What 911 Call Analysis Actually Is (and Why It's in Your Case)

A detective takes the stand, describes your client's 911 call as "staged," and offers no expert report, no methodology, no CV. That's not sloppy lawyering on the prosecution's side. It's the design.

911 call analysis is the practice of inferring a caller's guilt from how they speak during an emergency call. The best-known version, the Harpster Method, sorts callers against a proprietary checklist called the 911 COPS Scale©, built around claimed verbal "guilt indicators" like saying "blood" instead of "bleeding," or asking whether to touch the body. The problem is the foundation. The theory rests on two small studies of roughly 100 calls each, about 200 cases total, and the original 2006 thesis was expressly exploratory research. The data behind the COPS Scale has never been made public, contrary to standard academic practice.

The replication record is brutal. At least eight scientific studies have tried to test or replicate the method's indicators and found they don't hold up, including two FBI Behavioral Analysis Unit studies that found the indicators were inconsistent, sometimes pointing in the opposite direction, and warned the technique "may exacerbate bias." The FBI's own about-face has gone largely unnoticed: its 2008 bulletin spread the method to a readership of roughly 200,000 police readers without the "exploratory" label the underlying study carried.

So why is this still in courtrooms? Because prosecutors have learned to keep it out of admissibility fights. The Cornell Law Review calls this pattern "shadow forensics": the underlying methods are never disclosed and the founders are never proffered as experts, so the technique largely escapes Daubert or Frye challenge. Instead, a dispatcher or detective who took an eight-hour training (costing departments up to $3,500) testifies about the caller's tone and word choice as if it were ordinary lay observation. A Missouri prosecutor admitted in a 2016 email to Harpster that she presented the analysis in court "without calling it 'science'" to sidestep the hearings that would test it.

The scale of use is real. ProPublica documented more than 100 cases across 26 states where the method played a pivotal role in arrests, prosecutions, and convictions, and called that figure likely a fraction of the actual total. Yet awareness in the defense bar remains dangerously low; one Virginia public defender told ProPublica he'd "never heard of any of that claptrap" in his jurisdiction. That's the core problem this guide addresses: the evidence is already in your cases, often disguised as casual characterization of a call, and the first step in challenging it is recognizing it's there.

Practically, that means screening discovery for it. When you pull the 911 audio and transcript, look for who trained in call analysis, and compare the transcript against how officers later characterized it. Tools that produce timestamped, diarized transcripts with citations to the exact moment (for 911 call transcript analysis and jail call transcription alike) make those mismatches visible instead of buried in hours of audio.

The Scientific Record: 8+ Failed Replications and the FBI's Own Reversal

The scientific record against 911 call analysis is not thin. It is overwhelming. At least eight studies have attempted to replicate the method's claimed indicators of guilt, and every one has come up empty.

Start with the numbers behind the theory. The entire method rests on two small studies of roughly 100 calls each, per a defense-attorney analysis of the underlying research. That's about 200 calls total to support a technique now used in investigations across at least 26 states. The original 2006 thesis examined 100 homicide calls (50 guilty, 50 innocent) and was explicitly labeled exploratory research. It was never meant to be a courtroom tool.

Then the replication attempts started, and the indicators fell apart.

• Eight-plus failed replications. At least eight scientific studies have tested the method's guilt indicators and found they don't hold up, according to Innocence Project research. Six peer-reviewed studies alone could not replicate the "guilty/innocent behavior checklist."
• Two FBI BAU studies contradict the method. The FBI's own Behavioral Analysis Unit found the indicators inconsistent. Some pointed in the opposite direction of what the original study reported. The FBI warned that applying the method "may exacerbate bias."
• Cornell psychology research on trauma misreading. Four studies involving roughly 1,800 civilians and 300 law-enforcement officers found that 4 out of 5 participants cited emotion as the factor driving suspicion. Callers who didn't perform "expected" grief or urgency got flagged. Researcher Jessica Salerno notes there is no "one size fits all" response to trauma.

The FBI's reversal deserves special attention because almost nobody knows about it. In 2008, the FBI helped spread the method through a bulletin sent to police departments. That bulletin was not labeled exploratory, even though the underlying study was. It reached roughly 200,000 readers. When later FBI studies contradicted the findings, the correction never got the same distribution. As ProPublica's investigation documents, the reversal has gone largely unnoticed.

One FBI co-author of the original study, John Jarvis, now says the work doesn't "rise to the level of success by the scientific community." That's the agency's own scientist disowning the foundation of the method.

For your cross-exam prep, this record is the ammunition. When a detective or dispatcher testifies about a caller's tone, you can walk them through the failed replications, the FBI's contradictory findings, and the trauma research showing that flat affect often reflects neurodivergence or shock, not guilt. Getting the call audio and a clean, timestamped 911 call transcript analysis or jail call transcription makes that cross concrete instead of abstract, and it's the kind of grounded, citation-linked review work defense teams do in Evidence Copilot every day.

ctaText: Book a 20-minute walkthrough socialProofText: Grounded evidence review for criminal-defense teams. Every AI answer cites the exact moment it came from.

What It Costs When It Goes Unchallenged: The Wrongful Conviction Record

When 911 call analysis goes unchallenged, the consequences are not hypothetical. The documented record includes defendants who spent years, and in some cases decades, in prison before courts or exonerations caught up with a technique that has never validated itself scientifically.

