- Email: [email protected]
The safety observer effect: The effects of conducting safety observations
Journal of Safety Research 39 (2008) 365 – 373
www.elsevier.com/locate/jsr
www.nsc.org
The safety observer effect: The effects of conducting safety observations Alicia M. Alvero a,1 , Kristen Rost c,⁎, John Austin b,2 a
c
Queens College, CUNY, Department of Psychology, 65-30 Kissena Blvd., Flushing, NY 11367 b Western Michigan University, Department of Psychology, Kalamazoo, MI 49008 Queens College and The Graduate Center, CUNY, Department of Psychology, 65-30 Kissena Blvd., Flushing, NY 11367 Available online 26 July 2008
Abstract Introduction: Some research suggests that conducting safety observations of another's safety performance may serve as an effective tool in increasing the safety performance of the observer. The primary purpose of the present study was to assess the effects of conducting safety observations on the postural safety performance of observers engaging in an assembly task for short time periods. The secondary objectives of the study were: (a) to measure productivity, and (b) to measure the accuracy of participant safety observations. Method: An ABC (A: baseline, B: information, C: observation) multiple-baseline design counterbalanced across postural behaviors (back, shoulder, and feet position) was implemented with six participants. Results and Discussion: Substantial improvements in postural safety occurred after participants conducted safety observations, and these improvements did not appear to negatively affect productivity. Results also suggest that there is no relation between the accuracy of an observer's safety observation and their subsequent safety performance. Impact on Industry: This research provides evidence that a safety observation process can function to increase safe postural behavior of observers. Thus, the implementation of such a process may contribute to the prevention of musculoskeletal disorders and related costs in the workplace. © 2008 National Safety Council and Elsevier Ltd. All rights reserved. Keywords: Behavioral safety; Conducting observations; Observer performance; Postural behaviors; Accuracy of observations
1. Introduction In 2005, 1.2 million injuries and illnesses required days away from work3 in the private sector. Manufacturing reported 209,130 (17%) of these cases, while only accounting for 13% of private sector employment. Musculoskeletal disorders4 (MSDs) that often result from reaching, twisting, overexertion, or ⁎ Corresponding author. Tel.: +1 321 720 7051. E-mail addresses: [email protected] (A.M. Alvero), [email protected] (K. Rost), [email protected] (J. Austin). 1 Tel.: +1 718 997 3212. 2 Tel.: +1 269 387 4995. 3 Days away from work include those that result in days away from work with or without restricted work activity (U.S. Bureau of Labor Statistics, 2006). 4 Includes cases where the nature of injury is: sprains, strains, tears; back pain, hurt back; soreness, pain, hurt, except back; carpal tunnel syndrome; hernia; or musculoskeletal system and connective tissue diseases and disorders and when the event or exposure leading to the injury or illness is: bodily reaction/bending, climbing, crawling, reaching, twisting; overexertion; or repetition. Cases of Raynaud's phenomenon, tarsal tunnel syndrome, and herniated spinal discs are not included (U.S. Bureau of Labor Statistics, 2006).
repetition, accounted for 375,540 (30%) of injuries and illnesses with days away from work in 2005, and the manufacturing industry accounted for 69,130 (18%) of MSDs (U.S. Bureau of Labor Statistics, 2006). Undoubtedly, the manufacturing industry would benefit from the deterrence of such injuries and the application of preventative safety processes. Behavioral safety is an approach to improving safety within organizations that uses behavior analysis principles. Behavioral approaches typically include the following components: assessment and identification of behaviors; development and implementation of a behavioral observation process; evaluation of observation data; and execution of a behavioral feedback process. The observation process involves training employees to conduct safety observations using a behavioral checklist. When conducting observations, observers (i.e., trained employees) approach other employees, observe, and score their performance using the checklist. Recent research suggests that the observation process itself may serve as an effective tool in increasing the safety performance of the observer (Alvero & Austin, 2004; Sasson & Austin, 2005). The fundamental
0022-4375/$ - see front matter © 2008 National Safety Council and Elsevier Ltd. All rights reserved. doi:10.1016/j.jsr.2008.05.004
366
A.M. Alvero et al. / Journal of Safety Research 39 (2008) 365–373
question related to conducting behavioral observations is: “Do observers perform more safely as a result of conducting observations across various work settings and tasks?” Alvero and Austin (2004) was the first in a series of studies designed to assess the existence of the observer effect. The researchers used a simulated office setting to evaluate the effects of observation on the behavior of observers engaged in typical office tasks (e.g., typing, lifting, telephone use). During the intervention phase participants observed a video of a confederate performing office tasks and collected data on confederate safety performance using checklists. Results showed that observer safety performance increased substantially after conducting safety observations of confederate performance. Sasson and Austin (2005) examined the effects of conducting safety observations and performance feedback on the safety of clerical workers in a hospital. Sasson and Austin targeted body position while typing (i.e., neck, wrist, back/shoulder, and foot positions). Results showed that safety performance increases were more substantial and were maintained more frequently by participants who conducted safety observations versus those who did not. In addition, the researchers found a high correlation between the level of observer accuracy when conducting observations and the level of safety performance of the observer. The findings of Alvero and Austin (2004) and Sasson and Austin (2005) support the concept that conducting behavioral safety observations can increase the safety performance of observer body positions related to office tasks. It would be informative to replicate the findings of Alvero and Austin and Sasson and Austin with different tasks. Thus, the purposes of the present study were: (a) to replicate the observer effect with participants performing a parts-assembly task rather than office tasks; (b) measure how safety increases affect participant productivity and performance accuracy; and (c) measure how safety observation accuracy affects participant safety performance. 2. Methods 2.1. Participants, Setting, and Materials Four female and two male undergraduate university students participated. Participants ranged from 18 to 35 years of age. The study was conducted in a research laboratory located on a university campus. The laboratory was furnished to resemble an assembly-line workstation. The workstation consisted of a table, a chair, a plastic string approximately five feet in length, and six containers of plastic colored beads. The workstation was equipped with a video camera mounted in the corner of the room, which was used to record all sessions. All observation sessions were 15 minutes in length and each participant averaged about 22 sessions. Participants completed a maximum of two sessions per day. 2.2. Definition and Measurement of Dependent Variables 2.2.1. Safety performance Government ergonomic reports were reviewed in order to determine the appropriate definition for proper sitting position
(National Institute for Occupational Safety and Health [NIOSH], 1998; Office of Health and Safety Information System, 1998). The safe body positions were defined follows: 1. Back upright: back is parallel to the back of the chair, not leaning at an angle against it. 2. Shoulders aligned with back: shoulders are line with the back, not slouched forward. 3. Both feet on the floor: both feet should be flat on the floor, ball of foot and heel should touch floor. Each session was videotaped and a research assistant later scored each session using a 30-second momentary time sampling procedure. That is, every 30 seconds, data were collected for behaviors occurring at that moment. A body position was scored as “safe” when it satisfied the definition listed on the checklist. Research assistants were blind to experimental phase changes. 2.2.2. Productivity performance and accuracy Productivity performance was defined as the number of beads thread onto the string during the 15-minute work session. Productivity accuracy was defined as the percentage of beads thread in the correct color sequence. Data on these variables were collected to ensure that improvements in safety performance did not negatively affect productivity performance. 2.2.3. Accuracy of participant observations During the observation phase, participants were required to collect safety data on a confederate's safety performance using a safety checklist, which listed the relevant target body positions. Participants were instructed to use a 30-second whole interval measurement procedure to collect these data. A behavior was to be scored “safe” if the confederate performed a behavior safely throughout the 30-second interval. To calculate accuracy data, the primary researcher collected data on all confederate performance. Accuracy of participant observations was calculated as follows: number of agreements / number of agreements +disagreements × 100. An agreement was defined as any occurrence in which both the primary researcher and participant scored the same mark (safe or unsafe) for a body position. Observation accuracy was calculated for each target body position measured by the participant. 2.3. Interobserver Agreement Interobserver agreement was calculated on the target body positions. As a reliability check, the primary researcher scored 30% of all sessions using the same checklist as the research assistants. Interobserver agreement between the primary researcher and the research assistant was calculated as follows: number of agreements / number of agreements + disagreements × 100. An agreement was defined as any occurrence in which both researchers scored the same mark (safe or unsafe) for a body position for each corresponding interval.
