036_045_Cooper_0209:Layout 1 1/13/2009 10:19 AM Page 36
Safety Management
Safety Management
Behavioral Safety
Interventions
A review of process design factors
By. M. Dominic Cooper
B BEHAVIORAL SAFETY has many advocates and
many critics. Advocates have seen or experienced the
effects of a well-designed process on incident rates
(Cooper, 2003; Lyon, 2001; Veazie, 1999). Conversely,
critics do not believe it truly involves workers in the
overall safety process (Howe, 1998) and believe the
concept has run its course (Naso, 2002). The promo-
tion of operant theory (Skinner, 1953) within the
behavioral safety field (Geller, 1996; McSween, 2002;
Krause, 1997) has led many to believe that the
antecedent-behavior-consequence model focuses
almost exclusively on the psychology of safety.
In reality, like other safety management interven-
tions, behavioral safety processes require a concert-
ed effort by all to produce desired results. A simple
five-step management model (known as CLEAR)
shows there is no magic bullet involved:
•Clarify the objectives.
•Locate the problems.
•Execute the change strategy.
•Assess current progress.
•Review and adapt the process.
The purpose of a behavioral safety process is to
reduce incidents triggered by unsafe or at-risk behav-
iors. To achieve this, specific behavioral problems are
identified by focusing on incidents that result from the
interaction between people and
M. Dominic Cooper, Ph.D., C.Psychol., their working environment.
CFIOSH, is CEO of B-Safe Management This could include the pres-
Solutions Inc. in Franklin, IN. He has nearly ence, quality and functioning of
20 years’ experience in the SH&E field. Cooper various management systems
holds a B.S. in Psychology from University of (safety and nonsafety), the
East London, an M.S. in Industrial Psychology quality of leadership, resources
from the University of Hull and a Ph.D. in available (financial and nonfi-
Occupational Psychology from the University of nancial) and the overall safety
Manchester in the U.K. He is also a chartered culture (Cooper, 2000).
psychologist with the British Psychological Once these problems are
Society and a chartered fellow of the U.K.’s identified, attempts are made
Institution of Occupational Safety & Health. to discover which antecedents
Cooper emigrated to the U.S. in 2001 to join (e.g., unavailable equipment)
the faculty at Indiana University, Bloomington, are driving at-risk behavior
where he taught safety education and (e.g., using improvised tools),
industrial/organizational psychology. He is a and which consequences (e.g.,
member of ASSE’s Central Indiana Chapter. saves time) are reinforcing or
36 PROFESSIONAL SAFETY FEBRUARY 2009 www.asse.org
maintaining that behavior so that appropriate correc-
tive actions can be taken. Executing the change strat-
egy usually involves addressing the antecedents to
remove barriers while the associated safety behaviors
are placed on checklists so workers can conduct ob-
servations of ongoing behavior. Observation results
are used to facilitate corrective feedback (i.e., a conse-
quence) to those observed and to track overall
progress. The trends in observation data over a peri-
od of time are used to adapt the process to suit the
particular circumstances (e.g., change the behaviors
on the observation checklists, provide tailgate topics).
Evolutionary Changes
Since its inception and application in the mid-
1970s, behavioral safety has undergone a series of
evolutionary changes. The first approach, popular in
the early 1970s to mid-1980s, was largely a supervi-
sory, top-down-driven process. Based on operant
theory (Skinner, 1953), supervisors observed worker
behavior, gave feedback and provided some form of
positive or negative reinforcement. Behavior change
did not last once reinforcers were removed. Simple
and cheap to implement, this approach attracted
legitimate criticism (Howe, 1998) that has since been
hard to dispel.
Perhaps as a reaction to those criticisms, employ-
ee-led processes emerged during the early-1980s. In
these interventions, which are still common, em-
ployees develop the overall process, conduct peer-
to-peer or workgroup-based observations and
provide feedback. However, the downside was (and
is) the exclusion of management, leading to the com-
mon perception that behavioral safety processes
focus solely on employee behavior (Hopkins, 2006).
This led, in the 1990s, to the cultural approach
based on the concept of a managerial and employee
partnership. Employees monitor the behavior of all
members of a workgroup or work area, and man-
agers regularly monitor their own safety-related
leadership behaviors (e.g., whether they reviewed
and closed out corrective actions). Everyone in-
volved receives regular feedback, with some also
receiving tangible reinforcers or incentives (Chand-
036_045_Cooper_0209:Layout 1 1/13/2009 10:19 AM Page 37

