Eyewitnesses and Other Sources of Accident Data:
The Investigator's Dilemma

August 2003

The key to an incident investigation is data: the physical attributes of the site, the project and event artifacts, and eyewitness testimony. These three principle sources are needed to conduct a thorough construction accident investigation, but none is more worrisome than the recollections of witnesses. Investigators need to get the accounts quickly before they can be biased, follow leads, and approach each narrative with a healthy dose of skepticism.

by Ron Prichard, P.E. PhD.
Arcanum Professional Services

The key purpose of an incident investigation is to gather facts, like tiles of a mosaic, and then to assemble them into a coherent picture that describes the events leading up to, and which are part of the event of damage. For an investigation of any event to be complete, the investigator must answer who, what, why, when, and how. The order of addressing these is less important than the need to be as complete as possible with the description.

An earlier article in this series addressed the why aspect: why conduct an accident investigation and why did the event happen? How to conduct a systematic investigation was addressed in “On the Trail of Truth: Conducting an Accident Investigation.” This installment looks at the “who” and the “what” of the investigation.

Sources of Data

The key to an incident investigation is data. Without it, the process does not function, since there is nothing to evaluate. Thus, the mission of the investigator is to be thorough and complete in the search for data. Each piece of data is like a tile in a mosaic. The investigator cannot assemble a clear or coherent picture of the event without a sufficient number of pieces. The more data that can be collected, the more confidence that can be placed in the validity of the conclusions of the investigation. Yet, too much data can overcomplicate the picture and prevent assembly into an understandable picture of the event.

The investigator seeks to uncover as much information as practicable; however, even the most exhaustive search will still result in gaps in information. Some things may not be known by anyone, or may be unknowable. These gaps will either be left open as questions unanswered, or filled in through the experience, expertise, and insight of the investigator through assumptions, projections, educated “guesses,” or even conjecture. There is always the need to strike an appropriate balance between the effort required to ferret out information and the value of that information. This is related to the need for “stop rules” before commencing the investigation—the investigator needs to know when enough is enough.

There are three principal sources of data:

  • The physical attributes of the site of the event
  • The project and event artifacts
  • The eyewitnesses

The Physical Attributes of the Site. This relates to the components of the actual scene of the event. It involves an examination into the elements of space and things within that space, to evaluate how they might relate to the activities leading up to the event, and the actual occurrence. It includes a wide variety of information, such as the following.

  • Weather
  • Lighting
  • Visibility
  • Noise
  • Distractions
  • Time of day
  • Location within the project
  • The proximity of various aspects of the parts and pieces of the structures, equipment, and material (both installed and awaiting installation)
  • Interconnections with other components of the installed material and equipment
  • Access and constraints of the work area
  • The physical setting and the placement of people, machines, and other objects
  • The specific work tasks involved (and their purpose)
  • What other tasks were occurring around the event
  • What sequence of work was involved, its connectivity in terms of preceding and succeeding activities
  • Criticality of the work
  • The context of the work in relation to the overall construction of the project

This first set of data helps to develop facts that explain the context of the event, by defining the attributes of the physical environment in which the event occurred.

The Project and Event Artifacts. The second source of data represents physical items that you can hold see, touch, review, and study. They are items related to the event, but which are not necessarily components of the physical environment. The first category in this data set is the items directly involved, and thus to be found on the scene of the event, such as:

  • Personal protective equipment
  • Hand and power tools
  • The material or equipment being worked on or with
  • Tools and equipment, such as cranes and lifts, ladders or scaffolds, welding machines, compressors, cutting tools or other such machines or equipment designed to supplement or facilitate the physical labor of the workers being used in the work

The next category of artifacts is those items to be found in vehicles, on equipment, in job boxes, storage areas, and trailers, on bulletins boards, or within offices. The artifacts in this second category are primarily documents of various sorts. This includes documents such as the following.

  • Policies
  • Procedures
  • Instructions
  • Directions
  • Meeting minutes
  • Toolbox talks
  • Training manuals and material
  • Orientation briefings
  • Plans
  • Programs
  • Contract documents
  • Job memos
  • Other written material

By themselves, these artifacts will tell no story, and oftentimes, there might even be conflicts between the positions or information contained in them. However, they will provide valuable background material further defining the context of the event.

The role of the investigator is to evaluate the source and use of the documentation in this second set of artifacts, and query them for their reason and purpose before reaching conclusions. Artifacts do have value in explaining some of the activities leading up to an event, and are generally viewed to be more reliable than personal statements. They also provide insight into other influences on the sequence of actions through their description of the information available prior to the event. However, they can be intentionally misleading, or factually incorrect, and cannot be accepted as valid simply on their face.

The Eyewitnesses. The third source of data is people: the individuals who have information about the event, activities surrounding it, or actions leading up to it. This source consists of three groupings of individuals.

