April 16, 2007 20:3 WSPC/177-JCR 00049.tex Journal of Construction Research, Vol. 7, Nos. 1&2 (2006) 111-132 c World Scientic Publishing Company \u0001 ASSESSMENT OF RISK PERCEPTION OF IRONWORKERS JAVIER IRIZARRY Assistant Professor, Construction Management Department Southern Polytechnic State University 1100 South Marietta Parkway, Marietta, GA 30060 irizarry@spsu.edu DULCY M. ABRAHAM Associate Professor School of Civil Engineering, Purdue University 550 Stadium Mall Drive, West Lafayette, IN 47907-2051 dulcy@ecn.purdue.edu Received 25 January 2005 Accepted 14 July 2005 This study evaluated factors that could inuence the risk perception of ironworkers. Descriptive statistics and logistic regression were used to analyze data collected through a survey to ironworkers in the United States. Logistic regression was used to investigate the inuence of dierent variables studied on the probability of workers perceiving an action as dangerous, thereby providing an indication of how workers perceive risk in the tasks that they perform. Variables that were found to be signicant in the analysis were accident and injury experience, experience in steel erection and construction work, safety incentive programs, and type of tasks performed. Keywords: Safety; steel construction; falls; OSHA; risk perception; ironworkers; surveys; logistic regression. 1. Introduction Despite increased eorts in accident prevention, the construction industry continues to have the largest number of injuries and fatalities of all industries in the United States economy (Bureau of Labor Statistics, 2002). In 2002, one thousand one hundred and twenty-one (1,121) workers lost their lives in construction related accidents (Bureau of Labor Statistics, 2002), which is approximately 20% of all reported workplace fatalities. This high number of fatalities is no surprise if we consider that two of the most dangerous occupations in the United States are in the construction industry. Structural steel workers had the fourth highest fatality rate with 58.2 fatalities per 100,000 workers and construction laborers experienced fatalities at the rate of 27.7 fatalities per 100,000 workers (ninth highest) (BLS, 2002). Prior research has found that the causes of accidents can be attributed to 111 April 16, 2007 112 20:3 WSPC/177-JCR J. Irizarry & D. M. Abraham factors such as human error and unsafe behavior and the interaction of humans with materials, tools, and environmental factors (Lehto and Salvendy, 1991). Accident reports have been used to nd the causes of injuries and fatalities (Abdelhamid and Everett, 2000; Huang and Hinze, 2003; Edwards, 2003; Hide et al., 2003; Arboleda and Abraham, 2004; Chua and Goh, 2004). However, research based on the information obtained from these reports focuses mainly on after-the-fact information and stops at a premature level or misses important steps to identify the root causes of accidents (Abdelhamid and Everett, 2000). Brown (1995) suggests that accident investigation should be based on theories of accident causation and human error, resulting in a better understanding of the relation between the \"antecedent human behavior\" and the accident at a level that enables the root causes to be determined. The result could be more eective accident prevention strategies directed at the root causes of accidents and not at its symptoms. This paper will then evaluate factors that could inuence the risk perception of ironworkers engaged in the handling and installation of steel members. It will also analyze data regarding the characteristics of the ironworker population (i.e. work experience, safety training experience, accident and injury experience, age, and anthropometric characteristics) in order to better understand the issues that aect this population. The ndings will assist in the development of strategies that will help reduce worker exposure to hazards inherent in steel erection work by addressing worker risk perception as a root cause of possible accidents. The following sections are intended to provide a better understanding of the steel erection process, its safety related aspects, and the variables selected for analysis. 2. Literature Review 2.1. Steel erection Steel erection consists of the placement and installation of iron or steel girders, columns, and other structural elements. Ironworkers unload and stack the prefabricated steel and then hoist the steel by attaching cables (slings) to the steel and to a crane or derrick. During the hoisting operation, one worker directs the hoist operator with hand signals while another worker holds a tag line attached to the steel to prevent it from swinging. The piece is then hoisted into the installation position. Once the piece is hoisted, a team of two or more ironworkers position the steel with connecting bars and jacks. Workers using drift pins or the handle of a spud wrench (a long wrench with a pointed handle) align the holes in the steel with the holes in the framework. Before the bolts are permanently tightened, ironworkers check the vertical and horizontal alignment and then bolt or weld the piece permanently in place. Steel erection work is usually performed in all kinds of weather. Work may be suspended during wet, icy, or extremely windy conditions when work is to be performed at great heights. Due to the high risk of falls in steel erection, workers use safety devices such as fall arrest systems, safety harnesses, and nets. 00049.tex April 16, 2007 20:3 WSPC/177-JCR 00049.tex Assessment of Risk Perception of Ironworkers 113 Training for ironworkers is provided through apprenticeship programs. Most employers recommend a three to four year apprenticeship consisting of on-the-job training and evening classroom instruction. Labor organizations play an important role in the training of ironworkers. Apprenticeship programs are usually administered by committees composed of representatives of local unions of the International Association of Bridge, Structural, Ornamental, and Reinforcing Iron Workers (IABSORIW) or the local chapters of contractors' associations such as the Associated General Contractors of America. 