Safety Program

A

24 PROFESSIONAL SAFETY JULY 2004 www.asse.org

ASSE Foundation ResearchASSE Foundation Research

Safety Incentives

A study of their effectiveness in construction By Paul M. Goodrum and Manish Gangwar

A GREAT DEAL OF UNCERTAINTY exists regarding the effectiveness of safety incentive programs in construc- tion. Most research on incentives involves case studies and theoretical analyses of their advantages and disad- vantages. Using primary survey data from construction firms and craftworkers, this article examines the impact of incentives on the safety performance of U.S. construction firms. The study found that incentives are effective at improving many of the safety performance metrics used in construction. However, differences exist within the indus- try regarding perceptions of their effectiveness.

The performance of the construction industry has a tremendous effect on the U.S. economy. According to the U.S. Bureau of Economic Analysis, when one includes construction-related business involving design, equipment and materials manufacturing and supply, the construction industry accounts for 13 per- cent of the nation’s gross domestic product (GDP), making it the largest manufacturing industry in the U.S. Unfortunately, the construction industry does

not achieve this performance without sig- nificant cost to the safety and health of the workforce. According to NIOSH, 13.3 workers per 100,000 in construction were injured in 2001. Furthermore, the total number of U.S. fatalities in private con- struction in 2001 was 1,225—one-fifth of all workplace fatalities in the U.S.

Since the inception of OSHA in 1971, the safety performance of all U.S. indus- tries has improved. According to the Bureau of Labor Statistics, the number of fatalities in construction alone has been reduced by half. This improvement is attributed to many industry efforts, such as adopting safer technologies, improved work methods, better training and more thorough accident investigations.

Another measure some construction firms take to improve safety is the use of safety incentives. However, a great deal of debate surrounds the ability of incentives

to improve not only safety but also other construc- tion performance measures. In particular, there is concern about the viability of incentives as a provider of substantial long-term improvements (Prichard).

Literature Review Much has been written about the advantages and

disadvantages of safety incentive programs. Pro- ponents claim that worker behavior is affected by prevailing conditions and events. Behavior can be reinforced by positive feedback and discouraged by negative consequences (Geller 35). Furthermore, it is believed that incentives in the form of reward encourage and promote safe behavior and eventual- ly improve safety performance (Geller 34; Sims).

Skeptics argue that safety incentive programs do not provide long-term improvement of safety (AFL- CIO). They question the motivation provided by these programs, since working safely already deliv- ers significant intrinsic benefits to workers. Critics believe that attributing improved safety perform- ance to incentives is misleading, since anecdotal evi- dence suggests these improvements diminish or even fall below original levels once the incentive programs end (Geller 34).

Another major concern is that these programs do not actually improve the safety behavior targeted, but merely change the reporting of incidents; incentive programs may cause employees to not report acci- dents so that they can qualify for awards. Indeed, OSHA has addressed the concern of inaccurate acci- dent reporting due to incentives in its Voluntary Protection Programs (VPP) policies and procedures manual. “The on-site evaluation [of a company’s safe- ty incentive program] will focus on the incentive pro- gram’s potential impact on the accuracy of reporting injury and illnesses data” [OSHA(b)].

In 1998, OSHA worked with Dennison Associ- ates, an independent agency, to review the perform- ance of safety incentive programs across multiple industries. This study examined the results of 27 dif- ferent research projects on safety incentive programs

Paul M. Goodrum, Ph.D., P.E., is an assistant professor in civil

engineering at the University of Kentucky, where he teaches

undergraduate and graduate courses in construction engineering

and project management. Goodrum has worked

professionally in industry in both design and heavy construction.

His areas of specialty include construction productivity, applied technologies and

construction safety.

Manish Gangwar is a graduate student at the University of

Kentucky, where he is pursuing a master’s degree in construction

management. Gangwar worked in construction for five years before

enrolling at the university.

Goodrum Feature July 2004.qxd 6/16/04 11:08 AM Page 24

www.asse.org JULY 2004 PROFESSIONAL SAFETY 25

tiveness through regular tests and by providing two-way feedback.

These programs also help eliminate injury hiding by re- moving a direct link between an award and the number of accidents reported. B e h a v i o r - b a s e d observation can also provide data about equipment and facil- ities that put work- ers at risk for injury.

A downside of b e h a v i o r - b a s e d incentive programs is that they are comparatively diffi- cult to measure and monitor because em- ployee behavior is inherently more complex and diffi- cult to gauge (Geller 39). In addition, employee behavior changes constantly in reaction to exter- nal factors such as new facilities, new equipment and new workgroups.

Other Issues with Safety Incentive Programs

In both types of incentive programs, motivation is a criti- cal factor. Positive reinforcement, feed- back, and recogni- tion and reward are considered the four major components for motivation in an incentive program (Daniels). Positive r e i n f o r c e m e n t , which means any- thing that increases the desired behavior, is considered the weakest and most misunderstood link (Geller 37; Daniels). Positive reinforcement can be a small gift or simple praise. Everyone has different likes and dislikes, which change over time, making reinforcement even harder to identify (Daniels). Another important point about effective positive rein-

that existed from 1971 to 1987 among U.S. compa- nies [OSHA(a) 10-15]. Incentive programs were divided into two main categories: 1) programs that reward improved safe work practices; and 2) pro- grams that reward reductions in the number of injuries and illnesses reported [OSHA(a) 4-6]. The report found that all programs reviewed under these categories shared some improvement in safe work practices but not all safety-related work practices improved in all studies [OSHA(a) 7]. The study found no direct link between safety incentive pro- grams and the reported number of injuries.

