- Email: [email protected]
Increasing industrial safety practices and conditions through posted feedback
Journal of Safety Aeseorch, Vol. 15, pp. 7-21, 1984 0 1984 National Safety Council and Pergamon Press Ltd
0022-4375/&/ $3.00 + .OO Printed in the USA
Increasing Industrial Safety Practices and Conditions through Posted Feedback Denise J. Fellner and Beth Sulzer-Azaroff
This study examined the effects of posted feedback for improving safety in a paper mill. Data were taken once a week on 24 practices and 7 conditions. In addition, data on injuries with and without lost time were collected month-
ly. After posting feedback on safe and unsafe conditions for 6 months, more than half of the 17 divisions of the mill showed improvement. Similarly, safe practices increased after feedback was provided for them for 2 months. Most importantly, injuries were cut in half.
Accidents and their resulting injuries are a major public health problem. At the industrial level, the concern for occupational safety is clear. In 1981 alone, disabling injuries totalled 2,100,000, costing the nation 32.5 billion dollars (National Safety Council, 1982). The humanitarian, economic, and personal implications have resulted in a number of efforts aimed at improving safety. A review of the literature on efforts to promote safety reveals a number of different approaches. One area has focused on safety engineering research (i.e., the relationship between the worker, the machine, and the environment), with an emphasis on mechani-
Denise J. Fellner, PhD, is a Performance Manager at Raritan Bay Medical Center, 530 New Brunswick Avenue, Perth Amboy, NJ 08861. Beth Sulzer-Azaroff, PhD, is a Professor of Psychology at the University of Massachusetts, Amherst. This paper is based in part on a dissertation submitted by Dr. Fellner in partial fulfillment of the PhD degree at the University of Massachusetts. Special thanks are extended to Kate Cleary for her assistance in preparing this paper and to Robert Keating and Gail Spence for their assistance in data collection and analyses.
Spring 1984Nolume
15Akmber
1
cal guarding, protective equipment, and safe environments (Grimaldi & Simonds, 1975; Heinrich, Peterson, h Roos, 1980). Another has been research on the correlation between psychological factors such as stress and the worker’s age, experience level, personality characteristics, and accidents (Beehr & Newman, 1978; Grimaldi & Simonds, 1975; Kerr, 1957). A third approach has been the development of government standards and regulations enforced via external audits (Cohen & Margolis, 1973; Ellis, 1975). Finally, human motivation has been investigated as it relates to safety. Antecedents and consequences of behavior, such as performance feedback, have been contingently applied and analyzed in terms of their ability to promote safe practices and conditions. (See Sulzer-Azaroff, 1982, for a review.) Informational feedback on performance has been shown to be a simple, effective, and durable method for promoting safety. For example, Sulzer-Azaroff (1978) showed a dramatic and consistent decrease in the number of hazards in 30 university materials research laboratories when the supervisors were pro7
vided with a copy of the observation form. On these forms the location and nature of the hazard was noted, as well as suggestions for correcting the hazard(s). In a second study, in a plastic products manufacturing firm, Sulxer-Azaroff and deSantamaria (1980) found major reductions in both hazardous conditions and unsafe practices when written feedback was presented twice a week to the supervisors. This feedback included: (1) specific information as to the type and location of the hazard, (2) suggestions for improvement, and (3) positive evaluative comments. Zohar, Cohen, and Axar (1980) used individual feedback effectively to increase the use of ear protectors in a metal fabrication plant. The feedback entailed information to the workers on the extent of temporary hearing loss as a result of the use or nonuse of ear Plugs* While private feedback to both groups and individuals has been found effective for improving safety, Komaki and her colleagues examined the effects of public feedback as part of a package for increasing safe practices. An initial study (Komaki, Barwick, & Scott, 1978) was conducted in a wholesale bakery. The intervention consisted of: (1) training workers to discriminate between safe and unsafe practices, (2) setting a goal for safe practices, and (3) posting feedback on the percentage of incidents performed safely by the group. The results showed that the intervention package was effective for increasing safe practices. A subsequent study was conducted in a vehicle maintenance division of a city’s public works department in order to analyze the relative effects of training versus posted feedback and goal setting (Komaki, Heinzmann, & Lawson, 1980). The training, feedback, and goal-setting procedures were identical to the ones used in the previous study. In this study, however, training occurred prior to the implementation of goal setting and feedback. While there was a small increase in safe practices and conditions with the introduction of training, a more dramatic increase occurred when goal setting and posted feedback were implemented. The extent to which posted feedback, in and of itself, increases safe practices remains 8
to be determined. While Komaki et al. (1978, 1980) found further increases in safe performance with the introduction of posted feedback, the inclusion of goal setting with the feedback may have contributed to the improvements. The effectiveness of goal setting for improving performance in a number of areas is well documented. (See Latham and Yukl, 1975, for a review.) In addition, the training of workers may have enhanced the effects of posted feedback. Thus, the purpose of this study was to examine whether positive and specific feedback posted weekly, alone, would increase safe practices and conditions and consequently decrease injuries. METHOD
Participants and Setting The study was conducted in a paper mill employing approximately 500 workers. Approximately 150 hourly and 8 salaried workers, (i.e., foremen) from 17 rooms of the mill participated. The number of employees in each room ranged from 1 to 30. Consent was obtained from the Industrial Relations Officer of the plant. In addition, workers were informed of the nature of the study and offered the opportunity to decline participation. The plant manufactures products such as paper napkins and towels. Starting with wood pulp and recycled paper, the material is dehydrated, rolled, pressed, cut, printed, and packaged for distribution. Each operation requires specific skills and methods for maintaining a safe and healthy environment. For example, certain operations require guarding, the use of personal protective equipment, and/or specific procedures for operating a machine. Safety is a major priority in this plant. Regular safety audits are conducted and feedback given; safety meetings are held; signs encouraging safe performance are posted; and incentives are offered for intervals without lost time due to injuries. Despite accident rates below the national average, the management of this plant expressed interest and support for continuing to seek methods to improve safety. As part of the safety program at the mill, safety inspectors observe one third of the mill every 2 to 3 months. These inspections consist Journal of Safety Research
of recording any hazardous condition noted. (Neither a standard list of items to observe nor methods for conducting the inspections are provided.) The list of identified hazards is then sent to the foreman of the area inspected.
(i.e., greater than a “2”); (2) was amenable to observation; and (3) was not part of the standard plant safety activities (e.g., the plant has an ongoing system for detecting and fixing mechanical breakdowns; therefore, repairs were excluded from the list).
Personnel The first author (hereafter referred to as the experimenter) and one undergraduate student conducted the weekly inspections. Another student served as a reliability observer. Both students were naive as to the purpose of the study and the intended interventions. They were trained by the experimenter to conduct an inspection in the following manner: The students read the definitions and ob servational procedures. A short meeting was then held to answer any questions and review the definitions and procedures. The experimenter demonstrated how to conduct an inspection while verbally describing each step. Then both the experimenter and student conducted several inspections together. After the student accurately verbalized the procedure for conducting the observation, the experimenter and the student independently conducted an observation in each room. When comparisons of the independent scoring showed 80% interobserver agreement for each item, training was terminated.