2.2. Safety in steel erection Falls from elevations are considered to be one of the leading causes of fatalities in steel erection, with as high as 63% of fatalities being the result of falls (BLS, 2002). The safety standards that most directly address safety issues in steel erection are the OSHA steel erection standard (Subpart R-Steel Erection) and the \"Construction and Demolition Operations Steel Erection Safety Requirements (ANSI A.10.13)\" by the American National Standards Institute (ANSI). The purpose of these standards is to prevent the increasing number of injuries and fatalities that result from hazards in the steel erection process, including falls from elevations. The OSHA standard that covers steel erection work is OSHA Subpart R, the latest revision of which became eective on July 18, 2002. Some of the major changes made to Subpart R-Steel Erection are related to fall protection requirements, the duties of controlling contractors, site layout, planning of steel erection activities, specic requirements for hoisting and rigging activities, design requirements for anchor bolts, and design requirements for double connections (U.S. Department of Labor, 2001). In addition to the changes to Subpart R-Steel Erection, there were revisions to Subpart M-Fall Protection, for purposes of consistency. Prior to the revisions to the standard in 2000, Subpart M stated that the requirements related to fall protection in steel erection were provided in Section 1926.105 and in Subpart R. The revised standard changed Subpart M to include fall protection requirements for construction of towers and tanks only. This revision, which ensures safety of steel erection activities, is covered exclusively by Subpart R. Even with the strict enforcement of these safety standards, workers still experience accidents that result in injuries and fatalities. Understanding workers risk perception is an important step in the improvement of work practices that take into consideration not only the physical requirements of the steel erection process but the safety attitudes of workers doing these tasks. 2.3. Risk-taking behavior and accident causation There are numerous denitions of risk, among which are the existence of threats to life or health (Fischho et al., 1981), exposure to the chance of injury or loss (Hertz and Thomas, 1983), and the likelihood that harm will occur (Health and April 16, 2007 114 20:3 WSPC/177-JCR J. Irizarry & D. M. Abraham Safety Commission, 1995). Risk-taking can be dened as following a course of action selected at the end of a probabilistic process. Risk-taking behavior has been identied as a leading cause of accidents (Wagenaar, 1992). In many accident reports, the causes of accidents are attributed to irresponsible underestimation or acceptance of risk. This leads to the hypothesis that misperceived risk or consciously accepted risk is a major cause of accidents. Two risk theories relevant to this paper are the risk homeostasis theory (Wilde, 1982) and the zero-risk theory (N aatanen and Summala, 1974, 1976). The risk homeostasis theory states that an individual's behavior in risky situations is determined by a desire for cost minimization, which explains how behavior can be in accordance with risks, even subjectively perceived risks, without an ever-repeated process of conscious risk evaluation. This theory suggests that no safety measure will ever help to reduce risk and that risk control measures should be replaced by cost control measures. The zero-risk theory states that people seek situations in which there is no risk. Forces that play a role in this model are perceptual, experimental, and motivational. Both of these theories are important to the study of the risk perception of ironworkers because they relate dierent dimensions of risk perception with resulting behaviors in risky situations. Understanding these relationships can contribute to the development of safety training programs that target worker risk perception as a method of hazard prevention and avoidance. 2.4. Factors influencing risk perception The experience of construction workers and their knowledge of safety are important factors to consider in the evaluation of ironworkers' risk perception. In a study of hazard perception and risk estimation in accident causation, Zimolong (1985), found that acceptable risk levels are established as a result of previous experiences and cognition. Information about accident-causing factors was obtained from investigating working conditions and personal behavior in hazardous situations. He concluded that workers are more likely to underestimate high risk situations if they had a long-time experience with these hazards. Worker behavior regarding safety may be inuenced by the worker's perception of what is safe or unsafe. Based on this perception, decisions are made when to adopt or not adopt required safety precautions. This relationship was observed by Huang and Hinze (2003), who found that approximately 33.3% of fall accidents are caused by misjudgment of workers about hazardous situations. Other factors can aect safety on construction sites by increasing the probability of accidents. Toole (2002) found that the lack of proper training, decient enforcement of safety behavior, not using provided safety equipment, and poor attitudes toward safety were among the root causes of construction accidents. Lack of proper training can limit the ability of a worker to recognize and avoid a hazardous situation and hence increase the risk of accidents. Decient enforcement of safety can 00049.tex April 16, 2007 20:3 WSPC/177-JCR 00049.tex Assessment of Risk Perception of Ironworkers 115 increase the risk of accidents since workers have less direction regarding applicable safety standards and there is less control of unsafe behaviors. Improper use of safety equipment is a common cause of construction accidents. The risk of accidents is signicantly increased when safety equipment is not used eectively. In many instances, the use of safety equipment is reduced when the worker perceives that performance will be adversely aected. The perception that using safety equipment aects performance was investigated by Irizarry, Simonsen, and Abraham (2005). The study indicated that task durations in steel erection were increased by small levels when fall protection equipment was used. Poor attitudes toward safety involves worker beliefs, values, and work ethic. Workers may have been trained properly, but a \"tough-guy\" mentality prevents them from avoiding job hazards (Toole, 2002). Since the causes discussed are behavioral in nature, the factors that are involved in the causal process can be used to learn about the risk perception of workers. 