Think of these categories as three concentric circles radiating out from the physical point of the event. The first group is the inner circle. This includes only those individuals who were actual direct participants engaged in the event. It can, but does not always, include both individuals who are injured parties as a result of the event, and those coworkers who were actually directly engaged in the actions leading up to and involved in the event.

The second group, the middle circle of people, are those who were not directly involved in the activities under investigation, but who, due to their proximity to the activities in question, were able to see (and thus comment on) what they saw happen.

The third category, the outer circle, are those individuals who did not actually see or participate in the activities but can provide insight into influences, background, and other pertinent information relating to the event and activities leading up to it. While those people must be questioned, it needs to be recognized that they can seldom shed any light into the actual happenings surrounding the event due to their separation, both in the sense of time and from physical proximity, to the actual location of the event. As indirect participants, they seldom can speak to the details.

The most valuable information for the investigator involves the details related to what actually occurred. This information, if available, can only be acquired through discussions with individuals in the middle and inner circles. As either direct participants in the event, or through their proximity in space and time, these individuals can provide direct observations about what occurred. They are part of a particular classification of participant known as eyewitnesses. The balance of this article will focus on the features of this special classification and considerations that accompany it.

The Importance of Eyewitnesses

The eyewitness is the most valuable source of information an investigator can tap. Eyewitnesses can speak in a way that no amount of documentation or visits to the actual site can. They provide cues and clues, describe the actual actions and activities that led up to the event, describe recollections, and place things into context. These are all vital aspects for an investigator to use.

The Problems of Eyewitnesses. The principal caveat with an eyewitness is related to the memory process and the associated memory tasks. Human memory is better known for its fallibility and malleability than its reliability. Memory tasks include recognition, reconciliation, and recall, while the memory process involves encoding, storage, and retrieval. As a result, the conditions that trigger the creation of the memory are an important determinant in the veracity of it. Our brains don’t function like a computer. Memory is easily contaminated, and it becomes difficult to sort out what is actually known and what is derived from information provided by others.

The brain creates a memory (the encoding step) by noting a few facts. Often, memories relate hearsay from others, a transmission process fraught with distortion, exaggeration, overemphasis, supplementation, or omission. In extreme cases, memories can be completely made up, in a situation known as False Memory Syndrome. Fundamental human information processing limitations—the “compression effect”—is another problem. The brain’s “reducing valve” shrinks down the amount of detail about a situation to ease the storing of the event in memory. These are among the more common encoding errors.

Also, memory becomes fallible due to mistakes made in the mental storage process step. Our mental “filing system” is notoriously prone to mix-ups, with our mental “notes” (used for encoding) being put in one place initially, until the brain has had time to consider and compare it to other “notes,” wherein the memory is refiled elsewhere. Fragmentation occurs when different elements are “filed” in different places, with “indexing errors” resulting when the brain forgets where it put the various component pieces. “Entanglement” involves memories of different events getting intertwined. As our mental “notes” are pulled and refiled, these small errors made in the retrieval and restorage process means that with each recounting, the details of an event become less accurate.

What we remember depends on where we were, what we actually saw (and heard and felt) versus what we thought we saw (and heard). Through selective observation (wherein the brain seeks out confirming evidence of things it already knows), personal bias (wherein we tend to see only what we are prepared to see and ignore incongruous evidence, since it demands more mental energy and presence of mind to “hold”), and impressionistic recall (where the brain gravitates to the lesser effort task of accommodating confirming evidence) the input of memory can be distorted from the very start.

Not only are memories fallible, as we assemble memory out of the bits and pieces of what is seen, heard, and experienced, it is also malleable. Memories change over time, and are easily susceptible to modification. Additional bits are picked up from other sources, leading to inventions of details to fit the picture. Through this “integration” effect, separate, but related events (even if spread over a long period of time) will get merged into a single, indistinguishable memory.

The malleability of memory leads it to be particularly unreliable after a witness has either spoken to or exchanged notes with others. In these situations, details or elements that were not remembered in the original memory get modified as the individual incorporates the recall, insights, and observations of others. Gaps in memory get filled in, and memories get distorted through repeated telling and retelling, by a phenomenon known as the information distortion sequence.

Mistaken details are common, as generally, witnesses are not paying close attention to things that are happening outside of their immediate area, especially when working. It is usually the event which galvanizes attention, not the actions or activities immediately preceding it. The proximity of the individual to an event, in both space and time, and their degree of involvement or detachment will also have a high degree of influence on recollection. Five people at an event will usually see five different outcomes and each will have a different story, depending on their individual perspective. This is where the skill of the investigator, in teasing out details and reconciling discrepancies between recollections, comes into play.