ler & Huntebrinker, 2003). Surveys of behavioral To date, the impact of observation focus on injury
safety users show that all three approaches are reduction and behavior change has not been assessed.
widely used around the world (Cooper, 2008). Many processes adopt a one-on-one, peer-to-peer
observation approach, where a trained observer
Key Components of Behavioral Safety approaches a colleague, asks permission to observe
Regardless of the approach adopted (all are based the person while working, then provides on-the-spot
on the same psychological principles), many key feedback once the observation is complete (Geller, Abstract: Behavioral
structural components can affect the success of a 1996; Krause, 1997; McSween, 2002). This approach safety interventions
behavioral safety implementation. Many primary requires as many observers as possible and people vary greatly across the
academic research articles have addressed these willing to be observed. It also requires much time and facilities that have used
(DePasquale & Geller, 1999; Sulzer-Azaroff & Austin, effort to recruit observers and to sustain their motiva- them. This review of
2000) while others are based on received wisdom tion over the longer term (Whitney, 2006). published literature
(Hurst & Payla, 2003). Each has tried to address the Other processes adopt a workgroup observation examines the potential
most efficient way to design the process to produce approach, where one or more trained observers impact of process
positive results in a cost-effective manner. embedded within each workgroup monitor the designs and their com-
The components of an ideal process are: behavior of their colleagues during a single observa- ponents to determine
•Identify unsafe behaviors (obtained from injury tion (Cooper, 1998). Typically, permission to observe which are most effec-
and near-hit incident records). is not required. Feedback is provided immediately if tive. Injury reduction
•Develop appropriate observation checklists the observer is comfortable doing so. In addition, and behavior change
(which feature behaviors implicated in injuries). analyses of the workgroup observations for the are used as key effec-
•Educate everyone. Tell and sell to all, and train entire week are discussed at weekly workgroup tiveness indicators.
observers, facilitators and champions. meetings. To avoid observer fatigue, colleagues
•Assess ongoing safety behavior by conducting rotate into that role every few months. This way,
behavioral observations. everyone eventually becomes an observer. At the
•Provide limitless feedback—verbal, graphical same time, updating the behavioral checklists
and written—on results. ensures a focus on relevant safety behaviors.
Some processes go further and include goal-set- Self-observation approaches are often used for
ting, training and/or incentives. Sulzer-Azaroff and lone workers (e.g., drivers) who provide self-feed-
Austin (2000) stated that the effectiveness of the var- back when completing the observation checklist
ious approaches is often dictated by the purpose of (Olsen & Austin, 2001). Compiled data are often ana-
implementation. They cite variations in observation lyzed for all lone workers to provide information
frequency and mechanisms of feedback, priorities, about training needs and other corrective actions.
support structures and roles of key personnel. A less-common approach is to focus on outcomes
Variations in work settings (e.g., static and dynamic) of behavior. This approach is akin to weekly man-
and observation focus (e.g., individuals, work- agement walk-arounds focused on unsafe condi-
groups, outcomes) are also factors. tions. Typically, these will have a particular theme,
Therefore, many process designs exist, some of such as housekeeping, PPE use or hazard identifica-
which may be more effective in particular circum- tion. Again, employees receive immediate verbal
stances than others. For example, a behavioral feedback during these observation tours.
process might be configured differently for static
(e.g., manufacturing plant) or dynamic (e.g., con- Feedback Mechanisms
struction) work settings that experience constant Academic behavioral safety reviews (Cameron &
change in people or the working environment. Duff, 2007; Grindle, Dickinson & Boettcher, 2000;
McAfee & Winn, 1989; Sulzer-Azaroff, Harris &
Observation Processes McCann, 1994) indicate the importance of feedback,
Observations are the foundation of a behavioral the purpose of which is to allow people to adjust
safety process; they provide opportunities for verbal their performance. Many processes use various com-
feedback and coaching at the point of contact. In prin- binations of available feedback mechanisms (e.g.,
ciple, observations are analyzed to provide objective verbal, graphical, written, tokens). Verbal feedback
decision-making data (e.g., provide tailgate topics, between the observed and observer at the point of
identify training needs). Two aspects of observation contact is probably the most heavily used approach
processes might affect outcomes: frequency and focus. (Coplen, Ranney & Zuschlag, 2007). Corresponding-
In terms of observation frequency, most processes ly, behavioral safety steering committee members
monitor the contact rate—the rate of contacts between spend much time and effort recording and evaluat-
observers and those observed. While some processes ing the quality of these feedback interactions, while
advocate daily contact (Cooper, 1998) because it pro- simultaneously trying to keep the observations
vides a more reliable picture of ongoing safety per- anonymous (no names recorded).
formance, others recommend two or three times a Many also display charts in the workplace to
week (Komaki, Barwick & Scott, 1978), once per week visually report trends in behavioral performance. In
or a few times a year. Anecdotal evidence (Geller, effect, they highlight how close to 100% safe the
1996; McSween, 2002) suggests the greater the contact recorded observations indicate people are working.
rate, the larger the impact on incident rates. In some instances, a target (assigned by a steering
www.asse.org FEBRUARY 2009 PROFESSIONAL SAFETY 37
036_045_Cooper_0209:Layout 1 1/13/2009 10:19 AM Page 38
committee or set jointly by the workforce) is indicat-
ed on the chart as a motivational element of the
process (Cameron & Duff, 2007).
Some processes also use written feedback based
on an analysis of the compiled observation data. In
some processes, workgroups receive the data at ded-
icated weekly briefings. In others, steering commit-
tees report the summary data to management each
month or give it to focus groups with the explicit
intention of eliminating barriers to performance.
Some processes also provide tokens or incentives as
a form of feedback to reinforce good performance.
Process Design Structure
Some processes make use of goal-setting, safety
training, competition, incentives or a combination of
these along with the observation and feedback com-
ponents (Cameron & Duff, 2007). Goal-setting is
motivational as it focuses people’s attention and
commits people to a particular course of action
(Cooper, 1993).
In essence, there are three types of goals: implicit,
assigned or participative. Feedback-only interven-
tions use implicit goals, where the underlying
assumption is to improve performance to 100% safe.
Table
Table 1 1
Studies Reviewed
38 PROFESSIONAL SAFETY FEBRUARY 2009 www.asse.org
Assigned goals are usually determined and set by
those in authority, such as managers and steering
committees, with no input from the workforce.
Participative goals are set jointly and are agreed
upon by all involved with reaching the goal.
Average levels of behavioral performance meas-
ured during a baseline period at the beginning of a
process are used as the starting point to help set tar-
gets. In some cases, targets are changed as soon as
the original target has been reached; in others, the
target is set for a specific period to allow certain safe-
ty behaviors to become habitual.
Although not seen as an integral part of behavioral
safety, safety training is usually present in some form.
The initial training focuses on the at-risk behaviors
identified in incident analysis. Thereafter, the training
focuses on specific issues (e.g., manual handling)
highlighted by analysis of observation data.
The use of competition between workgroups or
departments and incentives in particular can be con-
troversial because it can be viewed as paying for
good safety performance (Gibb & Foster, 1996).
Arguments for (Sims, 1999) and against (Krause,
1998) these methods have been published. Regard-
less, many behavioral safety processes (and safety in
036_045_Cooper_0209:Layout 1 1/13/2009 10:19 AM
general) often use competition to reinforce and/or
reward good performance.
Given the complexity and variety of approaches,
it makes sense to try to identify the optimum design
of a behavioral safety process. Academic studies
have established that behavioral safety works.
However, to date no review has examined the poten-
tial impact of process designs and their components
to determine which are most effective. The author’s
study attempted to do so using injury reduction and
behavior change as key effectiveness indicators.
Method
Literature Search
A wide-ranging literature search located 106 pro-
fessional and academic behavioral safety articles.
These were examined and included in this review
only if they 1) focused solely on occupational safety;
2) quantified behavioral change and incident reduc-
tions; 3) stated observation contact rates; and 4) were
written in English. Seventeen studies met these cri-
teria. Of these, five reported the results of two or
more separate studies within the article. In total, this
provided 24 useable data sets (Table 1).
Common study characteristics were identified
and coded. The fundamental components included
1) the focus of monitoring (individuals, workgroups,
outcomes); 2) observation frequency; and 3) feed-
back mechanisms used (posted, verbal, written,
briefings). Component structures included safety
training, goal-setting and incentives or competition.
Study outcomes were the degree of injury reduction
and behavioral improvement.
Data Transformations
Many of the studies reported success in different
ways. To ensure equitable comparisons, several data
transformations were required:
•Behavioral change. The degree of behavioral
improvement was obtained directly from the report-
ed statistics or by subtracting the reported baseline
score from the final intervention score when the spe-
cific degree of improvement was not reported.
•Incident reduction. A similar procedure ascer-
tained the degree of injury reduction. In one study
containing three data sets (Reber, Wallin & Chokar,
1990), baseline injury figures encompassed the pre-
ceding 3 years, rather than the corresponding period
in the previous 12 months. This could inflate or
deflate the claimed degree of injury reduction. To
ensure correct comparison with the other studies,
the reported injury rate baseline was divided by 36
months to obtain an average monthly injury rate.
The product was multiplied by 12 to obtain an esti-
mate of the prior annual injury rate.
•Incident rates. The calculation of reported
injury rates also differed across the studies. Some
were based on 100,000 or 200,000 hours worked and
some on 1 million hours. All injury rates were recal-
culated to reflect the rate for 200,000 hours worked.
This did not affect the magnitude of change within
any individual study, it merely facilitated like-for-
like comparisons across the studies.
Page 39
Analytic Method
Meta-analysis is a quantitative review
Given the
method commonly used to synthesize
data from multiple studies (Hunter &
complexity of
Schmidt, 1990) to provide an estimate of
the mean treatment effect. This requires
the process and
the use of a common metric and is usual-
ly obtained by converting the results of
variety of
each study into a standardized difference
score (d) or correlation coefficient (r).
approaches used
Accumulated across all studies of interest
and corrected for error, the outcome pro-
in workplaces
vides statistically valid estimates of the
mean treatment effect.
to implement
Cohen’s standardized difference score behavioral safety,
(d) is the statistic used in this review. In
essence, this is obtained by dividing any it makes sense
gain in scores (Xpre - Xpost) resulting from
an intervention, by the spread of scores to try to identify
(SDpooled). In other words, Cohen’s d is
the difference in mean scores divided by the optimum design
the pooled standard deviation of the rel-
evant before and after scores. As such, d of a behavioral
represents standard deviation units.
When greater than 0.8 the effect is con- safety process.
sidered large (Cohen, 1992), but not
necessarily statistically significant. Con-
fidence intervals are used to determine statistical sig-
nificance. When the range between the upper and
lower limits exclude zero, the results are considered
statistically significant. Conversely, confidence inter-
vals that include zero (i.e., no effect) are not consid-
ered statistically significant.
Analysis of Study Characteristics
The injury and behavior change data for each
study were transformed into common treatment
effect sizes (Cohen’s d) using dedicated meta-analyt-
ic software (Borenstein & Rothstein, 1999). For each
component or process design, the studies were sep-
arated into groups containing the same characteris-
tics and analyzed to establish the degree of behavior
change and injury reduction.
For each analysis, the accumulated and averaged
treatment effect sizes of the various behavioral safe-
ty components form the basis for the reported
results. The meta-analytic software also produces
confidence intervals and significance tests. Those
failing to achieve significance are highlighted by ns.
All other results achieved statistical significance at
either 0.01 (99%) or 0.05 (95%) levels (Table 2, p. 40).
Results
Overall Results
The 17 studies were conducted in the U.S. (n = 12)
and Europe (n = 5). The total number of people
involved in all the studies was 25,852. The average
length of the studies was 60 weeks (range = +/- 30).
For all studies combined, Cohen’s d was 3.18 for
behavior change and 5.21 for injury rate reduction.
These large effect sizes show that behavioral safety
processes positively affect behavior and reduce inci-
dent rates.
www.asse.org FEBRUARY 2009 PROFESSIONAL SAFETY 39
036_045_Cooper_0209:Layout 1 1/13/2009 10:19 AM Page 40