2.5. Factors influencing safety performance Prior research studies have identied a number of factors that could impact safety at the construction site. Hinze and Gambatese (2003) identied several factors that inuence the safety performance of specialty contractors (mechanical and roong contractors). This study compared the factors that were believed to impact safety performance with the median injury rate of the specialty contractors surveyed. They concluded that factors such as turnover, drug testing programs, worker training, involvement of trade associations, and safety inspections signicantly inuence the safety performance of specialty contractors. Safety incentive programs were also considered in the study, but strong evidence to support their eectiveness in reducing injuries was not found. Recommendations that resulted from the ndings of this study included the minimization of turnover, implementation of drug testing, and training with the assistance of trade associations (Hinze and Gambatese, 2003). Ahmad and Gibb (2003) identied the presence of a safety ocer and tool-box talks as some of the safety control measures (SCMs) that aect safety performance on construction sites. The relationship between age and worker fatalities was addressed by Chen and Fosbroke (1998), Buskin and Paulozzi (1987), Kisner and Fosbroke (1994), and Kisner and Pratt (1997). These studies suggest that older construction workers have a higher risk of injury. Height and weight are two factors that were found to aect safety performance. Kelsey and Golden (1998) found that workers with less than or greater than optimal body mass index have a higher risk of back injuries. The cited studies provide the motivation for the selection of the variables to be studied, which are believed to inuence the risk perception of ironworkers. The following section describes the methodology used for the assessment of ironworker risk perception. April 16, 2007 116 20:3 WSPC/177-JCR J. Irizarry & D. M. Abraham 3. Research Methodology In order to assess the perception of risk of ironworkers and to better understand the factors that could inuence the behavior that results in accidents, a survey was conducted among ironworkers across the United States. The motivation to learn about such attitudes arises from the safety-related behaviors that were observed in a related study conducted by the authors to evaluate the eects of safety and environmental factors on task durations in steel erection [Irizarry, et al., (2005)]. From the analysis of the behaviors observed in the previous study, several work practices that could be considered unsafe were identied, for example, not using personal protective equipment (PPE) while moving from connection point to connection point, while making connections above 15 and 30 ft, and while unhooking the steel elements. In order to develop strategies that minimize the hazards encountered during steel erection, it is important to learn why workers engage in such unsafe practices. Since the goal of the survey is to learn about the attitudes of ironworkers toward safety, the target population was ironworkers involved in the tasks of rigging, connection of steel members, and detailing and welding. Crane operators were also included in the survey. These groups were selected since the workers in these groups are exposed to two signicant hazards found in the steel erection process, namely, falls from elevations and being struck by objects. The use of the survey questionnaire was a non-experimental approach to the investigation of variables that may inuence the risk perception of ironworkers. The results are descriptive in nature and are not intended to establish cause-eect relationships. What the survey intended to reveal is whether the risk perception of ironworkers is dependent on the selected variables. The questions included in this survey were divided into four groups: safety-related, job site-related, worker-related, and personal aspects. Each question is considered as a variable that is believed to aect the risk perception of ironworkers. Appendix A shows the questions included in the questionnaire. In addition to the four groups of questions, additional questions were included that asked respondents to classify several actions using a four point Likert-type scale (from 4 = \"very dangerous\" to 1 = \"not dangerous\") and to classify several safety-enhancing strategies using a similar scale (from 4 = \"very important\" to 1 = \"not important\"). These questions are shown in Appendix B. An observed action that was included in the questions and that merits explanation is \"riding\" a beam during hoisting. This action implies the worker sitting on the beam while it is being hoisted to the installation position. This practice is illegal in steel erection, however, it was observed during the data collection of the study by Irizarry, et al. (2005) and therefore it was included in the survey. The survey was sent to steel erection companies and local unions of the International Association of Bridge, Structural, Ornamental, and Reinforcing Iron Workers in twenty (20) states across the United States. A total of 895 questionnaires were sent and 235 responses were received (26.3% response rate) from 13 states (see Table 1). As shown in Table 2, ve geographic regions were included in the survey. 00049.tex April 16, 2007 20:3 WSPC/177-JCR 00049.tex Assessment of Risk Perception of Ironworkers 117 Table 1. States included in the survey to ironworkers. States Geographic region Questionnaires sent Responses received Response rate Percent of total California Colorado Georgia Illinois Indiana Maryland Massachusetts New York Oregon Pennsylvania Texas Washington Wisconsin West Coast Central East Coast Mid West Mid West East Coast East Coast East Coast West Coast East Coast Central West Coast Mid West 100 55 20 40 52 5 20 20 30 80 93 80 40 59 4 1 15 35 5 9 16 5 22 33 21 10 59.0% 7.3% 5.0% 37.5% 67.3% 100.0% 45.0% 80.0% 16.7% 27.5% 35.5% 26.3% 25.0% 11.2% 6.1% 2.2% 4.5% 5.8% 0.6% 2.2% 2.2% 3.4% 8.9% 10.4% 8.9% 4.5% Table 2. Geographic distribution of survey to ironworkers. Geographic region Questionnaires sent Responses received Response rate West Coast Central Mid West East Coast 210 228 222 235 85 37 60 53 40.5% 16.2% 27.0% 22.6% Total 895 235 26.3% Two major limitations of the study were the lack of responses from companies and union locals from several states and the fact that most of the respondents were union workers (90.8%). Information obtained from the International Association of Bridge, Structural, Ornamental, and Reinforcing Iron Workers (IABSORIW) indicates that the ironworker trade is highly unionized. Of the 130,000 members of the IABSORIW, approximately 50% are ironworkers (Kicielinsky, 2005). Responses were not received from companies and union locals in states such as Florida and New Mexico. These states include large minority work groups from diverse ethnic backgrounds. This limitation precluded the inclusion of variables such as race and gender for the analysis. 4. Data Analysis The analysis of the survey to ironworkers includes descriptive statistics from the sample, and logistic regression of the responses to survey questions. Logistic regression is a statistical analysis tool for modeling the relationship between a response variable and a set of explanatory variables when the response variable is categorical. April 16, 2007 118 20:3 WSPC/177-JCR J. Irizarry & D. M. Abraham It is most often used when the dependent variable is dichotomous (i.e. \"dangerous\" or \"not dangerous\"). The risk perception of the ironworker is quantied by the eect that the variables studied have on the odds of the worker responding that an action is \"dangerous\" or \"not dangerous.