The more time that passes between the occurrence of the event and the discussion with witnesses about their knowledge of it, the greater the probability that detail will be adversely impacted. Elements will have been forgotten, changed, or otherwise influenced in their recollection. Even vivid memories start to fade within a matter of days after the event. These mental notes also begin to “drift” from a tight congruency with the actual event over time. Both the forgetting and the drift are manifestations of the effects of entropy.

The majority of the effects noted above, and their negative consequences for the accounts of witnesses, are unintentional results stemming from the normal mental processes related to being human. However, occasionally, a witness can intentionally distort his recollection. This is usually out of a great fear of telling the truth. Individuals are frequently afraid that something they might say could either jeopardize their own position, or that of someone they care about. It is a normal human trait to hold back information, particularly in those situations where there is a large hierarchical gap between the witness and the investigator. Often individuals will not be forthcoming with their information because it will expose themselves or others to blame or reprisals. “I know nothing” might protect the witness or his friends, but it will not facilitate getting to the root cause of the event.

Proper Handling of Eyewitnesses

The value of eyewitness information can be enhanced, diminished, or even negated depending on the approach taken to finding and questioning them. For real value to be achieved through eyewitnesses, the investigator needs to master the art of the interview. The most important aspect of eyewitness questioning is related to the fleeting and deteriorating nature of memory. Thus, it is critical to identify and speak with witnesses as soon as possible after the event, and then follow-up with additional questioning later as appropriate. Not only do you want to query witnesses while the event is fresh in their memory, you want to get them to recount their story before they have had a lot of time to run their mind through the sequence of activities over and over again, particularly with others.

Setting the Tone. The next aspect is to set the appropriate tone from the start. In those situations where the organization has chosen to use the investigation as a way to place blame, no lasting benefit can accrue to the organization. Rather, the investigation should be used to identify systemic errors and learn lessons for the future, and thus prevent accidents from happening. It is important that the interviewer not transform the discussion with an eyewitness into an interrogation, with all of its attendant harshness in language, authoritarian and dominating body language, and oppressive setting. These verbal and non-verbal cues will immediately inhibit, if not intimidate, an eyewitness, even if they had nothing to do with the particular event in question. Rather than be open and honest and complete in their responses, each statement will be carefully weighed before being given in light of how it might affect themselves or others they either care about or are concerned for. Not only does this make the investigation more difficult than it needs to be, it also diminishes the value of any conclusions which might be reached based on the information developed.

In many cases, if this “blame placing” approach is a standard operating method in the organization, it is likely to lead people to deny that they have any knowledge about the events in question to spare themselves being drawn into the investigation. If the consequences for individuals in previous situations were harsh, the “blame placing” approach might even be viewed as an inquisition. Also, since witnesses can routinely be mistaken in their recollections, or might even fabricate facts under such an atmosphere, a witness might find testimony to be an easy way to invoke retribution against someone for a grievance, real or imaginary. By taking advantage of the direction of the inquiry to accuse someone of misconduct, an eyewitness can throw the entire investigation off track. All witnesses providing information in such an environment and their statements must be viewed with skepticism.

Selecting the Setting. The next aspect of the interview is to pick an appropriate setting. Witnesses should always be questioned on a one-to-one basis, preferably in a private setting. This helps to ensure an uninterrupted interview and to prevent others from influencing the statements of the witnesses. In those situations where there is a language barrier, the interpreter should be included in the discussions. While it is always preferable that all witnesses be questioned by the same investigator, there are occasions when this simply is not feasible. If there will be multiple parties conducting interviews, it is imperative that they agree on procedures, lines of questioning, and other general administrative issues prior to beginning.

It will also help facilitate the questioning if the witness can write down, sign, and date a statement of the actions surrounding the event. In this statement, the witness should describe, to the best of his or her knowledge, what happened. This statement should be read in advance of the interview to help the investigator frame the questions to be asked. The investigator should make some notes of points that require clarification, or threads of inquiry that should be pursued, to help guide the interview.

The method of questioning, and the investigator’s conduct during the course of the interview can also influence witness recollection. First, the investigator should see to the comfort of the witness and then provide an opening statement. This is a briefing of the witness to explain what is happening, why they are present, and what will occur in the course of the interview. This introduction should put the witness at ease so they will be willing to respond. Explain that notes will be taken during the interview, since some people may be disturbed by this if it is not adequately explained. Also tell the witness that it may seem like some questions are being asked more than once. This is to be certain that the interviewer fully understands the recollection. With these actions, the investigator has set the stage for the interview to begin.

As an aside, never underestimate the role of emotions and their influence on recollection. Fear, anxiety, and concern will constrain and inhibit witnesses, while anger and resentment might lead to the manufacture of elements or conjecture. Calmness is a virtue in getting a useful dialogue underway.

