Skip to Content
Construction Safety

Fundamentally, Safety Is a People Problem—An Overview

Peter Furst | March 1, 2013

On This Page
Group of construction workers of different ages

If you want to improve your company's safety performance, stop focusing on your accident statistics, loss source analysis, traditional interventions, and regulatory compliance and start thinking holistically. We need to look for and find our answers in the elemental aspects of the organization.

Most companies create an output (a product or service). They have internal systems (plant, tools, equipment, technology, policies, procedures, means, and methods) with which to create that output and people to energize, operate, and manage the systems. But first, we have to dispel some myths about the management of safety before tackling the improvement effort.

Figure 1: Elemental Aspects of an Organization

Elemental Aspects of an Organization

The belief that regulatory compliance ensures positive safety results is something of a myth. Complying with Occupational Safety and Health Administration (OSHA) standards does not necessarily ensure you have a safe workplace. Virtually all things covered in the safety standards are conditions focused and managed by site inspections and engineering controls. Herbert William Heinrich's maxim that the vast majority of accidents is caused by unsafe acts provides a clue to the problem with OSHA compliance.

H.W. Heinrich's research found that 10 percent of accidents were caused by unsafe conditions, while 88 percent were caused by unsafe acts. Since the standards concentrate on the 10 percent portion of the sources of accidents, there is little chance of truly improving safety performance. Dr. Heinrich's research was reaffirmed by a later study conducted by Frank E. Bird, which attributed 5 percent of accidents to conditions and 95 percent to unsafe acts. Given this, how much improvement do you think we can get with the traditional approach to safety management?

This does not mean that the safety standards have no value. Of course not! They have had some effect in improving the overall occupational safety statistics. This improvement was mostly achieved by making the worst companies better, but they have done little in improving the safety performance of the good companies. If we are truly serious about dramatically improving our safety outcomes, we need to identify and explore more effective means for achieving better results.

Improving Safety Conditions

The first axiom in Dr. Heinrich's book, Industrial Accident Prevention, first published in 1931, stated that accidents are caused by unsafe acts and conditions. Let's look at the conditions part of this statement. To get injured, there usually needs to be an unsafe condition to which the worker is exposed; then the worker needs to perform the unsafe act. Had the unsafe condition not existed, the worker might not have been injured. So, why was the unsafe condition present? Behind every unsafe condition, there is a management system that allowed it to occur and exist. More on this later.

Next, let's look at the premise that accidents are mostly caused by unsafe acts. This has resulted in focusing on the worker to improve safety performance outcomes since it is the worker who gets injured after all. This has spawned a number of interventions. If you look at the safety standards, there are numerous instances where training is required. This is to familiarize workers with the standards. The underlying premise is that, armed with this knowledge, the worker will perform the work in a safer manner. In practice, to this end, there are toolbox talks, safety meetings, training sessions, signs and posters as reminders, feedback, observations, etc. All of these are there to change the worker's behavior. Despite all these tools, workers still get injured. So there must be other factors at play that are not effectively addressed by these interventions.

To devise improvement, interventions loss data is usually analyzed. This information primarily comes from accident investigation reports submitted to the insurance company. These reports are reviewed by claims personnel, who assign a cause label and enter the information into their database. The focus of claims is quite different from that of accident prevention, so using claims data does not necessarily provide operationally effective intervention information. But safety improvement interventions usually use this information to formulate improvement strategies. Safety improvement strategies need information on why the person did what they did much more than how and when they were injured so as to come up with an operationally effective prevention strategy. By focusing on unsafe acts and conditions, organizations end up dealing with accidents on a symptomatic rather than at the causal level.

There are the metrics used to measure and manage safety performance. The prevailing measures of safety performance are the frequency of accidents and their severity (mainly, the total recordable and the "Day Away, Restricted, Transferred" rates). This sort of data might provide general information on performance compared to previous years or Bureau of Labor Statistics published values, but they have little correlation to business metrics. Namely, they provide no direct evaluation of the efficiency or value of any particular intervention or how any of this contributed to operational effectiveness or return on the invested resources. Because safety management has devised unique measures that aren't very meaningful to senior management, these executives have concluded that the techniques that work for improving quality, productivity, or partner relations are not applicable for safety improvement because it's somehow different.


