AMERICAN SOCIETY OF SAFETY ENGINEERS JULY 2000 19

nly a few years ago, manysafety and health profession-als knew little about behav-ior-based safety (BBS)—asystematic approach to pro-moting behavior supportiveof injury prevention. Today,

the media and many professional organi-zations are giving increased attention tothe topic. Safety professionals want tolearn more about the way this strategyoperates and what it accomplishes—inother words, does it really work?

That may not be the best question topose, however. This article briefly de-scribes behavioral safety systems andtheir fundamental elements. It discussesseveral (of myriad) factors that potential-ly influence the operation of these meth-ods in a given situation. Facts about itsimpact on incidence rates are also re-viewed to help readers judge the effec-tiveness of BBS and the appropriatenessof posing the question in this fashion.

FUNDAMENTAL ELEMENTSAlthough specific behavioral safety

systems may vary in form and complexi-ty, at their most basic level they share sev-eral common elements:

1) Identify (or target) behaviors thatimpact safety.

2) Define these behaviors preciselyenough to measure them reliably.

3) Develop and implement mechan-isms for measuring those behaviors inorder to determine their current statusand set reasonable goals.

4) Provide feedback.5) Reinforce progress.

WHAT QUALITIES DEFINE BBS TODAY?As any complex effort to prevent

threats to people’s wellness, BBS contin-ues to evolve. In many settings, behav-ioral safety is now an important facet ofthe total injury prevention package—integrated with areas such as ergonomics,engineering, training, and occupationalhealth and safety management.

To illustrate this integrated aspect, aninvestigation of a behavior-based strate-gy to prevent back injuries includedpinpointing of specific skills; trainingmaintenance workers in the ergonomicsof moving materials safely; bio-feedbackon muscle tension; and verbal feedbackand rewards for progress in practicingthose skills safely (Rosado). In anotherillustrative case, McCann and Sulzer-Azaroff selected a set of behavioral tar-gets for computer-terminal operatorsseeking to reduce their risk of injury;these targets were based on recommen-dations of an occupational health physi-cian (277+).

Safety management was a focus areaof a program implemented in a telecom-munications plant (Sulzer-Azaroff, Loaf-mann, et al 99+). Managerial pinpointsincluded providing weekly safety per-formance updates to the safety director;participating in the awards program; pro-viding reinforcement to supervisors ofdepartments that met or exceeded goals;helping to develop pinpoints; and en-couraging safety suggestions. Safety-related behaviors include not only thoseof workers, but also those of supervisors,managers and others within the systemwhose support is crucial.

BEHAVIOR-BASED SAFETY

Does BBS Work?Does BBS Work?BEHAVIOR-BASED SAFETY & INJURY REDUCTION:

A SURVEY OF THE EVIDENCEBy BETH SULZER-AZAROFF and JOHN AUSTIN

As behavioral safetycontinues to gain visibilitysafety professionals must

assess the efficacyof the methods. From a

conservative perspective,the question, “Does behavior-

based safety work to decreaseincidents in every case?”can be difficult to answer.

However, sufficient dataare available to demonstrate

that the approach canaccomplish that end.

This article also examinesa related question: “How can

safety professionals makebehavior-based safety

work better?” Depending on the purpose of the

intervention, severalsuggestions are offered.

OO

WHAT DOES “DOES BBS WORK” MEAN?Beyond its basic elements, BBS can vary

from time to time and place to place.Depending on an organization’s needs,resources and objectives, each system willhave uniquely customized features. So,when potential users ask, “Does BBSwork?” it is difficult to provide a definitiveanswer. Several variables account for thediversity of results one might encounter.

Target BehaviorsData on efficacy are influenced by the

behaviors targeted for change. An inter-vention’s ultimate value can only be asgood as the precise behaviors selected.These behaviors must closely relate toimproving well-being—in other words,when practiced consistently, they willreduce risks and improve the physicalwell-being of employees and manage-ment. Attempting to change what peoplesay, think or feel is not the same as chang-ing what they do. In addition, the de-scription of a safe practice must beobjective, correct and complete (i.e.,valid). No matter how much individualsimprove the way they perform a task, ifkey components of that task are irrele-vant, omitted or wrong, materials han-dlers will remain in danger.

Purpose of InterventionWhen setting behavioral targets, one

must consider the objective of the inter-vention. The purpose of an endeavor caninfluence the outcome as well as the man-ner in which findings are interpreted andreported. The form of evaluation candepend on whether the intervention ispart of a controlled laboratory investiga-tion, a systematic field study or a localassessment.

