Cognitive Bias in OHS Investigation: The Biasing Power of Source Identity

The influence of an information source's identity can significantly bias occupational health and safety investigators' judgments.

• By Carla L. MacLean, Surveer Boparai
• May 31, 2024

Social media influencers, marketers, and TV “experts” understand that who a person is, not just what they say, is influential. The persuasiveness of an information source has important implications for all decisions, including those made in occupational health and safety (OHS) investigations. This article explores the persuasive power of the identity of an information source and how it can shape an OHS investigator's understanding of events.

Information Source and Bias

Identity is ubiquitous and people rarely consider the impact it has on how information is processed. Equity, diversity and inclusion (EDI) training is on the rise globally, and this raises awareness of how identity shapes interactions. Nevertheless, an individual's characteristics remain a robust source of influence in decision-making, mainly because the bias that results from them usually occurs outside the decision-maker's awareness. Bias is not an ethical issue — honest, hard working and well-intentioned decision-makers can make biased decisions. 

Psychologists who study dual process theories of cognition offer an explanation of how and why the identity of an information source nudges our choices in one direction or another. As author Daniel Kahneman explains, people tend to think at two speeds, fast and slow. Thinking slow is deliberative, time-consuming and effortful. For instance, when people try to speak a foreign language. Thinking fast, on the other hand, relies on experience, association and expertise to simplify the decision-making task, make good assumptions and navigate with minimal cognitive effort. People experience fast thinking when they speak their native language.

The identity of a person is a piece of information that provides the decision-maker with a “fast” cognitive shortcut about the quality of the information being offered to them. For instance, an individual is more likely to accept car advice from a trusted mechanic than from a car salesman. If one’s understanding of information is altered because of who provided it, it is a biased judgment. Bias is the systematic deviation from what an evidenced-based, objective judgment of the information would determine. Bias is bidirectional, meaning it can lead to rendering a judgment moreso or less than what the evidence shows. 

Characteristics of the person sharing the information, such as affiliation with the decision-maker as an in-group, or cues that suggest credibility, such as experience can be biasing. Uncertainty, insufficient knowledge and limited cognitive resources because of such things as fatigue or time pressure can amplify the likelihood of bias as they lead decision-makers to depend more on easily accessible, yet potentially irrelevant, peripheral elements of a message.

OHS Investigation 

The research has not carefully considered how an information source’s identity shapes OHS investigators’ judgments. However, individual influence has been explored as a way to bring about positive change in occupational safety, and applied research from the construction sector has shown how the identity of a source shapes employees' judgments of risk. Stacey M. Conchie and Calvin Burns found that employees rated safety personnel in their work environment the most trustworthy (i.e., UK HSE and safety managers) compared to other personnel (i.e., project managers and work colleagues). Safety personnel were also found to have the greatest influence on employees' behavioral intentions regarding risk.

Demonstrating the Power of the Person

The previous research demonstrates how source identity influences intentions in real work settings. It also shows that the identity of an information source has the potential to specifically bias the judgments of OHS investigators. To directly test this hypothesis, 40 professional investigators voluntarily participated in an online investigative activity. Their participation was completely anonymous and they could withdraw from the activity at any time without negative consequences. 

Respondents were highly experienced. Prior to working in their current, high-profile, investigative role, all respondents had held investigative positions in other agencies. The team of investigators involved in this activity had a variety of backgrounds and training. The Public Safety Group were individuals with a law enforcement background (e.g., police, border services), the Industry Group had an industry background (e.g., occupational health and safety personnel from forestry, mining, fishing, etc. or prevention services), the Other Group was comprised of subject matter experts from disciplines such as hygiene, human factors and engineering. 

Respondents self-identified their group membership at the start of the activity and the number of years they had worked with the safety agency. Twenty of the respondents self-identified as Public Safety and twenty identified as Other. Three respondents identified themselves as Industry. These three individuals have been excluded from the presentation of the findings below, as there is little informational value in such a small sample size. The 40 investigators whose data are presented had been in their current investigative role with the agency for an average of 5.47 years (SD = 5.11). 

Materials and Methods

A subject matter expert (SME) helped the lead author develop and refine the materials for the investigation activity. The SME had 29 years of experience in police investigations, followed by 8 years in OHS investigations. 

Respondents began the activity with an overview of a fatal incident that had occurred in a remote location. They learned that they were in the role of Primary Investigator (PI), with two investigating officers reporting to them. Both officers had equivalent years of investigation experience, but one of the officers had a public safety background and the other had an industry background.

Participants then read the event summary about a boom boat which is used in logging operations. “This boom boat began taking on water,” the event summary read. “The boat operator radioed for help. The boat sunk shortly after the distress call. It is believed that an excess amount of water in the bilge caused the boat to sink. The boat operator perished in the event.” A photo of the boat accompanied the summary. 

Respondents then chose the next best step in the investigation from two options that addressed why there was excess water in the bilge, one offered by the Public Safety officer and one offered by the Industry officer. The two options were of equal value as a next step. To ensure that the identity of the officer—rather than the content of the option—was driving the results, the options were counterbalanced so that half of the participants received the first option from the Public Safety officer and half of the participants received the same opinion from the Industry officer, and vice versa for the second opinion. The final step of the activity was to answer the manipulation check questions, which asked respondents to identify the background of the two investigating officers, the type of event (e.g., near miss, fatality or severe injury) and whether they were familiar with this type of event. 

It was predicted that respondents would show bias by choosing the option offered by the investigator with the same background as themselves (i.e. Public Safety or Industry). Too few Industry investigators responded to test how their responses differed from the Public Safety Investigators.

Results

Respondents were biased by the identity of the officer providing the information but not as predicted. For clarity, if investigators made their judgments based solely on the evidence— regardless of the source—the results would show a relatively consistent pattern of choice for the two next-step opinions offered.

Rather, respondents preferred the option offered by the officer with the industry background (26/40 investigators or 65 percent) over the officer with the public safety background (14/40 investigators or 35 percent). This preference for the option offered by the industry investigator occurred regardless of the background of the respondent doing the activity. Discussions with the full investigation group after the activity confirmed that the Industry investigators have considerable credibility within the investigation team because of their nuanced understanding of work systems.

Conclusion

This shows how a contextual feature of the decision-making environment, the identity of the person providing the information, biased the judgments of professional OHS investigators. Recall that bias is the systematic deviation from what an impartial assessment of the evidence would support. 

This finding of bias is important because it shows that a biasing context can be subtle and still affect choice. In the activity, the credibility of the two officers giving opinions to the PI was relatively comparable. In real-world decision-making, however, it is easy to imagine how a more extreme context could significantly obfuscate quality decision-making. Imagine, for example, a sub-optimal opinion from a high credibility source, or superior ideas from a less than desirable employee. 

Professional investigators work hard to make the best possible judgments to improve the well-being of individuals in the workplace. To combat the subversive effect of context on decision-making, investigators are urged to be diligent and engage the “slow” system of thinking, to accept that context may bias their judgments without their awareness, and to take steps to mitigate bias. Investigators who incorporate bias management into their decision-making can collectively move the needle on workplace safety and provide more meaningful improvements to the lives of those in the workplace.

This article originally appeared in the September 2024 issue of Occupational Health & Safety.