I was asked to get involved in a situation where a residential framer, who was at a 140 percent loss ratio, vehemently disagreed with the loss control consultant's recommendations for reduction of its accident rate and, as a result, their cost of risk. The insured's most significant problem areas were a very high incidence of workers getting injured while carrying material, injuries from falls, and injuries from access issues. The loss data used by the insurance consultant came from his vast claims database.

Of greatest concern to the underwriter, who reviewed the claims data, was that this risk's manual material handling claims ran about twice that of its book for similar businesses. He dispatched the loss control consultant to address this issue. The consultant's recommendation was to use mechanical aids (such things as pallet jacks, dollies, etc.) in its material handling process. The insured maintained that these recommendations were impractical, which led to my involvement.

After reviewing and discussing the circumstances around a number of these accidents, it became evident that, while carrying materials (mostly beams, joist, and plywood), the workers tripped on debris on the floor. This resulted in a housekeeping recommendation that addressed the underlying problem and was operationally feasible. In fact, the framer stated that, even though cleanup was not his responsibility, its cost was by far much less than the savings he would realize in insurance costs as well as operational efficiencies, and he would have an avenue to either get the developer to do it more effectively or collect the cleanup cost from it.

Changing Processes

Accidents are rare events. There is research that shows a worker can work 5–7 years before having an accident. Engaging in unsafe acts does not routinely or necessarily result in accidents. This leads to acceptance of such acts as normal practice. For most organizations, particularly smaller ones, recordable injury rates have little statistical validity. They neither diagnose nor fix what went wrong. They also have very little value in providing "real-time" information leading to operational solutions. Moreover, injury statistics are downstream measures, and what is really needed are upstream operational measures that tell you what is going on or whether your systems, practices, procedures, and processes are getting any better going forward due to any changes instituted.

Another fundamental problem is the fact that organizations implement interventions (audits, inspections, training, incentives, feedback, punishment, etc.) based on an analysis of the previous year's loss data. After implementation, they have no good indication of what really worked or to what extent any of these interventions truly had an impact on the outcomes. There is little or no correlation between the inputs and the outcomes. This leads to repeating potentially ineffective improvement measures, such as priority or emphasis programs, which at best are ineffective or useless at worst. Organizations cannot afford to misuse time and valuable resources without positively impacting operational efficiency and/or the bottom line.

In larger organizations where there are multiples divisions, business units, sites, or projects, the safety performance of these entities usually varies. Some have lower losses than others despite the fact that they all adhere to the same organizational policies as well as use the same safety program. They all have safety meetings, perform job safety analyses, engage in safety planning, conduct daily site inspections, and require subcontractors to hold weekly toolbox talks. The sites with the higher losses get attention, while the others are pretty much left alone. This does not make any sense, as all locations can improve, and the good ones certainly can get better. The question should be why these techniques are more effective in some locations than in others. This apparent variability, to some extent, has to do with the effectiveness of planning and the execution of the program on the part of the people at the various locations.

Examining Personnel

Let's look at the people involved. There is the worker who gets injured and the supervisor (foreman) who oversees the crew. The other people involved are the superintendent and project manager responsible for the site. And, of course, there are all the people at the main office. Effectively, there are three management levels plus staff (supervision, middle managers, and executives) who can and should exert influence on the safety outcome. Yet, traditionally, it is the worker who is the focus of the effort and the safety (practitioner/professional) person who tries to remedy the situation. This does not make sound operational sense and adversely impacts the business.

The worker should try to work safely, use good judgment and practices, utilize relevant information, pay attention to the task at hand, certainly be aware of the work environment, and try to be efficient and effective as well as productive. Despite all that, the worker is still human and prone to make mistakes. If management changes (reduces) the risk picture, the resulting outcomes should be better. If the worker is somehow deficient, is not capable, or does not have the knowledge or motivation, then it is management who did the hiring. Having a robust hiring practice, which excludes "unacceptable" workers, is a management responsibility. So, if management hired a "deficient" worker, then they either should have remedied the deficiency before making the task assignment or should have assigned the worker to a task that he or she is capable of performing. Task assignment is a management responsibility.

If the worker has "bad" habits, is a risk taker, is not prone to follow directions, or has any other unacceptable traits that may not have been apparent during the hiring process, then it is the supervisor who should observe the worker's behavior and take action to remedy the situation or remove the worker from the task. If the worker was a "good" performer and somehow changed later (say, due to some personal issues), then again it is the supervisor who should do something about the adverse situation to correct it; otherwise, the occurrence of an accident is a foregone conclusion.

