Inconsistency is not Fraud

One of the quickest ways an investigation can move beyond the evidence is when an inconsistency is treated as proof of dishonesty.

A statement does not match a photograph. A timeline changes. An invoice describes something differently to the person who installed it. CCTV does not quite fit the version of events. A witness remembers something one way and another witness remembers it differently.

But they are not, by themselves, fraud.

An inconsistency is evidence of one thing first: two pieces of information do not align. Anything beyond that still has to be established.

In forensic investigations this is important because our reports can influence insurance decisions, litigation and, sometimes, allegations about a person’s conduct.

As experts, our job is to identify the inconsistency, investigate it and explain what effect it has on our technical opinion. It is not to fill the gap with an allegation the evidence has not established.

Conflicting evidence is normal

Fire investigations rarely give us a perfectly clean set of facts. We are reconstructing something that has already happened, usually after the fire itself has damaged or destroyed some of the evidence we would ideally want to examine.

Witness information is one source of evidence, but it has to be considered alongside physical evidence, photographs, records, fire patterns, electrical or mechanical evidence, CCTV, timelines and other available information.

Kirk’s Fire Investigation specifically discusses evaluating interview information by correlating it with other evidence and documenting both corroborating and conflicting information. The existence of a conflict is therefore something an investigator is expected to recognise and document. That is very different from automatically deciding why the conflict exists.

Two people can genuinely describe the same event differently. A witness may be describing the fire from a different location or at a different point in its development. Records may be incomplete. Terminology may be used differently. A person may simply be wrong about a time, sequence or technical detail. Of course, deliberate deception is also possible.

But possibility is not proof.

Report what the evidence actually shows

This is where wording matters. There is a significant difference between writing:

“The account provided is inconsistent with the CCTV footage.”

and:

“The insured has provided a false account.”

The first statement identifies an evidentiary conflict. The second assigns intent.

Likewise:

“The reported time of isolation cannot be reconciled with the available electrical records.”

is different from:

“The records appear to have been manipulated.”

One may ultimately lead investigators toward the other. But there needs to be evidence supporting that additional step.

NFPA 921 requires collected and available data to be analysed using the scientific method, with hypotheses developed and tested against that data. It also states that expert opinions should make the underlying data, hypothesis testing and known limitations transparent.

Let the inconsistency become a question

When I find conflicting evidence, my first response is not, someone must be lying. My response is usually, why don’t these two things match? That question creates somewhere useful for the investigation to go. Can the timeline be independently verified? Are there photographs?Is there CCTV? Are there phone records, access logs, alarm records or maintenance documents?

Can another witness corroborate either account?

Could fire development explain what the witness observed?Could the equipment have behaved differently from the way the person believed it behaved?Is one document recording an estimate while another records an exact time?

This is exactly what hypothesis testing is supposed to do. Kirk’s describes the scientific process as comparing hypotheses against all known facts, seeking additional information where necessary, considering alternative hypotheses and modifying the working hypothesis when the evidence requires it.

Sometimes the additional evidence resolves the inconsistency or make its more significant. Or it just remains unresolved. All three outcomes are legitimate investigative findings.

Experts also need to stay inside their lane

There is another problem with jumping from inconsistency to fraud: it can take an expert outside the question they were actually retained to answer.

A fire investigator may have been asked to determine the area of origin, fire cause, ignition mechanism or whether a particular failure contributed to the incident.

During that work, we may identify evidence that conflicts with a version of events. We absolutely should report it where it is relevant.

For example: “The physical evidence is not consistent with the reported sequence of events.” Then explain why. That may be highly significant to the insurer, solicitor or other party relying upon the report.

But there is an important difference between providing them with the technical evidence and deciding what someone’s motive was.

NFPA 921 recognises that an investigator’s assignment can extend to questions of responsibility, but the investigation still requires hypotheses to be developed and tested against available data. An expert report should also contain the facts and data relied upon and explain the reasoning by which the opinion was reached.

If the evidence demonstrates only a contradiction, call it a contradiction. If it demonstrates that a version of events is physically impossible, say that. If records cannot be reconciled, explain the discrepancy.

If additional investigation is required, recommend it. Those statements can be extremely powerful without adding the word fraud.

Suspicion is not a methodology

The danger starts when an investigator sees an inconsistency, forms a suspicion and then begins interpreting everything else through that suspicion.

Now the investigation is no longer asking, what does the evidence tell me? It is asking, what else can I find to support what I already think?

That is exactly where confirmation bias becomes a problem.

The scientific method exists, in part, to stop us doing this. A working hypothesis is supposed to remain open to review and modification. Alternative hypotheses are supposed to be considered. New evidence should be capable of changing the conclusion rather than simply being forced into the original theory.

The 2024 edition of NFPA 921 specifically expanded its guidance around confirmation bias, expert opinions and expressions of certainty. That is particularly relevant when the potential conclusion carries serious consequences for another person.

Sometimes the strongest expert opinion is narrower

There can be pressure in contentious matters to answer the bigger question.

Do you think they are lying?

Does this look suspicious?

Do you think this was fraudulent?

But the strongest expert opinion may actually be:

“I cannot determine why the accounts differ. I can only say that the reported account is inconsistent with the physical evidence for the following reasons…”

That is not avoiding the issue. That is doing the job properly.

A scientifically defensible opinion is not measured by how dramatic the conclusion sounds. It is measured by whether the evidence supports the conclusion that has been expressed.

Experts do not need to ignore inconsistencies. We should identify them clearly. Test and explain them where possibly. And tell the client when they cannot be resolved.

But there is a line between identifying conflicting evidence and assigning dishonest intent.

The evidence may justify the first without ever establishing the second. And knowing where that line sits is part of being an expert.