Forms of Safety Incentive Programs Safety incentive programs can be divided into

two categories: 1) injury/illness-based incentive pro- grams; and 2) behavior-based incentive programs.

Injury/Illness-Based Programs Injury/illness-based safety incentive programs

are based on the number of injuries and/or illnesses as a criterion to reward workers and crews. Individuals or groups are rewarded for avoiding or lowering accidents during predefined periods. These programs work on the underlying assump- tions that: 1) facilities and equipment are safe and do not cause any accidents; 2) workers have proper training and knowledge to use equipment; and 3) accidents are primarily the result of worker negli- gence or compromise on safety (Smith 44).

One problem with this approach is that it directly equates prizes with a number of injuries (Krause 28). Injury/illness-based programs present the tempta- tion for workers to not report an injury so they will not lose individual incentives or be the reason that the whole group does not receive an award (Geller 37; Flanders and Lawrence). Another concern is that these programs may become trivial and hard to dis- continue in the long-run because workers can view incentives as an entitlement; discontinuation may cause significant negative impact (Smith 44).

Injury/illness-based incentive programs may also provide false feedback and cause mistrust between workers and management (Krause and McCorquodale 34; Prichard; Smith 44). For example, suppose a crew makes a substantial effort to avoid injuries, yet unfortunately experiences an accident. As a result, this crew will not receive an incentive. Meanwhile, another crew that makes no effort to avoid injury may manage to do so and, thus, may still receive a reward.

Behavior-Based Programs Behavior-based safety incentive programs observe

worker behavior as a criterion for awarding incen- tives. Examples of rewarded behavior include partici- pating in safety meetings and training; offering suggestions about how to improve jobsite safety; and other behavior that can help prevent accidents. Although such programs solve the problem of erro- neous feedback and improve attendance in meetings and training, their effectiveness is still questioned. To address this problem, some sites gauge program effec-

Statistics Terms Pearson Correlation The correlation between two variables shows the degree to which the variables are related. Pearson correlation ranges from +1 to -1, where +1 represents a perfect linear correlation between the variables and -1 represents a per- fectly inverse linear relationship between the variables. Zero denotes there is absolutely no linear relationship between two variables.

P-Value/Significance Value P-value—sometimes called the significance value—represents the probability of getting something as rare or extreme as the given result. Therefore, the lower the probability, the less chance there is that two samples are from the same population. Statistically, a P-value of less than 0.05, which is five percent, is consid- ered acceptable to reason that the discrepancy between two samples is assumed to be a result of two different populations, or, in other words, that two samples represent two different population groups with different measured characteristics.

ANOVA Analysis of variance (ANOVA) determines the probability that two or more samples were drawn from the same parent population. The purpose of an ANOVA is to verify that the means of a measured variable for two or more samples are different enough to not have occurred by chance. In other words, if the group means do not differ significantly, it is inferred that the independent variable(s) did not have an effect on the dependent variable. The key statistics in ANOVA are the degrees of freedom (df) and the F-value that are used to identify the significance value.

Chi-Square analysis Chi-square is a non-parametric test of statistical significance for cross-tabular or discontinuous data. Like the F statistics in ANOVA, the chi- square indicates the degree of confidence one can have in accepting or rejecting a hypothesis. Chi-square statistics, along with the degree of freedom, provide similar information about nominal data—whether they belong to one group or to two different groups of dissimilar characteristics.

Goodrum Feature July 2004.qxd 6/16/04 11:08 AM Page 25

26 PROFESSIONAL SAFETY JULY 2004 www.asse.org

criterion rather than competition to avoid unneces- sary rifts among workers and crews (Opfer). It is also believed that it is better to reward many participants rather than an individual (Geller 38; Opfer). Likewise, a group should not be penalized for the failure of an individual’s action (Geller 38). This is reflective of the belief that safety is a team effort and individuals do not cause accidents. Rather, accidents are the collective failure of the group. Most impor- tantly, safety incentives are not a panacea to improve safety. Others have found that incentives cannot work without a comprehensive safety program that addresses training, culture, drug testing and other critical elements (Opfer; Hinze 82; Trahan).

Study Methodology Clearly, significant differences of opinion exist on

the effectiveness of safety incentives. The quantified effect of incentive programs on construction safety performance remains uncertain. Also, little is known about how different implementation schemes of safe- ty incentives affect their performance. Furthermore, little is known about the effectiveness of safety incen- tives based on the perceptions and experiences of con- struction craftworkers. This article examines the effects of safety incentive programs on construction safety performance using industry data. First, differ-

forcement is that it should immediately follow the desired behavior, which is why positive rein- forcement should be a daily affair (Dan- iels; Geller 38; Hinze 84). Ideally, peers are in the best position to deliver positive reinforcement.