Scientific research attempts unam-biguously to sort out competing explana-tions for results of given interventions.For example, an investigator may studythe differential influence of massed ver-sus spaced practice as a method of pro-moting rapid improvement in the wayhealthcare workers lift and transferpatients (Alavosius and Sulzer-Azaroff151+). Program evaluators, on the otherhand, may ask whether specific varietiesof BBS are effective tools for improvingsafe practices or reducing injury rateswithin their organizations (Alavosius inpress; Krause, et al 1+).

Of course, both researchers and practi-tioners are concerned that results be asclear as possible. However, the scientisttakes far greater care to ensure that com-peting explanations can be rejected, whilethe practitioner is usually more concernedwith effectiveness and cost-efficiency.

Variations in ApplicationBasic components, such as identifying

and defining safe practices, observing,

and delivering feedback and reinforce-ment may be applied differently from oneprogram to another, as may supplemen-tal elements. These variations can affectthe outcome. (It should be noted thatunless each individual element has beencarefully isolated from the rest, the extentof any one facet’s influence generallyremains ambiguous. In practice, thesevariations tend to be combined andviewed as elements of the total interven-tion “package.”)

Feedback strategies are one suchexample. Behavior-analysis textbooksoffer diverse definitions for at least 12 dif-ferent types of feedback (Sulzer-Azaroffand Mayer; J. Cooper, et al). The processof implementing feedback strategies canalso vary. For example, Balcazar, et alidentified six characteristics of feedbackthat varied widely between studies (65+).McAfee and Winn commented on thesefluctuations when, in a set of studies theyreviewed, they found feedback assumingmany different forms (7+).

One can also cite variations in howagents define, supplement and imple-ment these components. Consider thisbrief list of potential variations in thefeedback-delivery process.

•Communicate general approval ordisapproval in the form of a gesture, a spo-ken “good,” “okay” or “safe” for one ormore safe performance targets, or sharemore-specific information in words orvisually (e.g., graph) to describe whatabout the performance was safe or unsafe.

•Comment on positive or negativeactions or offer constructive suggestionsfor future behavior, or some combination.

•Offer praise, recognition and/orrewards to individuals or groups.

•Provide one-on-one feedback, or askpeers, supervisors, subordinates, safetypersonnel, managers or external sourcesto deliver it.

•Offer feedback publicly or privatelyto individuals, groups or both.

•Provide feedback verbally, in writtenform (e.g., graph or digitally), via tangi-ble reward or some combination.

•Deliver feedback according to a fixedor variable schedule (often, seldom orsomewhere in-between).

•Combine feedback with goal-setting(interim, final or both).

•Precede feedback by a baseline phasethat includes all elements except feedback.

Many researchers and practitionersalso build-in components beyond identi-fication, observation and feedback. Theymay incorporate goal-setting (Austin, etal 49+; M. Cooper, et al 219+; Fellner andSulzer-Azaroff 7+; Reber and Wallin544+; Reber, et al 51+) or problem-solving(Fellner and Sulzer-Azaroff 7+; Killimett209+; Krause; Laitinen, et al 35+; Walters34+). Often, training is a key aspect aswell (Sulzer-Azaroff, Fox, et al).

Again, researchers may define thesecomponents differently. Goals may be seton a participative or assigned basis (Fellnerand Sulzer-Azaroff 3+). Many do not spec-ify how goals were set (Cohen and Jensen125+). Some install highly structured sys-tems to support problem solving (Krause),while others provide less-formal support(Laitinen, et al 35+).

Variations in the Role of System CoordinatorThe role of the individual who designs

and coordinates the system may influ-ence results as well. To the applied behav-ioral researcher, the “does it work”question usually means, “Can a function-al (cause-and-effect) relation be demon-strated between a particular set of BBSvariables and those that make up safetyperformance?” Relying primarily onmeasured changes in rates of safe behav-ior and perhaps incidents, this researcherapproaches the question by tightly con-trolling those variables thought to affectthe targeted outcome measures.

In fact, however, many of the varia-tions in intervention strategies have moreto do with achieving experimental con-trol (e.g., establishing inter-rater reliabili-ty, using the researcher to observe andprovide feedback) than anything else. Ifthe effect of the well-controlled interven-tion can be shown by using valid, reliableoutcome measures, the researcher con-cludes, “That intervention had a demon-strable effect on safety performance” (inother words, BBS worked in that case).

One example of this is a studydesigned to explore the impact of a self-generated and/or an external feedbackpackage on changes in the body posturesamong a small group of volunteer com-puter terminal operators (McCann andSulzer-Azaroff 277+). Following the suc-cessful demonstration and follow-upassessments, these researchers concludedthat their particular BBS package workedfor this new class of behaviors. Insuffi-cient resources prevented a wide-scaleimplementation of the methods to test thefinding’s validity and generality.