Conducting the Interview. The investigator should open by asking the witness to tell them what happened. If the witness is reluctant to start, use open-ended questions to get them to begin recounting what they recall. Don’t rush or press. Do not interrupt, and never disagree. It is important to observe the witness and watch for body language, an unwillingness to make direct eye contact, and other verbal and nonverbal cues as a means to gauge the truthfulness of what they are telling you. If you have a question about something that is said, note it and when an opportunity presents itself, ask. The most appropriate time to begin asking questions is after the witness has finished the initial recounting of the events.

If the eyewitness was directly involved in the event, the interview should be conducted in two stages. The first stage should be the private interview. In addition to talking to the witness about their recollections, the interviewer should ask the witness to draw diagrams or maps to help explain their position in relation to the event or to show what happened. The second stage, where (or when) feasible, is to take the witness out to the actual scene, and then have them describe where they were and what happened, with the investigator taking notes. This will provide three perspectives—the original statement, the private interview, and the site discussion—from the same individual. In comparing these various stories, at the points of similarity, the investigator will begin to detect, through the consistency, some core facts.

An investigator should always conduct a post-interview review of his or her notes, the witness statement, and any other information made known during the course of the discussion. While the witness and his statements are fresh, questions should be asked of the witness. A healthy sense of skepticism is a valuable trait of an investigator. Does what was said “make sense” with what is known or has been seen? Does it sound genuine, or are conclusions or conjectures being drawn in order to assemble a coherent picture of events in their own mind? Could the witness really see what was described, or have they made projections to bridge gaps in knowledge? Might there be something that was not revealed, and if so, why not? Does it appear that the witness is telling the truth, or is there a hidden agenda at work? Just as a fear of retribution might lead a witness to withhold information, “bad-blood” between parties might lead someone to “shade” the truth or modify a story to implicate or cast blame on another.

These and other questions ought to be used to guide the investigator in his review. The investigator has to be sensitive to the possibility that other forces might bear on the witness accounts. If there are still issues in the investigator’s mind after this review, it might be appropriate to schedule a second interview to clear up questions.

Interviewing Mistakes

It is easy to make mistakes when interviewing witnesses. There are three very common errors that diminish the value of witness statements.

  • Failing to secure individual recollections.
  • Failing to follow leads.
  • Placing too much faith in witness accounts.

Failing To Secure Individual Recollections. Normally, to expedite the securing of witness statements, everyone with information relating to an event is collected in a room while each is interviewed. While this facilitates the data collection aspect, it means that during the wait for their turn, people will talk to each other about what happened. As noted above, it has been found that others’ recollections have a profound influence—conscious or unconscious—on witness memories. Alternatively, again in the interests of speeding up the data collection process, an investigator might initially conduct group interviews, followed by focusing on particular individuals identified with special insight or particularly advantageous involvement.

Failing To Follow Leads. It is seldom that a single individual can provide all the needed information. Instead, they often generate clues or threads of inquiry that should be followed. If the investigator fails to question the stories, asking “So what?” and other questions of the recollections, they will fail to leverage the information available from those around the event. Investigators need to remember that occasionally, years after the event, legal proceedings might be involved and, at that point, memories will have faded or people will become unavailable. All the “threads” of the investigation must be pursed and run down while the event is still fresh. It is not enough to get a statement from an individual, as there will be many questions left unanswered. Any person who prepares a statement should also be interviewed to “flesh-out” their written statement.

Placing Too Much Faith in Witness Accounts. The final, and most common, error is to place too much faith in the veracity of the statements of the witnesses. Routinely, investigators fall victim to the air of confidence in a witness’s recollection, believing that this sense of certainty must carry over into the truthfulness of the story of the witness. Before succumbing to a confident witness, an investigator must inquire as to the source of the confidence. Does it truly stem from a knowledgeable eyewitness? Is it built from a witness having told and retold his story so many times that he has created a solid, coherent tale? Does it originate from a strong personal sense of knowing?

People can be certain that they are absolutely right about what they remember, and still be completely wrong. Reliance on invalid statements serves an investigation no better than an absence of information. Witnesses are most routinely the source of valuable insight into the activities and occurrences surrounding an incident. However, investigators must carefully examine and question witness statements, for while they are a most valuable source of incident information, they can frequently mislead and misdirect, undermining the intended purpose of the investigation.


The physical setting of an accident can provide important information during an accident investigation. However, by itself, this data points nowhere. Artifacts help fill in gaps, but can be misleading and create distortions, depending on their source or the reason for their creation in the first place. Witnesses are the best source of insight into incidents, due to their actual presence at the incident. Still, as the caveat about the fallibility and malleability of human memory demonstrates, great care must be taken in gathering the information witnesses can provide. And recollections should be viewed with caution.

Over-reliance on the accounts of witnesses is fraught with danger. To not gather the data is to preclude altogether the drawing of any conclusions; however, each source of information has traps that can trip up the unsuspecting or unprepared. Skepticism and caution—along with balance and consideration—are crucial elements to the collection of reliable information for use in an accident investigation.