\" 4.1. Descriptive statistics Information on important characteristics (i.e. work experience, safety training experience, accident and injury experience, age, and anthropometric characteristics) of ironworkers can be obtained from the survey data. The information obtained can increase the understanding of issues aecting ironworkers, such as training, accidents and injuries, ergonomics, an aging workforce, and work experience. Learning about these issues can contribute to the improvement of safety and health in the steel erection process. 4.1.1. Age of respondents The minimum age of the respondents was 19 years and the maximum age was 73 years, with a mean age of 40.4 years (n = 220). The data indicates that over 50% of the ironworkers surveyed are 40 years of age or older (see Fig. 1). It is possible that steel erection, due to its intense physical requirements, is less attractive for younger workers. This is of particular interest since it reveals the problem of an aging work force that will have increasing health issues and could be at a higher risk of injury. In a study of the lifetime risk of occupational fatal injury from age-specic rates, it was found that older workers (over 45 years) are at a higher risk of injury (See and Bailer, 1998). Since a signicant number of ironworkers from the sample studied are Fig. 1. Age distribution of respondents. 00049.tex April 16, 2007 20:3 WSPC/177-JCR 00049.tex Assessment of Risk Perception of Ironworkers 119 older than 40 years, they can be considered to be at a signicantly higher risk of injury. 4.1.2. Anthropometric characteristics The data collected regarding the anthropometric characteristics of ironworkers showed that the minimum height of the respondents was 1.55 m (5 ft1in) and the maximum height was 2.11 m (6 ft11in). The average height was 1.78 m (5 ft10in) (n = 222). The variance of ironworker height was 0.057. This information is useful when considering the ergonomic factors of the design of PPE for ironworkers. The minimum weight of the respondents was 58.97 kg (130 pounds) and the maximum weight was 149.68 kg (330 pounds). The mean weight was 87.08 kg (192 pounds) (n = 222). Knowing the range and average weight of ironworkers is also useful in the design of PPE. The required strength of materials and conguration of the equipment can be determined and optimized to reduce the additional weight that the workers must carry. 4.1.3. Experience in construction and steel erection process The experience of the respondents in construction work ranged from 1 to 54 years with a mean of 14.4 years (n = 218, std. error = 11). The work experience in steel erection also ranged from 1 to 54 years with a mean of 15.6 years (n = 217, std. error = 11.2). From the data it can be observed that there is large variability in the level of experience of ironworkers. Most of the ironworkers surveyed have spent most of their careers in steel erection (82% on average). 4.1.4. Worker classication Approximately eighty percent (79.1%) of the respondents indicated that they worked as ironworkers (n = 225). Supervisors composed 8% of the respondents and 13% of workers had the dual function of ironworker and supervisor. An ironworker performs any of the tasks required for the installation of structural steel members. A supervisor (foreman) is in charge of directing steel erection operations. The worker who has the dual function of ironworker and supervisor can perform any of the tasks performed by an ironworker and also directs specic parts of the steel erection operation, and he is referred to commonly as the crew leader. 4.1.5. Union aliation The data shows that a signicant number of the ironworkers surveyed are aliated with a labor union (90.8%, n = 228). Dissemination of safety research ndings through such organizations is a potential strategy for reaching a greater number of ironworkers. The established training program that unions provide to their members provides a strong vehicle for dissemination of safety information. April 16, 2007 120 20:3 WSPC/177-JCR 00049.tex J. Irizarry & D. M. Abraham 4.1.6. Tasks performed As shown in Table 3, ironworkers perform multiple tasks. The most frequent tasks are rigging (60.8%), detailing (54%), and connecting (49%). Ironworkers perform other tasks, such as crane operation, safety supervision and training, and welding. For this and other questions in the survey the percentages do not total 100% because a response could fall into more than one category. 4.1.7. Safety training As shown in Table 4, almost all ironworkers surveyed have received some form of OSHA-related training (97%, n = 230). Over half (60.3%, n = 214) of the respondents stated that this training took place over six months prior to the survey, 18.7% (n = 214) three months prior, and 21% (n = 214) one month prior. When asked about specic safety training, 87.5% (n = 232) of respondents stated that they had received OSHA 10-hr training versus 36.4% (n = 225) who stated that they received OSHA 30-hr training. A total of 87.2% (n = 235) of respondents indicated that they had received PPE training. The majority of ironworkers surveyed (58%, n = 205) received PPE training six months or more prior to the survey. Almost the same proportion of ironworkers received PPE training three months and one month prior to the survey (20.5% and 21.5% respectively, n = 205). The signicant number of ironworkers who have received safety-related training over six months prior to the survey may indicate the need for refresher training, especially when new equipment or methods are Table 3. Frequency of ironworker tasks. Task % of total who responded yes (n = 235) Rigging Detailing Connecting Crane Operator Welding Safety Ocer Trainer 60.8 54.0 48.9 2.5 0.8 0.4 0.4 Table 4. Safety training received by ironworkers. Type of training OSHA Related OSHA 10 hr OSHA 30 hr PPE % of total who responded yes 96.9 (n = 230) 87.5 (n = 232) 36.4 (n = 232) 87.2% (n = 235) April 16, 2007 20:3 WSPC/177-JCR 00049.tex Assessment of Risk Perception of Ironworkers 121 implemented. Although a signicant number of ironworkers have received some type of safety training, its eectiveness in reducing accidents could not be determined from the data. 4.1.8. Accidents and injuries in steel erection According to the data, 29% of the respondents (n = 229) had been injured in a steel erection accident. In addition, 10.5% of the respondents (n = 229) had been involved in a steel erection accident without being injured and 13.1% (n = 229) of the respondents had been injured while performing steel erection tasks without being involved in an accident. This nding suggests that some workers may suer minor injuries while performing steel erection tasks and not report the incidents to supervisors. Fewer than half of the respondents (46.7%, n = 229) were neither involved in steel erection accidents nor injured while performing steel erection tasks. 