Researchers also want to be as certainas possible that their purported “treat-ments” actually occur as planned. Conse-quently, they or trained assistants deliverintervention components instead of inter-nal organizational personnel (Alavosiusand Sulzer-Azaroff 151+; Lingard andRowlinson 243+; Ludwig and Geller 253+;Reber, et al 51+; Saarela 177+). This causesone to question whether the organizationcould obtain similar results by applyingthose methods independently.

To the safety professional or businessmanager, however, the “does it work”question can mean something entirely dif-ferent. Typically, publishing their results isnot a high priority. Nor do they have timeto wait for scientific confirmation of the

20 PROFESSIONAL SAFETY

JULY 2000 21

results. Immediate decisions must bemade to address human and cost issues.

Therefore, managers must make thebest judgment based on the informationavailable. Their questions often encom-pass a wider domain of practical con-cerns: Will it work in this organization, inthis work culture? Will all levels of theworkforce buy into the program’s con-cepts? Are the resources available tomake this work over the long term? Ifemployees buy-in and resources areavailable, should the firm expect bothimmediate and continuous improve-ment—not only in safe performance, butalso in fewer incidents and injuries?

A business-oriented intervention typi-cally calls on the organization’s work-force to deliver system components,frequently with each level playing a keyrole. In these cases, individuals whochampion the program provide the nec-essary training and guidance (Alavosiusin press; Krause, et al 1+; Krause). Thatperson or group is then challenged tomaintain the system’s integrity—to verifythat it is transpiring as planned.

Variations in PrioritiesWhile the universal goal is to reduce

incident and severity rates, other specificoutcomes can be affected, depending onthe role of the person/group within theorganization.

For example, executives, stockholdersand directors focus on the bottom line:How much does it cost and what is thereturn? BBS data are beginning to showshort- and long-term savings due toreduced injuries and their associated costs(insurance rates, direct medical, equip-ment replacement and repair, down-timeand lost-time costs).

Of course, figures can diverge, de-pending on methods used to calculatecosts. For example, Veltri offered a direct-cost impact model designed to demon-strate how reducing accidents can lowerthe degree of operating leverage (67+).Sulzer-Azaroff, Loafmann, et al used aless-formal, more-conservative basis toestimate that in three small departmentswithin a large manufacturing plant,$55,000 was saved due to reduced injuryrates over a six-month period (99+).

Production and quality managers maybe concerned with production rates andquality, the amount of time employeeswould be away from work for training,observing and providing feedback, and

the effects of performing less-expedi-tiously but more safely. Others may seekchange in safety climate or culture.

Individuals with power or control ofreinforcers will focus on those outcomesof concern to them. This fact, alone, willinfluence results because what gets mon-itored gets reinforced and what gets rein-forced is repeated more frequently(Komaki 270+). The opposite is alsotrue—those elements monitored lesscarefully or ignored tend to diminish.

Variations in Support StructuresThe organizational support structures

in place also influence long-term results.Many experts agree that the following fac-tors can affect the durability of any BBSeffort: clear, visible, ongoing senior execu-tive commitment; institutionalized mech-anisms such as actively participatingsafety/steering committees; data collec-tion (e.g., minutes of meetings; observingand recording numbers of observationsconducted, feedback charts up-to-date);feedback routines supporting systemmaintenance; encouragement of involve-ment by all personnel (Sulzer-Azaroff andLischeid 31+).

WHAT ABOUT RESULTS?What would many consider the bot-

tom-line measure when attempting toanswer the “does it work” question? Inmost cases, incidence rates. Therefore, theauthors surveyed literature on BBS to dis-cover what its impact has been on “acci-dents” or incidence and injury rates.

The Network on Behavioral Safety ofthe Assn. for Behavior Analysis main-tains an exhaustive list of reports on thistopic. Eighty-three of those listed de-scribe case demonstrations or experi-ments that contained hard numericaldata. “Promote safe performance” wasthe main focus of most of these studies.

Many reports omitted correlated dataon accident/injury rates, the measure ofconcern in this article. Several factors mayhave driven these omissions: 1) the investi-gation or implementation was conductedover too short a time period to producemeaningful differences; 2) the number ofparticipants was too small to generatemeaningful injury data (McCann andSulzer-Azaroff 277+); 3) the base rate ofinjuries was too low to be meaningful(Sulzer-Azaroff 11+); and 4) data was keptconfidential for business purposes orbecause researchers had misgivings about

its reliability. It is also possible that someinconclusive or negative incidence/injurydata simply were not submitted or accept-ed for publication. [It should be noted thatin most fields involving intervention (e.g.,clinical health, social service, psychothera-py), there is potential bias because poor,negative or non-results may not be submit-ted for publication—or when they are,they are often rejected.]