Dependent Variables
Preliminary Assessment Prior to collecting baseline data, it was necessary to specify safe and unsafe practices and conditions to generate as complete a list of dependent variables as feasible. (See Sulzer-Azaroff and Fellner, 1983, for complete details.) This was accomplished by: (1) reviewing records, (2) conducting interviews, (3) observing a safety inspection, and (4) setting priorities for the items to observe. These activities yielded a list of 88 items. This list was then given to the security officer, safety manager, resident managers, and general foreman. They were requested to rate the relative importance of each item on a scale of 0 to 3, with “3” extremely important, “2” neutral, “‘1” not critical at this time, and “0” not applicable. After these ratings, 31 items were selected. An item was selected if it: (1) had resulted in an injury and/or received an overall high rating from employees at the mill Spring 1984Nolume
kWJumber 1
Safe and unsafe practices and conditions. This measure reflected the 31 items (24 practices and 7 conditions) generated in the preliminary assessment. Practices refer to behaviors of employees while operating a machine or conducting a task. Conditions refer to the placement and/or storage of materials and equipment. These items were operationally defined and grouped into the following five categories: (1) practices, (2) hoses, (3) skids, (4) fire and electrical hazards, and (5) obstructions. The selection of these categories was based on the way employees at the mill conceptualize and group practices and conditions. (See Table 1 for a sample list of the definitions of these measures.) Injuries. This reflected any accident that resulted in an incident report being filed, ranging from paper cuts to broken knees. Also included were those accidents that accounted for lost time. These data were summarized by the four worker units (manufacturing, shipping and yard, mill maintenance, and specialty products) that staffed the 17 rooms.
Recording Procedures Forms were developed for observing practices and conditions in each of the 17 rooms in the mill. This involved touring the plant to: (1) draw a map of each room, (2) specify which of the 31 practices and conditions could be observed in the room, and (3) divide the room into a number of zones. (See Figure 1 for an example of an observation form.) The inspection was conducted in one room at a time for a brief observational period. Each room was inspected between 7 a.m. and 3 p.m. once a week by one of the observers, Safe and unsafe practices were observed in the following manner: Each employee working on a machine or task was observed sequentially. All practices were observed for a given employee before another was ob9
TABLE 1 SAMPLE LIST OF DEF~~ITI~~S
_ DEFINITIONS
Practice3 1. Guards: All guards are put down in place and are closed completely when machine is operating. 2. Visual contact: When o erating a machine, or conducting a task, an employee has visual contact with the machine or task at han B. 3. Air hoses: An employee using an air hose should: a. Use safety goggles if it is being used on a paper machine. h. Have a safety nozzle on the end of the hose. c. Not aim the hose in the direction of any employee or themseives, including their feet, body, and head. 4. Hoists: Hoists are used when lifting a roll of paper onto the machine; the sling of the hoist should be wrapped around the roll. 5. Towmotors: Towmotor operators should blow their horn prior to entering the following hazardous intersections: a. Flapping doors. b. Going around comers (a 90’ turn around a wall or any stack of materials that exceeds the height of the towmotor). c. Entering the roll-wrap room. d. Going to the dock from the roll-wrap room. 6. Hardhats: Hardhats are to worn while: a. Operating a towmotor. b. In the yard. 7. Pro$e&ue equipment: When working with chemicals or cutting wire: a. Gloves are worn. b. Safety goggles are worn. Hoses 1. Air and w&r hoses: Hoses are coiled totally on their han ers, and/or totally Iyin within 6 inches of walls or machines. A hose is totally coiled if the end cannot be Purther wrapped aroun (4!the hanger. If any art extends 6 inches then it is considered a hazard unless the end of the hose cannot be further coiled. If tK e hose is attached at both ends or is in use, it is not a hazard. Skids 1. Knockdown arranged securely: Only score those that are stacked flat on top of one another. Do not score those stacked on end against a wall or beam. The knockdowns are not stacked for more than 6 feet and/or are not tipping over by producing a 20” angle or more. Knockdowns strapped together by one cord strap are not a hazard. Fire and Electrical I. Fire ~n~~~~: All fire extinguishers are on their bangers. 2. Electrical panels: Panels with a series of switches or controlling the on/off operation of machines are accessible, They are not blocked by barrels, cases, or skids. To assess, one should be able to waik straight to the panel (i.e., not have to walk sideways between an object and a panel) and fully open the door of the panel without moving obstructions. A cluster of blocked panels, two or more, constitutes one hazard.
served. Each practice was scored with a “yes” if it was conducted safely, a “no” if it was conducted unsafely, or an “N.A.” if the employee did not have the oppo~un~ty to engage in the particular practice during the observation. During a single inspection each practice was scored only once for each employee. However, practices such as operating the napkin press biade, sealer, and towmotor were observed for one minute for each employee. Each time the employee engaged in the practice during the one-minute period, a 10
“yes” or a “no” was scored depending on whether or not it was performed safely. These practices were observed for one minute because: (1) they were discrete acts with a definite beginning and end, and (2) there were several opportunities for the worker to engage in them for a given period. As they occurred relatively infrequently, it was difficult to observe practices such as the use of an air hose or a hoist. Thus, observers were instructed to immediately suspend other obse~ation~ activities and focus on the employees engaging in these practices. An asterJournal of Safety Research
SAMPLE Date
A.
FIGURE 1 OBSERVATION
FORM
TiDe
Inspector
PrSctices I. Gwrds ......................... 2. Eye Contsct .................... *3. Air Xoses . . SJfctY nozzle .............. Aimed away from pspcr ...... b. *4. ttoistr......................... *5. Feeding pSper into rollc s. Fingers outside of feed rolls ...................... b. Fingers outside of Qmpsn rolls ...................... c. Fingers outside of embossing rolls ............ *6. HSchine is shut off vhen: a. CleSning ..............*.... b. DislodgingoSpkins . . . . . . . . . 7. Ear PrOteCtiOn . . . . . . . . . . . . . . . . . P. Fingers not on rupkin press
blsdes ..........~.............. 9. One employee 0perStespress
bSnd ..........*................ 10. Lift with legs .................
TOTAL
II
t
TALLY B.
Skids xockdowns 1. SrrSnged securely ... 2. Skids sre flat . . . . . . . . . . . . . . . . . Skids of paper. caaea Sod rolls. 3.
C.
Fire Snd ElectricsltIStsrd6 1. Fire Extinguishers............. 2. ElectricSlPSnelS ..............
D.
Obstructions P*sssgeways 1.
IL co)oIENTs
. . . . . . . . . . . . . . . . ..-. TOTAL
Spring 1984NoEume
15hVumber
1
11
isk was placed next to each of these items on the observation forms. After practices were scored, conditions were scored in the following manner: The observer audited one zone at a time in the room (zones were indicated by the heavy lines on the map in Figure 1), focusing on one item at a time. If the item was deemed safe, nothing was recorded. If the item was not safe, the letter and number that identified the item on the list was placed on the map where it was found. Each time an item was found to be unsafe, its location was noted. The same procedure was followed for all conditions in a zone. If there were no unsafe items in a zone, the word “safe” was written across that zone. The remaining zones in the room were audited in the same manner. Interobserver agreement checks were taken several times throughout the study and at least once during each phase for each room observed. For practices, an agreement was defined as both observers scoring the occurrence or nonoccurrence of an unsafe practice. For conditions, an agreement was defined as both observers scoring the occurrence or nonoccurrence of an unsafe condition in the same location of a particular zone. Interobserver agreement was calculated for each practice and condition by the formula: Agreements Agreements + Disagreements
x 100
Interobserver agreement scores were 85 % for practices, with a range of 25%-100% and 87 % for conditionswith a rangeof 75 %-98%. Experimental
Conditions
and Design
A multiple baseline across classes of behaviors (Komaki, 1977) -first conditions, then practices - was used to evaluate the effects of posted feedback. After baseline rates were established for conditions and practices, feedback was posted for conditions only. Thus, a change in conditions, with little or no change in practices, would suggest the effectiveness of posted feedback. Once performance was stable for both conditions and practices, posted feedback was implemented for practices. A change in practices coincident with posted feedback would further support the effectiveness of the feedback. By obtaining repeated 12
measures on both behaviors, replicating the effects of posted feedback and demonstrating a change in the behavior only when posted feedback was implemented, the reliability of the independent variable would be supported (Komaki, 1977). This design also controls for confounds such as maturation and subject selection. Phase 1 -Baseline. Baseline consisted of the observers auditing the rooms, using the inspection forms developed in this study. For conditions, each data point represented the number of nonhazardous zones summed across conditions, divided by the total number of zones scored. Practices were computed by dividing the total number of safe practices by safe plus unsafe practices. Feedback was not provided on the results of these inspections. The regular safety activities at the plant continued in this phase and throughout the study. Phase 2 -Posted feedback on conditions. Each week, feedback was posted in every room of the mill, in an area that was visible to all the employees of that room. After an inspection, the scores for each room were summarized on a standardized form (Figure 2). The percentage of nonhazardous zones for each item was noted in the appropriate blank, and a graph for each condition showed the percentage of nonhazardous zones summed across items. These graphs were updated following each inspection and reflected performance over time. (Initially, there were four points on each graph reflecting the last four inspections.) Finally, the location of each hazardous condition was noted on the map. Brief meetings of approximately 10 minutes were held with the hourly and salaried workers to explain and discuss the feedback. After 4 months, Phase 2 was initiated in which feedback was posted on practices. As with conditions, this entailed posting feedback in every room of the mill after each weekly inspection. The form used (Figure 3) showed the percentage of times that each practice was observed to be conducted safely. There was space for 3 months permitting an ongoing account, as well as a graph showing the total percentage of safe practices. (As in Phase 2, four points reflecting the last four Journal of Safety Research
FIGURE 2 FORMUSEDFORFEEDBACKONCONDITIONS Date
Boom
Tire
The higher the score, the better A.