4.1.9. Productivity and PPE When asked if they perceived PPE as a factor aecting the productivity of the steel erection process, 85.7% (n = 231) of respondents stated that it did, 12.1% (n = 231) percent stated that it did not, and 2.2% (n = 231) did not know. This shows the extent of the belief among ironworkers that PPE aects productivity. 4.1.10. Job site practices Pre-work safety meetings, tool box meetings, and mandatory drug testing are job site practices that may inuence safety performance. The majority of ironworkers surveyed indicated that pre-work safety meetings and tool box meetings were conducted at their job sites (92.6% (n = 230) and 91.2% (n = 227) respectively). As shown in Table 5, the frequency of pre-work safety meetings and tool box meetings are similar. Mandatory drug testing was less frequent compared to pre-work safety meetings and tool box meetings. However, a signicant number of ironworkers (74.7%, n = 225) indicated that mandatory drug testing was conducted at their job sites. The remaining 21.3% (n = 225) indicated that mandatory drug testing was not Table 5. Frequency of pre-work safety meetings and tool box meetings. Frequency Weekly Daily Monthly Do not know Pre-work safety meeting (n = 208) Tool box meeting (n = 201) 75.5% 20.7% 2.4% 1.4% 71.6% 24.4% 2.5% 1.5% April 16, 2007 122 20:3 WSPC/177-JCR 00049.tex J. Irizarry & D. M. Abraham Table 6. Safety incentives and disciplinary action programs. Safety incentive programs (n = 234) Disciplinary action programs (n = 230) 47.8% 40.7% 11.5% 73.5% 16.1% 10.4% Yes No Do not know implemented at their job sites. Only a small number of ironworkers did not know if their employer had mandatory drug testing (4%, n = 225). 4.1.11. Safety incentive and disciplinary action programs As shown in Table 6, ironworkers indicated that disciplinary action programs are more common than safety incentive programs, which could be inuenced by variables not considered in this study, such as company size, safety management strategy, type of projects, etc. It appears that some companies choose a negative reinforcement approach to safety management by adopting disciplinary action programs. Such programs promote safety by penalizing unsafe behavior rather than rewarding safe behavior. 4.1.12. Importance of safety-enhancing strategies When asked to rate the importance of safety-enhancing strategies, the majority of the respondents stated that safety training was the most important strategy. Similarly important were the use of PPE, the presence of a competent person at the job site, and the availability of PPE that did not aect productivity. 4.1.13. Danger perceived in steel erection tasks Figure 3 shows the number of respondents who stated that the presented action was dangerous. It was observed that the actions that most workers perceived as dangerous were not using PPE while riding a beam during the hoisting operation and not using PPE while unhooking the steel member. As elevation increased from 15 ft to 30 ft, the number of workers who perceived the action of connecting as \"dangerous\" was almost identical. 4.2. Logistic regression Logistic regression models can be used when the response variable is qualitative and has two possible outcomes. In logistic regression the logistic response function is used to describe the nature of the relationship between the mean response (e.g. the probability of perceiving an action as dangerous or not dangerous) and one or more predictor variables. It can also be used to make predictions about the April 16, 2007 20:3 WSPC/177-JCR 00049.tex Assessment of Risk Perception of Ironworkers 123 Fig. 2. Importance given to safety-enhancing strategies. Fig. 3. Perceived danger of steel erection tasks. response to survey questions based on other questions and characteristics of the respondent. A complete discussion of logistic regression is available in Chapter 14 of Neter et al. (1996). The sections that follow describe the results of the logistic regression models applied to the survey data. Results are presented for models where independent variables had a statistically signicant eect on the dependent variable (action presented to respondents). 4.2.1. Eect of safety-related variables Only two of the safety-related variables signicantly aected the probability of respondents stating that the action of \"riding\" a beam during hoisting was dangerous. As shown in Table 7, the odds that workers perceived this action as dangerous are almost ve (4.9) times higher for workers who have been injured during steel erection activities than for workers who have not been injured. On the other hand, the probability that a worker perceives the described action as dangerous decreases for workers who have been involved in steel erection accidents. Table 8 shows that April 16, 2007 20:3 124 WSPC/177-JCR 00049.tex J. Irizarry & D. M. Abraham Table 7. Logistic regression results for safety-related variables. Eect Maximum likelihood estimate Odds ratio Pr > ChiSq 0.6354 0.7906 0.281 4.861 0.0189* 0.0133* Accident in steel erection (Yes vs. No) Injured in steel erection (Yes vs. No) *Statistically signicant at = 0.05, n = 191. Table 8. Logistic regression results for accident experience Eect Riding beam while hoisting w/o PPE (Dangerous vs. Not Dangerous) PPE available that does not aect productivity (Important vs. Not Important) Injured in steel erection (Yes vs. No) Years of experience in steel erection (per each 10 years) Pr > ChiSq Maximum likelihood estimate Odds Ratio 0.5774 0.315 0.0317 1.1101 0.109 0.0907 1.0454 8.091 < 0.0001 0.0649 1.914 0.0003 *Statistically signicant at = 0.05. **Statistically signicant at = 0.10, n = 195. workers who have been involved in steel erection accidents are 71.9% (10.281) less likely to perceive as dangerous the action of \"riding\" a beam during hoisting than workers who have not. It is interesting to note the dierent eects that these two variables have on the dependent variable of danger perception. The correlation between the two variables (r = 0.51539, p < 0.0001) suggests that a worker who has been involved in a steel erection accident has also been injured. However, the dierence in responses by the two groups suggests that workers who have suered injuries have a lower tolerance of risk since they perceive the same action as dangerous more often (93.6%) than workers who have been involved in steel erection accidents but who have not