Organizational Holistic View

In looking at the figure, one can see a distinct link between the various systems at the different levels of the organization and the people who devised and/or oversaw them, leading to the resulting unacceptable or unanticipated outcomes. In construction, work in place is the output. Inefficiency, poor quality, and injuries are outcomes, not outputs. They are a side effect of the operational process that is influenced by the organizational systems and management's goals and expectations. Unsafe conditions as well as the unsafe acts may be caused by the worker's perception or understanding of what is required or expected by management.

If the operation is conducted in a different way (change the task design, modify the production goal, alter the task demand, revise the process, or adjust the means or methods), or if management evaluates the level of the risk and does something to reduce its severity, the outcome could be different. Much of the process is devised and controlled by management. This is not to say that all risk can be eliminated, but, with a robust risk assessment and planning process, either a large percentage of the hazards can be eliminated or their adverse outcome modified to acceptable levels. Those hazards that cannot be eliminated or modified should be identified and discussed before any task starts to see what additional measures to reduce the exposure or modify the work process, task design, assignment, or any other mitigating procedure to ensure minimization of loss can be taken.

To manage, you need to measure, and to measure, you need to have goals. All companies have production, qualilty, and safety goals. It is a fact that organizational metrics drive behavior. So, if there are multiple goals and metrics, which there usually are, then there is the potential for confusion or conflict. Some metrics are easier to come by than others are. Production metrics become readily apparent and are easily observed. This can be measured on a daily, even hourly, basis. Quality results are not as readily apparent, and safety results are even more difficult to effectively measure on a daily basis.

When it comes to goals, there is the belief that "stretch" goals are a good thing. They motivate the worker to try harder or use ingenuity to do better. This is true to some extent, but successive "stretch" goals eventually put the worker at a point where he or she reaches maximum capacity, and no amount of discretionary effort will increase output. This is the point where the worker may have to make a choice between production and safety. Since the worker generally needs to work, and a shortcut (unsafe act) may increase productivity, the choice is a no-brainer! This is a system-driven risk!

Behind every unsafe act, there is a reason that caused the worker to make a decision that resulted in the unsafe behavior. In most cases, there is a management policy or practice that leads the worker to perceive he or she is making a choice that management wants or expects. This may or may not be the case, but perception becomes reality, which can potentially cause unexpected or unwanted outcomes. To be "successful," the worker has to function within the systems of the organization. Many of the behaviors and risks taken are a result of this symbiotic relationship. Therefore, to structurally eliminate risks (hazards) from the organization's systems, management must ensure that all the systems are integrated and the practices are aligned with the business goals.

To solve the safety problem, we primarily have to look at the "management" people much more than we do at the "worker" people. We have to find our solutions at the organizational level and not the unsafe condition or act level. It is the management people who define the values and create the culture that delineates the norms, symbols, rituals, artifacts, heroes, and stories. We must assess the climate to see what is deemed acceptable and what is not. We have to look at the people in the organization and the systems they devised to drive it. How much emphasis is placed on leadership as opposed to management? Is safety an intrinsic part of everything that is planned and done? It is leadership that inspires excellence while management stresses achieving goals. Are people put into situations where they have to choose between production goals and safe work practices? We have to determine why workers do what they do and how the organizational systems allow or even encourage such actions. Until we get to that causal level, we are not going to get very far in improving our safety results.

There is another issue to note. Unless we eliminate the systems that allowed risk to enter into the operations as well as allowed it to exist, intervention focused on changing or removing a worker will not effectively solve the problem. The system will eventually "get" the next worker and then the next and the next and so on. Therefore, management must be vigilant to ferret out any possible situation that may cause confusion or is open to misinterpretation. Clear communication of objectives can play a big part in avoiding confusion. The organizational metrics, how people are measured and rewarded, the culture of the organization, and the work climate all may lead the worker to thinking that unsafe behavior is okay. Excess pressure to achieve goals or to meet the schedule increases stress and may lead to errors resulting in inferior work quality or worker injuries.

Opinions expressed in Expert Commentary articles are those of the author and are not necessarily held by the author's employer or IRMI. Expert Commentary articles and other IRMI Online content do not purport to provide legal, accounting, or other professional advice or opinion. If such advice is needed, consult with your attorney, accountant, or other qualified adviser.