Table
Table 2 2
Degree of Behavior Change & Injury Reduction

Work Settings and weekly. Six studies reported daily observations,

The studies included 13 static (e.g., automotive, with data for 10 interventions (Table 2). Eight stud-
manufacturing, metal refining, postal services, vehi- ies observed two or three times per week, with 10
cle maintenance) and 11 dynamic (e.g., construction, data sets. Three studies observed weekly with four
offshore oil and gas, police services, shipbuilding sets of data.
and transit operations) work settings. As shown in The results (Figure 2) reveal that daily contact led
Figure 1, processes implemented in static settings to greater injury reduction (d = 7.0) than intermittent
reduced the average incident rate by 6.6 SD units, contact (d = 5.8), but the latter exerted a slightly larg-
compared to 3.6 SD units in dynamic settings. er impact on behavioral change. For weekly obser-
Behavior changes were also twice as large in static vations, the impact on injury reduction and behavior
(d = 4.2) settings compared to dynamic settings change was not statistically significant (d = 0.7), sig-
(d = 1.9). Thus, in these reviewed studies, behavioral naling minimal contact rates will not always change
safety appears to work better in static settings, with behavior or reduce injuries.
stable workforces and/or in stable environments.
Observation Focus
Observation Frequency Eight studies encompassing 11 data sets contained
Contact rate was grouped into three observation workgroup observations. Six studies with eight data
conditions: daily, intermittent (2 or 3 days per week) sets used individual one-on-one observations, while
40 PROFESSIONAL SAFETY FEBRUARY 2009 www.asse.org
036_045_Cooper_0209:Layout 1 1/13/2009 10:19 AM Page 41

three studies with five data sets observed the out- change (d = 6.7) was much greater than other combi-
comes of behavior (e.g., housekeeping). nations. Perhaps this reflects the increased opportuni-
Figure 3 (p. 42) shows that injury reduction was ties for discussion between all involved to address
greater for workgroup based observations
(d = 7.1) than a focus on outcomes (d = 5.5)
or individuals (d = 2.5). Similarly, greater
Figure 1
Figure 1
behavioral change was reported for work- Changes in Behavior & Injury
groups (d = 3.9) followed by individual one-
on-one observations (d = 2.8) and Reduction by Work Setting
outcome-based approaches (d = 2.2). As
outcome-based approaches explicitly focus
on unsafe conditions where behavior
change is indirect, greater behavior change
for workgroup and individual approaches
is not surprising. As a whole, these results
show that workgroup observations are
more effective than one-on-one, peer-to-
peer observations.
Because many case studies suggest that
one-on-one observations are effective, the
studies were further analyzed by observa-
tion focus and work setting to assess pos-
sible situational impacts. Incident rate
reductions (Figure 4, p. 42) show a clear
effect of situational influences on observa-
tional approaches.
Workgroup observations were far more
effective in static (d = 10.87) than dynamic
(d = 0.47, ns) settings. Explanations may
reside in ideal culture change conditions
arising from long-term stable workgroups
in static settings, compared to transient
workforces in dynamic settings.
One-on-one approaches reduce injuries
significantly more in dynamic (d = 5.02)
Figure 2
Figure 2
than static (d = 1.65) settings. Little differ-
ences were found for behavior change,
Changes in Behavior & Injury
suggesting that one-on-one observation
approaches lend themselves to dynamic
Reduction by Contact Rate
settings. Both one-on-one and outcome-
based observations (d = 5.47) were better
than workgroup (d = 0.47) observations in
dynamic settings.
Feedback Mechanisms
All studies included verbal feedback,
with 14 (providing 19 data sets) using
posted feedback charts, six studies (with
seven data sets) providing a written analy-
sis of observation results and five studies
(with six data sets) using weekly briefings
with employees. Separated into feedback
method groups, three groupings contained
only one study, the results of which are
viewed solely as indicative, not definitive.
Figure 5 (p. 43) shows the impact the
various feedback mechanisms exerted on
incident reduction and behavior change. Of
the groups containing more than one study,
a combination of posted, verbal and written
feedback presented at weekly briefings was
the most effective method. Effect sizes for
injury reduction (d = 10.5) and behavior
www.asse.org FEBRUARY 2009 PROFESSIONAL SAFETY 41
036_045_Cooper_0209:Layout 1 1/13/2009 10:19 AM Page 42

Figure
Figure 3 3
Changes in Behavior & Injury
Reduction by Observation Focus
those using one to two or three to four
mechanisms. Analyses revealed that more
feedback mechanisms led to greater injury
reduction (d = 8.9 vs. 3.7) and larger
behavioral improvement (d = 5.51 vs.
2.21). These results (Figure 6) hold for
injury reduction in both static and dynam-
ic settings. This is also mostly true for
behavior change, except one to two feed-
back mechanisms (d = 2.66) was slightly
better than three to four mechanisms (d =
2.4) in dynamic settings.
Design Structure
In terms of goal-setting, eight of the
studies made use of implicit goals, with 10
data sets. Six studies used assigned goals
providing 10 data sets, while three studies
used participative goals, providing four
data sets. Six studies included training in
their interventions, providing nine data
sets, with only one making use of compe-
tition and incentives.
The analysis of the various design
Figure
Figure 4 4
structures produced interesting results
(Figure 7, p. 44). First, the effects of design
structure are on injury reduction rather
Changes in Behavior & Injury than behavior change. Ignoring the two
designs with only a single study, goal-set-
Reduction by Observation Focus ting and feedback (d = 8.7) reduced
injuries more than training, goal-setting
& Work Setting and feedback (d = 5.8) and feedback only
(d = 3.5) designs.
At first glance, this suggests that the
motivational effects of explicit safety im-
provement goals are neutralized by safety
training. However, six of the eight studies
in the training, goal-setting and feedback
condition used assigned goals, which
proved to be the least effective for injury
reduction (Figure 8, p. 44). Both participa-
tive (d = 9.92) and implicit (d = 6.6) goals
were more effective at reducing injuries
than assigned goals (d = 2.3), which pro-
duced nonsignificant reductions. Behavior-
al change was also greater when
participative goals were used (d = 5.6) com-
pared to implicit (d = 3.2) and assigned
goals (d = 2.16).

reasons for particular unsafe behaviors. Verbal feed-
back was effective for injury reduction (d = 4.8) but did
not lead to statistically significant behavior changes
(d = 1.9). Conversely, posted feedback produced
behavioral changes (d = 2.2) but nonsignificant injury
reductions (d = 2.9).
Due to the limited number of studies in three
feedback groupings, studies were subdivided into
42 PROFESSIONAL SAFETY FEBRUARY 2009 www.asse.org
Discussion
This behavioral safety review revealed
some interesting findings, some of which
support popular opinion, others which con-
tradict it. Based on the studies reviewed,
behavioral safety does improve safety behavior and
reduce injuries. However, the studies also show that
some process designs are more effective than others.
Injury reductions and behavior changes in static
settings were twice those of dynamic settings.
Within static settings, a workgroup approach re-
duced injuries about six times more than a one-on-
one observation focus on individuals. Since about
036_045_Cooper_0209:Layout 1 1/13/2009 10:19 AM Page 43