To provide posi- tive reinforcement, incentives can be awarded in different forms. Efficient incentives need to have more personal value than a significant dollar value. The dollar value of incentives is unimportant as long as the incentive is meaningful and a positive reinforcer to the worker (Toft). In addition, incentives should be distributed separately from normal com- pensation (Opfer). Many successful programs rely on low-cost gifts with high perceived value for this very reason. Leboeuf identified 10 categories of incentives: 1) recognition; 2) time off; 3) stock ownership; 4) spe- cial assignments; 5) advancement; 6) increased auton- omy; 7) training and education; 8) social gatherings; 9) prizes; and 10) money (Sims; Toft). Some experts suggest that an incentive preceded by a celebration gives worker the opportunity to relive the event and further reinforce the behavior (Geller 37). Incentives and rewards should be specified and should be per- ceived as achievable (Geller 39). Finally, incentives should be based on long-term progress rather than on short-term achievement (Opfer).

Who will be rewarded is also an important con- sideration. Some have found that everyone who meets the criteria should be rewarded (Geller 39; Opfer). This gives workers a sense of belonging and makes them feel they are part of the safety initiative. Furthermore, incentives should be based on absolute

Breakdown by State Total

State Responses Percentage

Kentucky 35 21.21 North Carolina 25 15.15 Tennessee 20 12.12 Ohio 16 9.70 Georgia 16 9.70 Virginia 16 9.70 Indiana 15 9.09 Florida 11 6.67 Alabama 7 4.24 Others 4 2.42

Table 1Table 1 Figure 1Figure 1 Breakdown by Work Type

Differences in Annual Volume of Work, 2001

Number of Mean Pearson Companies Volume F-Value Significance Correlation

With SIP 85 $101 mil 4.82 0.03 -0.18 Without SIP 54 $21 mil

Note: SIP denotes safety incentive program.

Table 2Table 2

Goodrum Feature July 2004.qxd 6/16/04 11:08 AM Page 26

www.asse.org JULY 2004 PROFESSIONAL SAFETY 27

both heavy and building construction, and 12 per- cent were specialty contractors (Figure 1).

Survey results were coded into SPSS (statistical analysis software) for detailed statistical analysis. Frequency tables, box plot, chi-square and analysis of variances (ANOVA) were used to analyze the sur- vey results (see sidebar on pg. 25).

Research Sample for Craftworkers’ Survey The second survey was a one-page survey for

construction craftworkers of companies with safety incentive programs. The researchers contacted six construction firms in Kentucky, Ohio and Tennessee who agreed to administer and return the completed survey. To help protect respondent anonymity, the survey requested no information regarding the respondent or his/her employer. The total sample size for this survey was 252 workers. The mean years of experience in the construction industry of all workers surveyed was 16.14. Fifty percent of the workers had less than 15 years of construction expe- rience, and 10 percent had more than 31 years of construction experience.

Measures of Safety Performance To quantify the effectiveness of safety incentive

programs, four different measures of safety per- formance were collected in the managers’ survey:

1) OSHA recordable cases. Cases when workers, due to an injury/illness sustained at work, must visit a doctor for more than first aid.

2) Lost-time workday cases. Cases when work- ers, due to an injury/illness sustained at work, are

ences in safety per- formance data be- tween construction firms with safety incentive programs and those without are analyzed. Then, the impact of differ- ent implementation schemes is assessed. Finally, craftwork- ers’ experiences with these programs are examined.

Data Collection A great deal of previous

research on safety incentives is based on anecdotal evidence. To collect empirical data on the use of safety incentives and company safety performance, two surveys were developed: a managers’ survey and a craft- workers’ survey. The man- agers’ survey was designed for company safety directors and other SH&E professionals in charge of their company’s pro- grams. This survey examined how companies administered their safety incentive program, safety performance data, and the percep- tions and experiences that each company had with its incentive program. The craftworkers’ survey was administered only to workers who were currently employed by a construction firm that used safety incentives. This survey examined their personal experiences and perceptions of participating in a safety incentive program.

Research Sample for Managers’ Survey In the process of creating the managers’ survey, an

extensive literature review was performed. Based on previous findings and the research objectives, a pilot survey was created and administered to three gener- al contractors in Kentucky and Tennessee. Their com- ments and suggestions were incorporated into the actual survey. Next, the managers’ survey was administered by mail to several construction firms in the Midwest and southeastern U.S. Of the 165 sur- veys received, 22 percent were collected from con- tractors in Kentucky; 14 percent from North Carolina; 12 percent from Tennessee; 10 percent each from Ohio, Georgia and Virginia; and the remaining 22 per- cent from Indiana, Florida, Alabama, Illinois, West Virginia, Michigan, Missouri and Texas (Table 1).

In 2001, the average number of workers for the sampled companies was 199. Fifty percent of the companies had less than 67 workers in 2001, and 10 percent had more than 368 workers. Of the 165 sam- pled companies, 42 percent were engaged in only building construction, 29 percent were involved in heavy construction, 14 percent were involved in

Objectives Behind Implementing an SIP Objective Primary (3) Secondary (2) Tertiary (1) Total Weight

To change workers’ behavior 37 26 15 178 To improve workers’ awareness 29 37 9 170 To reduce recordable accidents 21 11 21 106 To minimize losses 4 5 20 42 To minimize safety-related claims 1 7 9 26 To maintain good safety records 3 2 12 25

Table 3Table 3

Safety Performance Based on 2001 SIP With SIP Without SIP Mean N Mean N df F-Value Sig.