SKIDS 1.
2. 3.
B.
In
of
the
100
tohes,
knobs were not found hazardous. In of the tones, #kids VcrZZ upright. In_ of the roses, skids of paper, cases and rolls were not found hazardous.
so 0
I_-
PIREhELEcTBIcALBAaBDs 1.
2.
In of the zones, fire extinguishers were not off their henSerr In of the zones, electrical panel8 were not blocked.
t
P100 w
. 00 C.
OBSTBUCTIONS 1.
In of the tones, uer~obstructions.
1
O-I 10 ImPnCTlm
Spring 1984Nolume
IS/Number
1
20
20
WuuDm
13
FIGURE 3 FORM USED FOR FEEDBACK ON PRACTICES
The scores below represent the percent of times that each practice was observed to be conducted safely. The higher the score, the better.
DATE:
PRACTICES I
I
I
I
I
I
I
1.
2. 3.
4.
Machine Shut Off:
5. 6. 7. 8. 9. 10.
TOTAL X OF SAFE PRACTICES 100-r
SO-
OI
I
I
10
20
30
INSPECTION
14
NUMBER
Journal of Safety Research
inspections were included on this graph.) This form was updated following each inspection and reflected performance over time. For some observations, however, employees did not have the opportunity to engage in practices. Thus, for these observations, no feedback was provided on safety practices. RESULTS
Practices and Conditions After feedback was posted on conditions, workers showed an improvement in 9 of the 17 rooms, remained the same in 6, and showed a reduction in 1. Those in one room were already performing at the 100% level during baseline. Improvements ranged from 3.9 to 18 % (M = 6 %) from baseline to posted feedback. A correlated dtest was performed using the last 11 observations of both the baseline and feedback phases after performance had stabilized. The difference between baseline and posted feedback on nonhazardous zones across all 17 rooms was significant (t [163 = 4.32, p< 901). Safe practices also consistently increased for 10 out of the 17 rooms with improvements ranging from 4 to 30 % . Workers in three of the remaining rooms were already engaging in practices safely 100 % of the time observed; no data were available for one room; one room remained the same; and one room showed a decrease in safe practices. Overall, safe practices increased by 8 % . Each of the last eight observations during baseline and feedback was used to evaluate the effectiveness of the feedback system on safe practices. This was especially important for evaluating the impact of feedback on practices, since the percentage of safe practices was higher at the end of baseline. Using a correlated t test, the difference for practices between baseline and posted feedback was significantly different (t [ 151~ 2.48, p < .025). Figure 4 represents the safety performance of all 17 rooms combined. As can be seen, the percentages of nonhazardous zones and safe practices increased consistently with the introduction of posted feedback. The means during baseline and posted feedback, respectively, were 79 % and 85 % for conditions and 78% and 86% for practices. Spring 1984Nolume
lS/Number
1
An example of a room (Room A) in which workers showed a dramatic improvement in conditions (e.g., nonhazardous zones) is shown in Figure 5. The mean percentages during baseline and feedback were 74 % and 88 % for nonhazardous zones. Figure 6 (Room B) reflects a room in which safe practices dramatically increased from a mean of 89 % during baseline to 96% during feedback. Injuries
Rates of injuries can fluctuate as a function of the time of year. For example, during the -Christmas season in this mill production is often at its peak, so injuries may occur more frequently. Thus, the rate of injuries per 100 workers that occurred when feedback was provided on both practices and conditions during November, December, and Januarywas compared to the rate of injuries that occurred during November, December, and January of the previous 3 years (when no feedback was provided). The rate of injuries was calculated by dividing the number of injuries in a given month by the number of individuals working that month times 100. Figure 7 indicates that when feedback was delivered, injuries decreased for three of these four units. The total group had 21 injuries in 1979 and 1980 (rates of 13.1 and 12.4, respectively), 13 injuries in 1980 before feedback (a rate of 6.9) and 9 during feedback (a rate of 4.9). A correlated t test demonstrated a modest significant difference in the number of injuries between baseline and posted feedback (t [3] = 1.64, p< .lO). Cost-Benefit Analysis In addition to the social and humanitarian aspects of a program that produces a decrease in injuries, the economic implications are also striking. Table 2 shows the amount of time and total costs for developing and maintaining this safety program. Using research assistants as personnel, developing the entire program cost approximately $575. The actual operation of the system, using the average hourly rate of workers at the present plant, would cost about $14 per week. While the development of the program is responsible for the bulk of the costs, over the long term the plant would benefit not only from a humanitarian point of view but also an economic 15
PERCENTAGES
FIGURE 4 OF SAFE PRACTICES AND OF HAZARDOUS ZONES DURING BASELINE AND FEEDBACK
ENTIRE 100
100
PLANT : I
BASELINE
POSTED
FEEDBACK
I
r
----e-w-
I
9
I
I
I I
I
IO
20
30
40
50
60
WEEKS Note. -For a zone to be scored as nonhazardous, ardous conditions.
