As human beings, it is natural to rely on our experience when assessing day-to-day situations and making decisions. Fire investigators are no different. Experiences developed through investigations, if left unchecked, can become a double-edged sword as cognitive biases. These cognitive biases can adversely affect methodology in a fire investigation.

This is where NFPA 921: Guide for Fire and Explosion Investigations (2024 ed.) provides a valuable resource for origin-and-cause investigations. This approach guides investigators to follow an objective, scientific-method approach. One that incorporates collecting data, developing, and testing reasonable hypotheses, and reaching conclusions based on facts.

This article highlights two forms of bias, which can undermine that process, damage findings, and has the potential to create legal and professional implications.

Confirmation bias

Confirmation bias looks at an investigator’s tendency to lean towards a hypothesis throughout the investigation giving more weight to one hypothesis while discounting or failing to consider other hypotheses with the same level of accuracy and scientific method. NFPA 921 cautions that investigator to be aware of potential confirmation bias emphasizing a scientific-method approach that requires developing and equally testing reasonable alternative hypotheses against the facts [1, §4.3.10].

Potential consequences of confirmation bias include the following:

  • Lack of equal focus on all reasonable possibilities: An investigation focuses on details that fit a preferred theory. An example of this is observing a fire effect that appears to be consistent with a recognized “drop down” [1, §3.3.52], and focus on that interpretation without confirming whether ventilation, suppression or fire spread characteristics support this conclusion. Alternative explanations may receive limited analysis, if at all.
  • Premature elimination of competing hypotheses: When an investigator feels the observed effect is a fit, consideration of competing theories may cease. Even if other evidence or fire patterns suggest a competing hypothesis, the investigator dismisses those without following the same scientific methodology. This can include discounting conflicting witness timelines or fire patterns that are not convenient for the initial hypothesis without evidence-based rationale.
  • Biased evidence collection: Collection of evidence may only relate to the preferred hypothesis. Artifacts or information related to competing theories may be left unexamined or collected.
  • Opens door for later challenges to theory: The bias can become even more glaring in cases where legal proceedings become necessary. The investigator can be caught overstating the preferred hypothesis while having not used the same methodology for other hypotheses. This can undermine the credibility of the investigator’s opinion.

Expectation bias

Expectation bias occurs when prior expectations formed by experience, early witness statements, or case context influence what is noticed and guides the investigation from an early stage in the process. NFPA 921 cautions against reaching conclusions or revealing early details to guide the investigation without examining and considering all relevant data [1, §4.3.9]. An example is a witness statement revealed early or video surveillance that gives you an early “aha” moment without even stepping foot on a scene.

Potential consequences of expectation bias include the following:

  • Improper pattern interpretation: Instead of using the scientific method, a context is applied strictly based on expecting that an individual, for example, was in a tenant and landlord dispute, therefore, they must have left the stove on as retribution. In this example, this can push neutral observations toward an “incendiary” interpretation.
  • Early statements or information becomes the basepoint: Initial statements from witnesses/authorities/stakeholders shape later perception and credibility assessments, in many cases before any evidence is ever examined.
  • Biased evidence interpretation and reporting: when an arson expectation exists, a negative ignitable-liquid test result may be rationalized as a “miss”, whereas it would otherwise be treated as meaningful when an accidental expectation exists. This would also be reflected in a report where it becomes “obvious” that the information provided is suggestive of what ended up happening. Often times, this is not supported by scientific data.

To avoid bias, an investigator must only allow the scientific method, evidence and facts to dictate the direction of the investigation. This would avoid the methodology and documentation from becoming unbalanced and the investigative method from being compromised.

Further consequences that may arise relate to litigation matters. Investigator bias can become a focal point impacting both the case at hand and the integrity of the investigator. In this situation the reliability of the method is put into question, as well as whether the scientific-method approach was consistently used throughout and ultimately if the opinion of the investigator was based on this scientific method, and did not exceed the limits of the available data.

If the investigator is found to have been biased in the investigation, this can harm the investigator standing as an independent expert, affecting future reliance in court and reputation between clients and peers.

The following practices can help reduce the influence of bias and strengthen the reliability of the investigation:

  • Consciously remain neutral and avoid early conclusions: Remain neutral in assessments and non-conclusive notes by following the scientific method, and delay opinion forming until data collection and hypothesis testing is complete.
  • Generate and any reasonable hypotheses: Even when one hypothesis appears to be more viable, identify what observations would validate or negate each hypothesis.
  • Make the connections only through scientific methods: Conduct assessments relying on evidence to help connect the ignition sources, fuel loads, fire spread, ventilation assessments, and accounting for suppression or overhaul effects, rather than fitting observations to a favoured narrative.
  • Independent technical review: Use peer reviews to identify unsupported hypothesis, exclusions, missing alternatives, and overstatements before conclusions are finalized.

 

Conclusion

Confirmation bias and expectation bias are inherent human tendencies that can quietly erode the objectivity of an otherwise well‑intentioned investigation, if not consciously aware of their influences. The consequence is not only weaker findings, but also increased vulnerability under legal scrutiny.

Following the scientific method creates a mechanism that requires assessing alternative hypotheses, and making reasoning transparent. This helps reduce the influence of bias, while preserving both investigative reliability and the investigator’s independence and reputation from the beginning of the investigation to the conclusion.

 

References

  1. National Fire Protection Association (NFPA), NFPA 921: Guide for Fire and Explosion Investigations, 2024 ed., §3.3.52, §4.3.9, §4.3.10. Quincy, MA, USA: NFPA, 2024.