Lost-time workday incidence rates 1.45 81 4.99 56 136 9.35 0.00 Restricted workday incidence rates 1.26 74 2.53 54 127 3.65 0.06 OSHA recordable incidence rates 4.20 79 5.46 54 132 1.68 0.20

Note: Sample size does not equal 165 due to nonresponse.

Table 4Table 4

Goodrum Feature July 2004.qxd 6/16/04 11:08 AM Page 27

28 PROFESSIONAL SAFETY JULY 2004 www.asse.org

not able to perform work fully or partly; this is a subset of total OSHA recordable cases.

3) Restricted workday cases. Cases when workers are not able to work to their full capaci- ty due to an injury/illness sus- tained at work, and are assigned a lower workload; these cases are part of lost-time workday cases.

4) Experience modification rate (EMR). EMR is primarily used to establish workers’ compensation (WC) insurance premium rates. EMR calcula- tions are based on each em- ployer’s compensation claim

experience over its last three years as com- pared to the average of the industry. EMR takes into account the number of accidents as well as the severity of cases. The man- agers’ survey also collected total employ- ee hours from each company; this allowed recordable, lost-time and restricted inci- dence rates to be calculated for each com- pany using the following equation:

Incidence rate = Number of cases x 200,000

Total employee hours per year (1)

Data Analysis Breakdown of Companies With & Without Safety Incentive Programs

Of the 165 companies sampled, 59 per- cent had a safety incentive program in 2001 and 41 percent did not. The mean safety incentive program was 5.87 years old. Fifty percent of the companies that had an incentive program had the pro- gram for less than four years, while 10 percent of the companies with a safety incentive program had used the program for more than 12 years. Significant correla- tion was found between the annual vol- ume of work and safety incentive programs. Companies with a safety incen- tive program also had a substantial and statistically significant larger mean vol- ume of work compared to those with no incentive program. Those with an incen- tive program had a mean volume of work of $101 million compared to $21 million for companies without a safety incentive program (Table 2). An F-value of 4.82 shows the difference in work volume to be statistically significant.

To begin to understand the motivation of why construction companies implement a safety incentive program, the managers’

survey asked each respondent whose company had such a program to rate six objectives, in order of pref-

EMR for Different Years Based on SIP With SIP Without SIP Mean N Mean N df F-Value Sig.

2001 EMR with SIP 0.773 53 0.849 62 114 6.05 0.02 implemented in 1999 2001 EMR with SIP 0.760 41 0.844 74 114 6.69 0.01 implemented in 1998 2001 EMR with SIP 0.734 36 0.851 79 114 12.74 0.00 implemented in 1997 2000 EMR with SIP 0.763 40 0.861 72 111 6.25 0.01 implemented in 1998 2000 EMR with SIP 0.730 35 0.870 77 111 10.53 0.00 implemented in 1997 1999 EMR with SIP 0.764 34 0.890 69 102 3.81 0.00 implemented in 1997

Table 6Table 6

Differences in Change of EMR Based on Various Timeframes for SIP

With SIP Without SIP Mean N Mean N df F-Value Sig.

Difference in EMR -0.039 29 -0.008 56 84 0.04 0.50 between 2001 and 1997 with SIP 1997 Difference in EMR -0.022 34 -0.047 56 89 0.41 0.52 between 2001 and 1998 with SIP 1998 Difference in EMR -0.025 48 -0.043 54 101 0.34 0.56 between 2001 and 1999 with SIP 1999

Table 7Table 7

Differences in Performance Change Based on 2000 SIP

With SIP Without SIP Mean N Mean N df F-Value Sig.

Lost-time workday incidence -1.02 60 1.41 55 114 4.82 0.03 difference between 2001 and 1999 Restricted workday incidence 0.53 57 0.80 54 110 0.15 0.70 difference between 2001 and 1999 OSHA recordable incidence -0.92 61 -0.18 53 113 0.74 0.39 difference between 2001 and 1999

Table 5Table 5

Goodrum Feature July 2004.qxd 6/16/04 11:08 AM Page 28

www.asse.org JULY 2004 PROFESSIONAL SAFETY 29

incidence rate in 2001 for companies with a safety incentive program was 1.26 compared to 2.53 for those without such a program. The difference was statistically significant at the 94 percent confidence interval with an F-value of 3.65 (Table 4).

Finally, of the 133 companies that indicated their OSHA recordable incidence rates, a difference was found in mean OSHA recordable incidence rates.

erence, for implementing the program. The objectives were to: 1) reduce recordable accidents; 2) improve safety awareness among workers; 3) change workers’ behav- ior to adopt safer work practices; 4) main- tain good safety records; 5) minimize safety-related claims; and 6) minimize loss- es. Table 3 shows the number of companies rating each objective as primary, secondary and tertiary.