the entire section of the room must be completely free of any haz-
Journal
of Safety Research
one. The small cost required to maintain the system was accompanied by an over 50% reduction in injury rate compared to pre-feedback levels. While it is not possible in the present case to specify savings in dollars, an outline of potential savings can be given. Money can be saved by a decrease in: (1) insurance premiums, (2) medical expenses, (3) compensation to the injured worker, (4) time away from work by the injured individual and his co-workers who attend to the injury, (5) costs of training a worker to replace the injured individual, and (6) time for filling out accident forms. As can be seen, the costs of operating the system are minimal, while the potential savings are substantial. DISCUi%ION
Posted feedback on 31 conditions and practices was found to be both a socially and a
SAMPLE
statistically significant method for increasing safe practices and conditions in a paper mill. Though injuries were not dealt with directly, they did subsequently decrease when feedback was implemented. Hence, the practices and conditions targeted in this study appear to have been tied to injury prevention while their measurement served as a reasonably valid estimate of safety on the job. Finally, the economic implications were striking. The findings of this study support the conclusions of Cohen (1977) and Smith, Cohen, Cohen, and Cleveland (1978) on the elements that constitute successful safety programs. For example, Cohen (1977) found that management in plants with low accident rates made provisions for immediately recognizing unsafe practices and conditions by formally conducting regular safety inspections and providing the written results of these inspections, In addition, these findings are consis-
FIGURE 5 ROOM A DURING BASELINE AND FEEDBACK: NONHAZARDOUS ZONES
ROOM 100
r
I
BASELINE
IO
A
POSTED FEEDBACK
20
30
40
50
60
WEEKS Spring 1984Nolume
ISNumber
1
17
FIGURE 6 SAMPLE
ROOM B DURING BASELINE SAFE PRACTICES
ROOM
AND FEEDBACK:
8
POSTED IO
20
30
40
FEEDBACK 50
60
WEEKS
tent with previous research on the effectiveness of feedback for improvingsafety (Komaki et al., 1978, 1980; Sulzer-Azaroff, 1978; Sulzer-Azaroff & desantamaria, 1980). This study, however, represents an extension of previous research in a number of ways. First, the feedback was public rather than private and did not specifically require that any positive comments or suggestions be given other than generally encouraging foremen to praise improvements. Furthermore, posted feedback was evaluated in and of itself. Goal setting was not used nor were workers trained to discriminate safe from unsafe practices prior to the feedback as in the Komaki et al. (1978,1979) studies. Thus, this study reflects a clear demonstration of the efficacy of simple posted feedback for improving the safety performance of workers. Several variables may account for the im18
provement in the safety performance in some rooms and not in others. First, where there was improvement, workers often approached the data collector about their performance and examined the feedback immediately after it was posted. This attention to the feedback was not observed in rooms in which no improvement was evident. In addition, where improvements occurred, foremen reported frequently discussing the feedback with their workers. Thus, it may he that examining and discussing the feedback contributes to the effectiveness of this procedure. Future studies might systematically measure the extent to which those involved examine and discuss feedback. If performance is superior when data are examined and discussed, then feedback systems might provide features to ensure that this happens. Finally, differences in improvementswould Journal of Safety Research
Cossairt, Hall, & Hopkins, 1976). Thus, future studies should measure rates of delivery of praise and then contrast feedback plus praise with feedback alone as they influence safety on the job. Finally, the number of workers to whom the feedback is applied may relate to the effectiveness of posted feedback. For example, in this study only four rooms contained more than 10 workers, and only in one of these did safe practices and conditions improve. The remaining 13 rooms involved fewer than 10 workers. More than half of these rooms improved on safe conditions and 80 % of them improved on safe practices. Perhaps in units with many employees, posted feedback is not
not likely be due to demographic variables. For example, improvements were found in areas where there were predominantly males and in areas with predominantly females. Also, the level of experience and age of the workers were equally distributed across all areas. The second variable relates to delivery of praise for improvements. In the present study, while foremen were encouraged to praise improvements, the extent to which this occurred is unknown. Measuring the degree to which praise has been delivered may be important because evidence has indicated that feedback accompanied by praise is often superior to feedback alone (Brown, Willis, & Reid, 1981;
FIGURE 7 OF INJURIES PRIOR TO AND DURING FEEDBACK FOR THE FOUR WORKER UNITS SHIPPING & YARD, MILL MAINTENANCE, AND SPECIALTY STAFFING THE 17 ROOMS
NUMBER (MANUFACTURING,
‘79 ‘BO‘91 ‘82
‘79 ‘BO ‘(II ‘82
SH,Yd
MFG
Note. -Horizontal
‘79 ‘BO ‘81 ‘82
‘79 ‘60 ‘01 ‘82
MM
SP
‘79 ‘80 II TOTAL
PRODUCTS)
‘e2 ONOV.
O&JAN-NO
q NOV, BEC. JAN
FEEDBACK
-FEEDBACK
lines indicate the different months.
Spring 1984Nolume
1MVumber 1
19
TABLE 2 COSTS OF DEVRLOPING AND OPERATING DEVELOPMENT
-
SAFETY PROGRAM
COSTS
Activity
Time Spent
Total Cost (at $5/hour)
40 hours
$200.~
Record reviews (includes development of forms, reviewing records, training, and summarizing information) Interviews (includes developing format, training, interviewing and summarizing information) Observing safety inspection Setting priorities and identifying items Defining safety practices Developing recording forms Developing feedback and goal setting procedures
20 hours
1OOQO
4 hours 10 hours 20 hours 16 hours 5 hours
20.00 50.00 100.00 80.00 25.00
115 hours
$575.00
Time/Week (for entire mill)
Total Cost (at $7lhour)
Total OPR%ATiNC
COSTS
Activity
45 30 30 15
Conducting inspections Posting feedback S~mar~ng and graphing data Xerozing
REFERENCES Beehr, T. A., & Newman, J. Il. (1978). Jobstress, employee health, and organizational effeetiivengs: A facet analysis, model and literature review. Personnel Pq~~o~ogy~31, 665-699. Brown, K. M., Willis, B. S., & Reid, 0. II. (1980). Differential effects of supervisor verbal feedback and feedback plus approval on institutional staff performance. Journal of Organizational Behavior Management, 3. 57-65. Cohen, A. (1977). Factors in successful occupational safety programs. Journal of Safety Research, 9, 16% 178, Cohen, A., ik Ma&is B. (1973). Initialpsychological research related to the Occupational Safety and He&b Act of 1970. Am&can Psycho&g&, 28, 600-606. 20
5.25 3.50 3.50 1.75
2 hours
Total
readily visible. In such cases, feedback may need to be posted in several locations, delivered to smaller groups, or given indi~du~y. Future research might explore these issues. In conclusion, this study demonstrated the effectiveness of pinpointing and directly observing safe practices and conditions and posting feedback for improving safety among workers in an industrial setting.
minutes minutes minutes minutes
$14.00
Cossairt, A., Hall, R. C., & Hopkins, B. L. (1973). The effects of experimenter’s instructions, feedback, and praise on teacher praise and student attending behavior. ,lournal A~~~ B&&r Angle, 6,89-100. Ellis, L. (1975). A review of research on efforts to promote occupational safety. loumal of Safety Research, 7, 180-189. Grim~d~, J. V., &rSimonds, R. H. (1975). Sajetq management. Homewood, IL: Richard D. Irwin, Inc. Heinrich, H. S., Peterson, D., & Roes, N. (1980). Industrial accident prevention: A safety management approach. New York: McGraw-Hill, Kerr, W. (1957). Complementary theories of safety psychology. The Journal of Social P~chology, 45, 3-9. Komaki, J. (1977). Alternative evaluation strategies in work settings: Reversal and multiple baseline designs. ,$uyI of Organizations Behavior Management, 1,
oj
-
.
Komaki, J., Barwick, K. D., & Scott, behavioral approach to occupational ing and reinforcing safe performance facturing plant. Journal of Applied
L. R. (1978). A safety: Pinpointin a food manuPsychology,
63,
424-445.
Komaki, J., Heinzmann, A. T., & Lawson, L. (1980). Effects of training and feedback: A component analysis of a behavioral safety program. Journal of Applied Psychology,
6.5, 261-270.
Latham, G. P., & Yukl, G. A. (1975). A review of research on the application of goal setting in organizations. Ace&my of Mrmagement Jounurl, 18, SM.
]ournal of Safety &search
National Safety Council (1982). Accidentfacts. Chicago: Author. Smith, M. J., Cohen, H. H., Cohen, A., & Cleveland, Ft. J. (1978). Characteristics of successful safety programs. Journal of Safety Research, 10, 5-15. SuIzer-Azaroff, B. (1978). Behavioral ecology and accident prevention. Journal of Organixational Behador Management, 2, 11-44. Suhzer-Azaroff, B, (1982). Behavioral approaches to health and occupational safety. In L. Frederiksen (Ed.), Handbook of organizational behavior management. New York: John Wiley & Sons.
Spring
1984Nolume
15LVumbet
1
Sulzer-Azaroff, B., C deSantamaria, M. C. (1980). Industrial safety hazard reduction through performance feedback. 287-295.
Journal of Applied
Behauior Analysis,
13,
SuIzer-Azaroff, B., t FeIlner, D. J. (1983, May). Searching for performance targets in the behavioral anaIysis of occupational health and safety: An assessment strategy. Paper presented at the Ninth Annual Association for Behavior Analysis Convention, Milwaukee, WI. Zohar, D., Cohen, A., & Azar, N. (1980). Promoting increased use of ear protectors in noise through information feedback. Human Factors, 22, 69-79.
