by collecting baselines of varying lengths, then applying the identical procedures to different people or in different places or with different behaviors. If change occurs reliably, regardless of when the procedure is implemented, and only then, the influence of the procedure is convincingly documented. (cf. Fig. 3 for an example). As an alternative, if behavior changes systematically as a function of the sequential presentation and withdrawal of the procedure, the case is also quite convincing. These and other experimental analytic methods particularly well suited to applied settings are discussed in detail in Komaki (1982)) Sulzer-Azaroff and Mayer (1977), Hersen and Barlow (1976)) and elsewhere. CONCLUSIONS

Through its technology documenting an important

and discoveries, behavioral psychology contribution in the area of occupational

has begun safety and

196

health. Progress has been made in identifying methods for increasing safe and decreasing unsafe performances and conditions and ways to cope with behaviors that are complex or resistant to change are starting to be examined. An extensive array of behavioral procedures have been analyzed in areas of application outside of the occupational health and safety setting. These, reported in the over thirty professional behavioral journals and in a burgeoning number of texts, often have relevance for this field and should be investigated. Researchers of occupational health and safety will expand their knowledge and competence in the area by continuing their educations in behavior analysis and applying what they learn. The ultimate beneficiaries will be the working public, for whom more effective accident and injury prevention should be achieved.

REFERENCES Accident Facts, 1984. National Safety Council, Chicago. Alavosius, M.P. and Sulzer-Azaroff, B.. 1985. An on-the-job method to evaluate patient lifting technique. Appl. Ergonom., 16: 305-311. Alavosius. M.P. and Sulzer-Azaroff, B., 1986. The effects of performance feedback on the safety of client lifting and transfer. J. Appl. Behav. Anal., 19: 261-267. Alavosius. M.P. and Sulzer-Azaroff, B., 1987. An examination of the effects of two schedules of feedback on the endurance of behavior change. Unpublished manuscript. Azrin. N.H. and Holz, W.C, 1966. Punishment. In: W.H. Honig (Ed.), Operant Behavior: Areas of Research and Application. Appleton-Century-Croft, New York, pp. 380-447. Baer, D.M., Williams, J.A., Osnes, P.G. and Stokes, T.F., 1984. Delayed reinforcement as an indiscriminable contingency in verbal/nonverbal correspondence training. J. Appl. Behav. Anal.. 17: 429-440. Baer, D.M.. Wolf, M.M. and Risely, T.R., 1968. Some current dimensions of applied behavior analysis. J. Appl. Behav. Anal., 1: 91-97. Carter, N. and Menckel, E., 1985. Near-accident reporting: A review of Swedish research. J. Occup. Accid., 7: 41-64. Daniels, A.C. and Rosen, T.A., 1982. Performance Management. Performance Management Publications, Inc., Tucker, GA. Davis, H. and McIntire, R.W., 1969. Conditioned suppression under positive, negative, and no contingency between conditioned and unconditioned stimuli. J. Exp. Anal. Behav., 12: 633-640. Dowrick, P.W. and Dove, C., 1980. The use of self-modeling to improve the swimming performance of spina bifida children. J. Appl. Behav. Anal., 13: 51-56. Fairbank, J.A. and Prue, D.M., 1982. Developing performance feedback systems. In: L. Frederiksen (Ed. ) , Handbook of Organizational Behavior Management. John Wiley & Sons, New York, pp. 281-299. Fellner, D. and Sulzer-Azaroff, B., 1984a. A behavioral analysis of goal setting. J. Organ. Behav. Manage., 6 (1): 33-51. Fellner, D.J. and Sulzer-Azaroff, B., 198413.Increasing industrial safety practices and conditions through posted feedback. J. Saf. Res., 15: 17-21. Fellner, D.J. and Sulzer-Azaroff, B., 1985. Occupational safety: Assessing the impact of adding assigned or participative goal setting. J. Organ. Behav. Manage., 7: 3-24. Fox. C.J. and Sulzer-Azaroff, B., in press. Instructions and written feedback to improve the com-