51% of behavioral safety processes observe individ- that safety is important and taken seriously by the
uals (Cooper, 2008), this is an important finding. company.
Workgroup observations are probably more effec- Although feedback alone reduces injuries and
tive because of peer pressure resulting from social changes behavior, processes incorporating explicit
dynamics within a workgroup (Mullen &
Copper, 1994). This helps people perform
safely even when not being observed.
Figure 5
Figure 5
Similarly, group feedback is more effective Changes in Behavior & Injury
than one-on-one feedback to individuals
(Alvero, Bucklin & Austin, 2001). Togeth- Reduction by Feedback Mechanism
er, these factors help to positively reset
group norms, which in turn positively
influences the prevailing safety culture
(Cooper, 2000).
That behavioral safety works in dynam-
ic settings is good news, as implementa-
tions can be very difficult in settings where
people, the working environment or both,
change regularly. In these environments,
observing outcomes (e.g., unsafe condi-
tions) reduced injuries slightly more than
monitoring individuals, both of which sig-
nificantly reduced injuries more than work-
group-focused approaches. This shows
workgroup observations are more effective
in static settings, while one-on-one observa-
tions are more suited to dynamic settings.
Thus, a focus on unsafe conditions in con-
junction with one-on-one observations
have an important role to play in settings
where the people and/or the environment
are subject to continual change.
Supporting the anecdotal evidence in
the field, daily or intermittent contact Figure 6
Figure 6
rates produced larger effects on both
behavior and injury reduction than week- Changes in Behavior & Injury
ly contact rates, which produced non-
significant changes. Approximately 47% Reduction by Number of
of behavioral safety processes use a week-
ly contact rate (Cooper, 2008), which sug- Feedback Mechanisms
gests that increased contact rates would
benefit these users. Greater contact rates
for line management’s weekly inspections
would also help to maintain a consistent
focus on safety, which, in turn, may lead to
further injury reductions.
Processes that use three to four feed-
back mechanisms had more than twice the
impact on injuries and behavior than
those with one to two mechanisms, in
both static and dynamic settings. The
overall message, therefore, is to “use it is
to make it useful.” Behavioral safety
processes should use as many feedback
mechanisms as possible to facilitate any
necessary adjustments in performance
(Algera, 1990). Different people pay atten-
tion to different forms of feedback and by
using several mechanisms people may
perceive they are part of the safety
improvement process, not the problem.
More feedback also transmits the message
www.asse.org FEBRUARY 2009 PROFESSIONAL SAFETY 43
036_045_Cooper_0209:Layout 1 1/13/2009 10:20 AM Page 44

goals and/or training about the behavioral targets
were between one and two times more effective,
suggesting these designs helped to maximize the
benefits of the process. However, participative goals
Figure
Figure 7 7
Changes in Behavior & Injury
Reduction by Design Structure
were superior to assigned goals, which produced
nonsignificant injury reductions. The likely reason
participative goals are more effective is that employ-
ees jointly work out the best ways to achieve their
goals (Locke & Latham, 2002), which
induces commitment to goal achievement
(Cooper, 1992; Ludwig & Geller, 1997).
Limitations
One limitation of this review was the
relatively small number of published
research studies that fit the established cri-
teria for inclusion. Excluding some pub-
lished articles simply because they did not
report the appropriate change statistics or
failed to specify the observation contact
rate meant some groupings of design
characteristics were limited to only one or
two studies.
For example, only one study was avail-
able in the training and feedback group,
so it was not possible to draw meaningful
conclusions about the effectiveness of this
design. This also restricted the analyses of
other possible design combinations. Many
companies and providers have imple-
mented behavioral safety using various
designs. These could and should be ana-
lyzed and published as case studies so
that behavioral safety can evolve to the
point where it always works, in all types
of work settings.

Figure
Figure 8 8
Conclusion
The results of this review show that the
Changes in Behavior & Injury design of a behavioral safety process is as
important as the psychology of behavioral
Reduction by Goal Type safety (all studies reviewed are based on
the same psychological research). Simply
measuring behavior and providing conse-
quences is not enough to sustain incident
reduction. Designs that incorporate daily
observations, focus on workgroups in
static settings and use participative goals
with multiple feedback mechanisms will
reduce injuries more than others. All
behavioral safety processes require strong
management support to help deliver the
intended benefits and sustain them over
the longer term (Cameron & Duff, 2007;
Cooper, 2006a; 2006b). Without it, even the
best designed process can fail. 䡲