necessarily been injured (86.46%). The data were used to explore which variables might inuence the odds of a worker having an accident or being injured while performing steel erection tasks. In this analysis the two variables related to accident and injury experience were designated as dependent variables and the remaining as independent variables. Logistic regression models were developed for the two dependent variables to determine the eect on the odds of an ironworker having an accident or being injured. The results April 16, 2007 20:3 WSPC/177-JCR 00049.tex Assessment of Risk Perception of Ironworkers 125 Table 9. Logistic regression results for injury experience. Eect Riding beam while hoisting w/o PPE (Dangerous vs. Not Dangerous) Accident in steel erection (Yes vs. No) Safety incentives (Yes vs. No) Safety incentives (Do not know vs. No) Years of experience in steel erection (per each 10 years) Maximum likelihood estimate Odds ratio Pr > ChiSq 0.7220 4.238 0.0229 0.9630 6.862 < 0.0001 0.7200 0.9382 1.652 0.315 0.0168 0.0388 0.0496 1.641 0.0090 *Statistically signicant at = 0.05, n = 164. of the analysis showed that the odds of a worker having an accident or being injured in steel erection are most inuenced by the variables shown in Tables 8 and 9. The results show that workers who perceived that the action of \"riding\" a beam during hoisting as dangerous are 68.5% (10.315) less likely to be involved in a steel erection accident. However, for workers who stated that they have been injured while performing steel erection tasks, the odds that they perceive the same action as dangerous are 4.2 times higher (see Table 9). For workers who stated that the availability of PPE that does not aect productivity was important, the odds of stating that they have been involved in a steel erection accident are 89.1% (10.109) lower. The correlation between the variables of accident and injury experience suggested a relationship between these two variables. This relationship is clearer from the logistic regression results shown in Table 8. The odds of a worker stating that he had been involved in a steel erection accident are 8.1 times higher for workers who had been injured while performing steel erection tasks. Similarly, the odds of a worker stating that he had been injured while performing steel erection tasks are 6.9 times higher for workers who had been involved in an accident in steel erection (see Table 9). Another variable that signicantly aected the response variable of injury experience was the variable related to safety incentives. The odds of a worker stating that he had been injured are 1.6 times higher for workers who stated that their employer had an incentive program for safe work practices. For workers who did not know if their employer had a safety incentive program, the odds of being injured while performing steel erection tasks are 68.5% (10.315) lower. This result points to the possible ineectiveness of safety incentive programs in reducing injuries among workers in steel erection. The variable of years of experience in steel erection had a similar eect on the odds of a worker being involved in a steel erection accident or being injured April 16, 2007 126 20:3 WSPC/177-JCR 00049.tex J. Irizarry & D. M. Abraham Table 10. Logistic regression results for worker-related variables. Eect Maximum likelihood estimate Odds ratio Pr > ChiSq 0.3905 0.4294 0.458 2.361 0.0776** 0.0477* Connecting (Yes vs. No) Detailing (Yes vs. No) *Statistically signicant at = 0.05. **Statistically signicant at = 0.10. while performing steel erection tasks. The odds of a worker being involved in a steel erection accident were almost double per each ten years of experience in steel erection and over one and one-half times for being injured per each ten years of experience. For example, a worker with twenty years of experience in steel erection is almost four times more likely to be involved in a steel erection accident and almost three times more likely to be injured than a worker with only ten years of experience in steel erection. 4.2.2. Eect of worker-related variables One of the worker-related variables signicantly aected the dependent variable of danger perception for the action of \"riding\" a beam during hoisting. As shown in Table 10, the odds that the action of \"riding\" a beam during hoisting is perceived as dangerous are 54.2% (10.458) lower for workers who perform the task of connecting steel members. For workers who perform the task of detailing, the odds that they will perceive the same action as dangerous are 2.4 times higher. 5. Implications for Construction The methodology employed in this study can be extended to other trades in the construction industry. Its usefulness lies in the opportunity to evaluate the risk that workers perceive in the tasks they perform. Workers who misjudge risk could have a higher probability of injury from accidents. Therefore, a reduction in accidents could be achieved by addressing worker perception of risk as a cause of accidents. Improvements to present safety practices are possible in the areas of safety management, PPE, and process design. Safety management can be improved if site managers consider the characteristics of workers which inuence the risk they perceive in the tasks they perform. The result of this strategy could be a reduction of accidents and injuries by proactive rather than reactive safety management. Learning about the anthropometric characteristics of ironworkers can help in the development of PPE that has greater functionality and meets the ergonomic requirements of the work for which it is used. This could result in increased use by workers since the physical restrictions of using the equipment could be minimized. The design of job site operations is another area that can benet from the ndings of the study April 16, 2007 20:3 WSPC/177-JCR 00049.tex Assessment of Risk Perception of Ironworkers 127 and the methodology employed. Investigating the factors that inuence worker perception of risk can help to develop new ways of performing tasks. These new or redesigned processes would consider the physical limitations of the workers and the interaction between the worker, the materials and tools on the site, and the safety equipment used. 6. Conclusions The high number of construction-related injuries and fatalities remains a major problem for the construction industry. This problem is of particular importance for the steel erection industry since ironworkers have one of the most dangerous occupations in the United States. Finding ways to reduce the number of accidents that result in injuries and fatalities in steel erection should be a priority for construction safety research. This study contributes to the goal of reducing accidents by exploring factors