To identify the three most preferred objectives for implementing a safety incen- tive program, different weights (3 for pri- mary, 2 for secondary and 1 for tertiary) were assigned to the number of responses for each objective. For example, 21 compa- nies weighted “to reduce recordable accidents” as their primary objective; 11 companies weighted the same as their secondary objective; and 21 companies weighted the same as their tertiary objec- tive. Therefore, the total weight for this cri- terion was (21 x 3) + (11 x 2) + (21 x 1)=106 (Table 3). Based on this calculation, the top three objectives for implementing a safety incentive program were to:

1) change worker behavior to adopt safer work practices;

2) improve safety awareness among workers;

3) reduce recordable accidents. As seen, many companies use a safety

incentive program not only to reduce acci- dent rates but also to have an impact on worker behavior. By changing worker behavior and safety awareness, safety per- formance should improve. The next sec- tion of the analysis examined whether this expectation is actually being achieved.

Effectiveness of Safety Incentive Program in Terms of Various Safety Performance Indicators

To estimate the impact of incentives on safety performance, the lost-time, restrict- ed and recordable rates were compared between companies that did and did not have a safety incentive program in 2001 (Table 4). Among the 137 companies that indicated their lost-time workday inci- dence rates, a significant difference was seen in the mean lost-time workday inci- dence rates between companies with and without a safety incentive program in 2001. Companies with a program had a mean lost-time incidence rate of 1.45 compared to 4.99 for those without a program, which was statis- tically significant with an F-value of 9.35 (Table 4).

Of the 128 companies that indicated their restrict- ed workday incidence rates, there was also a differ- ence in mean restricted workday incidence rates between companies with and without a safety incen- tive program in 2001. The mean restricted workday

Differences Due to Crew or Worker Performance

Worker Only Crew* Performance Mean N Mean N df F-Value Sig.

Lost-time workday 1.27 35 1.68 38 72 0.31 0.58 incidence 2001 Restricted workday 1.09 31 1.34 35 65 0.32 0.58 incidence 2001 OSHA recordable 4.08 33 4.14 38 70 0.00 0.96 incidence 2001 EMR 2001 0.79 32 0.81 32 63 0.20 0.66

*Workers and supervisors

Table 8Table 8

Differences Due to Measuring Criteria Injuries Behavior Both Mean N Mean N Mean N Sig.

Lost-time workday 1.24 30 1.57 25 1.77 18 0.84 incidence 2001 Restricted workday 1.39 27 0.80 21 1.74 18 0.27 incidence 2001 OSHA recordable 5.06 29 4.06 24 3.88 18 0.63 incidence 2001 EMR 2001 0.84 27 0.79 22 0.78 15 0.51

Table 9Table 9

Differences Due to Incentive Period Less than Monthly to More than One Month Quarterly Six Months Mean N Mean N Mean N Sig.

Lost-time workday 1.07 26 1.34 21 1.99 30 0.52 incidence 2001 Restricted workday 1.07 22 1.68 19 1.04 29 0.43 incidence 2001 OSHA recordable 3.81 25 5.51 21 3.66 29 0.32 incidence 2001 EMR 2001 0.796 22 0.829 18 0.813 28 0.85

Table 10Table 10

Goodrum Feature July 2004.qxd 6/16/04 11:08 AM Page 29

30 PROFESSIONAL SAFETY JULY 2004 www.asse.org

ferences in OSHA recordable incidence rates were also not statistically significant with a F-value of 0.74 (Table 5).

Next, the research examined differences in EMR. Since EMR reflects a three-year average of a compa- ny’s safety performance, EMR for particular years was compared with respect to when each company’s safety incentive program was implemented in previ- ous years (Table 6). The mean 2001 EMR for compa- nies that implemented a safety incentive program in 1999 was 0.773 compared to 0.849 for companies with no incentive program in 1999, a difference of 0.076. This difference was statistically significant with an F-value of 6.05.

The difference in EMR between companies with and without a safety incentive program increased with the age of the program. The 2001 EMR for com- panies that implemented a safety incentive program

The mean rate of companies with a safety incentive program was 4.20 compared to 5.46 for companies without a program. However, the difference was only statistically significant at the 80 percent confi- dence interval with an F-value of 1.68 (Table 4).

Change in Safety Performance from 1999 to 2001 To identify whether safety performance changed

differently over time as a result of safety incentive programs, differences in the change of various inci- dence rates from 1999 to 2001 were compared between companies with and without a safety incen- tive program in 2000. (Year 2001 was the latest inci- dence rates available for the study. Year 2000 was chosen to ensure that affects of a safety incentive program were reasonably reflected in the 2001 data.)

For companies that had such a program in 2000, the mean lost-time workday incidence rate from 1999 to 2001 decreased by 1.02, representing a decline of 44.16 percent from the 1999 mean lost-time workday incidence rate of 2.31. In comparison, for companies without a safety incentive program in 2000, the mean lost-time workday inci- dence rate from 1999 to 2001 increased by 1.41, representing an increase of 41.84 percent from the 1999 lost-time work- day incidence rate of 3.37. The differences were statistically significant with an F-value of 4.82 (Table 5).