21
0022-4375/&/ $3.00 + .OO Printed in the USA
Increasing Industrial Safety Practices and Conditions through Posted Feedback Denise J. Fellner and Beth Sulzer-Azaroff
This study examined the effects of posted feedback for improving safety in a paper mill. Data were taken once a week on 24 practices and 7 conditions. In addition, data on injuries with and without lost time were collected month-
ly. After posting feedback on safe and unsafe conditions for 6 months, more than half of the 17 divisions of the mill showed improvement. Similarly, safe practices increased after feedback was provided for them for 2 months. Most importantly, injuries were cut in half.
Accidents and their resulting injuries are a major public health problem. At the industrial level, the concern for occupational safety is clear. In 1981 alone, disabling injuries totalled 2,100,000, costing the nation 32.5 billion dollars (National Safety Council, 1982). The humanitarian, economic, and personal implications have resulted in a number of efforts aimed at improving safety. A review of the literature on efforts to promote safety reveals a number of different approaches. One area has focused on safety engineering research (i.e., the relationship between the worker, the machine, and the environment), with an emphasis on mechani-
Denise J. Fellner, PhD, is a Performance Manager at Raritan Bay Medical Center, 530 New Brunswick Avenue, Perth Amboy, NJ 08861. Beth Sulzer-Azaroff, PhD, is a Professor of Psychology at the University of Massachusetts, Amherst. This paper is based in part on a dissertation submitted by Dr. Fellner in partial fulfillment of the PhD degree at the University of Massachusetts. Special thanks are extended to Kate Cleary for her assistance in preparing this paper and to Robert Keating and Gail Spence for their assistance in data collection and analyses.
Spring 1984Nolume
15Akmber
1
cal guarding, protective equipment, and safe environments (Grimaldi & Simonds, 1975; Heinrich, Peterson, h Roos, 1980). Another has been research on the correlation between psychological factors such as stress and the worker’s age, experience level, personality characteristics, and accidents (Beehr & Newman, 1978; Grimaldi & Simonds, 1975; Kerr, 1957). A third approach has been the development of government standards and regulations enforced via external audits (Cohen & Margolis, 1973; Ellis, 1975). Finally, human motivation has been investigated as it relates to safety. Antecedents and consequences of behavior, such as performance feedback, have been contingently applied and analyzed in terms of their ability to promote safe practices and conditions. (See Sulzer-Azaroff, 1982, for a review.) Informational feedback on performance has been shown to be a simple, effective, and durable method for promoting safety. For example, Sulzer-Azaroff (1978) showed a dramatic and consistent decrease in the number of hazards in 30 university materials research laboratories when the supervisors were pro7
vided with a copy of the observation form. On these forms the location and nature of the hazard was noted, as well as suggestions for correcting the hazard(s). In a second study, in a plastic products manufacturing firm, Sulxer-Azaroff and deSantamaria (1980) found major reductions in both hazardous conditions and unsafe practices when written feedback was presented twice a week to the supervisors. This feedback included: (1) specific information as to the type and location of the hazard, (2) suggestions for improvement, and (3) positive evaluative comments. Zohar, Cohen, and Axar (1980) used individual feedback effectively to increase the use of ear protectors in a metal fabrication plant. The feedback entailed information to the workers on the extent of temporary hearing loss as a result of the use or nonuse of ear Plugs* While private feedback to both groups and individuals has been found effective for improving safety, Komaki and her colleagues examined the effects of public feedback as part of a package for increasing safe practices. An initial study (Komaki, Barwick, & Scott, 1978) was conducted in a wholesale bakery. The intervention consisted of: (1) training workers to discriminate between safe and unsafe practices, (2) setting a goal for safe practices, and (3) posting feedback on the percentage of incidents performed safely by the group. The results showed that the intervention package was effective for increasing safe practices. A subsequent study was conducted in a vehicle maintenance division of a city’s public works department in order to analyze the relative effects of training versus posted feedback and goal setting (Komaki, Heinzmann, & Lawson, 1980). The training, feedback, and goal-setting procedures were identical to the ones used in the previous study. In this study, however, training occurred prior to the implementation of goal setting and feedback. While there was a small increase in safe practices and conditions with the introduction of training, a more dramatic increase occurred when goal setting and posted feedback were implemented. The extent to which posted feedback, in and of itself, increases safe practices remains 8
to be determined. While Komaki et al. (1978, 1980) found further increases in safe performance with the introduction of posted feedback, the inclusion of goal setting with the feedback may have contributed to the improvements. The effectiveness of goal setting for improving performance in a number of areas is well documented. (See Latham and Yukl, 1975, for a review.) In addition, the training of workers may have enhanced the effects of posted feedback. Thus, the purpose of this study was to examine whether positive and specific feedback posted weekly, alone, would increase safe practices and conditions and consequently decrease injuries. METHOD
Participants and Setting The study was conducted in a paper mill employing approximately 500 workers. Approximately 150 hourly and 8 salaried workers, (i.e., foremen) from 17 rooms of the mill participated. The number of employees in each room ranged from 1 to 30. Consent was obtained from the Industrial Relations Officer of the plant. In addition, workers were informed of the nature of the study and offered the opportunity to decline participation. The plant manufactures products such as paper napkins and towels. Starting with wood pulp and recycled paper, the material is dehydrated, rolled, pressed, cut, printed, and packaged for distribution. Each operation requires specific skills and methods for maintaining a safe and healthy environment. For example, certain operations require guarding, the use of personal protective equipment, and/or specific procedures for operating a machine. Safety is a major priority in this plant. Regular safety audits are conducted and feedback given; safety meetings are held; signs encouraging safe performance are posted; and incentives are offered for intervals without lost time due to injuries. Despite accident rates below the national average, the management of this plant expressed interest and support for continuing to seek methods to improve safety. As part of the safety program at the mill, safety inspectors observe one third of the mill every 2 to 3 months. These inspections consist Journal of Safety Research
of recording any hazardous condition noted. (Neither a standard list of items to observe nor methods for conducting the inspections are provided.) The list of identified hazards is then sent to the foreman of the area inspected.
(i.e., greater than a “2”); (2) was amenable to observation; and (3) was not part of the standard plant safety activities (e.g., the plant has an ongoing system for detecting and fixing mechanical breakdowns; therefore, repairs were excluded from the list).
Personnel The first author (hereafter referred to as the experimenter) and one undergraduate student conducted the weekly inspections. Another student served as a reliability observer. Both students were naive as to the purpose of the study and the intended interventions. They were trained by the experimenter to conduct an inspection in the following manner: The students read the definitions and ob servational procedures. A short meeting was then held to answer any questions and review the definitions and procedures. The experimenter demonstrated how to conduct an inspection while verbally describing each step. Then both the experimenter and student conducted several inspections together. After the student accurately verbalized the procedure for conducting the observation, the experimenter and the student independently conducted an observation in each room. When comparisons of the independent scoring showed 80% interobserver agreement for each item, training was terminated.
Dependent Variables
Preliminary Assessment Prior to collecting baseline data, it was necessary to specify safe and unsafe practices and conditions to generate as complete a list of dependent variables as feasible. (See Sulzer-Azaroff and Fellner, 1983, for complete details.) This was accomplished by: (1) reviewing records, (2) conducting interviews, (3) observing a safety inspection, and (4) setting priorities for the items to observe. These activities yielded a list of 88 items. This list was then given to the security officer, safety manager, resident managers, and general foreman. They were requested to rate the relative importance of each item on a scale of 0 to 3, with “3” extremely important, “2” neutral, “‘1” not critical at this time, and “0” not applicable. After these ratings, 31 items were selected. An item was selected if it: (1) had resulted in an injury and/or received an overall high rating from employees at the mill Spring 1984Nolume
kWJumber 1
Safe and unsafe practices and conditions. This measure reflected the 31 items (24 practices and 7 conditions) generated in the preliminary assessment. Practices refer to behaviors of employees while operating a machine or conducting a task. Conditions refer to the placement and/or storage of materials and equipment. These items were operationally defined and grouped into the following five categories: (1) practices, (2) hoses, (3) skids, (4) fire and electrical hazards, and (5) obstructions. The selection of these categories was based on the way employees at the mill conceptualize and group practices and conditions. (See Table 1 for a sample list of the definitions of these measures.) Injuries. This reflected any accident that resulted in an incident report being filed, ranging from paper cuts to broken knees. Also included were those accidents that accounted for lost time. These data were summarized by the four worker units (manufacturing, shipping and yard, mill maintenance, and specialty products) that staffed the 17 rooms.