that may inuence the risk perception of ironworkers by considering that risk perception is a root cause of accidents. The methodology employed in the study included a questionnaire survey of ironworkers across the United States. A varied sample was received from 13 states for a 26% response rate. The data collected was analyzed by obtaining descriptive statistics and applying logistic regression to specic survey questions that were intended to explore the risk perception of ironworkers. Logistic regression was used to determine the impact of the variables studied on the probability of workers perceiving an action as dangerous, thereby providing an indication of which variables inuence their perception of the risk involved in their work. The major limitations of the study were the lack of responses from companies and union locals from states that traditionally employ diverse ethnic groups and non-union workers. These limitations should be addressed in the future to obtain a more representative sample of the population. Following are the most relevant results reported in the paper: (1) The age of the ironworker population is a cause for concern. The data shows that a signicant number of ironworkers are 40 years of age or older, which reveals that ironworkers can be susceptible to the problems facing an aging population that is required to perform physically-demanding tasks and that are at a higher risk of injury. (2) Greater importance should be given to the anthropometric characteristics of ironworkers for the design of PPE. The ndings show that these characteristics widely vary. If these variations are considered in the design of PPE, increased usage could result. One common reason given by ironworkers for not using the provided PPE was the limitation of movement that PPE causes. This problem could be addressed by implementing more ergonomic designs for PPE. (3) Ironworkers perform multiple tasks and each of these tasks has unique hazards. This must be considered in the development of tasks-specic safety training. April 16, 2007 20:3 128 WSPC/177-JCR J. Irizarry & D. M. Abraham (4) A signicant number of ironworkers have received some type of safety training. However, the eectiveness of each type of training in preventing accidents and injuries cannot be determined from the data. Most of the workers surveyed had received training six months or more prior to the survey, which indicates a need for refresher training, especially since ironworkers can perform multiple tasks that have specic safety issues that are particular to each task. (5) Job site safety practices appear to be similar at most job sites since a signicant number of ironworkers indicated that weekly pre-work safety and tool box meeting took place at their job sites. However, drug testing was less frequent than pre-work safety and tool box meetings. It is important to determine why this occurs since it has been shown that drug testing on construction sites positively inuences safety. (6) It was also found that disciplinary action programs are more frequent than safety incentive programs. This indicates that negative reinforcement by penalizing unsafe behavior is a preferred practice by safety management in steel erection projects. Its eectiveness in reducing accidents should be compared to positive reinforcement strategies such as incentive programs for safe behavior. (7) When asked to indicate whether a given action was dangerous, a higher number of ironworkers indicated that actions considered as extreme (i.e. \"riding\" a beam) were dangerous. A lower number of ironworkers indicated that other routine actions (i.e. connections) were dangerous. This result gives an indication that safety management should be vigilant of workers performing tasks that they consider as dangerous less frequently since there is the risk of underestimation of the danger involved. (8) The results also showed the eect that safety-related variables, such as the accident and injury experience of ironworkers, have on their perceptions of risk in the tasks they perform. Workers who perceived specic actions as dangerous were less likely to be involved in steel erection accidents and workers who had been injured while performing steel erection tasks are more likely to perceive the same actions as dangerous, which indicates that ironworkers who have experienced an accident or an injury in steel erection may be more aware of the risks involved in their work. It is important that safety training consider the accident or injury experience of workers. This strategy can help to increase the awareness of workers to the risks involved in their work before an accident occurs. (9) Analysis of worker-related variables showed that workers who perform the task of detailing perceive tasks as dangerous more often than workers who perform the task of connecting. Again this indicates that safety training that considers the type of tasks performed by ironworkers could be used to increase awareness of the dangers involved in steel erection. 00049.tex April 16, 2007 20:3 WSPC/177-JCR 00049.tex Assessment of Risk Perception of Ironworkers 129 (10) Most workers stated that safety training was the most important safetyenhancing strategy. Safety training programs could be improved by considering the characteristics of ironworkers that impact their perception of risk. By incorporating these characteristics, safety training would address the needs of specic groups of workers. The ndings of this study reveal important information about the perception of risk of ironworkers. In addition to the desire for cost minimization as the factor that determines an individual's behavior in risky situations (Wilde, 1982), there are other factors that can inuence the risk perception of workers. It is clear that construction workers do not always conform to traditional risk theories, such as the zero-risk theory (N a atanen and Summala, 1974; 1976) since they do not always avoid risky situations. If improved safety is to be achieved, these factors should be incorporated into safety training programs that target worker risk perception as a method of hazard prevention and avoidance. Acknowledgements The authors would like to acknowledge the valuable assistance of Mr. Frank Migliaccio, Executive Director for Safety of the International Association of Bridge, Structural, Ornamental, and Reinforcing Iron Workers in the deployment of the survey to ironworkers. Also, we acknowledge the assistance of Mrs. Gayla Hobbs, statistical consultant at Purdue University for her assistance in the analysis of the survey data. References Abdelhamid, TS and Everett, JG (2000). Identifying root causes of construction accidents. J. Constr. Engrg. Mgmt., ASCE 126(1), 52-60. Ahmad, RK and Gibb, AGF (2003). Measuring safety