However, the mean restrict- ed workday incidence rate from 1999 to 2001 increased for both categories of com- panies. For companies that had a safety incentive program in 2000, the increase was 0.53, representing an increase of 29.43 percent from the 1999 mean restricted workday incidence rate of 1.80.

For companies with no safety incentive program in 2000, the increase was 0.80, representing an increase of 55.96 percent from the 1999 mean restricted workday incidence rate of 1.43. However, the differ- ences in incidence rates were not statisti- cally significant with an F-value of only 0.15 (Table 5).

Furthermore, the mean OSHA record- able incidence rates for companies with a safety incentive program from 1999 to 2001 reduced by 0.92, representing a decline of 15.78 percent from the 1999 mean OSHA recordable incidence rate of 5.83. In comparison, the mean OSHA recordable incidence rate for companies with no safety incentive program from 1999 to 2001 reduced by 0.18, a decline of 3.08 percent from the 1999 mean OSHA recordable incidence rate of 5.84. The dif-

Differences Due to Award Type Tangible Both

Award Type Mean N Mean N df F-Value Sig.

Lost-time workday 1.13 54 2.35 24 77 2.73 0.10 incidence 2001 Restricted workday 1.04 50 1.87 21 70 2.94 0.09 incidence 2001 OSHA recordable 3.93 53 5.24 23 75 1.30 0.26 incidence 2001 EMR 2001 0.82 46 0.78 24 69 1.04 0.31

Table 11Table 11

Figure 2Figure 2 Perceptions about SIPs: Overall Value

Note: Chi-square value of difference is 52.75 with 4 df and a P-value of 0.00.

Safety incentives

cannot work

without a safety program

that addresses training,

culture and other

critical elements.

Goodrum Feature July 2004.qxd 6/16/04 11:08 AM Page 30

www.asse.org JULY 2004 PROFESSIONAL SAFETY 31

other time periods examined in the study (Table 6). The study then examined whether changes in EMR were related to the use of a safety incentive program. As Table 7 shows, there was no statistically signifi- cant difference in EMR changes between 1997 and

2001 for the companies studied. For example, the mean EMR decreased by 0.039 for those with a safety incentive pro- gram as compared to a decrease of 0.008 for those without such a program. This difference was not statistically significant with an F-value of 0.04.

Descriptive Statistics of Companies with a Safety Incentive Program

To assess whether differences existed regarding how safety incentives are implemented, the research examined dif- ferences in safety performance among just those companies that had a safety incen- tive program in 2001. The research exam- ined several factors to determine whether they made any difference in safety meas- ures among those companies with a safe- ty incentive program: 1) who received incentives; 2) which type of incentives (injury- or behavior-based) were used; 3) the time period for awarding incentives; and 4) the type of award presented.

Forty-six companies—46.9 percent of those with a safety incentive program— evaluated only individual worker per- formance as an award criterion for the program. However, 44 companies—45 per- cent—evaluated whole-crew performance to award safety incentives. Meanwhile, eight companies—8.2 percent—did not respond to this question. The research examined whether differences existed based on who received incentives. Although companies that awarded incen- tives to workers only had slightly lower ratings, none of the differences were statis- tically significant (Table 8).

Thirty-eight companies—38.8 percent of those with a safety incentive program— based their program on injuries only, while 28 companies—28.6 percent of those with an incentive program—based their safety incentive program on behavior only. Twenty-three companies (23.5 per- cent) based their program on both injuries and behavior. However, nine companies, 9.2 percent of the companies with a safety incentive program, did not respond to this question (Table 9). There appeared to be little to no relation in accident rates and whether award criterion was based on injuries or behavior. Various incidence rates and EMR for 2001 were not statisti- cally different for companies that meas- ured injuries, behavior or both (Table 9).

in 1997 was 0.734 compared to 0.851 for companies that had not implemented a safety incentive pro- gram in 1997, a difference of 0.117. In fact, firms with a safety incentive program consistently had a lower and statistically significant different EMR for all

Figure 4Figure 4 Perceptions about SIPs: Impact on Worker Behavior

Note: Chi-square value of difference is 16.41 with 4 df and a P-value of 0.00.

Figure 3Figure 3 Perceptions about SIPs: Reducing Accidents

Note: Chi-square value of difference is 29.27 with 4 df and a P-value of 0.00.

Goodrum Feature July 2004.qxd 6/16/04 11:08 AM Page 31

32 PROFESSIONAL SAFETY JULY 2004 www.asse.org

started, in comparison to 35.7 percent that had not altered their program since its inception (3.1 percent of companies with a safety incentive program in 2001 did not respond to this question). The study also examined whether the size of the incentive—as measured by its cost—had an impact on its effec- tiveness. In this sample of safety incentive programs, no such relation was found.

Next, the research examined whether the time period during which incentives were awarded made any difference. Of 98 companies reported to have a safety incentive program in 2001, 30.6 percent awarded incentives each month; 26.5 percent rewarded employees quarterly to bi-annually; and 36.7 percent rewarded employees bi-annually or less often. Furthermore, 6.1 percent of the participants did not respond to this question. Again, various incidence rates and EMR for 2001 were not statistically different among companies with different timeframes for awarding safety incentives (Table 10).