A.M. Alvero et al. / Journal of Safety Research 39 (2008) 365–373
367
information phase was to eliminate demand characteristics that are often displayed by participants taking part in a laboratory study (Kazdin, 1992).
2.4. Procedures 2.4.1. Baseline At the start of each baseline session, all participants were handed a list of instructions that described an assembly task, which they were then asked to perform. The assembly task involved stringing beads onto a plastic string in a specific color order. This task was repeated throughout the 15-minute session, thus simulating the work a person might perform on a manufacturing assembly line. Participants remained in this phase until safety performance stabilized; that is, data did not show an upward trend or extreme variability.
2.4.3. Observation phase At the start of each session during the observation phase, participants were asked to observe and score a 5-minute video of a confederate performing the same assembly task the participant performed during each session. Participants were asked to collect data on the same target behavior(s) introduced in the information phase (group A: feet position; group B: back and shoulder position) while they observed the video. Participants scored confederate safety behavior using a checklist comprised of the relevant target behaviors and definitions of how to perform them safely. After conducting the 5-minute observation, participants were given a list of assembly task instructions to perform, and for the remainder of the session procedures were the same as baseline. After safety performance stabilized on the first target behavior(s) introduced to the observation phase, the remaining target behavior(s) were added to the observation checklist. Therefore, participants were asked to collect data for all three behaviors during the second portion of the observation phase.
2.4.2. Information phase During the first session of the information phase, all participants were informed about the general nature of the study (to measure safety performance). At the start of each session throughout this phase participants were given a handout containing definitions for one, or two, of the three target behaviors and how to perform them safely. Three participants comprised “Group A” and received information on one target behavior (feet position when sitting). The remaining three participants comprised “Group B” and received information on the other two target behaviors (back and shoulder position when sitting). Information was not provided to all participants for all target behaviors in order to examine the effects of the observation phase without a preceding information phase. The information provided was counterbalanced across Groups A and B. Participants were required to review the information for five minutes before they were handed the list of tasks to perform. The remainder of the session followed the same procedures as those during baseline. Participants remained in this phase until performance stabilized. The purpose of the
2.5. Integrity of the Independent Variables The videos shown to the participants were kept in a specific order to ensure that all participants were exposed to the same video sequence. Participants were required to initial and date the information sheets and collect confederate safety performance data directly onto the observation checklists. This provided verification of exposure to each intervention.
Table 1 A summary of mean percent safe, standard deviation, and the number of sessions for all participants across all experimental phases Group
Participant
Behavior
Baseline M
SD
n
M
SD
n
M
SD
n
A
1
Feet Back Shoulders Feet Back Shoulders Feet Back Shoulders Feet Back Shoulders Feet Back Shoulders Feet Back Shoulders
13.3 1 0.4 0 1 1 33.3 0.8 0.3 8.4 6.3 0.5 6.7 2.4 0 73.2 2.5 2.5
32.7 2 1.1 0 3.3 3.3 45.3 2.1 0.9 26.5 8.6 0 25.8 2.5 0 42.8 3.3 3.3
6 14 14 3 16 16 4 12 12 17 6 6 15 5 5 16 4 4
95.9 79 99 98.6 95 4.3 0 12.4 12.4
4.4 35.9 1.7 2.7 7.1 5.1 0 15.4 15.4
5 7 3 7 7 3 3 7 7
100 54.1 48.6 95.7 89.8 89.8 99.8 99.2 99.5 59.4 98.9 98.9 92 82.5 78.5 100 98.1 98.1
0 35.48 35.2 12.5 14.3 14.3 0.8 1.4 1.2 54.2 2.4 2.4 7.6 17.4 17.5 0 3.8 3.8
11 8 8 14 8 8 16 11 11 5 9 9 5 12 12 4 9 9
2
3
B
4
5
6
Note. Dashes indicate the data were not collected for that measure.
Information
Observation
368
A.M. Alvero et al. / Journal of Safety Research 39 (2008) 365–373
2.7. Data Analysis For between phase comparisons of safety performance, standardized effect size was calculated using Cohen's (d) formula for effect size (Cohen, 1988). In addition, Pearson correlation coefficient (r) was calculated to determine the correlation between safe postural performance and observation accuracy. 3. Results Overall, the observation phase produced the highest safety levels compared to baseline for all participants across all postural behaviors (see Table 1). For the most part, once participants began conducting observations, the increases in safety performance maintained throughout the remainder of the study (see Figs. 1–6). 3.1. Safety Performance Table 1 summarizes the means and standard deviations of safety performance for all participants across experimental phases. The top portions of Figs. 1 through 6 display participant
Fig. 1. Performance data for participant 1A.
2.6. Experimental Design A within-subjects, ABC (A: baseline, B: information, C: observation) multiple-baseline design was used. The observation phase was first implemented for one target behavior for participants in group A and for the remaining two target behaviors for participants in group B. After performance on the first behavior(s) stabilized, the next behavior(s) were exposed to the observation phase for both groups.
Fig. 2. Performance data for participant 2A.