ARTICLE ANALYSISOf the 83 data-based evaluations of

behavioral safety programs, the authorsidentified 33 that reported data onchanges in incidence rates. Table 1 pro-vides details of these studies.

Two researchers then independentlycategorized changes in incidence rates ineach study. In all but five cases, entriesmatched, producing an agreement score of85.7 percent [(agreements)/(agreements +disagreements) x 100]. Three disagree-ments occurred because one reviewer con-verted the reported incidence frequenciesto percentage of changes in incidence rates,while the other reviewer did not. In addi-tion, in three cases, one reviewer over-looked a portion of the reported data.Therefore, to portray the results as accu-rately as possible, the disagreements werediscussed and ultimately reclassified.

ANALYSIS OF RESULTSParticipants & Settings

The number of participants rangedfrom five (Haynes, et al 407+; Larson, et al571+) to 39,664 that one organizationtracked (Krause, et al 1+). The number ofsites evaluated ranged from single loca-tions to 73 (Krause, et al 1+). Settingsincluded construction sites, grocerydistributors, electrical and gas utilities,manufacturing plants, mines, policedepartments, railroads, shipyards andtransit systems. In addition, althoughmost studies were conducted in the U.S.,some involved sites in Chile, Cuba,Finland, Hong Kong, Spain and the U.K.

Injury OutcomesOf the 33 articles reviewed, 32 report-

ed reductions in injuries. In many cases,however, the reporting format differed.Some listed numbers of lost-workdays;others, numbers of accidents. In addition,some accident rates were calculated onthe basis of hours worked or miles dri-ven. (In the future, researchers should beencouraged to report rates—not just raw

In many settings, behavioral safety is now an important facetof the total injury prevention package—integrated with areas such asergonomics, training, and occupational health and safety management.

22 PROFESSIONAL SAFETY

STUDY AUTHOR(S) NUMBER OF PARTICIPANTS SETTING REDUCTION IN ACCIDENT/INCIDENT RATES

M. Alavosius (in press) 5 – 500 50 small companies Lost-workdays per 100 workers: 184 pre-intervention; 111 during; 84 post-intervention (six months) and 58 (12 months).

M.D. Cooper, et al (1994)

540 Construction industry From 6.33 prior to 3.88 at end; from 3.3 to 0.56 on checklisted items.

D.J. Fellner and B. Sulzer-Azaroff (1984)

158 Paper mill Significant difference between pre- and during-feedback—from 6.9 percent to 4.9 percent.

F. Fiedler (1987) 500 Mine Baseline = 226 percent; follow-up two percent over industry average.

D.K. Fox, et al (1987) 1,754 Coal mine Range: 15 to 32 percent.

D. Harshbarger and T. Rose (1991)

a) 100 b) 350 – 400

a) Bedding b) Footwear

Lost-time accidents a) 95 percent; b) 87 percent.

R.S. Haynes, et al (1982) 100 Urban transit 24.9 percent.

B.S. Karan and R.E. Kopelman (1987)

Not reported Vehicular and industrial 2.2 percent and 4.0 percent.

J.L. Komaki, et al (1978) 38 Food manufacturing plant Injuries fell to “. . . less than 10 lost-time accidents per million hours worked, a relatively low number” (pg. 441).

J.L. Komaki, et al (1980) 55 Vehicle maintenance Decline from 3.0 lost-time injury rate per month preceding to 0.4 during and 1.8 following intervention.

T.R. Krause, et al (1999) 51 to 3,000 per site (39,664 across 73 sites)

73 facilities participating for up to five years

Year 1: 26 percent. Year 2: 42 percent. Year 3: 50 percent. Year 4: 60 percent. Year 5: 69 percent.

H. Laitinen, et al (1998) 300 Engineering workshop 46 percent reduction in absenteeism.

B. Loafmann (1998) Not reported Utility company Treatment group about 78 percent; control group had a 50-percent increase.

L. Lopez-Mena and J.V. Antidrian (1990)

914 Forestry (2) and cement factory (1)

62.8 percent; maintained for three years.

L. Lopez-Mena and R. Baynes (1988)

41 Electrical distribution system

84.9 percent in one setting; 60.8 percent in a second setting.

L. Lopez-Mena, et al (1988)

191 Electrical energy distribution system

34.3 percent.

M. Mattila and M. Hyödynmaa (1988)

100 Building construction Accident rate per 100 workers at control site higher during (166) and after (55) than experimental site—94 and 47, respectively.