References
Algera, J.A. (1990). Feedback systems in organi-
zations. In C.L. Cooper and I.T. Robertson (Eds.),
International review of industrial and organizational psy-
chology (pp. 169-193). New York: John Wiley & Sons.
Alvero, A.M., Bucklin, B.R. & Austin, J. (2001).
An objective review of the effectiveness and essential
characteristics of performance feedback in organiza-
tional settings (1985-1998). Journal of Organizational
Behavior Management, 21, 3-29.
44 PROFESSIONAL SAFETY FEBRUARY 2009 www.asse.org
036_045_Cooper_0209:Layout 1 1/13/2009 10:20 AM
Borenstein, M. & Rothstein, H. (1999). Comprehensive meta-
analysis: A computer program for research synthesis. Englewood, NJ:
BioStat.
Cameron, I. & Duff, A.R. (2007). A critical review of safety
initiatives using goal-setting and feedback. Construction Manage-
ment and Economics, 25, 495-508.
Chandler, B. & Huntebrinker, T.A. (2003, June). Multisite suc-
cess with systematic BBS: A case study. Professional Safety, 48(6),
35-42.
Cohen, J. (1992). Statistical power analysis. Current Directions
in Psychological Science, 1, 98.
Cooper, M.D. (1992). An examination of assigned and partici-
pative goal-setting in relation to the improvement of safety on
construction sites. Unpublished doctoral dissertation, University
of Manchester Institute of Science & Technology, U.K.
Cooper, M.D. (1993, Nov.). Goal-setting for safety. Safety &
Health Practitioner, 32-38.
Cooper, M.D. (1998). Improving safety culture: A practical guide.
Chichester, U.K.: John Wiley & Sons.
Cooper, M.D. (2000). Towards a model of safety culture. Safety
Science, 32, 111-136.
Cooper, M.D. (2002, June). Safety culture: A model for under-
standing and quantifying a difficult concept. Professional Safety,
47(6), 30-36.
Cooper, M.D. (2003, April). Behavior-based safety still a viable
strategy. Safety+Health, 46-48.
Cooper, M.D. (2006a). Exploratory analyses of the effects of
managerial support and feedback consequences on behavioral
safety maintenance. Journal of Organizational Behavior Management,
26, 1-41.
Cooper, M.D. (2006b). The impact of management’s commit-
ment on employee behavior: A field study. Proceedings of the 7th
Professional Development Conference & Exhibition, Kingdom of
Bahrain, 40-47.
Cooper, M.D. (2008). User survey report. Franklin, IN: Behav-
ioral-Safety.com. Retrieved March 13, 2008, from http://www
.behavioral-safety.com/survey-results/finalsurveybscom1.html.
Cooper, M.D., Phillips, R.A., Sutherland, V.J., et al. (1994).
Reducing accidents with goal-setting and feedback: A field study.
Journal of Occupational and Organizational Psychology, 67, 219-240.
Cooper, S.E. & Newbold, R.C. (1994). Combining external
and internal behavioral system consultation to enhance plant safe-
ty. Consulting Psychology Journal, 46, 32-41.
Coplen, M., Ranney, J. & Zuschlag, M. (2007, Feb.). Behavior-
based safety at Amtrak-Chicago associated with reduced injuries
and costs. U.S. Department of Transportation Research Results, 1-4.
DePasquale, J.P. & Geller, E.S. (1999). Critical success factors
for behavior-based safety: A study of 20 industry-wide applica-
tions. Journal of Safety Research, 30, 237-249.
Fellner, D.J. & Sulzer-Azaroff, B. (1984). Increasing industrial
safety practices and conditions through posted feedback. Journal of
Safety Research, 15, 17-21.
Geller, E.S. (1996). The psychology of safety: How to improve
behaviors and attitudes on the job. Boca Raton, FL: CRC Press.
Gibb, A.G.F. & Foster, M. (1996). Safety motivation: evalua-
tion of incentive schemes. In L.M. Alves Dias and R.J. Coble (Eds.)
Implementation of safety and health on construction sites (pp. 405-415).
Rotterdam, Netherlands: Balkema.
Grindle, A.C., Dickinson, A.M. & Boettcher, W. (2000). Behav-
ioral safety research in manufacturing settings: A review of the liter-
ature. Journal of Organizational Behavior Management, 20, 29-68.
Haynes, R.S., Pine, R.C. & Fitch, H.G. (1982). Reducing acci-
dent rates with organizational behavior modification. Academy of
Management Journal, 25, 407-416.
Hodson, S. & Gordon, T. (2000, May). Tenneco’s drive to
become injury-free. Occupational Hazards, 5, 85.
Hopkins, A. (2006). What are we to make of safety behavior
programs? Safety Science, 44, 583-597.
Howe, J. (1998). A union critique of behavior safety. Paper pre-
sented at the ASSE Behavioral Safety Symposium, Orlando, FL.
Hunter, J.E. & Schmidt, F.L. (1990). Methods of meta-analysis:
Correcting error and bias in research findings. London: Sage.
Hurst, P.W. & Payla, W.L. (2003, Sept.). Selecting an effective
BBS process. Professional Safety, 48(9), 39-41.
Komaki, J., Barwick, K. & Scott, L. (1978). A behavioral
approach to occupational safety: Pinpointing and reinforcing safe-
Page 45
ty performance in a food manufacturing plant. Journal of Applied
Psychology, 63, 434-445.
Komaki, J.L., Heinzmann, A.T. & Lawson, L. (1980). Effect of
training and feedback: Component analysis of a behavioral safety
program. Journal of Applied Psychology, 65, 261-270.
Krause, T.R. (1997). The behavior-based safety process (2nd ed.).
New York: Van Nostrand Reinhold.
Krause, T.R. (1998, May). Safety incentives from a behavioral
perspective. Professional Safety, 43(5), 24-28.
Krause, T.R. (2001, May). Moving to the second generation in
behavior-based safety. Professional Safety, 46(5), 20-25.
Larson, L.D., Schnelle, J.F., Kirchner, R., et al. (1980). Reduc-
tion of police vehicle accidents through mechanically aided super-
vision. Journal of Applied Behavior Analysis, 13, 571-581.
Locke, E.A. & Latham, G.P. (2002). Building a practically use-
ful theory of goal-setting and task motivation: A 35-year odyssey.
American Psychologist, 57(9), 705-717.
Ludwig, T.D. & Geller, E.S. (1997). Assigned versus participa-
tive goal-setting and response generalization: Managing injury
control among professional pizza deliverers. Journal of Applied
Psychology, 82, 253-261.
Lyon, B.K. (2001, Oct.). Behavioral sampling of theme park
operators. Professional Safety, 46(10), 35-42.
McAfee, R.B. & Winn, A.R. (1989). The use of incentives/
feedback to enhance workplace safety: A critique of the literature.
Journal of Safety Research, 20, 7-19.
McSween, T.E. (2002). The values-based safety process: Improving
your safety culture with a behavioral approach (2nd ed.). New York:
John Wiley & Sons.
Mattila, M. & Hyödynmaa, M. (1988). Promoting job safety in
building: An experiment on the behavior analysis approach.
Journal of Occupational Injury, 9, 255-267.
Mullen, B. & Copper, C. (1994). The relation between group
cohesiveness and performance: An integration. Psychological
Bulletin, 115, 2210-2227.
Naso, M. (2002, Dec.). Change in behavior: The decline of
behavior-based safety gives way to safety management initatives.
Safety+Health, 33-39.
Näsänen, M. & Saari, J. (1987). Effects of positive feedback on
housekeeping and accidents at a shipyard. Journal of Occupational
Accidents, 8, 237-250.
Olsen, R. & Austin, J. (2001, Nov.). ABCs for lone workers: A
behavior-based study of bus drivers. Professional Safety, 46(11), 20-25.
Reber, R.A. & Wallin, J.A. (1984). Reducing industrial
injuries: A behavioral experiment. Employee Relations, 23, 119-124.
Reber, R.A. & Wallin, J.A. (1994). Utilizing performance man-
agement to improve offshore oilfield diving safety. International
Journal of Organizational Analysis, 1, 88-98.
Reber, R.A., Wallin, J.A. & Chokar, J. (1990). Improving safe-
ty performance with goal-setting and feedback. Human Perform-
ance, 3, 51-61.
Saari, J. (1994). When does behavior modification prevent
injuries? Leadership and Organization Development Journal, 15, 11-15.
Sims, W. Jr. (1999, April). Successful safety incentive pro-
grams. Professional Safety, 44(4), 37-39.
Skinner, B.F. (1953). Science and human behavior. New York:
Free Press.
Sulzer-Azaroff, B. & Austin, J. (2000, July). Does BBS work?
Behavior-based safety and injury reduction: A survey of the evi-
dence. Professional Safety, 45(7), 19-24.
Sulzer-Azaroff, B., Harris, T.C. & McCann, K. (1994). Beyond
training: Organizational performance management techniques.
Occupational Medicine: State of the Art Reviews, 9, 321-339.
Veazie, R. (1999, April). Foundation principles: Keys to suc-
cess of a behavior-based safety initiative. Professional Safety, 44(4),
24-26.
Whitney, P. (2006). How behavior-based safety can work in
different national cultures. Paper presented at the Behavioral
Safety Now Conference, Kansas City, MO.
Winn, G.L., Fredrick, L.J. & Church, G.M. (1999, Dec.).
Unions and behavior-based safety: Always the odd couple?
Professional Safety, 44(12), 32-34.
Zhu, Z., Wallin, J.A. & Reber, R.A. (2000). Safety improve-
ments: An application of behavior modification techniques.
Journal of Applied Management Studies, 9, 135-140.
www.asse.org FEBRUARY 2009 PROFESSIONAL SAFETY 45
 What’s wrong with behavior-based safety?
By Thomas A. Smith, CHCM, CPSM

Published in Professional Safety, September 1999

Introduction

It is not at all unusual these days to pick up a copy of any current safety magazine and find articles and
advertisements celebrating the principles, methods and application of behavior-based safety. There are obviously a
large number of advocates of behavior-based safety. Indeed, the only thing that seems to be larger than the number
of BBS advocates is the number of articles and advertisements in print.

There comes a time when an idea is so prevalent it is accepted and applied without question. When this happens we
are so conditioned to the correctness of it we fail to examine its basic premise. I believe we are at that point when it
comes to behavior-based safety. At the risk of invoking the wrath of those safety professionals who advocate its use I
am going to suggest it’s time to re-examine the behavior-based safety (BBS) model. I will also propose that in the
present and future workplace BBS is not just partially but a totally wrong solution to preventing accidents at work.

The Behavior Based Safety Model

First let’s examine the basic theory advocated by the behavior-based proponents. The goal of behavior-based safety
is to change the behavior of employees from “at risk” behaviors to “safe” behaviors. They use what is called the ABC
model to change human behavior. Most of today’s behavioral safety efforts are based on this theory that says all
behaviors are a result of antecedents and consequences. This theory, promoted by B.F. Skinner, is that Antecedents
serve as triggers to observable Behaviors. Consequences either enforce or discourage repetition of the behaviors.