A.M. Alvero et al. / Journal of Safety Research 39 (2008) 365–373
369
3.3. Accuracy of Observations Table 3 summarizes the overall means and standard deviations of observation accuracy for each participant across the three target behaviors. All participants averaged the highest observation accuracy for feet, followed by shoulders and back, respectively. Correlations were calculated between the percent safe score and the accuracy score per session and per behavior (e.g., the participant's percent safe for feet position during session 1 was correlated with the accuracy score obtained on the observation for feet position conducted during session 1). The Pearson correlation coefficient (r) between percent safe and accuracy was r = 0.2 for participant 1A; r = − 0.1 for participant 2A; r = 0.1 for participant 3A; r = − 0.1 for participant 4B; r = − 0.2 for participant 5B; and r = − 0.1 for participant 6B. See Fig. 7 for scatter plots of these correlations. 4. Discussion 4.1. Effects of Safety Information The magnitude of the effect sizes from baseline to safety information varied across participants and behaviors. Participants
Fig. 3. Performance data for participant 3A.
safety performance during the course of the experiment in a multiple baseline fashion. For further comparison of the safety levels between phases, effect sizes were calculated. Table 4 summarizes the effect sizes between each phase for all participants and behaviors. Effect sizes that range from 0.2 to 0.49 are considered small; effect sizes that range from 0.5–0.79 are considered medium; and effect sizes of 0.8 or greater are considered large (Cohen, 1988). In this study, effect sizes ranged from very small (0.12) to very large (94.16). Shoulder position for participant 5B remained at 0% safe from baseline to information, and was the only instance in which behavior change was not observed between phases. 3.2. Productivity Performance and Accuracy Table 2 provides a summary of the overall means and standard deviations for the productivity and accuracy performance of the assembly task. Each participant averaged 99% or higher accuracy when performing the task. Productivity ranged from a mean of 270 (participant 5B) to 405 (participant 4B) beads strung per session.
Fig. 4. Performance data for participant 4B.
370
A.M. Alvero et al. / Journal of Safety Research 39 (2008) 365–373
Fig. 5. Performance data for participant 5B.
Fig. 6. Performance data for participant 6B.
1A, 2A, and 3A were exposed to safety information for feet position, which resulted in large effect sizes for all three participants. The consistent and large effect sizes from baseline to safety information for feet position corroborate previous research findings (Alvero & Austin, 2004). It is likely that these sizeable improvements occurred because it may be easy for participants to modify feet position. Unlike feet position, the effect sizes from baseline to safety information for back and shoulder positions were not consistent across participants. For participant 4B, safety information resulted in large effect sizes for back and shoulder positions. For participant 5B, safety information resulted in a medium effect size for back position and no effect for shoulder position. For participant 6B safety information resulted in medium effect sizes for both back and shoulder positions. With the exception of participant 4B, participants who were exposed to safety information for back and shoulder positions did not modify their behavior so that it was consistently safe.
tion. For all three participants, the effect sizes from baseline to the observation phase for back and shoulder positions were large, and the effect sizes from the safety information to the observation phase for feet position were medium to large. Participants 4B, 5B, and 6B conducted observations for feet position directly after baseline, and conducted observations for back and shoulder positions after exposure to safety information. For participants 4B and 5B, effect sizes from baseline to observation for feet position were large, and for participant 6B the effect size from baseline to observation for feet position was medium. Although conducting observations resulted in a large
4.2. Effects of Conducting Observations Participants 1A, 2A, and 3A conducted observations for back and shoulder positions directly after baseline, and conducted observations for feet position after exposure to safety informa-
Table 2 A summary of the overall mean and standard deviation of productivity and accuracy, and the number of sessions for each participant Group
A
B
Participant
1 2 3 4 5 6
Productivity Performance
Productivity Accuracy
M
SD
n
M
SD
n
403 308 301 405 270 287
42.4 30.8 20.3 34 50.5 24.5
22 24 23 22 20 20
99.9 99.8 99.9 99.9 99.7 99.9
0.3 0.4 0.3 0.2 0.7 0.3
22 24 23 22 20 20
A.M. Alvero et al. / Journal of Safety Research 39 (2008) 365–373 Table 3 A summary of the overall mean and standard deviation of observation accuracy, and the number of sessions per observation phase for each participant across all behaivors Group
A
B
Participant
1 2 3 4 5 6
Feet
Back
Shoulders
M
SD
n
M
SD
n
M
SD
n
91.8 96.9 96.9 96 94 97.5
14 8.3 7 8.9 8.9 5
11 14 16 5 5 4
81.3 77.5 82.7 76.7 76.7 90
22.3 36.6 28 39.4 32 26.5
8 8 11 9 12 9
83.8 90 90 81.1 78.3 100
14 21.4 19.5 33 29.8 0
8 8 11 9 12 9
effect size over baseline for participant 4B's feet position, the performance trend during the observation phase varies from all other targeted behaviors throughout the study and merits some explanation. During the first two sessions within the observation
371
phase, 4B's safe feet position averaged 0% and then jumped to 97% during the third session (session #20). Although such a dramatic increase during the observation phase is not unusual, it typically occurs during the first session. Before the start of session 20, participant 4B asked an experimenter to help her lower her chair because she could not remember how to do so. Once the participant was helped, her safe feet performance dramatically increased and remained high through the completion of the study. 4.3. Productivity Performance and Accuracy Productivity performance and accuracy data were primarily collected to ensure that an inverse relationship did not exist between safety and productivity. The results do not indicate the existence of any such inverse relationship. Implementation of the information phase and the observation phase appeared to have no effect on productivity performance and accuracy. The productivity performance and accuracy data collected
Fig. 7. The relationship between observer accuracy and percent safe scores for participants 1A, 2A, 3A, 4B, 5B, and 6B.