T. McSween (1995) Not reported Gas pipeline company 35 percent lost-time accidents.

T. McSween (1995) Not reported Chemical company (union-coordinated)

From four to zero the next 18 months.

R. Montero (1996) Not reported Industry (general) “Rate dropped almost to zero.”

M. Näsänen and J. Saari (1987)

32 Shipyard 50- to 75-percent reduction in accidents.

D. Petersen (1984) Not reported Railroad “Experimental groups had fewer injuries than control [groups].”

R.A. Reber, et al (1984) 105 Farm machinery manufacturing

53.83 percent.

R.A. Reber, et al (1990) 44 Manufacturing 50 percent.

K.L. Saarela (1989) 2,800 Shipyard Modest, non-significant reduction in accident frequency. This intervention involved a poster campaign, not a full behavioral program, and feedback to supervisors.

K.L. Saarela (1990) 24 Shipyard 20 percent during; about 40 percent after.

K.L. Saarela, et al (1989) 58 Shipyards No significant differences (poster campaign; general subject feedback).

K.L. Saarela (in press) >900 Shipyard 60+ percent.

J. Saari and M. Näsänen (1989)

24 Shipyard 25-percent reduction in accidents; 30-percent reduction in injuries.

R. Schwartz (1989) 110 Grocery distribution workers

39.4 percent.

M. Smith, et al (1978) 44 Shipyard Average decrease in eye injuries of 7.48 per 100 workers; control group average reduction of 1.16.

B. Sulzer-Azaroff, et al (1986)

140 Paper mill From 19 recordable incidents during baseline to two after feedback given for three behaviors.

B. Sulzer-Azaroff, et al (1990)

225 Telecommunications parts manufacturing plant

Number of OSHA recordables dropped by 17 from prior to during intervention; lost-time from 14 to 1.

TABLE 1 BBS Studies Reporting Impact on Injury Rates

JULY 2000 23

numbers—as well as the basis for the ratecalculation.) Differences in experimentaldesigns also affected data reports. Thoseusing group comparisons reported infor-mation on statistical significance, whilethose using repeated measure or inten-sive designs (e.g., multiple baselines)relied, according to convention, primarilyon graphic presentations. To simplify thedescriptive analysis for this article, when-ever feasible, average changes were cal-culated from pre- to during and/orpost-intervention.

To illustrate the difficulties encoun-tered in drawing unambiguous conclu-sions about the efficacy of a given BBSintervention, consider this example ofSaarela’s (in press) thwarted efforts tomaintain tight experimental controls dur-ing an intervention at a shipyard.

Following a two-hour training seminar,safety teams from each production depart-ment identified departmental safety prob-lems, then set goals, solved problems,implemented improvements, monitoredresults and provided feedback. Accordingto the original research plan, only threedepartments were to be involved, withothers joining later; however, when thosenot included saw the results, they de-manded immediate participation.

The success of those tactics was under-scored when results were analyzed. Bythe time the 900 workers in all 10 depart-ments were actively participating in theprocess, the shipyard’s accident rate wasmore than halved. Subsequently, the pro-gram was continued under the coordina-tion of the shipyard’s safety personnel.

Beyond experimental-design issues,other cautions must be considered. Forexample, how much room remained forimprovement? Sites with high injuryrates have a larger potential for improve-ment than those with low rates. One mustalso consider problems in the reliabilityof reporting of incidence rates.

THE REAL CHALLENGE:HOW TO MAKE BBS WORK BETTER

Despite the limitations noted, the datagathered support the conclusion that BBSsystems appear to have helped reduceinjuries on many occasions. In light ofthat finding, the next question is, “Giventhese particular purposes and circum-stances, how can this organization makeBBS work as well as possible?”

The answer depends on the main pur-pose of the intervention. If it is researchseeking to demonstrate a clear causalrelation between a given class of behav-iors and a particular change method, thenfinely honed, tightly controlled experi-ments must be designed. These include:

•valid, reliable measures of perfor-mance and side effects, under baseline,treatment and afterward;

•change strategies that hold the most

promise for attaining improvement in be-haviors of concern within a given context;

•data which demonstrate that theexperiment was conducted as described.

However, when the purpose is toimplement a process in the most-effec-tive, efficient way within a given organ-ization, one must look beyond the basicsof the technology itself to organizationalissues that impact how the technology isimplemented. Are mechanisms in placeto sustain longevity? Is the method com-patible with the site implementing it?Does a reasonable balance exist betweenobjectives and resources?