culture with SPMT eld data. J. of Construction Research, 4(1), 29-44. Arboleda, CA and Abraham, DM (2004). Fatalities in Trenching Operations Analysis Using Models of Accident Causations. J. Constr. Engrg. Mgmt., ASCE 130(2), 273-280. Brown, ID (1995). Accident reporting and analysis. Evaluation of human work, JR Wilson and EN Corlett, eds., Taylor & Francis, London. Bureau of Labor Statistics BLS, (2002). National Census of Fatal Occupational Injuries in 2001, U.S. Department of Labor, Washington, DC. Buskin, SE and Paulozzi, LJ (1987). Fatal injuries in the construction industry in Washington State. Am J. Industrial Medicine, 11, 453-460. Chen, GX and Fosbroke, DE (1998). Work-related fatal-injury risk of construction workers by occupation and cause of death. Human and Ecological Risk Assessment, 4(6), 1371-1390. Chua, DKH and Goh, YM (2004). Incident Causation Model for Improving Feedback of Safety Knowledge. J. Constr. Engrg. Mgmt., ASCE 130(4), 542-551. Edwards, DJ (2003). Accident Trends Involving Construction Plant: An Exploration Analysis. J. of Construction Research, 4(2), 161-173. Fischho, B, Lichtenstein, S, Slovic, P, Derby, SL and Keeney, RL (1981). Acceptable risk. New York: Cambridge University Press. April 16, 2007 130 20:3 WSPC/177-JCR J. Irizarry & D. M. Abraham Health and Safety Commission (1995). Designing for health and safety in construction, HSE Books, London. Hertz, D and Thomas, H (1983). Risk analysis and its applications, Wiley, New York. Hide, S, Gibb, A, Haslam, R, Gyi, D, Pavitt, T, Atkinson, S and Du, R (2003). Tools and Equipment Their role in accident causality. CIB W99 International Conference on Construction Project Management Systems: The Challenge of Integration., Sao Paulo, Brazil, March 2003. Hinze, J and Gambatese, J (2003). Factors that inuence safety performance of specialty contractors. J. Constr. Engrg. Mgmt., ASCE 129(2), 159-164. Huang, X, Hinze, J (2003). Analysis of Construction Worker Fall Accidents. J. Constr. Engrg. Mgmt., ASCE 129(3), 262-271. Irizarry, J, Simonsen, KL and Abraham, DM (2005). Eect of Safety and Environmental Variables on Task Durations in Steel Erection J. of Construction Engineering and Management, ASCE, 131(12), 1310-1319. Kelsey, JL and Golden, AL (1998). Occupational and workplace factors associated with low back pain. Occupational Medicine: State of Arts Reviews, 3(1), 7-16. Kicielinsky, T (2005). Personal communication with editor of The Ironworker Magazine of the International Association of Bridge, Structural, Ornamental, and Reinforcing Iron Workers. January 12, 2005. Kisner, SM and Fosbroke, DE (1994). Injury hazards in the construction industry. J. Occupational Medicine, 36, 137-143. Kisner, SM and Pratt, SG (1997). Occupational fatalities among older workers in the United States: 1980-1991. J. Occupational and Environmental Medicine, 39, 715-721. Lehto, M and Salvendy, G (1991). Models of accident causation and their application: Review and reappraisal. J. Eng. Technol. Manage., 8, 173-205. N aa tanen, R and Summala, H (1974). A model for the role of motivational factors in drivers' decision making. Accident and Prevention, 6, 243-261. N aa tanen, R and Summala, H (1976). Road user behavior and trac accidents. Amsterdam: NorthHolland. Neter, J et al. (1996) Applied Linear Statistical Models, 4th Edition, McGraw-Hill Burr Ridge, Illinois. See, K and Bailer, AJ (1998). Estimates of lifetime risk of occupational fatal injury from age-specic rates. Human and Ecological Risk Assessment, 4(6), 1309-1319. Toole, TM (2002). Construction site safety roles. J. Constr. Engrg. Mgmt., ASCE, 128(3), 203-210. U.S. Department of Labor. (2001). Occupational safety and health standards for the construction industry. 29 CFR 1926, Occupational Safety and Health Administration, U.S. Government Printing Oce, Washington, DC. Wagenaar, WA (1990). Risk Evaluation and the Causes of Accidents. In Borcherding, K., Laricher, O.I. and Messick (eds), Contemporary Issues in Decision Making. Amsterdam: North Holland. Wilde, GJS (1982). The theory of risk homeostasis: Implications for safety and health. Risk Analysis, 2, 209-225 Zimolong, B (1985). Hazard perception and risk estimation in accident causation. In Eberts, R & Eberts, C (Eds.), Trends in ergonomics/human factors II. pp. 463-470. Amsterdam: Elsevier. 00049.tex April 16, 2007 20:3 WSPC/177-JCR 00049.tex Assessment of Risk Perception of Ironworkers 131 Appendix A. Survey Questions Regarding Characteristics of Ironworkers Variables Questions Safety-Related Have you received OSHA related safety training? When was the last time you received safety training? Have you received OSHA 10 hour training? Have you received OSHA 30 hour training? Have you been trained in proper use of Personal Protective Equipment? When was the last time you received PPE training? Do you know what Personal Protective Equipment is? Do you know what the OSHA standard that applies to steel erection is? Do you know what a competent person is? Have you ever been involved in a steel erection accident? Have you ever been injured while performing steel erection related tasks? Do you think that Personal Protective Equipment aects the productivity of the steel erection process? Job Site-Related Does your company use the practice of pre-work safety meetings? If pre-work safety meetings are used, what is their frequency? Does your company use the practice of toolbox meetings? If toolbox meetings are used, what is their frequency? Is there a competent person at your job site? Does your company have an incentive program for safe work practices? Does your company have a disciplinary action program for un-safe work practices? Does your company have a mandatory drug testing program as part of its hiring requirements? Worker-Related Classication: Do you belong to a Labor Union: Task that you perform most of the time: Years of experience in construction: Years of experience in steel erection: Personal Aspects Do you have a personal computer at home? Do you have access to the internet? Height Weight April 16, 2007 20:3 132 WSPC/177-JCR 00049.tex J. Irizarry & D. M. Abraham Appendix B. Survey Questions for Assessment of Risk Perception and Importance of Safety-Enhancing Strategies Items Questions Observed Practice Personal protective equipment (PPE) not used when moving from connection point to connection point PPE not used when making connections below 15 ft PPE not used when making connections above 15 ft PPE not used when making connections above 30 ft PPE not used while \"riding\" beam during hoisting operation PPE not used while unhooking element Safety-Enhancing Strategies Safety Training Use of Personal Protective Equipment Presence of a Competent Person Availability of Personal Protective Equipment that does not aect productivity