The research also examined whether the type of award—tangible versus intan- gible—made any difference. Sixty-six companies (67.3 percent of those with an incentive program) awarded tangible items such as cash, lottery, gifts and prizes. No company reported to award only intangible items such as trophies, cer- tificates, time off and parties. However, 29 companies (39.6 percent) awarded both tangible and intangible awards (Table 11).

Companies that distributed only tangi- ble awards had lower lost-time and restricted workday incidence rates than those which distributed both tangible and intangible awards. Firms that gave only tangible awards had a mean lost-time inci- dence rate of 1.13 and a restricted inci- dence rate of 1.04 compared to 2.35 and 1.87, respectively, for companies that also gave intangible awards. The statistical sig- nificance for both these differences was at or above the 90 percent confidence level (Table 11). Although companies that gave only tangible awards also had lower recordables and EMR, these differences were not statistically significant.

The managers’ survey collected addi- tional data about whether companies had other incentive programs (outside of safe- ty); whether safety performance evalua- tion of supervisors was linked to safety incentives; whether a safety incentive pro- gram was changed after its inception; who attended award ceremonies; and who received incentives. For companies that had a safety incentive program in 2001, 49 percent had other incentive programs, compared to 43.9 percent with no other incentive program (7.1 percent did not respond to the question). In addition, 72.4 percent of companies with a safety incen- tive program in 2001 linked supervisor’s performance to safety incentives, com- pared to 22.4 percent that did not (5.2 per- cent of companies with a safety incentive program did not respond).

In addition, 61.2 percent had changed their safety incentive program since it

Figure 5Figure 5 Perceptions about SIPs: Increasing Safety Awareness

Note: Chi-square value of difference is 18.13 with 4 df and a P-value of 0.00.

Figure 6Figure 6 Perceptions about SIPs: Long-Term Improvement

Note: Chi-square value of difference is 12.55 with 4 df and a P-value of 0.01.

Craft workers

have a more

favorable opinion

regarding the effec-

tiveness of safety

incentives than do

safety directors.

Goodrum Feature July 2004.qxd 6/16/04 11:08 AM Page 32

www.asse.org JULY 2004 PROFESSIONAL SAFETY 33

Management & Craftworker Perceptions

The study also examined differences in opinion between management and workers about various issues regarding safety incentive programs. The sur- veys asked 165 company safety directors and other management personnel, as well as 252 craftworkers, their opinions on: 1) the overall value of a safety incentive program; 2) its ability to reduce recordable accidents; 3) its impact on worker behavior; 4) the program’s ability to increase safety awareness among workers; and 5) its ability to provide long- term improvements in safety (Figures 2-6).

Overall, craftworkers have a more favorable opinion regarding the effective- ness of safety incentives than do safety directors and other managers who oversee the programs. For example, 42 percent of surveyed craftworkers indicated there was an extreme overall value in safety incentive programs compared to five percent of sur- veyed SH&E managers. Craftworkers had a greater overall value of safety incentive programs than was perceived by manage- ment in charge of the programs. The differ- ence was significantly different with a chi-square value of 52.76 with 4 degrees of freedom and a P-value of 0.00.

Likewise, workers’ perception about the ability of safety incentive program to reduce recordable incidence rates was more positive than that of management. The dif- ference was significantly different with a chi-square value of 29.27 with 4 degrees of freedom and a P-value of 0.00. Workers’ perception about a program’s ability to change their safety behavior was also more positive than management’s perception. This difference was significantly different with a chi-square value of 16.41 with 4 degrees of freedom and a P-value of 0.00.

Employees’ perception about the pro- grams’ ability to increase safety awareness among workers was more positive than management’s. The difference was signifi- cantly different with a chi-square value of 18.13 with 4 degrees of freedom and a P-value of 0.00. Finally, workers’ percep- tion about the ability of safety incentives to improve long-term safety performance was more positive than was manage- ment’s perception. Again, this difference was significantly different with a chi- square value of 12.55 with 4 degrees of freedom and a P-value of 0.01.

Advantages & Disadvantages of Safety Incentive Programs: Worker Perceptions

To further examine the experience that workers have with safety incentive pro- grams, craftworkers were asked to identify both advantages and disadvantages of

Of those with a safety incentive program in 2001, 10.2 percent awarded safety incentives to their work- ers only; 11.2 percent awarded safety incentives to everyone but workers; 13.3 percent awarded safety incentives to workers and foremen only; 62.2 percent awarded safety incentives to everyone—workers, foremen, superintendents, safety personnel, field engineers and even managers in some cases. None of the cited factors influenced safety performance, which was measured in terms of OSHA recordable cases, lost-time workday cases, restricted workday cases and EMR, of the companies surveyed in a sta- tistically significant way.

Figure 7Figure 7 Craftworkers’ Perspective of SIPs: Advantages

Figure 8Figure 8 Craftworkers’ Perspective of SIPs: Disadvantages

Goodrum Feature July 2004.qxd 6/16/04 11:08 AM Page 33

34 PROFESSIONAL SAFETY JULY 2004 www.asse.org

References AFL-CIO. “Safety Incentive and Injury Discipline Policies: The

Bad, the Worse and the Downright Ugly.” Washington, DC: AFL- CIO, Dec. 1999. <http://www.aflcio.com/yourjobeconomy/safe ty/issues/upload/factsheet1.pdf>.