Others share a similar perspective.After evaluating the outcomes of 41efficacy studies of safety training, John-ston, et al concluded that effective trainingprograms shared four common character-istics: 1) needs assessment; 2) programdevelopment; 3) goal setting; and 4) knowl-edge of results through feedback (147+).In a related analysis, Hidley identifiedseven factors critical to the success of BBSimplementations:

1) Use a process blueprint character-ized by structure and rigorous implemen-tation, while allowing flexibility to adapttechnologies to site-specific needs.

2) Create buy-in through communica-tion. Highest levels of leadership, manage-ment and labor must be committed toimprovement—in both actions and words.

3) Demonstrate leadership and activesupport for the change process.

4) Ensure implementation-team com-petence through training and follow-up.

5) Provide action-oriented skills train-ing in a structured, safe environment inwhich trainees can practice new skills.

6) Use feedback data based on safebehavior under the control of employeesto remove barriers.

7) Ensure that adequate technical assis-tance (internal or external) is availablefrom individuals who make practical, fea-sible and well-considered recommenda-tions for process success (30+).

A similar list describing a qualitybehavioral safety system was detailed in asurvey of behavioral safety specialists(Sulzer-Azaroff and Lischeid 31+). That listincluded: well-defined, correct, relevanttarget behaviors; worker participation;managerial involvement; observationaldata collection; training/education; posi-tive reinforcement; interventions based onsound behavioral technology; behavioralfeedback; and organizational systemsstructured to support the effort.

CONCLUSIONThe studies evaluated suggest that

despite some reservations about how accu-rately the published literature reflects real-ity, in many cases, incidence rates havebeen reported to decline following imple-mentation of BBS system. Presumably the

growing interest in this technology is justi-fied by the gains many companies havereportedly achieved. Most essential tothose gains are the training, organizational,managerial, follow-through and other fac-tors discussed. To ensure that the systemworks most effectively, these factors, alongwith the findings generated from ongoingand future research, should be consideredwhen designing any BBS system. �

REFERENCESAlavosius, M. “Behavioral Approaches to

Organizational Safety.” In The Handbook ofApplied Behavior Analysis, J. Austin and J.E.Carr, eds. Reno, NV: Context Press (in press).

Alavosius, M. and B. Sulzer-Azaroff.“Acquisition and Maintenance of Health-care Routines as a Function of FeedbackDensity.” Journal of Applied Behavior Analy-sis. 23(1990): 151-162.

Austin, J., et al. “Using Feedback andReinforcement to Improve the Performanceand Safety of a Roofing Crew.” Journal ofOrganizational Behavior Management. 16(1996):49-75.

Balcazar, F., et al. “A Critical, ObjectiveReview of Performance Feedback.” Journalof Organizational Behavior Management.7(1985/6): 65-89.

Cohen, H.H. and R.C. Jensen. “Measuringthe Effectiveness of an Industrial Lift TruckSafety Training Program.” Journal of SafetyResearch. 15(1984): 125-135.

Cooper, J.O., et al. Applied BehaviorAnalysis. Columbus, OH: Merrill, 1987.

Cooper, M.D., et al. “Reducing AccidentsUsing Goal Setting and Feedback: A FieldStudy.” Journal of Occupational and Organiza-tional Psychology. 67(1994): 219-240.

Fellner, D.J. and B. Sulzer-Azaroff.“Increasing Industrial Safety Practices andConditions Through Posted Feedback.”Journal of Safety Research. 15(1984): 7-21.

Fellner, D.J. and B. Sulzer-Azaroff.“Occupational Safety: Assessing the Impactof Adding Assigned or Participative Goal-Setting.” Journal of Organizational BehaviorManagement. 7(1986): 3-24.

Fiedler, F. “Structured ManagementTraining in Underground Mining: FiveYears Later.” Proceedings of Bureau of MinesTechnology Transfer Seminar. Pittsburgh:Bureau of Mines, 1987. 149-153.

Fox, D.K., et al. “The Long-Term Effectsof a Token Economy on Safety Performancein Open-Pit Mining.” Journal of AppliedBehavior Analysis. 20(1987): 215-224.

Harshbarger, D. and T. Rose. “NewPossibilities in Safety Performance and theControl of Workers’ Compensation Costs.”Journal of Occupational Rehabilitation. 1(1991):133-143.

Haynes, R.S., et al. “Reducing AccidentRates with Organizational Behavior Modifi-cation.” Academy of Management Journal.25(1982): 407-416.

Hidley, J.H. “Critical Success Factors forBehavior-Based Safety: Avoiding CommonPitfalls and Achieving Real Gains.” Profes-sional Safety. July 1998: 30-34.