Recording Procedures Forms were developed for observing practices and conditions in each of the 17 rooms in the mill. This involved touring the plant to: (1) draw a map of each room, (2) specify which of the 31 practices and conditions could be observed in the room, and (3) divide the room into a number of zones. (See Figure 1 for an example of an observation form.) The inspection was conducted in one room at a time for a brief observational period. Each room was inspected between 7 a.m. and 3 p.m. once a week by one of the observers, Safe and unsafe practices were observed in the following manner: Each employee working on a machine or task was observed sequentially. All practices were observed for a given employee before another was ob9
TABLE 1 SAMPLE LIST OF DEF~~ITI~~S
_ DEFINITIONS
Practice3 1. Guards: All guards are put down in place and are closed completely when machine is operating. 2. Visual contact: When o erating a machine, or conducting a task, an employee has visual contact with the machine or task at han B. 3. Air hoses: An employee using an air hose should: a. Use safety goggles if it is being used on a paper machine. h. Have a safety nozzle on the end of the hose. c. Not aim the hose in the direction of any employee or themseives, including their feet, body, and head. 4. Hoists: Hoists are used when lifting a roll of paper onto the machine; the sling of the hoist should be wrapped around the roll. 5. Towmotors: Towmotor operators should blow their horn prior to entering the following hazardous intersections: a. Flapping doors. b. Going around comers (a 90’ turn around a wall or any stack of materials that exceeds the height of the towmotor). c. Entering the roll-wrap room. d. Going to the dock from the roll-wrap room. 6. Hardhats: Hardhats are to worn while: a. Operating a towmotor. b. In the yard. 7. Pro$e&ue equipment: When working with chemicals or cutting wire: a. Gloves are worn. b. Safety goggles are worn. Hoses 1. Air and w&r hoses: Hoses are coiled totally on their han ers, and/or totally Iyin within 6 inches of walls or machines. A hose is totally coiled if the end cannot be Purther wrapped aroun (4!the hanger. If any art extends 6 inches then it is considered a hazard unless the end of the hose cannot be further coiled. If tK e hose is attached at both ends or is in use, it is not a hazard. Skids 1. Knockdown arranged securely: Only score those that are stacked flat on top of one another. Do not score those stacked on end against a wall or beam. The knockdowns are not stacked for more than 6 feet and/or are not tipping over by producing a 20” angle or more. Knockdowns strapped together by one cord strap are not a hazard. Fire and Electrical I. Fire ~n~~~~: All fire extinguishers are on their bangers. 2. Electrical panels: Panels with a series of switches or controlling the on/off operation of machines are accessible, They are not blocked by barrels, cases, or skids. To assess, one should be able to waik straight to the panel (i.e., not have to walk sideways between an object and a panel) and fully open the door of the panel without moving obstructions. A cluster of blocked panels, two or more, constitutes one hazard.
served. Each practice was scored with a “yes” if it was conducted safely, a “no” if it was conducted unsafely, or an “N.A.” if the employee did not have the oppo~un~ty to engage in the particular practice during the observation. During a single inspection each practice was scored only once for each employee. However, practices such as operating the napkin press biade, sealer, and towmotor were observed for one minute for each employee. Each time the employee engaged in the practice during the one-minute period, a 10
“yes” or a “no” was scored depending on whether or not it was performed safely. These practices were observed for one minute because: (1) they were discrete acts with a definite beginning and end, and (2) there were several opportunities for the worker to engage in them for a given period. As they occurred relatively infrequently, it was difficult to observe practices such as the use of an air hose or a hoist. Thus, observers were instructed to immediately suspend other obse~ation~ activities and focus on the employees engaging in these practices. An asterJournal of Safety Research
SAMPLE Date
A.
FIGURE 1 OBSERVATION
FORM
TiDe
Inspector
PrSctices I. Gwrds ......................... 2. Eye Contsct .................... *3. Air Xoses . . SJfctY nozzle .............. Aimed away from pspcr ...... b. *4. ttoistr......................... *5. Feeding pSper into rollc s. Fingers outside of feed rolls ...................... b. Fingers outside of Qmpsn rolls ...................... c. Fingers outside of embossing rolls ............ *6. HSchine is shut off vhen: a. CleSning ..............*.... b. DislodgingoSpkins . . . . . . . . . 7. Ear PrOteCtiOn . . . . . . . . . . . . . . . . . P. Fingers not on rupkin press
blsdes ..........~.............. 9. One employee 0perStespress
bSnd ..........*................ 10. Lift with legs .................
TOTAL
II
t
TALLY B.
Skids xockdowns 1. SrrSnged securely ... 2. Skids sre flat . . . . . . . . . . . . . . . . . Skids of paper. caaea Sod rolls. 3.
C.
Fire Snd ElectricsltIStsrd6 1. Fire Extinguishers............. 2. ElectricSlPSnelS ..............
D.
Obstructions P*sssgeways 1.
IL co)oIENTs
. . . . . . . . . . . . . . . . ..-. TOTAL
Spring 1984NoEume
15hVumber
1
11
isk was placed next to each of these items on the observation forms. After practices were scored, conditions were scored in the following manner: The observer audited one zone at a time in the room (zones were indicated by the heavy lines on the map in Figure 1), focusing on one item at a time. If the item was deemed safe, nothing was recorded. If the item was not safe, the letter and number that identified the item on the list was placed on the map where it was found. Each time an item was found to be unsafe, its location was noted. The same procedure was followed for all conditions in a zone. If there were no unsafe items in a zone, the word “safe” was written across that zone. The remaining zones in the room were audited in the same manner. Interobserver agreement checks were taken several times throughout the study and at least once during each phase for each room observed. For practices, an agreement was defined as both observers scoring the occurrence or nonoccurrence of an unsafe practice. For conditions, an agreement was defined as both observers scoring the occurrence or nonoccurrence of an unsafe condition in the same location of a particular zone. Interobserver agreement was calculated for each practice and condition by the formula: Agreements Agreements + Disagreements
x 100
Interobserver agreement scores were 85 % for practices, with a range of 25%-100% and 87 % for conditionswith a rangeof 75 %-98%. Experimental
Conditions
and Design
A multiple baseline across classes of behaviors (Komaki, 1977) -first conditions, then practices - was used to evaluate the effects of posted feedback. After baseline rates were established for conditions and practices, feedback was posted for conditions only. Thus, a change in conditions, with little or no change in practices, would suggest the effectiveness of posted feedback. Once performance was stable for both conditions and practices, posted feedback was implemented for practices. A change in practices coincident with posted feedback would further support the effectiveness of the feedback. By obtaining repeated 12
measures on both behaviors, replicating the effects of posted feedback and demonstrating a change in the behavior only when posted feedback was implemented, the reliability of the independent variable would be supported (Komaki, 1977). This design also controls for confounds such as maturation and subject selection. Phase 1 -Baseline. Baseline consisted of the observers auditing the rooms, using the inspection forms developed in this study. For conditions, each data point represented the number of nonhazardous zones summed across conditions, divided by the total number of zones scored. Practices were computed by dividing the total number of safe practices by safe plus unsafe practices. Feedback was not provided on the results of these inspections. The regular safety activities at the plant continued in this phase and throughout the study. Phase 2 -Posted feedback on conditions. Each week, feedback was posted in every room of the mill, in an area that was visible to all the employees of that room. After an inspection, the scores for each room were summarized on a standardized form (Figure 2). The percentage of nonhazardous zones for each item was noted in the appropriate blank, and a graph for each condition showed the percentage of nonhazardous zones summed across items. These graphs were updated following each inspection and reflected performance over time. (Initially, there were four points on each graph reflecting the last four inspections.) Finally, the location of each hazardous condition was noted on the map. Brief meetings of approximately 10 minutes were held with the hourly and salaried workers to explain and discuss the feedback. After 4 months, Phase 2 was initiated in which feedback was posted on practices. As with conditions, this entailed posting feedback in every room of the mill after each weekly inspection. The form used (Figure 3) showed the percentage of times that each practice was observed to be conducted safely. There was space for 3 months permitting an ongoing account, as well as a graph showing the total percentage of safe practices. (As in Phase 2, four points reflecting the last four Journal of Safety Research
FIGURE 2 FORMUSEDFORFEEDBACKONCONDITIONS Date
Boom
Tire
The higher the score, the better A.