372
A.M. Alvero et al. / Journal of Safety Research 39 (2008) 365–373
Table 4 Effect size between experimental phases across all participants and behaviors Group Participant Behavior
A
1
2
3
B
4
5
6
Feet Back Shoulders Feet Back Shoulders Feet Back Shoulders Feet Back Shoulders Feet Back Shoulders Feet Back Shoulders
d Baseline to Information
Baseline to Observation
Information to Observation
3.36 2.54 1.87 15.05 18.02 0.53 0 0.78 0.78
4.59 2.53 2.31 8.22 10.43 10.43 3.59 24.57 94.16 1.5 16.37 52.26 3.7 5.36 5.24 0.69 26.05 26.05
1.74 0.74 0.84 0.12 0.78 4.85 4.88 8.18 8.18
Note. Dashes indicate the data were not collected for that measure.
during this experiment may be interpreted as a measure of social validity. An intervention, such as conducting safety observations within a behavioral safety process, may not be accepted by organizations if increases in safety resulted in decreases in performance. These productivity data provide some empirical evidence to suggest that productivity remains unchanged by improvements in postural safety behavior. These findings are specific to the postural behavior observed in this study, however. Future research could examine whether productivity performance is impeded by increases in other safety-related behaviors for which performing safely is more likely to result in increased response effort and/or task completion time. 4.4. Accuracy of Participant Observations The accuracy of participant observations observed does not appear to be correlated with safety performance (See Fig. 7). This general conclusion was drawn based on the lack of observable trends between variables and the results of the Pearson product correlations, but no formal deduction can be made based on these observations because no experimental manipulations were made to address this issue. Despite this, the data do present interesting implications for the implementation of behavioral safety processes in applied settings. Typical behavioral safety applications (Krause, 1997; McSween, 2003) require the safety observer to deliver safety performance feedback to the person observed. The observation accuracy data collected throughout this research seem to indicate that, for some behaviors, participants may be unable to correctly identify differences between safe and unsafe performance, yet are able to perform the behaviors safely themselves. In other words, the person being observed may receive inaccurate safety performance feedback despite the fact that the observer can correctly
perform the behaviors. The extent to which participants could accurately score the safety performances of others may have not been a skill problem, however. No contingencies were in place that required accurate scoring, thus it is not clear if participants lacked the skills to discriminate between safe and unsafe behaviors or if they lacked the motivation to do so. 4.5. Strengths and Limitations The present study successfully replicated the safety observer effect using a manufacturing-like task and produced substantial improvements of the targeted behaviors. Other strengths of the present research are the data collected on productivity performance and observation accuracy. The data collected on the productivity performance provide social validity and the data on the accuracy of participant observations provide some ideas for future research. For example, future research should more directly examine the effects of an observer's data collection accuracy on the observer's own performance. As previously mentioned, the observation accuracy data seem to indicate that participants may be unable to correctly discriminate between safe and unsafe performance, yet are able to improve their own performance. This seems to indicate that safety observers may be delivering unreliable safety performance feedback to those being observed. Therefore, determining if improvements to an observer's accuracy further improve their own safety performance would be of practical value. Sasson and Austin (2005) explored this issue indirectly and found a strong relationship between accuracy and performance improvement for some, but not all, observers. The limitations of the study are related to the parameters of the laboratory setting and participants. The experimental sessions lasted only 15 minutes, a work-time duration that is not analogous to a typical workday in a manufacturing setting. In addition, the participants were college students in a laboratory setting and it is not clear if such effects would generalize to a real-world setting where there are many competing contingencies (e.g., productivity deadlines, peer pressures, environmental issues). Sasson and Austin (2005) did replicate these effects on postural safety performance in an applied office setting; however, it would be beneficial if future applied research in this area targeted non-postural behaviors or postural behaviors in non-office settings. Another difference between an applied setting and the laboratory is participant knowledge that they are constantly being observed via a camera. Although Sasson and Austin (2005) replicated the effects in an applied setting, the participants were aware when their behavior was being observed. The safety observer effect occurs when an observer's safety performance increases as a result of conducting a safety observation and then they themselves are being observed. In other words, would this observer effect occur if the safety observer, or participant, were unaware that their performance was being monitored? This is an important question that should be addressed in order to strengthen the validity of this effect. The improvements in safety observed in this study should be considered with all of the above limitations in mind.
A.M. Alvero et al. / Journal of Safety Research 39 (2008) 365–373
In conclusion, the current research helps support the existing research on the effects of conducting safety observations on the performance of the observer. The results of the present study also provide some suggestions for interesting and noteworthy future research. Such research could help safety practitioners modify their present approach toward improving safety to ensure the development of the most efficient and effective process.
373
National Institute for Occupational Safety and Health [NIOSH]. (1998). OSHA Ergonomics Report DT931018 [Data file]. Available from: http://www.osha. lc.gov/SLTC/ergonomics/ergo-hiosh.html Office of Health and Safety Information System. (1998). Safety manuals [Data file]. Available from: http://www.cdc.gov/od/ohs/default.htm Sasson, J. R., & Austin, J. (2005). The effects of information, feedback, and conducting behavioral safety observations on office ergonomic behavior. Journal of Organizational Behavior Management, 24(4), 1−30. U.S. Bureau Labor of Statistics. (2006). Nonfatal occupational injuries and illnesses requiring days away from work, 2006 [Data file]. Available from: http://www.bls.gov
4.6. Impact on Industry In general, the results of this research suggest that if employees conduct safety observations of peer safety performance, then their own safety performance may improve. Specifically, the results of this study show that the safety of at-risk body positions associated with the development of MSDs improve after conducting safety observations. Employees in the manufacturing sector are at particular risk for developing MSDs, and may benefit from the implementation of behavioral safety programs that involve peer safety observations. In addition, these results suggest that safety professionals and employers can be confident that increases in postural safety will not decrease worker productivity. References Alvero, A. M., & Austin, J. (2004). The effects of conducting behavioral observations on the behavior of the observer. Journal of Applied Behavior Analysis, 37, 457−468. Cohen, J. (1988). Statistical power analyses for the behavioral sciences, 2nd ed. Hillsdale, NJ: L. Erlbaum Associates. Kazdin, A. E. (1992). Methodological issues and strategies in clinical research. Washington, DC: American Psychological Association. Krause, T. R. (1997). The behavior-based safety process. New York: Van Nostrand Reinhold. McSween, T. E. (2003). The values-based safety process, 2nd ed. New York: John Wiley & Sons, Inc.
Dr. Alvero is an assistant professor of psychology at Queens College, the City University of New York, where she teaches courses in organizational behavior management and behavioral safety. She holds a B.A. from Florida International University and an M.A. and Ph.D. from Western Michigan University. Dr. Alvero is an editorial board member of the Journal of Organizational Behavior Management and the Journal of Safety Research, and has consulted with various organizations in the areas of training and behavioral safety. Kristen Rost is a doctoral candidate in the Learning Processes and Behavior Analysis program at Queens College and the Graduate Center, the City University of New York, where she teaches courses in experimental psychology and industrial/organizational psychology. She holds a B.S. from Western Michigan University and an M.S. from Florida Institute of Technology. Kristen is the Queens College student committee program representative for the Association for Behavior Analysis, and conducts research in the area of behavioral safety. Dr. Austin is an associate professor of psychology at Western Michigan University, where he teaches courses in performance management and consults with large and small business on behavioral safety and performance improvement systems. He holds a B.A. from the University of Notre Dame and an M.S. and Ph.D. from Florida State University. Dr. Austin is co-editor of the Journal of Organizational Behavior Management, an editorial board member for the Journal of Applied Behavior Analysis, and the director of the OBM Network (www.obmnetwork.com).