The basic behavior-based process consists of identifying observable safe behaviors upstream in the process. Then
you need to identify the antecedents (activators) that encourage these safe behaviors and encourage them. You
should also identify those antecedents that discourage safe behaviors and remove them. Behaviorist theory says
those consequences (reinforcement) that are positive, immediate and certain (rewards) will keep employees working
safely. Negative consequences which are immediate and certain (punishment) discourage unsafe behaviors. The
goal for management is to set up a system to control the antecedents and consequences so workers will increase
their safe behaviors. The theory being that by setting up a system of well -planned antecedents and consequences
you can control the unsafe behaviors of employees thus accidents and injuries will be reduced. [1]

Furthermore the behaviorists believe that consequences are the driving force to changing people’s behavior. The
tools of positive and negative reinforcement are what is needed to make people behave in the prescribed manner.
Consequences are those events that occur as a result of behavior. Positive reinforcement rewards a person for
behaving in a certain way. Negative reinforcement provides unwanted or unpleasant consequences. The theory is
punishment decreases the probability a behavior will be repeated. Some behaviorist believe that negative
reinforcement prompts only a minimal level of compliance. However, positive reinforcement encourages employees
to exceed the minimum.

What’s in a theory?

At this point I think we should review the concept of theories. To start with all theories are correct. Some are just more
useful than others. The fact is you can never really prove a theory. You can test it to see if it is valid. The more time s
the theory works the more confidence you have in it. However, no number of examples proves the theory. The next
test of the theory could provide an exception. One exception to your theory and you must either abandon or adjust it
so it you can explain and account for the exception.

1
The psychological theory of behaviorism

The behavior-based safety depends on the psychological theory of behaviorism. This theory attempted to explain all
human behavior in terms of being driven by external consequences.

B. F. Skinner was one of the most famous psychologists of all times and he contributed the concept of operant
conditioning (shaping the behavior of an organism using positi ve and negative reinforcement) as an explanation of
why people behave the way they do. His concept of re-enforcers as a learning technique developed into
“programmed learning.”

What does the research on behaviorism reveal?

Many of the articles that defend and advocate behavior-based safety point out the enormous amount of research
substantiating it. They do not mention the large amount of scientific research that refutes it.

A major problem with behavior-based safety is the fact that when behaviorism was held up to the scrutiny of the
scientific method it failed. “As behavior research accumulated it was apparent to even the most ardent followers of
the theory that the animals being studied frequently acted in ways the theory couldn’t explain. [2] One major problem
was different animals failed to conform to supposedly universa l principles of conditioning. Skinner said “Pigeon, rat,
monkey, which is which? It doesn’t matter,” but it did matter. The researchers found they could easily train a pigeon to
peck at a disc for food but it was next to impossible to get it to flap its wi ngs for the same reward. They could also get
rats to press a bar for food but to get a cat to do it was quite a different matter. There were just too many comparable
findings in their research that forced behaviorists to admit that each species had its ow n built-in processes that made
it learn some things easier than others and some things not at all. Their own research showed the laws of learning
could not be applied universally.

Behaviorism denies internal processing that goes on in human beings. The behaviorist research could not explain for
example the behavior in rats when at the beginning of an extinction trial an animal would respond to the stimulus with
more vigor than it had during a long series of reinforcements. If a rat that had been getting a food pellet each time it
pressed a bar was deprived of the pellet, it would press the bar with more force repetitively. The strict behaviorist
theory predicted that the absence of the reward should have weakened the response, not strengthened it.

Behaviorism failed other tests of scientific method. For instance it could not provide an adequate explanation of
memory. Reward and repetition provided only a partial explanation of rat behavior and an even less satisfactory one
for human behavior. The internal workings of the mind were more or less ignored by behaviorist and explained away
as insignificant. Their use of the stimulus-response bond for example did not account for memory and how it works.
Even the behaviorist realized that memory was more than a chain of mathematical terms. Memory has different forms
such as short term and long term that behaviorism could not address.

There were many other things that psychologists needed to explore such as motivation, perception, creativity,
problem-solving, experience and interpersonal relations. Eventually, new data was gathered on these subjects and
raised questions that behaviorism couldn’t explain. This brought about a paradigm shift and led the way to a new
theory of psychology - known as “cognitive science”, in the 1960's. [3]

The field of psychology may have advanced beyond behaviorism but in the field of management it is quite a different
story. Behaviorism is still applied with a vengeance by managers. The fact is it is very useful in a command and
control management system. We have applied it with such force and magnitude we now accept its premise without
question. Behavior-based safety is a perfect example of this blanket application of behaviorism to the workers.

From the start behaviorism tried to make use of positive reinforcement as the tool to shape workers behavior.
Obviously it would not have been a very good idea to stress negative reinforcement as the tool. (No doubt
behaviorism would not have enjoyed the acceptance it has if negative re inforcement was its method of choice.)

However, if you examine it carefully you soon realize that positive and negative reinforcement are different sides of
the same coin. And that coin doesn’t buy much that’s worthwhile for motivating people. Both are really saying to
employees do this and you’ll get that. In the case of BBS the consequence is positive reinforcement handed out by
managers or peers. In actuality the purpose of the reward or punishment is to control or manipulate behavior of the

2
employee. All this takes place while management ignores or doesn’t comprehend the real reason safety problems
exist.

Research has shown that positive reinforcement does very little to alter the attitudes that underlie behaviors. You may
get a short term change but over time employees come to realize they are being manipulated while the safety system
stays the same. Using positive reinforcement, no matter what the intentions are by the person administering it, can
and does cause resentment between management and the workers. A bad situation is only made worse when
workers realize management is ignoring real safety issues while trying to appease employees by getting them to
change their behavior using positive reinforcement - carrots or bribes.

Let’s look at just a few of the problems with the positive reinforcement approach. There is a basic assumption that
positive reinforcement is an enjoyable and greatly appreciated consequence handed out by management. It is
believed passing out positive reinforcement will change the at risk behavior and replace with one that is safe
therefore no possible harm could result from its use. Hence we should use it to motivate workers to do a good job or
to work safely. This sounds good but it has some real drawbacks.

Research has shown using rewards on people to get them to do what you want them to is not such a clear cut
solution to the question of “How do you change people’s behavior?” As early as the 1960's research projects were
revealing that rewards (positive reinforcement) were not all that effective when used to try and improve skills or work
performance. Some examples were:

1. A researcher asked 128 undergraduates to solve a problem in a lab setting. They were given matches,
thumbtacks, and the boxes they came in and asked to mount a candle on a wall using only these materials.
Some students were offered from $5 to $20 if they succeeded; others weren’t promised anything. Those
who were working for the financial incentive took 50% longer to solve the problem.
2. In another project a researcher asked undergraduates to “select a pattern on each page that was least like
the two other patterns on that page. “The students “who were not offered money performed significantly
better than those who were paid.”
Obviously, these two simple examples aren’t enough to negate the system of behavior-based safety. But they expose
a major flaw in the theory that people in general will do what you want if you use the tool of positive reinforcement. In
fact it can have just the opposite effect.

I invite all proponents of behavior-based safety to do their own research on the psychology of behaviorism as they
staunchly advocate in their writings. You will find an enormous amount of research refutes the behaviorist ABC
theory. Their first step should be to read the research on the other side of the coin such as Alfie Kohn’s thoroughly
researched book, Punished by Rewards. [4]

The basic problem with BBS approach

The cornerstone of behavior-based safety is the principle that the majority of work related accidents are caused by
the unsafe actions of the workers. The traditional safety management theory (Developed by Heinrich with no real
scientific proof) is you should focus on unsafe actions since they are believed to be the majority (85 -95%) of the
reason accidents occur. If you accept this premise then it is just a small step to the assumption to improve safety you
must concentrate on changing the behavior of the worker.

