Case File 072 · Investigation and Psychology

How Investigators Compare Multiple Accounts

How investigators compare statements from different people, and why neither agreement nor disagreement settles what really happened.

By Murder Book Editorial TeamPublished 16 September 2026Updated 16 September 2026

Investigators compare multiple accounts by treating each one as a separate piece of evidence. They keep witnesses apart as early as possible, record what each person remembers in their own words, and only then set the accounts side by side against a timeline and the physical evidence. The goal is not to count votes. It is to work out where accounts agree, where they differ, and why they differ. Agreement is not automatically truth, because people who talk to each other can end up sharing details that only one of them saw. Disagreement is not automatically lying either, because honest witnesses notice different things, forget different things and sometimes remember more later. Good comparison asks what each difference could mean, then looks for independent evidence that can tell the explanations apart.

Key takeaways

Accounts are most useful when they are independent, so investigative guidance says to separate witnesses and discourage them from discussing the incident.
In one well-known experiment, 71% of witnesses who discussed an event with a co-witness later reported details they had picked up in the discussion.
Differences between accounts have many innocent causes, including viewpoint, attention, stress and the passage of time.
Matching accounts can reflect a shared source, such as a conversation or media coverage, rather than accurate memory.
Differences are leads to check against timelines and physical evidence, not verdicts on their own.

Why do investigators keep witnesses apart?

Comparing accounts only works if each account reflects one person's memory. Once witnesses start talking to each other, that independence can be lost.

The U.S. National Institute of Justice's guide Eyewitness Evidence: A Guide for Law Enforcement (1999) makes this a basic step at the scene. It tells officers to separate witnesses and instruct them to avoid discussing details of the incident with other witnesses. It also advises encouraging witnesses to avoid media accounts of the incident, since news coverage is another way outside information can enter a memory.

Research on memory conformity explains why. In a study by Fiona Gabbert, Amina Memon and Kevin Allan (2003), pairs of participants each watched a slightly different video of the same criminal event. Some pairs discussed what they had seen before recalling it individually. According to the study's summary, 71% of the witnesses who discussed the event went on to mistakenly recall items they had acquired during the discussion.

A follow-up study by Gabbert, Memon and Wright (2006) analysed the conversations themselves. The witness who started the discussion was the most likely to influence the other person's report. Witnesses were also most likely to be influenced when the discussion introduced an extra item that they had not seen themselves.

None of this requires anyone to lie. A witness can sincerely report a detail as their own memory when it actually came from someone else.

How are accounts recorded before they are compared?

How an account is gathered shapes what ends up in it, so comparison starts with careful recording.

The NIJ guide recommends open-ended questions, such as "What can you tell me about the car?", supplemented by closed questions, and it says to avoid leading questions such as "Was the car red?". A leading question can plant the very detail an investigator later finds "confirmed" in a second account.

The UK Ministry of Justice guidance Achieving Best Evidence in Criminal Proceedings makes a related point for interviewers who speak to more than one witness. It warns against asking a witness questions based on what previous witnesses said, because this could contaminate the witness's account. It also notes that the potential for memory contamination increases with delay.

For more on why question wording matters, see Open-Ended Versus Leading Questions.

What do investigators look for when accounts are side by side?

Once accounts are recorded independently, the comparison is detailed and methodical. The table below shows common types of difference and some of the explanations each one can have. None of these explanations is assumed in advance.

Type of difference	What it looks like	Possible explanations
Omission	One person mentions a detail and another does not	Did not see it, did not think it mattered, was not asked, forgot it
Addition	One account includes something no one else reports	Unique viewpoint, delayed recall, outside information, invention
Sequence	Events are told in a different order	Normal uncertainty about order, stress, different starting points
Timing	Estimates of time or duration differ	Estimates vary between honest people; clocks and records can check them
Direct contradiction	Two statements cannot both be true	Honest error by one or both people, mistaken identity, deception
Identical wording	Accounts match closely, even in phrasing	Shared conversation or media exposure, rehearsal, or simply the same event described plainly

The most important question for each row is whether the difference concerns a central fact of the case or a peripheral detail. A disagreement about the colour of a jacket is not the same as a disagreement about whether someone was present at all.

Why is agreement not the same as truth?

It is tempting to treat two matching accounts as corroboration. That only holds if the accounts were formed independently.

The memory conformity studies show that discussion can make witnesses converge on details that one of them never saw. The NIJ guide's advice on media exposure reflects the same risk. If several people read the same report, their accounts may agree because they share a source, not because they share a memory.

Agreement also does not tell you that the shared detail is correct. Two people can make the same mistake.

Why is disagreement not the same as lying?

Honest accounts differ for ordinary reasons. People stand in different places, pay attention to different things and bring different expectations. Stress also matters. A meta-analysis by Deffenbacher and colleagues (2004) found that high stress impaired both eyewitness identification and recall of crime details.

Memory also changes over time. People forget, and sometimes they remember more. Oeberst (2012) studied reminiscence, the recall of details that were not mentioned in an earlier account. Participants judged these late details to be unreliable, but in her studies their actual accuracy was consistently high and comparable to immediate recall.

This matters because a new detail in a second interview can look like a change of story. Sometimes it is. Sometimes it is simply a memory that surfaced later. The article Why Truthful Accounts Can Still Be Inconsistent looks at this problem within a single person's statements.

Consistency is also not the only thing people weigh. In a mock-jury study, Brewer and Burke (2002) found that a witness's confidence strongly influenced judgments, while the effect of testimonial consistency was weak and non-significant. People evaluating accounts may not weigh inconsistencies the way they think they do.

How does other evidence settle the differences?

Comparing accounts with each other only goes so far. The stronger step is to compare every account with evidence that does not depend on anyone's memory: records, recordings, physical traces and documented times.

This is where an investigative timeline becomes useful. Placing each account on the same timeline shows which differences actually conflict and which are just different slices of the same events. It also shows which claims can be tested independently.

The danger at this stage is deciding too early which account is "right" and reading everything else in that light. Saul Kassin (2005) reviewed research suggesting that investigators can make false-positive errors in pre-interrogation interviews, presuming innocent suspects guilty. An early judgement about who is lying can shape how every later difference is interpreted. How Confirmation Bias Can Affect an Investigation covers this in more detail.

How is comparing accounts different from finding inconsistencies?

The two ideas overlap, but they are not the same task. Finding inconsistencies usually means examining one story for internal gaps, contradictions or changes, which is covered in How Investigators Find Inconsistencies in a Story. Comparing accounts is about the relationship between stories from different people, and about whether they are independent enough to be compared at all.

How does Murder Book use conflicting accounts?

Murder Book: The Art of Interrogation is an upcoming narrative interrogation game from Gungrounds, an independent studio in Zagreb, Croatia. Its cases are built around unreliable stories, narrative gaps and contradictions. Development screenshots show an evidence board that links statements to each other, such as a flat denial of ever meeting someone linked to a fabricated description.

The research above is a useful way to approach that board. A contradiction tells you where to look closer, not who is guilty. The press kit also names behaviours such as Avoidance and Emotional Leakage, and development screenshots show a casebook entry for Stonewalling. In the game, as in research, behaviour is a reason to look closer, not proof. Cases are designed for focused 5–10-minute sessions.

Murder Book is in development for PC (Steam) and mobile, with release currently planned for the end of Q2 2027.