www.elsevier.com/locate/jsr
www.nsc.org
The safety observer effect: The effects of conducting safety observations Alicia M. Alvero a,1 , Kristen Rost c,⁎, John Austin b,2 a
c
Queens College, CUNY, Department of Psychology, 65-30 Kissena Blvd., Flushing, NY 11367 b Western Michigan University, Department of Psychology, Kalamazoo, MI 49008 Queens College and The Graduate Center, CUNY, Department of Psychology, 65-30 Kissena Blvd., Flushing, NY 11367 Available online 26 July 2008
Abstract Introduction: Some research suggests that conducting safety observations of another's safety performance may serve as an effective tool in increasing the safety performance of the observer. The primary purpose of the present study was to assess the effects of conducting safety observations on the postural safety performance of observers engaging in an assembly task for short time periods. The secondary objectives of the study were: (a) to measure productivity, and (b) to measure the accuracy of participant safety observations. Method: An ABC (A: baseline, B: information, C: observation) multiple-baseline design counterbalanced across postural behaviors (back, shoulder, and feet position) was implemented with six participants. Results and Discussion: Substantial improvements in postural safety occurred after participants conducted safety observations, and these improvements did not appear to negatively affect productivity. Results also suggest that there is no relation between the accuracy of an observer's safety observation and their subsequent safety performance. Impact on Industry: This research provides evidence that a safety observation process can function to increase safe postural behavior of observers. Thus, the implementation of such a process may contribute to the prevention of musculoskeletal disorders and related costs in the workplace. © 2008 National Safety Council and Elsevier Ltd. All rights reserved. Keywords: Behavioral safety; Conducting observations; Observer performance; Postural behaviors; Accuracy of observations
1. Introduction In 2005, 1.2 million injuries and illnesses required days away from work3 in the private sector. Manufacturing reported 209,130 (17%) of these cases, while only accounting for 13% of private sector employment. Musculoskeletal disorders4 (MSDs) that often result from reaching, twisting, overexertion, or ⁎ Corresponding author. Tel.: +1 321 720 7051. E-mail addresses: [email protected] (A.M. Alvero), [email protected] (K. Rost), [email protected] (J. Austin). 1 Tel.: +1 718 997 3212. 2 Tel.: +1 269 387 4995. 3 Days away from work include those that result in days away from work with or without restricted work activity (U.S. Bureau of Labor Statistics, 2006). 4 Includes cases where the nature of injury is: sprains, strains, tears; back pain, hurt back; soreness, pain, hurt, except back; carpal tunnel syndrome; hernia; or musculoskeletal system and connective tissue diseases and disorders and when the event or exposure leading to the injury or illness is: bodily reaction/bending, climbing, crawling, reaching, twisting; overexertion; or repetition. Cases of Raynaud's phenomenon, tarsal tunnel syndrome, and herniated spinal discs are not included (U.S. Bureau of Labor Statistics, 2006).
repetition, accounted for 375,540 (30%) of injuries and illnesses with days away from work in 2005, and the manufacturing industry accounted for 69,130 (18%) of MSDs (U.S. Bureau of Labor Statistics, 2006). Undoubtedly, the manufacturing industry would benefit from the deterrence of such injuries and the application of preventative safety processes. Behavioral safety is an approach to improving safety within organizations that uses behavior analysis principles. Behavioral approaches typically include the following components: assessment and identification of behaviors; development and implementation of a behavioral observation process; evaluation of observation data; and execution of a behavioral feedback process. The observation process involves training employees to conduct safety observations using a behavioral checklist. When conducting observations, observers (i.e., trained employees) approach other employees, observe, and score their performance using the checklist. Recent research suggests that the observation process itself may serve as an effective tool in increasing the safety performance of the observer (Alvero & Austin, 2004; Sasson & Austin, 2005). The fundamental
0022-4375/$ - see front matter © 2008 National Safety Council and Elsevier Ltd. All rights reserved. doi:10.1016/j.jsr.2008.05.004
366
A.M. Alvero et al. / Journal of Safety Research 39 (2008) 365–373
question related to conducting behavioral observations is: “Do observers perform more safely as a result of conducting observations across various work settings and tasks?” Alvero and Austin (2004) was the first in a series of studies designed to assess the existence of the observer effect. The researchers used a simulated office setting to evaluate the effects of observation on the behavior of observers engaged in typical office tasks (e.g., typing, lifting, telephone use). During the intervention phase participants observed a video of a confederate performing office tasks and collected data on confederate safety performance using checklists. Results showed that observer safety performance increased substantially after conducting safety observations of confederate performance. Sasson and Austin (2005) examined the effects of conducting safety observations and performance feedback on the safety of clerical workers in a hospital. Sasson and Austin targeted body position while typing (i.e., neck, wrist, back/shoulder, and foot positions). Results showed that safety performance increases were more substantial and were maintained more frequently by participants who conducted safety observations versus those who did not. In addition, the researchers found a high correlation between the level of observer accuracy when conducting observations and the level of safety performance of the observer. The findings of Alvero and Austin (2004) and Sasson and Austin (2005) support the concept that conducting behavioral safety observations can increase the safety performance of observer body positions related to office tasks. It would be informative to replicate the findings of Alvero and Austin and Sasson and Austin with different tasks. Thus, the purposes of the present study were: (a) to replicate the observer effect with participants performing a parts-assembly task rather than office tasks; (b) measure how safety increases affect participant productivity and performance accuracy; and (c) measure how safety observation accuracy affects participant safety performance. 2. Methods 2.1. Participants, Setting, and Materials Four female and two male undergraduate university students participated. Participants ranged from 18 to 35 years of age. The study was conducted in a research laboratory located on a university campus. The laboratory was furnished to resemble an assembly-line workstation. The workstation consisted of a table, a chair, a plastic string approximately five feet in length, and six containers of plastic colored beads. The workstation was equipped with a video camera mounted in the corner of the room, which was used to record all sessions. All observation sessions were 15 minutes in length and each participant averaged about 22 sessions. Participants completed a maximum of two sessions per day. 2.2. Definition and Measurement of Dependent Variables 2.2.1. Safety performance Government ergonomic reports were reviewed in order to determine the appropriate definition for proper sitting position
(National Institute for Occupational Safety and Health [NIOSH], 1998; Office of Health and Safety Information System, 1998). The safe body positions were defined follows: 1. Back upright: back is parallel to the back of the chair, not leaning at an angle against it. 2. Shoulders aligned with back: shoulders are line with the back, not slouched forward. 3. Both feet on the floor: both feet should be flat on the floor, ball of foot and heel should touch floor. Each session was videotaped and a research assistant later scored each session using a 30-second momentary time sampling procedure. That is, every 30 seconds, data were collected for behaviors occurring at that moment. A body position was scored as “safe” when it satisfied the definition listed on the checklist. Research assistants were blind to experimental phase changes. 2.2.2. Productivity performance and accuracy Productivity performance was defined as the number of beads thread onto the string during the 15-minute work session. Productivity accuracy was defined as the percentage of beads thread in the correct color sequence. Data on these variables were collected to ensure that improvements in safety performance did not negatively affect productivity performance. 2.2.3. Accuracy of participant observations During the observation phase, participants were required to collect safety data on a confederate's safety performance using a safety checklist, which listed the relevant target body positions. Participants were instructed to use a 30-second whole interval measurement procedure to collect these data. A behavior was to be scored “safe” if the confederate performed a behavior safely throughout the 30-second interval. To calculate accuracy data, the primary researcher collected data on all confederate performance. Accuracy of participant observations was calculated as follows: number of agreements / number of agreements +disagreements × 100. An agreement was defined as any occurrence in which both the primary researcher and participant scored the same mark (safe or unsafe) for a body position. Observation accuracy was calculated for each target body position measured by the participant. 2.3. Interobserver Agreement Interobserver agreement was calculated on the target body positions. As a reliability check, the primary researcher scored 30% of all sessions using the same checklist as the research assistants. Interobserver agreement between the primary researcher and the research assistant was calculated as follows: number of agreements / number of agreements + disagreements × 100. An agreement was defined as any occurrence in which both researchers scored the same mark (safe or unsafe) for a body position for each corresponding interval.