197

pletion of accident reports by foremen in a paper mill. Paper presented at the annual convention of the Association for Behavior Analysis, Columbus, OH, May 24-27. J. Saf. Res. Foxx, R.M. and Azrin, N.H., 1973. The elimination of autistic self-stimulatory behavior by overcorrection. J. Appl. Behav. Anal., 6: I-14. Guevremont, D.C., Osnes, P.G. and Stokes, T.F., 1986. Programming maintenance after correspondence training interventions with children. J. Appl. Behav. Anal., 19: 215-219. Heinrich, H.W., Petersen, D. and Roos, N., 1980. Industrial Accident Prevention. McGraw-Hill, New York. Hersen, M.H. and Barlow, D.H., 1976. Single Case Experimental Designs. Pergamon Press, New York. Hopkins, B.L., Conrad, R.J., Dangel, R.F., Fitch, H.G., Smith, M.J. and Anger, K.W., 1986. Behavioral technology for reducing occupational exposures to styrene. J. Appl. Behav. Anal., 19: 3-11. Kazdin, A.E., 1975. Behavior Modification in Applied Settings. Dorsey, Homewood, IL. Kim, J.S. and Hamner, W.C., 1976. Effects of performance feedback and goal setting on productivity and satisfaction in an organization setting. J. Appl. Psychol., 61: 48-57. Komaki, J.L., 1982. The case for the single case: Making judicious decisions about alternatives. In: L. Frederiksen (Ed.), Handbook of Organizational Behavior Management. John Wiley & Sons, New York, pp. 145-176. Komaki, J., Barwick, K.K. and Scott, L.R., 1978. A behavioral approach to occupational safety: Pinpointing and reinforcing safe performance in a food manufacturing plant. J. Appl. Psychol., 63: 434-445 Komaki, J., Heinzmann, A.T. and Lawson, L., 1980. Effect of training and feedback: Component analysis of a behavioral safety program. J. Appl. Psychol., 65: 261-270. Laner, S. and Sell, R.G., 1960. An experiment on the effects of specially designed safety posters. Occup. Saf., 34: 153-169. Mackintosh, N.J., 1977. Stimulus control: Attentional factors. In: W.K. Honigand J.E.R. Staddon (Eds.), Handbook of Operant Behavior. Prentice-Hall, Englewood Cliffs, NJ, pp. 481513. Morse, W.H. and Kelleher, R.T., 1977. Determinants of reinforcement and punishment. In: W.K. Honig and J.E.R. Staddon (Eds.), Handbook of Operant Behavior. Prentice-Hall, Englewood Cliffs, NJ, pp. 174-200. Rhoton, W.W., 1980. A procedure to improve compliance with coal mine safety regulations. J. Organ. Behav. Manage., 2: 243-249. Ritschl, E.R., Mirman, RI., Hall, R.V., Sigler, J.R. and Hopkins, B.L., 1976. Personal safety program A behavioral approach to industrial safety. Unpublished manuscript. Skinner, B.F., 1938. The Behavior of Organisms. Appleton-Century-Crofts, New York. Sulzer-Azaroff, B,. 1978. Behavioral ecology and accident prevention. J. Organ. Behav. Manage., 2: 122-130. Sulzer-Azaroff, B., 1982. Behavioral approaches to occupational safety and health. In: L. Frederiksen (Ed.), Handbook of Organizational Behavior Management. John Wiley & Sons, New York, pp. 505-538. Sulzer-Azaroff, B. and Fellner, D.J., 1984. Searching for performance targets in the behavior analysis of occupational health and safety. J. Organ. Behav. Manage., 6 (2): 53-65. Sulzer-Azaroff, B. and Mayer, G.R., 1977. Applying Behavior Analysis Procedures with Children and Youth. Holt, Rinehart and Winston, New York. Sulzer-Azaroff, B., Foxx, C., Davis, J.M. and Moss, SM., 1987. Social support and safety on the job. Submitted for publication. Sulzer-Azaroff, B. and de Santamaria, C., 1980. Industrial safety hazard reduction through performance feedback. J. Appl. Behav. Anal., 13: 287-295. Zohar, D., Cohen, A. and Azar, N., 1980. Promoting increased use of ear protectors in noise through information feedback. Hum. Factors, 22: 69-79.