SKIDS 1.
2. 3.
B.
In
of
the
100
tohes,
knobs were not found hazardous. In of the tones, #kids VcrZZ upright. In_ of the roses, skids of paper, cases and rolls were not found hazardous.
so 0
I_-
PIREhELEcTBIcALBAaBDs 1.
2.
In of the zones, fire extinguishers were not off their henSerr In of the zones, electrical panel8 were not blocked.
t
P100 w
. 00 C.
OBSTBUCTIONS 1.
In of the tones, uer~obstructions.
1
O-I 10 ImPnCTlm
Spring 1984Nolume
IS/Number
1
20
20
WuuDm
13
FIGURE 3 FORM USED FOR FEEDBACK ON PRACTICES
The scores below represent the percent of times that each practice was observed to be conducted safely. The higher the score, the better.
DATE:
PRACTICES I
I
I
I
I
I
I
1.
2. 3.
4.
Machine Shut Off:
5. 6. 7. 8. 9. 10.
TOTAL X OF SAFE PRACTICES 100-r
SO-
OI
I
I
10
20
30
INSPECTION
14
NUMBER
Journal of Safety Research
inspections were included on this graph.) This form was updated following each inspection and reflected performance over time. For some observations, however, employees did not have the opportunity to engage in practices. Thus, for these observations, no feedback was provided on safety practices. RESULTS
Practices and Conditions After feedback was posted on conditions, workers showed an improvement in 9 of the 17 rooms, remained the same in 6, and showed a reduction in 1. Those in one room were already performing at the 100% level during baseline. Improvements ranged from 3.9 to 18 % (M = 6 %) from baseline to posted feedback. A correlated dtest was performed using the last 11 observations of both the baseline and feedback phases after performance had stabilized. The difference between baseline and posted feedback on nonhazardous zones across all 17 rooms was significant (t [163 = 4.32, p< 901). Safe practices also consistently increased for 10 out of the 17 rooms with improvements ranging from 4 to 30 % . Workers in three of the remaining rooms were already engaging in practices safely 100 % of the time observed; no data were available for one room; one room remained the same; and one room showed a decrease in safe practices. Overall, safe practices increased by 8 % . Each of the last eight observations during baseline and feedback was used to evaluate the effectiveness of the feedback system on safe practices. This was especially important for evaluating the impact of feedback on practices, since the percentage of safe practices was higher at the end of baseline. Using a correlated t test, the difference for practices between baseline and posted feedback was significantly different (t [ 151~ 2.48, p < .025). Figure 4 represents the safety performance of all 17 rooms combined. As can be seen, the percentages of nonhazardous zones and safe practices increased consistently with the introduction of posted feedback. The means during baseline and posted feedback, respectively, were 79 % and 85 % for conditions and 78% and 86% for practices. Spring 1984Nolume
lS/Number
1
An example of a room (Room A) in which workers showed a dramatic improvement in conditions (e.g., nonhazardous zones) is shown in Figure 5. The mean percentages during baseline and feedback were 74 % and 88 % for nonhazardous zones. Figure 6 (Room B) reflects a room in which safe practices dramatically increased from a mean of 89 % during baseline to 96% during feedback. Injuries
Rates of injuries can fluctuate as a function of the time of year. For example, during the -Christmas season in this mill production is often at its peak, so injuries may occur more frequently. Thus, the rate of injuries per 100 workers that occurred when feedback was provided on both practices and conditions during November, December, and Januarywas compared to the rate of injuries that occurred during November, December, and January of the previous 3 years (when no feedback was provided). The rate of injuries was calculated by dividing the number of injuries in a given month by the number of individuals working that month times 100. Figure 7 indicates that when feedback was delivered, injuries decreased for three of these four units. The total group had 21 injuries in 1979 and 1980 (rates of 13.1 and 12.4, respectively), 13 injuries in 1980 before feedback (a rate of 6.9) and 9 during feedback (a rate of 4.9). A correlated t test demonstrated a modest significant difference in the number of injuries between baseline and posted feedback (t [3] = 1.64, p< .lO). Cost-Benefit Analysis In addition to the social and humanitarian aspects of a program that produces a decrease in injuries, the economic implications are also striking. Table 2 shows the amount of time and total costs for developing and maintaining this safety program. Using research assistants as personnel, developing the entire program cost approximately $575. The actual operation of the system, using the average hourly rate of workers at the present plant, would cost about $14 per week. While the development of the program is responsible for the bulk of the costs, over the long term the plant would benefit not only from a humanitarian point of view but also an economic 15
PERCENTAGES
FIGURE 4 OF SAFE PRACTICES AND OF HAZARDOUS ZONES DURING BASELINE AND FEEDBACK
ENTIRE 100
100
PLANT : I
BASELINE
POSTED
FEEDBACK
I
r
----e-w-
I
9
I
I
I I
I
IO
20
30
40
50
60
WEEKS Note. -For a zone to be scored as nonhazardous, ardous conditions.