A.M. Alvero et al. / Journal of Safety Research 39 (2008) 365–373
367
information phase was to eliminate demand characteristics that are often displayed by participants taking part in a laboratory study (Kazdin, 1992).
2.4. Procedures 2.4.1. Baseline At the start of each baseline session, all participants were handed a list of instructions that described an assembly task, which they were then asked to perform. The assembly task involved stringing beads onto a plastic string in a specific color order. This task was repeated throughout the 15-minute session, thus simulating the work a person might perform on a manufacturing assembly line. Participants remained in this phase until safety performance stabilized; that is, data did not show an upward trend or extreme variability.
2.4.3. Observation phase At the start of each session during the observation phase, participants were asked to observe and score a 5-minute video of a confederate performing the same assembly task the participant performed during each session. Participants were asked to collect data on the same target behavior(s) introduced in the information phase (group A: feet position; group B: back and shoulder position) while they observed the video. Participants scored confederate safety behavior using a checklist comprised of the relevant target behaviors and definitions of how to perform them safely. After conducting the 5-minute observation, participants were given a list of assembly task instructions to perform, and for the remainder of the session procedures were the same as baseline. After safety performance stabilized on the first target behavior(s) introduced to the observation phase, the remaining target behavior(s) were added to the observation checklist. Therefore, participants were asked to collect data for all three behaviors during the second portion of the observation phase.
2.4.2. Information phase During the first session of the information phase, all participants were informed about the general nature of the study (to measure safety performance). At the start of each session throughout this phase participants were given a handout containing definitions for one, or two, of the three target behaviors and how to perform them safely. Three participants comprised “Group A” and received information on one target behavior (feet position when sitting). The remaining three participants comprised “Group B” and received information on the other two target behaviors (back and shoulder position when sitting). Information was not provided to all participants for all target behaviors in order to examine the effects of the observation phase without a preceding information phase. The information provided was counterbalanced across Groups A and B. Participants were required to review the information for five minutes before they were handed the list of tasks to perform. The remainder of the session followed the same procedures as those during baseline. Participants remained in this phase until performance stabilized. The purpose of the
2.5. Integrity of the Independent Variables The videos shown to the participants were kept in a specific order to ensure that all participants were exposed to the same video sequence. Participants were required to initial and date the information sheets and collect confederate safety performance data directly onto the observation checklists. This provided verification of exposure to each intervention.
Table 1 A summary of mean percent safe, standard deviation, and the number of sessions for all participants across all experimental phases Group
Participant
Behavior
Baseline M
SD
n
M
SD
n
M
SD
n
A
1
Feet Back Shoulders Feet Back Shoulders Feet Back Shoulders Feet Back Shoulders Feet Back Shoulders Feet Back Shoulders
13.3 1 0.4 0 1 1 33.3 0.8 0.3 8.4 6.3 0.5 6.7 2.4 0 73.2 2.5 2.5
32.7 2 1.1 0 3.3 3.3 45.3 2.1 0.9 26.5 8.6 0 25.8 2.5 0 42.8 3.3 3.3
6 14 14 3 16 16 4 12 12 17 6 6 15 5 5 16 4 4
95.9 79 99 98.6 95 4.3 0 12.4 12.4
4.4 35.9 1.7 2.7 7.1 5.1 0 15.4 15.4
5 7 3 7 7 3 3 7 7
100 54.1 48.6 95.7 89.8 89.8 99.8 99.2 99.5 59.4 98.9 98.9 92 82.5 78.5 100 98.1 98.1
0 35.48 35.2 12.5 14.3 14.3 0.8 1.4 1.2 54.2 2.4 2.4 7.6 17.4 17.5 0 3.8 3.8
11 8 8 14 8 8 16 11 11 5 9 9 5 12 12 4 9 9
2
3
B
4
5
6
Note. Dashes indicate the data were not collected for that measure.
Information
Observation
368
A.M. Alvero et al. / Journal of Safety Research 39 (2008) 365–373
2.7. Data Analysis For between phase comparisons of safety performance, standardized effect size was calculated using Cohen's (d) formula for effect size (Cohen, 1988). In addition, Pearson correlation coefficient (r) was calculated to determine the correlation between safe postural performance and observation accuracy. 3. Results Overall, the observation phase produced the highest safety levels compared to baseline for all participants across all postural behaviors (see Table 1). For the most part, once participants began conducting observations, the increases in safety performance maintained throughout the remainder of the study (see Figs. 1–6). 3.1. Safety Performance Table 1 summarizes the means and standard deviations of safety performance for all participants across experimental phases. The top portions of Figs. 1 through 6 display participant
Fig. 1. Performance data for participant 1A.
2.6. Experimental Design A within-subjects, ABC (A: baseline, B: information, C: observation) multiple-baseline design was used. The observation phase was first implemented for one target behavior for participants in group A and for the remaining two target behaviors for participants in group B. After performance on the first behavior(s) stabilized, the next behavior(s) were exposed to the observation phase for both groups.
Fig. 2. Performance data for participant 2A.