Johnston, J.J., et al. “The Efficacy of

Training for Occupational Injury Control.”Occupational Medicine: State of the Art Reviews.9(1994): 147-158.

Karan, B.S. and R.E. Kopelman. “TheEffects of Objective Feedback on Vehicularand Industrial Accidents: A Field Experi-ment Using Outcome Feedback.” Journalof Organizational Behavior Management.8(1987): 45-56.

Killimett, P.T. “Completing the Loop:Action Planning for Continuous SafetyImprovement.” Tappi Journal. 75(1992): 209-211.

Komaki, J.L. “Toward Effective Super-vision: An Operant Analysis and Compari-son of Managers at Work.” Journal of AppliedPsychology. 71(1986): 270-279.

Komaki, J.L., et al. “Effect of Trainingand Feedback: Component Analysis of aBehavioral Safety Program.” Journal ofApplied Psychology. 65(1980): 261-270.

Krause, T.R. The Behavior-Based SafetyProcess: Managing Involvement for an Injury-Free Culture. 2nd ed. New York: VanNostrand Reinhold, 1997.

Krause, T.R., et al. “Long-Term Evalua-tion of a Behavior-Based Method forImproving Safety Performance: A Meta-Analysis of 73 Interrupted Time-SeriesReplications.” Safety Science. 32(1999): 1-18.

Laitinen, H., et al. “Improving Physicaland Psycho-Social Working ConditionsThrough a Participatory Ergonomic Proc-ess.” International Journal of Industrial Ergo-nomics. 21(1998): 35-45.

Larson, L.D., et al. “Reduction of PoliceVehicle Accidents Through MechanicallyAided Supervision.” Journal of AppliedBehavior Analysis. 13(1980): 571-581.

Lingard, H. and S. Rowlinson. “Behav-ior-Based Safety Management in HongKong’s Construction Industry.” Journal ofSafety Research. 28(1997): 243-256.

Loafmann, B. “Behavior-Based Safety:Power & Pitfalls.” Professional Safety. Aug.1998: 20-23.

Lopez-Mena, L. and J.V. Antidrian.“Applicaciones Del Refuerzo Positivo A LaReduccion De Accidents En El Trabajo.[Applications of Positive Reinforcement toReduction of Work Accidents.]” RevistaLatinoamericana de Psicologia. 22(1990): 357-371.

Lopez-Mena, L. and R. Baynes. “Preven-cion de Riesgos en el Trabojo: Effectos de laRetroalimentacion y La Participacion. [Pre-vention of Work Accidents: Effects of Feed-back and Participation.]” Avances en PsicologiaClinica Latinamericana. 6(1988): 53-65.

Lopez-Mena, L., et al. “Beneficios Econó-micos Pbtenidos con Un Programa Conduct-ual en Seguridad Del Trabajo. [EconomicBenefits Obtained with a Behavioral WorkSafety Program.]” Psicologia del Trabajo y deLas Organizaciones. 4(1988): 74-86.

Ludwig, T.D. and E.S. Geller. “AssignedVersus Participative Goal-Setting and Re-sponse Generalization: Managing InjuryControl Among Professional Pizza Deliver-ers.” Journal of Applied Psychology. 82(1997):253-261.

Mattila, M. and M. Hyöynmaa. “Pro-

moting Job Safety on Building: AnExperiment on the Behavior AnalysisApproach.” Journal of Occupational Acci-dents. 9(1988): 255-267.

McAfee, R.B. and A.R. Winn. “The Useof Incentives/Feedback to Enhance WorkPlace Safety: A Critique of the Literature.”Journal of Safety Research. 20(1989): 7-19.

McCann, K.B. and B. Sulzer-Azaroff.“Cumulative Trauma Disorders: BehavioralInjury Prevention at Work.” Journal ofApplied Behavioral Science. 32(1996): 277-291.

McSween, T. The Behavior-Based SafetyProcess. New York: Van Nostrand Reinhold,1995.

Montero, R. “How Much ‘Participation’Does a Behavior Modification Program Needfor Improving Safety?” Presentation at theInternational Conference on OccupationalHealth, Estocolmo, Suecia, Sept. 1996.

Näsänan, M. and J. Saari. “Effects ofPositive Feedback on Housekeeping andAccidents at a Shipyard.” Journal of Occupa-tional Accidents. 8(1987): 237-250.

O’Toole, M. “The Relationship BetweenEmployees Perceptions of Safety andOSHA Lost-Time Injury Rates.” Fundedproject in progress (2000).

Petersen, D. “An Experiment in PositiveReinforcement.” Professional Safety. May1984: 30-35.