the entire section of the room must be completely free of any haz-
Journal
of Safety Research
one. The small cost required to maintain the system was accompanied by an over 50% reduction in injury rate compared to pre-feedback levels. While it is not possible in the present case to specify savings in dollars, an outline of potential savings can be given. Money can be saved by a decrease in: (1) insurance premiums, (2) medical expenses, (3) compensation to the injured worker, (4) time away from work by the injured individual and his co-workers who attend to the injury, (5) costs of training a worker to replace the injured individual, and (6) time for filling out accident forms. As can be seen, the costs of operating the system are minimal, while the potential savings are substantial. DISCUi%ION
Posted feedback on 31 conditions and practices was found to be both a socially and a
SAMPLE
statistically significant method for increasing safe practices and conditions in a paper mill. Though injuries were not dealt with directly, they did subsequently decrease when feedback was implemented. Hence, the practices and conditions targeted in this study appear to have been tied to injury prevention while their measurement served as a reasonably valid estimate of safety on the job. Finally, the economic implications were striking. The findings of this study support the conclusions of Cohen (1977) and Smith, Cohen, Cohen, and Cleveland (1978) on the elements that constitute successful safety programs. For example, Cohen (1977) found that management in plants with low accident rates made provisions for immediately recognizing unsafe practices and conditions by formally conducting regular safety inspections and providing the written results of these inspections, In addition, these findings are consis-
FIGURE 5 ROOM A DURING BASELINE AND FEEDBACK: NONHAZARDOUS ZONES
ROOM 100
r
I
BASELINE
IO
A
POSTED FEEDBACK
20
30
40
50
60
WEEKS Spring 1984Nolume
ISNumber
1
17
FIGURE 6 SAMPLE
ROOM B DURING BASELINE SAFE PRACTICES
ROOM
AND FEEDBACK:
8
POSTED IO
20
30
40
FEEDBACK 50
60
WEEKS
tent with previous research on the effectiveness of feedback for improvingsafety (Komaki et al., 1978, 1980; Sulzer-Azaroff, 1978; Sulzer-Azaroff & desantamaria, 1980). This study, however, represents an extension of previous research in a number of ways. First, the feedback was public rather than private and did not specifically require that any positive comments or suggestions be given other than generally encouraging foremen to praise improvements. Furthermore, posted feedback was evaluated in and of itself. Goal setting was not used nor were workers trained to discriminate safe from unsafe practices prior to the feedback as in the Komaki et al. (1978,1979) studies. Thus, this study reflects a clear demonstration of the efficacy of simple posted feedback for improving the safety performance of workers. Several variables may account for the im18
provement in the safety performance in some rooms and not in others. First, where there was improvement, workers often approached the data collector about their performance and examined the feedback immediately after it was posted. This attention to the feedback was not observed in rooms in which no improvement was evident. In addition, where improvements occurred, foremen reported frequently discussing the feedback with their workers. Thus, it may he that examining and discussing the feedback contributes to the effectiveness of this procedure. Future studies might systematically measure the extent to which those involved examine and discuss feedback. If performance is superior when data are examined and discussed, then feedback systems might provide features to ensure that this happens. Finally, differences in improvementswould Journal of Safety Research
Cossairt, Hall, & Hopkins, 1976). Thus, future studies should measure rates of delivery of praise and then contrast feedback plus praise with feedback alone as they influence safety on the job. Finally, the number of workers to whom the feedback is applied may relate to the effectiveness of posted feedback. For example, in this study only four rooms contained more than 10 workers, and only in one of these did safe practices and conditions improve. The remaining 13 rooms involved fewer than 10 workers. More than half of these rooms improved on safe conditions and 80 % of them improved on safe practices. Perhaps in units with many employees, posted feedback is not
not likely be due to demographic variables. For example, improvements were found in areas where there were predominantly males and in areas with predominantly females. Also, the level of experience and age of the workers were equally distributed across all areas. The second variable relates to delivery of praise for improvements. In the present study, while foremen were encouraged to praise improvements, the extent to which this occurred is unknown. Measuring the degree to which praise has been delivered may be important because evidence has indicated that feedback accompanied by praise is often superior to feedback alone (Brown, Willis, & Reid, 1981;
FIGURE 7 OF INJURIES PRIOR TO AND DURING FEEDBACK FOR THE FOUR WORKER UNITS SHIPPING & YARD, MILL MAINTENANCE, AND SPECIALTY STAFFING THE 17 ROOMS
NUMBER (MANUFACTURING,
‘79 ‘BO‘91 ‘82
‘79 ‘BO ‘(II ‘82
SH,Yd
MFG
Note. -Horizontal
‘79 ‘BO ‘81 ‘82
‘79 ‘60 ‘01 ‘82
MM
SP
‘79 ‘80 II TOTAL
PRODUCTS)
‘e2 ONOV.
O&JAN-NO
q NOV, BEC. JAN
FEEDBACK
-FEEDBACK
lines indicate the different months.
Spring 1984Nolume
1MVumber 1
19
TABLE 2 COSTS OF DEVRLOPING AND OPERATING DEVELOPMENT
-
SAFETY PROGRAM
COSTS
Activity
Time Spent
Total Cost (at $5/hour)
40 hours
$200.~
Record reviews (includes development of forms, reviewing records, training, and summarizing information) Interviews (includes developing format, training, interviewing and summarizing information) Observing safety inspection Setting priorities and identifying items Defining safety practices Developing recording forms Developing feedback and goal setting procedures
20 hours
1OOQO
4 hours 10 hours 20 hours 16 hours 5 hours
20.00 50.00 100.00 80.00 25.00
115 hours
$575.00
Time/Week (for entire mill)
Total Cost (at $7lhour)
Total OPR%ATiNC
COSTS
Activity
45 30 30 15
Conducting inspections Posting feedback S~mar~ng and graphing data Xerozing
REFERENCES Beehr, T. A., & Newman, J. Il. (1978). Jobstress, employee health, and organizational effeetiivengs: A facet analysis, model and literature review. Personnel Pq~~o~ogy~31, 665-699. Brown, K. M., Willis, B. S., & Reid, 0. II. (1980). Differential effects of supervisor verbal feedback and feedback plus approval on institutional staff performance. Journal of Organizational Behavior Management, 3. 57-65. Cohen, A. (1977). Factors in successful occupational safety programs. Journal of Safety Research, 9, 16% 178, Cohen, A., ik Ma&is B. (1973). Initialpsychological research related to the Occupational Safety and He&b Act of 1970. Am&can Psycho&g&, 28, 600-606. 20
5.25 3.50 3.50 1.75
2 hours
Total
readily visible. In such cases, feedback may need to be posted in several locations, delivered to smaller groups, or given indi~du~y. Future research might explore these issues. In conclusion, this study demonstrated the effectiveness of pinpointing and directly observing safe practices and conditions and posting feedback for improving safety among workers in an industrial setting.
minutes minutes minutes minutes
$14.00
Cossairt, A., Hall, R. C., & Hopkins, B. L. (1973). The effects of experimenter’s instructions, feedback, and praise on teacher praise and student attending behavior. ,lournal A~~~ B&&r Angle, 6,89-100. Ellis, L. (1975). A review of research on efforts to promote occupational safety. loumal of Safety Research, 7, 180-189. Grim~d~, J. V., &rSimonds, R. H. (1975). Sajetq management. Homewood, IL: Richard D. Irwin, Inc. Heinrich, H. S., Peterson, D., & Roes, N. (1980). Industrial accident prevention: A safety management approach. New York: McGraw-Hill, Kerr, W. (1957). Complementary theories of safety psychology. The Journal of Social P~chology, 45, 3-9. Komaki, J. (1977). Alternative evaluation strategies in work settings: Reversal and multiple baseline designs. ,$uyI of Organizations Behavior Management, 1,
oj
-
.
Komaki, J., Barwick, K. D., & Scott, behavioral approach to occupational ing and reinforcing safe performance facturing plant. Journal of Applied
L. R. (1978). A safety: Pinpointin a food manuPsychology,
63,
424-445.
Komaki, J., Heinzmann, A. T., & Lawson, L. (1980). Effects of training and feedback: A component analysis of a behavioral safety program. Journal of Applied Psychology,
6.5, 261-270.
Latham, G. P., & Yukl, G. A. (1975). A review of research on the application of goal setting in organizations. Ace&my of Mrmagement Jounurl, 18, SM.
]ournal of Safety &search
National Safety Council (1982). Accidentfacts. Chicago: Author. Smith, M. J., Cohen, H. H., Cohen, A., & Cleveland, Ft. J. (1978). Characteristics of successful safety programs. Journal of Safety Research, 10, 5-15. SuIzer-Azaroff, B. (1978). Behavioral ecology and accident prevention. Journal of Organixational Behador Management, 2, 11-44. Suhzer-Azaroff, B, (1982). Behavioral approaches to health and occupational safety. In L. Frederiksen (Ed.), Handbook of organizational behavior management. New York: John Wiley & Sons.
Spring
1984Nolume
15LVumbet
1
Sulzer-Azaroff, B., C deSantamaria, M. C. (1980). Industrial safety hazard reduction through performance feedback. 287-295.
Journal of Applied
Behauior Analysis,
13,
SuIzer-Azaroff, B., t FeIlner, D. J. (1983, May). Searching for performance targets in the behavioral anaIysis of occupational health and safety: An assessment strategy. Paper presented at the Ninth Annual Association for Behavior Analysis Convention, Milwaukee, WI. Zohar, D., Cohen, A., & Azar, N. (1980). Promoting increased use of ear protectors in noise through information feedback. Human Factors, 22, 69-79.
21