A.M. Alvero et al. / Journal of Safety Research 39 (2008) 365–373
369
3.3. Accuracy of Observations Table 3 summarizes the overall means and standard deviations of observation accuracy for each participant across the three target behaviors. All participants averaged the highest observation accuracy for feet, followed by shoulders and back, respectively. Correlations were calculated between the percent safe score and the accuracy score per session and per behavior (e.g., the participant's percent safe for feet position during session 1 was correlated with the accuracy score obtained on the observation for feet position conducted during session 1). The Pearson correlation coefficient (r) between percent safe and accuracy was r = 0.2 for participant 1A; r = − 0.1 for participant 2A; r = 0.1 for participant 3A; r = − 0.1 for participant 4B; r = − 0.2 for participant 5B; and r = − 0.1 for participant 6B. See Fig. 7 for scatter plots of these correlations. 4. Discussion 4.1. Effects of Safety Information The magnitude of the effect sizes from baseline to safety information varied across participants and behaviors. Participants
Fig. 3. Performance data for participant 3A.
safety performance during the course of the experiment in a multiple baseline fashion. For further comparison of the safety levels between phases, effect sizes were calculated. Table 4 summarizes the effect sizes between each phase for all participants and behaviors. Effect sizes that range from 0.2 to 0.49 are considered small; effect sizes that range from 0.5–0.79 are considered medium; and effect sizes of 0.8 or greater are considered large (Cohen, 1988). In this study, effect sizes ranged from very small (0.12) to very large (94.16). Shoulder position for participant 5B remained at 0% safe from baseline to information, and was the only instance in which behavior change was not observed between phases. 3.2. Productivity Performance and Accuracy Table 2 provides a summary of the overall means and standard deviations for the productivity and accuracy performance of the assembly task. Each participant averaged 99% or higher accuracy when performing the task. Productivity ranged from a mean of 270 (participant 5B) to 405 (participant 4B) beads strung per session.
Fig. 4. Performance data for participant 4B.
370
A.M. Alvero et al. / Journal of Safety Research 39 (2008) 365–373
Fig. 5. Performance data for participant 5B.
Fig. 6. Performance data for participant 6B.
1A, 2A, and 3A were exposed to safety information for feet position, which resulted in large effect sizes for all three participants. The consistent and large effect sizes from baseline to safety information for feet position corroborate previous research findings (Alvero & Austin, 2004). It is likely that these sizeable improvements occurred because it may be easy for participants to modify feet position. Unlike feet position, the effect sizes from baseline to safety information for back and shoulder positions were not consistent across participants. For participant 4B, safety information resulted in large effect sizes for back and shoulder positions. For participant 5B, safety information resulted in a medium effect size for back position and no effect for shoulder position. For participant 6B safety information resulted in medium effect sizes for both back and shoulder positions. With the exception of participant 4B, participants who were exposed to safety information for back and shoulder positions did not modify their behavior so that it was consistently safe.
tion. For all three participants, the effect sizes from baseline to the observation phase for back and shoulder positions were large, and the effect sizes from the safety information to the observation phase for feet position were medium to large. Participants 4B, 5B, and 6B conducted observations for feet position directly after baseline, and conducted observations for back and shoulder positions after exposure to safety information. For participants 4B and 5B, effect sizes from baseline to observation for feet position were large, and for participant 6B the effect size from baseline to observation for feet position was medium. Although conducting observations resulted in a large
4.2. Effects of Conducting Observations Participants 1A, 2A, and 3A conducted observations for back and shoulder positions directly after baseline, and conducted observations for feet position after exposure to safety informa-
Table 2 A summary of the overall mean and standard deviation of productivity and accuracy, and the number of sessions for each participant Group
A
B
Participant
1 2 3 4 5 6
Productivity Performance
Productivity Accuracy
M
SD
n
M
SD
n
403 308 301 405 270 287
42.4 30.8 20.3 34 50.5 24.5
22 24 23 22 20 20
99.9 99.8 99.9 99.9 99.7 99.9
0.3 0.4 0.3 0.2 0.7 0.3
22 24 23 22 20 20
A.M. Alvero et al. / Journal of Safety Research 39 (2008) 365–373 Table 3 A summary of the overall mean and standard deviation of observation accuracy, and the number of sessions per observation phase for each participant across all behaivors Group
A
B
Participant
1 2 3 4 5 6
Feet
Back
Shoulders
M
SD
n
M
SD
n
M
SD
n
91.8 96.9 96.9 96 94 97.5
14 8.3 7 8.9 8.9 5
11 14 16 5 5 4
81.3 77.5 82.7 76.7 76.7 90
22.3 36.6 28 39.4 32 26.5
8 8 11 9 12 9
83.8 90 90 81.1 78.3 100
14 21.4 19.5 33 29.8 0
8 8 11 9 12 9
effect size over baseline for participant 4B's feet position, the performance trend during the observation phase varies from all other targeted behaviors throughout the study and merits some explanation. During the first two sessions within the observation
371
phase, 4B's safe feet position averaged 0% and then jumped to 97% during the third session (session #20). Although such a dramatic increase during the observation phase is not unusual, it typically occurs during the first session. Before the start of session 20, participant 4B asked an experimenter to help her lower her chair because she could not remember how to do so. Once the participant was helped, her safe feet performance dramatically increased and remained high through the completion of the study. 4.3. Productivity Performance and Accuracy Productivity performance and accuracy data were primarily collected to ensure that an inverse relationship did not exist between safety and productivity. The results do not indicate the existence of any such inverse relationship. Implementation of the information phase and the observation phase appeared to have no effect on productivity performance and accuracy. The productivity performance and accuracy data collected
Fig. 7. The relationship between observer accuracy and percent safe scores for participants 1A, 2A, 3A, 4B, 5B, and 6B.
372
A.M. Alvero et al. / Journal of Safety Research 39 (2008) 365–373
Table 4 Effect size between experimental phases across all participants and behaviors Group Participant Behavior
A
1
2
3
B
4
5
6
Feet Back Shoulders Feet Back Shoulders Feet Back Shoulders Feet Back Shoulders Feet Back Shoulders Feet Back Shoulders
d Baseline to Information
Baseline to Observation
Information to Observation
3.36 2.54 1.87 15.05 18.02 0.53 0 0.78 0.78
4.59 2.53 2.31 8.22 10.43 10.43 3.59 24.57 94.16 1.5 16.37 52.26 3.7 5.36 5.24 0.69 26.05 26.05
1.74 0.74 0.84 0.12 0.78 4.85 4.88 8.18 8.18
Note. Dashes indicate the data were not collected for that measure.