Consider the concrete outcomes. Russ Faria was wrongfully convicted of murdering his wife after a dispatch supervisor trained in call analysis testified that his 911 call sounded staged. He spent 3.5 years in prison before his acquittal at retrial, according to a ProPublica investigation built on 80+ open records requests and roughly 120 interviews. In Michigan, 16-year-old Riley Spitler's conviction was overturned because 911 call analysis and other statement analysis techniques should never have been admitted. Huwe Burton, convicted at 16, served more than 20 years before his 2019 exoneration.

Then there is Robert Roberson, who has spent more than two decades on Texas death row and narrowly avoided execution three times. Years after his conviction, he was diagnosed with autism spectrum disorder. His flat affect, which investigators read as suspicious, was neurodivergence, not guilt. That distinction is exactly what Cornell psychologist Jessica Salerno's research addresses: people respond to trauma in radically different ways due to culture, personality, neurodivergence, and shock, so there is no "one size fits all" for how innocence should sound.

The scale of the problem is documented, not anecdotal:

• ProPublica documented more than 100 cases in 26 states where call analysis played a pivotal role in arrests, prosecutions, or convictions, and called that figure "likely a fraction of the actual number."
• At least eight scientific studies have failed to validate the method's guilt indicators.
• Two FBI Behavioral Analysis Unit studies found the indicators inconsistent, with some pointing in the opposite direction of what was previously claimed.

For defense practice, this record is citable ammunition. When you move to exclude the analysis or impeach a trained detective's lay testimony, the wrongful-conviction history functions as evidence of the technique's demonstrated error rate and its prejudice under Rule 403. Faria, Spitler, Burton, and Roberson are not just cautionary tales; they are precedent for arguing that admission has already produced documented miscarriages of justice.

Practical detection matters here, because the technique often hides inside casual-sounding testimony about tone and word choice. Reviewing the actual call audio against the detective's characterization, with timestamped 911 call transcript analysis, gives you the record you need to show the gap between what the caller said and what the analyst claims it means. The same grounded-review workflow applies to jail call transcription and other statement evidence that statement-analysis enthusiasts may target next. Every citation, every quoted second, keeps your challenge anchored to the source rather than to memory.

The Defense Playbook: 5 Moves to Challenge 911 Call Analysis in 2026

Prosecutors don't need to call Tracy Harpster to put 911 call analysis in front of a jury. They just need a detective or dispatcher who took an eight-hour class. That's the evasion pattern, and it's why you have to build detection into discovery review before anything else.

Start with the discovery dump itself. Flag any detective, dispatcher, or investigator whose training records show attendance at a Harpster Method or COPS Scale class. ProPublica documented more than 100 cases in 26 states where the technique drove arrests and convictions, and Harpster has consulted on at least 2,500 homicide investigations as of March 2024, so the odds your case crosses this evidence are higher than most defenders assume. When you see a witness list that includes anyone trained in call analysis, pull the 911 audio immediately and run it through 911 call transcript analysis so you can pinpoint the exact moments the prosecution will highlight: word choice, pauses, the caller's affect.

Once you know the method is in play, attack the foundation. Demand disclosure of the COPS Scale, the underlying data, and the training materials. The Cornell Law Review frames this as "shadow forensics": the technique "largely escapes Daubert or Frye evidentiary challenge" because the methods are never disclosed and the founders are never proffered as experts. Force the disclosure, and you force a gatekeeping hearing under your state's standard. The data behind the COPS Scale has never been made public, contrary to standard academic practice, which is itself a Daubert problem.

Then load the replication record into your cross-exam outline. At least eight scientific studies have failed to validate the method's guilt indicators, including two FBI Behavioral Analysis Unit studies that found indicators went "in the opposite direction of what was previously found" and warned application "may exacerbate bias." The FBI's reversal has gone largely unnoticed, even though the original 2008 bulletin reached roughly 200,000 law enforcement readers. FBI co-author John Jarvis has already gone on record: "I don't see where that work rises to the level of success by the scientific community."

For demeanor testimony, reframe it as confirmation bias and trauma misreading. Cornell's peer-reviewed research found that callers who fail to show "expected levels" of emotion or urgency may trigger suspicion, described as "potentially the first step toward a wrongful conviction." Robert Roberson, who has spent more than two decades on Texas' death row, was diagnosed with autism spectrum disorder years after his conviction. That is the kind of evidence that turns a flat affect into reasonable doubt.

Finally, name the wrongful convictions. Russ Faria spent 3.5 years in prison before acquittal after a trained dispatch supervisor testified his call sounded staged. Riley Spitler's Michigan conviction was overturned because 911 call analysis should not have been admitted. Huwe Burton served more than 20 years before exoneration in 2019. These are not hypotheticals; they are the documented error rate.

The workflow layer matters here. Timestamp-aware transcripts let you cite the exact second a detective's "guilt" opinion rests on, and jail call transcription lets you compare what a defendant said after the call against the prosecution's narrative of staged emotion. When the method depends on subjective inference, precision is your best weapon.

If you want a 20-minute walkthrough of how timestamped 911 and jail call review fits into a suppression or cross-exam workflow, book at evidencecopilot.com/book.