The fact is accidents are like most bacteria. They are all around u s, and most aren’t harmful. You can be involved in
many kinds of accidents and there is no real harm. For example you pick up your clothes from the dry cleaners and
find the button you requested to be replaced is still missing. This is an accident but ther e is really no harm done. You
mention it to the manager and he blames it on the lazy workers who did not follow instructions. He says he will follow -
up with the worker to make certain it doesn’t happen the next time. But is this the workers fault? Or are there other
things that could have affected why the task wasn’t performed. The basic assumption on the manager’s part is
someone messed up and making them pay is the proper corrective action.

A similar simplistic view was held by American managers about quality problems in the 1970's and 80's. Their lament
was if only employees would follow management’s specifications the products workers made would have no defects,
scrap, rework or production problems. They really believed all that was necessary to achie ve high quality was for the
workers to “do it right the first time.” This same logic is applied when using behavior-based safety. Since
accidents/incidents are caused by unsafe actions on the part of the worker the key to preventing accidents is to
change their “at risk behavior.”

3
Using this logic behavior-based safety misses the mark much the same as managers did for quality in the 1970's. A
quality management system requires action on the system, not the workers, by management, to solve quality
problems. Quality management knows the problems of output from the system are built into it. You cannot achieve
more quality then what the system was designed to deliver. This is also true for safety. Most quality and safety
problems are not created by the individual worker. If your goal is to reduce defects and improve quality blaming the
worker for problems built into the system will not get you any improvement. If you want to improve the outcomes of a
system you must focus on how you manage the system, Enormous gains have been made to reduce the defects,
scrap and rework inherent in work systems using quality management theories.

What does cause accidents?

A basic shift is necessary in your understanding of why things go wrong that lead to accidents, scr ap, rework and
defects in a work system. New knowledge of why accidents/incidents happen must be applied. In a quality
management system managers understand accidents result from common causes and special causes. Common
causes are faults in the elements of a system. Special causes are causes that are unusual and not normally found in
the system. They can come and go without any warning. In a quality management system common causes account
for 85-95% of your problems and special causes are responsible for remaining 5-15%. This is the theory of what
causes accidents in a quality management system. It is much more useful as it will help everyone understand how to
fix the system instead of fixing the blame for accidents.

This requires a major shift in the accident causation model. The simple explanation that the unsafe actions and
unsafe conditions cause accidents will not suffice. The quality theory again can be used to provide a much better
understanding of why do accidents happen at work? Quality theory has taught us that defects, mistakes, scrap and
rework are caused by the way we manage the work system. Not the miscues of the workers. This same knowledge
must be applied in safety. Accidents are caused by how we manage the work system.

The Power Shift

Quality systems created a major shift in how companies manage their work processes. W. Edwards Deming taught
us all work is a system and a system has to be managed. (He would often say the shouting slogans or putting up
posters encouraging workers to “do it right the first time or “safety is your (the employee’s) job” were a waste of time.)
Companies have learned that in order to improve productivity and customer satisfaction they must improve quality -
period. Improving quality requires a new management theory. Until recently, the management system of choice was
to manage for quantity using the command and control model. In today’s world managing for quantity will not
succeed.

Quantity methods used from the 1950's - 70's relied on building the product, inspecting and selling it. Companies
were more concerned with production than quality. They could sell anything they made. Now they are faced with the
reality that competition may come from anywhere in the world. To compete their products and services must have
value and lower costs than their competitors. They must also anticipate the needs of their customers and fulfill these
needs even though the customer often doesn’t know what they want.

Quality management methods also require the Voice of the Custom ers be integrated into your system. If you don’t do
this your competitors will and you will lose customers. To understand and incorporate the Voice of the Customer in
your system you have to continually research their needs and reinvent products and servic es to meet these needs.
Customers are people who benefit from your product and service.

In the command and control method there is a division of labor. Managers do all the thinking and workers do the
work. They just punch in, catch hell and punch out. Good parts are separated from the bad by inspecting them at the
end of the production line. They are reworked or scraped. Emphasis is placed on inspections for achieve quality.
Catch the bad stuff and ship the good stuff. This is a very inefficient way to m ake things but it worked and it paid off
handsomely.

Quality management systems

In the 1970's a quality revolution was started in the US by the imports of products from Japan. This included but was
not limited to products previously manufactured in the US. (Television sets, radios, calculators and automobiles). This

4
revolution focused on how to manage a work system to improve quality instead of concentrating on mass quantity as
the goal.

The revolutionary idea in this management system is everyone wh o works in the system is needed to study the
problems in it and then make improvements. The goal of this system is to reduce waste, scrap, rework and accidents
so you can make good products. In this system workers are hired for their mental labor as well a s their manual labor.
It does not require command and control of the workers. The goal is to get control of the system. Control is not the
goal but the effect of good management.

In a quality management system the mental labor of all employees is needed to fix the system. Everyone must learn
the tools to identify system problems and take the proper action to fix the system instead of fixing the blame. (Fixing
the blame is very costly and destroys teamwork.) In this work setting the role of managers is changed. Workers “on
the line” now learn how to record data about their processes. They work in teams instead of for the boss. In some
companies there is no boss. The teams study their operations and look for better ways to do the job, eliminate waste,
scrap and rework before it happens - not after. In short, teams replace the old front line supervisor whose job it was to
make workers do their jobs correctly.

This management system is slowly evolving. The new model of how we will manage in the 21 st century is not
complete and will probably continue to evolve over the next decade. One thing for certain, it will not look anything like
the command and control model we have used and are presently dismantling.

The point is all workers roles change in a quality management system. In the command and control model the line
workers ideas, creativity and motivation were not required. Management took care of these things. The new
management model will require that everyone’s ideas, creativity, ingenuity and motivation is paramount to the
success of the organization.

Problems for BBS in the new quality management system

We seem to be moving away from the roots of behaviorism in the new quality management models being used by
companies. The following are just some of the problems the BBS model creates when applied in the new quality
management system:

 A major problem for behavior-based safety is it requires and relies on external motivators (antecedents and
consequences) to change people’s behavior. As explained earlier, these techniques did not work all that well
on rats and other animals let alone human beings. The reality is they are more likely to be de -motivators in
the new quality management system described above. Studies have shown that the large st factors to cause
dissatisfaction on the job are “company policy and administration.” [5] These are the very things BBS tries to
enforce when trying to change employee behaviors.

 Which leads to the next problem for behavior-based safety and that is extrinsic motivators destroy the
intrinsic motivation which is inherent in people to do good work and work safely. The antecedent-behavior-
consequences model assumes an extrinsic motivator is required to make the employee behave a certain
way. I have yet to meet an employee that sets out in the morning with a goal of trying to get injured on the
job. Yet, the basic premise of behavior-based safety is employees need to be motivated to exercise “safe
behaviors.” A system of re-enforcement and incentives is applied to accomplish this motivational goal. The
re-enforcement ranges from the negative (reprimands) to the positive (rewards -compliments, bribes) with the
goal of having the individual see the error of his ways and then changing the at risk behavior so accidents
will be avoided. (It also begs the question, if behaviorism works so well for safety, why don’t why isn't it used
in the quality arena? All one has to do is read any serious work on quality by any of the quality gurus and
you will realize that the workers weren’t responsible for quality problems - it was management.)

 As mentioned earlier there is a large body of research that contradicts the widely held common sense belief
that carrots and sticks will result in a permanent attitudinal change. Research has shown positive
reinforcement, no matter what form it is administered by BBS methods, will only provide a temporary
change. There is even a greater danger due to the fact that a short term success will result in a perception
the real safety problem has been solved. In the long run positive reinforcement may c ause more harm than
good. When workers eventually have to face the safety problems after their heroic efforts to avoid an injury
are defeated by the hazards built into the system an ever deeper chasm is created between them and
management.