during this experiment may be interpreted as a measure of social validity. An intervention, such as conducting safety observations within a behavioral safety process, may not be accepted by organizations if increases in safety resulted in decreases in performance. These productivity data provide some empirical evidence to suggest that productivity remains unchanged by improvements in postural safety behavior. These findings are specific to the postural behavior observed in this study, however. Future research could examine whether productivity performance is impeded by increases in other safety-related behaviors for which performing safely is more likely to result in increased response effort and/or task completion time. 4.4. Accuracy of Participant Observations The accuracy of participant observations observed does not appear to be correlated with safety performance (See Fig. 7). This general conclusion was drawn based on the lack of observable trends between variables and the results of the Pearson product correlations, but no formal deduction can be made based on these observations because no experimental manipulations were made to address this issue. Despite this, the data do present interesting implications for the implementation of behavioral safety processes in applied settings. Typical behavioral safety applications (Krause, 1997; McSween, 2003) require the safety observer to deliver safety performance feedback to the person observed. The observation accuracy data collected throughout this research seem to indicate that, for some behaviors, participants may be unable to correctly identify differences between safe and unsafe performance, yet are able to perform the behaviors safely themselves. In other words, the person being observed may receive inaccurate safety performance feedback despite the fact that the observer can correctly
perform the behaviors. The extent to which participants could accurately score the safety performances of others may have not been a skill problem, however. No contingencies were in place that required accurate scoring, thus it is not clear if participants lacked the skills to discriminate between safe and unsafe behaviors or if they lacked the motivation to do so. 4.5. Strengths and Limitations The present study successfully replicated the safety observer effect using a manufacturing-like task and produced substantial improvements of the targeted behaviors. Other strengths of the present research are the data collected on productivity performance and observation accuracy. The data collected on the productivity performance provide social validity and the data on the accuracy of participant observations provide some ideas for future research. For example, future research should more directly examine the effects of an observer's data collection accuracy on the observer's own performance. As previously mentioned, the observation accuracy data seem to indicate that participants may be unable to correctly discriminate between safe and unsafe performance, yet are able to improve their own performance. This seems to indicate that safety observers may be delivering unreliable safety performance feedback to those being observed. Therefore, determining if improvements to an observer's accuracy further improve their own safety performance would be of practical value. Sasson and Austin (2005) explored this issue indirectly and found a strong relationship between accuracy and performance improvement for some, but not all, observers. The limitations of the study are related to the parameters of the laboratory setting and participants. The experimental sessions lasted only 15 minutes, a work-time duration that is not analogous to a typical workday in a manufacturing setting. In addition, the participants were college students in a laboratory setting and it is not clear if such effects would generalize to a real-world setting where there are many competing contingencies (e.g., productivity deadlines, peer pressures, environmental issues). Sasson and Austin (2005) did replicate these effects on postural safety performance in an applied office setting; however, it would be beneficial if future applied research in this area targeted non-postural behaviors or postural behaviors in non-office settings. Another difference between an applied setting and the laboratory is participant knowledge that they are constantly being observed via a camera. Although Sasson and Austin (2005) replicated the effects in an applied setting, the participants were aware when their behavior was being observed. The safety observer effect occurs when an observer's safety performance increases as a result of conducting a safety observation and then they themselves are being observed. In other words, would this observer effect occur if the safety observer, or participant, were unaware that their performance was being monitored? This is an important question that should be addressed in order to strengthen the validity of this effect. The improvements in safety observed in this study should be considered with all of the above limitations in mind.
A.M. Alvero et al. / Journal of Safety Research 39 (2008) 365–373
In conclusion, the current research helps support the existing research on the effects of conducting safety observations on the performance of the observer. The results of the present study also provide some suggestions for interesting and noteworthy future research. Such research could help safety practitioners modify their present approach toward improving safety to ensure the development of the most efficient and effective process.
373
National Institute for Occupational Safety and Health [NIOSH]. (1998). OSHA Ergonomics Report DT931018 [Data file]. Available from: http://www.osha. lc.gov/SLTC/ergonomics/ergo-hiosh.html Office of Health and Safety Information System. (1998). Safety manuals [Data file]. Available from: http://www.cdc.gov/od/ohs/default.htm Sasson, J. R., & Austin, J. (2005). The effects of information, feedback, and conducting behavioral safety observations on office ergonomic behavior. Journal of Organizational Behavior Management, 24(4), 1−30. U.S. Bureau Labor of Statistics. (2006). Nonfatal occupational injuries and illnesses requiring days away from work, 2006 [Data file]. Available from: http://www.bls.gov
4.6. Impact on Industry In general, the results of this research suggest that if employees conduct safety observations of peer safety performance, then their own safety performance may improve. Specifically, the results of this study show that the safety of at-risk body positions associated with the development of MSDs improve after conducting safety observations. Employees in the manufacturing sector are at particular risk for developing MSDs, and may benefit from the implementation of behavioral safety programs that involve peer safety observations. In addition, these results suggest that safety professionals and employers can be confident that increases in postural safety will not decrease worker productivity. References Alvero, A. M., & Austin, J. (2004). The effects of conducting behavioral observations on the behavior of the observer. Journal of Applied Behavior Analysis, 37, 457−468. Cohen, J. (1988). Statistical power analyses for the behavioral sciences, 2nd ed. Hillsdale, NJ: L. Erlbaum Associates. Kazdin, A. E. (1992). Methodological issues and strategies in clinical research. Washington, DC: American Psychological Association. Krause, T. R. (1997). The behavior-based safety process. New York: Van Nostrand Reinhold. McSween, T. E. (2003). The values-based safety process, 2nd ed. New York: John Wiley & Sons, Inc.
Dr. Alvero is an assistant professor of psychology at Queens College, the City University of New York, where she teaches courses in organizational behavior management and behavioral safety. She holds a B.A. from Florida International University and an M.A. and Ph.D. from Western Michigan University. Dr. Alvero is an editorial board member of the Journal of Organizational Behavior Management and the Journal of Safety Research, and has consulted with various organizations in the areas of training and behavioral safety. Kristen Rost is a doctoral candidate in the Learning Processes and Behavior Analysis program at Queens College and the Graduate Center, the City University of New York, where she teaches courses in experimental psychology and industrial/organizational psychology. She holds a B.S. from Western Michigan University and an M.S. from Florida Institute of Technology. Kristen is the Queens College student committee program representative for the Association for Behavior Analysis, and conducts research in the area of behavioral safety. Dr. Austin is an associate professor of psychology at Western Michigan University, where he teaches courses in performance management and consults with large and small business on behavioral safety and performance improvement systems. He holds a B.A. from the University of Notre Dame and an M.S. and Ph.D. from Florida State University. Dr. Austin is co-editor of the Journal of Organizational Behavior Management, an editorial board member for the Journal of Applied Behavior Analysis, and the director of the OBM Network (www.obmnetwork.com).