Daniels, A. “Incentives, Safety and Performance Manage- ment.” May 20, 2002.

Flanders, M.E. and T.W. Lawrence Jr. “Warning! Safety Incentive Programs Under OSHA Scrutiny.” Professional Safety. Dec. 1999: 29-31.

Geller, E.S. “The Truth about Safety Incentives.” Professional Safety. Oct. 1996: 34-39.

Hinze, J. “Safety Incentives: Do They Reduce Injuries?” Practice Periodical on Structural Design and Construction. May 2002: 81-84.

Krause, T.R. and S.J. Hodson. “A Close Look at Safety Incentives.” Occupational Health & Safety. June 1998: 28-30.

Krause, T.R. and R.J. McCorquodale. “Transitioning Away from Safety Incentive Programs.” Professional Safety. March 1996: 33-36.

Opfer, N.D. “Construction Safety Improvement Through Incentive Compensation.” 1998 AACE International Transactions. Morgantown, WV: AACE International, 1998.

OSHA(a). “Review of the Literature on Safety Incentives.” Docket No-S-777 Ex. 502-281. Washington, DC: OSHA, 1998.

OSHA(b). Voluntary Protection Programs: Polices and Procedures Manual. Washington, DC: U.S. Dept. of Labor, OSHA, March 25, 2003. VI-8.

Prichard, R. “Safety Incentive Programs: A Critical Assess- ment.” Dallas: International Risk Management Institute, April 2001. May 20, 2002. <http://www.irmi.com/expert/Articles/ 2001/Prichard04.asp>.

Sims, B. Jr.(a) “Effective Motivation.” May 21, 2002. <http:// www.billsims.com/oshmag2.php>.

Sims, B. Jr.(b) “How Successful Safety Incentive Programs Reduce Injuries Without Injury Hiding.” May 8, 2002. <http://www.billsims .com/a1.php>.

Smith, T.A. “What’s Wrong With Safety Incentives?” Professional Safety. May 1997: 44.

Trahan, G. Phone interview. Aug. 14, 2002.

Toft, D. “The Secrets of a Successful Safety Incentive Program.” Safety + Health. <http://www .billsims.com/a2.php>.

working under a safety incentive program. The most popular advantage was making the workplace safer—indicated in 41.1 percent of total responses. This was followed by receiving the award as a bene- fit in and of itself (22.1 percent); reduction in acci- dents (17.9 percent); and increase in safety alertness (16.3 percent) (Figure 7).

Meanwhile, workers identified the most popular disadvantage as the fact that many accidents—and the resulting loss of incentive—are not the fault of workers themselves, but are the result of factors beyond their control. In addition, 6.7 percent believed safety incentive programs consume too much of a worker’s time and slow production; 4.2 percent believed the programs are a slow way to improve safety; and 4.2 percent believed that these programs increase nonreporting of accidents (Figure 8). Meanwhile, 71.4 percent of surveyed craftwork- ers said that the use of safety incentive programs posed no disadvantages.

Conclusions As a result of these findings, the following con-

clusions can be drawn. 1) Among the companies surveyed, those that

have a safety incentive program have lower lost- time incidence rates, restricted incidence rates and EMRs compared to companies that do not.

2) There is some indication, as measured by dif- ferences in lost-time incidence rates, that the compa- nies with a safety incentive program experienced a greater improvement in safety between the studied periods 1997 and 2001 compared to the companies with no safety incentive program.

3) Rewards based on crew versus individual per- formance, injury versus behavior performance and different time periods for giving the awards made no difference in effectiveness of the programs among the sampled companies. However, companies that used only tangible awards versus those that used both tangible and intangible awards had slightly better safety performance measures.

4) Craftworkers have a more favorable opinion of the effectiveness of safety incentive program than do company mangers.

5) While craftworkers recognize that safety incen- tive programs have some disadvantages, most feel there are no drawbacks and believe their greatest advantage is improving jobsite safety.

While these findings support the use of safety incentive programs to improve jobsite safety, such a program should not exist by itself. Instead, it should be part of an overall comprehensive SH&E program that not only involves workforce training but also engineers safety into the construction process. The study found that safety is improved by the use of incentives based on traditional outcome measures. The study did not examine the impact of safety incentives on worker behavior during the construc- tion process. This is worthy of future research and could be accomplished by incorporating behavior- based safety measures. �

Your Feedback Did you find this article interesting and useful? Circle the corresponding number on the reader service card.

RSC# Feedback 25 Yes 26 Somewhat 27 No

Practical Applications It is evident that companies which want to reduce their experience modification rates, lost- time workday incidences and restricted work- day incidences can use safety incentive programs successfully. The study also found that various other factors, such as injury- or behavior-based incentive programs, period of incentives and kind of awards, do not change the effectiveness of safety incentive programs in a significant way. It was also found that craft- workers have a much more positive reaction to incentives than do their managers. While safety incentive programs do lead to awards being given to craftworkers, most craftworkers think that the greatest advantage of these programs is the improvement in jobsite safety.

Goodrum Feature July 2004.qxd 6/16/04 11:08 AM Page 34