5
 There is little thought given to another problem with behavior-based safety. That is the managers who use it
look upon employees only as a suppliers of manual labor in the production system. One behaviorist points
out companies are only renting the behavior of the employee. Therefore all that is required by management
is to control the behavior through the use of positive and negative re -enforcement. It is the major conceit of
command and control managers that they must motivate the workers since they are not already motivated to
do a good job or control their own unsafe actions. [6]

 The problem is in a quality management system, employees are both a customer and a supplier in the work
system. If safety management is going to achieve the awesome levels of improvement seen in quality it will
have to treat workers as customers first and suppliers second. If you look at safety as a system used to
manage the work system you must first examine who are the people that will benefit from the activities and
processes of safety? The obvious answer is the workers doing the work. They a re the first customers of
safety management and should be handled as you would any customer.

 To satisfy customers you must answer the following questions: Who are your customers? What do they
want and need? How do you know they are satisfied? The BBS model pays no attention to this exercise.
Indeed the workers are treated as though they don’t know anything about what they want or need when it
comes to safety. They should just do what they’re told and accept the safety situation as it is. The best
employees are ones who comply with the prescribed methods already designed by management. In the
world of quality the customer defines quality. The same is true for safety. Since employees know the job and
hazards associated with it better than anyone else they should be the ones telling safety management about
the problems with the methods. Now some will think I’m advocating the inmates should take control of the
asylum. This is not at all what I’m saying. Workers will gladly help management when it comes to making the
job safer. Given the opportunity they often provide solutions less costly then the injuries built into the system.
Ergonomic problems come to mind immediately but other less obvious safety problems are equally attacked
and results attained.

 Another problem with behavior-based safety is the assumption there is one “safe” way to do the job and it
can be applied to all workers. BBS assumes you can observe the employee’s behaviors, compare them to
the prescribed safe behavior and correct the employees at risk behavior using activators and consequences.
This means the “at risk behavior” must be changed. No thought or attempt is made to get any input into
whether or not worker, the main safety customer, defines the job as “safe.”

This approach is in direct conflict with the theory of a quality system in which the employees are given much
more respect. The individual employee knows more about the issues of quality than management because
they are closest to the job. That is why quality m anagement systems make use of operational definitions.
Workers control the process and must have a voice in what specifications mean. The only way a
specification can have real meaning is to develop an operational definition. When the BBS observation
teams pass judgment on the workers being observed they typically use safety specifications provided by
non-elective suppliers such as OSHA, consultants, or management created job safety analysis.. What does
“at risk behavior” mean. At risk because it doesn’t met OSHA regulations? or Corporate Job Safety
Analysis? or for the comfort of the worker doing the job? Each one of these definitions would be different
and produce different results. Without an operational definition of what is “safe.” a safety specificati on is
meaningless.

In the past if a quality problem existed management was prone to blame the workers for not following safety
instructions. This mistake is repeated in behavior-based safety. The fact is, the workers know more about
safety problems on the job than anyone else in the system. They do not create them. Most safety problems
are built into work systems and it will take action by management with the help of the workers to reduce or
eliminate them.

 The BBS approach sees cooperation as workers meeting the instructions for safety create by management
and regulators. Cooperation therefore equals compliance by the workers with prescribed management
activities. That is not the type of cooperation we are talking about. Cooperation desired in a quality
management system is to have management and hourly people work together to change the system so it
meets the needs of both external and internal customers. I think everyone has experienced the anxiety and
confusion of having someone watch you to see if you are doing your job properly. It’s not a very pleasant
way to learn and the long term consequences are usually negative. First the anxiety created from the

6
 process can be really upsetting to the person being monitored. Second, it perpetuates the myth that people
must be managed or they will mess up.

 And finally, the BBS process is what the Japanese call a form of “muda” - waste. Paying supervision and
peers to watch others work and then applying an outmoded theory to the outcome is wasteful a nd
contradicts continual improvement methods. [7] Focusing on the unsafe actions of the individual has nothing
to do with the real upstream factors that create them - the command and control methods prevalent in the
BBS management system. Managers who truly understand quality management systems have much better
ways of spending their time. Mainly studying, improving and fixing the safety system with the help of the
workers. Not by asking them to comply with safety rules but by setting up a system in which their critical
thinking can be applied to their work environment to change the system.

Systems thinking applied to safety

The way in which employees are interviewed, trained, treated by management on the job, the design of work
methods, materials, machinery and equipment (the system) have a great deal more to do with why accidents happen
at work than “unsafe actions”. The simple idea that unsafe actions by workers are the driving force that create
incidents and accidents is now outdated. This technique of focusing on the event (the unsafe action) in the hope that
it will lead to understanding why it happened is doomed to failure. The reality is, the system creates the behavior of
both mangers and workers. We know you can replace an entire workforce wi th new people and they will produce the
same results. This is true for quality and safety. Systems can be very complex. It is the complexity and variation
[8]
inherent in all components of the system that generate the defects, scrap, rework, incidents and acci dents.

When you attain the knowledge that the system itself causes accidents BBS is no longer useful. It is even detrimental
because workers have already learned about quality management techniques. You create a situation of
organizational schizophrenia. On one hand management freely engages and accepts the assistance of line workers
to study and improve quality in their work systems. With the other hand management enlists behaviorist principles to
manipulate employees actions instead of using quality management methods to fix safety problems built into the work
system. Employees now know and understand systems thinking. They’ve applied it in quality. Once you’ve learned a
new and better way of doing something you aren’t going back to the inferior system.

The key is to apply systems thinking to safety management. With quality systems we learned that work is a system. A
system is a series of events that must have an aim. When you know the aim of the system everyone must work
together in a cooperative manner to achieve the aim. Businesses must have an aim of satisfying and taking care of
the customer. If you don’t someone else will and you will lose business. This principle must be applied in safety
management.

How you manage a system is a system in itself. I’ve described earlier the command and control system. Behavior-
based safety fits well with this model. It does not fit at all with a quality management model. The reason being the
employees are considered both customers and then suppliers in the quality system. It is true employees do get paid
to do their jobs. However in a quality management system work previously reserved for managers is now assigned to
line employees. They solve problems of quality on the line. They have the authority to make changes and spend
money to improve the system. This changes everything. Workers are required to think on the job. They can come up
with solutions using their own ideas about what should be changed to improve quality and productivity. BBS all but
eliminates any intrinsic motivation employees have and for this reason it will fail in the new quality management
systems. When it comes to how you manage a work system what is true for quality is als o true for safety. For the
reasons pointed out in this article I believe quality management systems, not behavior based safety, will drive the
new safety management model we will be using in the 21 st century and beyond.

7
Thomas A. Smith, CHCM, CSPM, is President of MOCAL,INC., Lake Orion, M i. 48360. He holds a BS degree
from Northern M ichigan University and is a member of ASSE’s Greater Detroit Chapter. He is a past chair of
the SE M ichigan Safety Council’s Insurance Division and safety committee chair of the Associated General
Contractors Crime Prevention and Safety Committee, Detroit Chapter. He is a leading authority on loss control
and safety practices, and has provided safety process consulting to such companies as Callaway Golf, Ford
M otor, Becton Dickinson, Kaiser Aluminum, Pfizer, Consumers Energy and Jackson Aluminum Supply. He can
be reached at 1-248-391-1818. M ocal WEB site is at www.mocalinc.com You can E-M ail M r. Smith:
tsmith@mocalinc.com

[1]
Reynolds, Stephen H., The Importance of Learning the ABC’s of Behavioral Safety, Professional Safety, February
1997. P 23-25.

[2]
Hunt, Morton, The Story of Psychology. Anchor Book, p. 275

[3]
Hunt, Morton, The Story of Psychology, Anchor Books, p 276.

[4]
Kohn, Alfie, Punished by Rewards. Houghton Mifflin, p 42-43.

[5]
Herzberg, Frederick, One More Time: How do you motivate employees ?, Harvard Business Review.
[14][6]
Fournies, Ferdinand F., Coaching for Improved Work Performance, Van Nostrand Reinhold Co., 1978 p 44.

[7]
Womak, James P. and Jones, Daniel, Lean Thinking, Simon and Schuster, 1996 p.48.

[8]
Scholtes, Peter R., The Leaders Handbook, McGraw Hill, 1998 p 82-85.