Reber, R.A., et al. “Improving Safety Per-formance with Goal Setting and Feedback.”Human Performance. 3(1990): 51-61.

Reber, R.A. and J.A. Wallin. “The Effectsof Training, Goal Setting and Knowledge ofResults on Safe Behavior: A ComponentAnalysis.” Academy of Management Journal.27(1984): 544-560.

Reber, R.A., et al. “Reducing IndustrialAccidents: A Behavioral Experiment.”Industrial Relations. 23(1984): 119-125.

Rosado, W.T. “The Application of Be-havioral Principles to Promote Safe Practicesand Reduce Accidents and Injuries.” Unpub-lished master’s thesis. Miami: Florida Inter-national University, 1999.

Saarela, K.L. “A Poster Campaign forImproving Safety on Shipyard Scaffolds.”Journal of Safety Research. 20(1989): 177-185.

Saarela, K.L. “An Intervention ProgramUtilizing Small Groups: A ComparativeStudy.” Journal of Safety Research. 21(1990):149-156.

Saarela, K.L. “An Intervention Study forImproving Safety at a Shipyard.” In People atWork, H. Laitinen and J. Saari, eds. Finland:Finnish Dept. of Occupational Health, Div. ofOccupational Safety (in press).

Saarela, K.L., et al. “The Effects of anInformation Safety Campaign in the Ship-building Industry.” Journal of OccupationalAccidents. 10(1989): 255-266.

Saari, J. And M. Näsänen. “The Effect ofPositive Feedback on Industrial House-keeping and Accidents: A Long-Term Studyat a Shipyard.” International Journal of Indus-trial Ergonomics. 4(1989): 201-211.

Schwartz, R. “Cognition and Learning inIndustrial Accident Injury Prevention: AnOccupational Therapy Perspective.” HealthPromotion Prevention Programs. 6(1989): 67-85.

Smith, M., et al. “Behavioral Modifica-tion Applied to Occupational Safety.”Journal of Safety Research. 10(1978): 87-88.

Sulzer-Azaroff, B. “Behavioral Ecologyand Accident Prevention.” Journal of Organi-zational Behavior Management. 2(1978): 11-44.

Sulzer-Azaroff, B. Who Killed My Daddy?A Behavioral Safety Fable. Concord, MA: Cam-bridge Center for Behavioral Studies, 1998.

Sulzer-Azaroff, B., C. Fox, et al. “Feed-back and Safety: Involving Workers.”Final report to National Institute of Occupa-tional Safety and Health. Grant #1 RO1-OHO1928-01, 1986.

Sulzer-Azaroff, B. and W.E. Lischeid.“Assessing the Quality of Behavioral SafetyInitiatives.” Professional Safety. April 1999:31-36.

Sulzer-Azaroff, B., B. Loafmann, et al.“Improving Occupational Safety in a LargeIndustrial Plant: A Systematic Replication.”Journal of Organizational Behavior Manage-ment. 11(1990): 99-120.

Sulzer-Azaroff, B. and G.R. Mayer. Be-havior Analysis for Lasting Change. FortWorth: Harcourt, Brace, Jovanovich, 1991.

Sulzer-Azaroff, B. and K.E. McCann.“Promoting Health and Safety on the Job.”In Performance Management: Behavior Analy-sis in Organizations, C.M. Johnson, W. Red-mon and T. Mawhinney, eds. New York:Springer Press (in press).

Veltri, A. “An Accident Cost ImpactModel: The Direct Cost Component.”Journal of Safety Research. 21(1990): 67-73.

Walters, H.A. “Identifying & RemovingBarriers to Safe Behaviors.” ProfessionalSafety. Jan. 1998: 34-36.

Beth Sulzer-Azaroff, Ph.D., is a professor emeri-tus at the University of Massachusetts and anadjunct professor at Florida International Univer-sity, Miami. In addition, as a performance manage-ment consultant, she develops and writes aboutbehavioral systems and how they promote perform-ance on the job. Her Ph.D. was earned at theUniversity of Minnesota.

John Austin, Ph.D., is an assistant professor in theDept. of Psychology at Western Michigan Univer-sity in Kalamazoo. At the university, he teachesorganizational and behavioral psychology, andbehavior-based safety courses. Austin is also co-edi-tor of the Journal of Organizational BehaviorManagement and serves on the editorial board ofthe Journal of Applied Behavior Analysis. Austinholds a B.A. from the University of Notre Dame andan M.S. and Ph.D. from Florida State University.

24 PROFESSIONAL SAFETY

READER FEEDBACKDid you find this article interestingand useful? Circle the correspondingnumber on the reader service card.

YES 28SOMEWHAT 29NO 30