What 911 Call Analysis Actually Is, and Why It's Under Fire

You likely have a 911 call in your discovery dump right now. If the prosecution plans to use it, you need to know what they’re actually doing with it.

911 call analysis is a controversial forensic technique developed by former Deputy Chief Tracy Harpster. It claims to identify guilt or deception by analyzing a caller’s word choice, cadence, and grammar.

The method relies on a proprietary framework called the COPS Scale. This scale uses 38 specific indicators to judge a statement's validity. University of Cincinnati Law Review analysis notes that these indicators range from "Defendant Mentality" to unintelligible comments.

The scientific foundation is thin. The original research relied on a small sample of 200 calls. Half of those callers had already been convicted or confessed. The underlying data has never been made public.

This lack of transparency has triggered a major credibility crisis in 2025 and 2026. ProPublica investigation found the technique used in over 100 cases across 26 states. Researchers attempting to corroborate the claims failed to find sufficient scientific backing.

The technique often bypasses standard evidentiary hurdles. Proponents frequently present it as lay testimony rather than expert opinion. This strategy allows the analysis to "largely escape Daubert or Frye evidentiary challenge," according to Cornell Law Review.

This approach is now labeled "shadow forensics." It sits alongside other junk disciplines like bitemark evidence and hair microscopy. These fields have led to hundreds of exonerations.

Institutional backlash is accelerating. North Carolina issued a statewide warning to defense attorneys. The Illinois Supreme Court is reviewing a conviction where this method played a key role. Governing.com reports that judges and prosecutors are pushing for bans and post-conviction reviews.

You need to spot this technique early in your workflow. It often arrives disguised as simple lay testimony. Accurate, timestamped transcripts are critical for identifying the specific indicators the prosecution might highlight.

Consider these risks when reviewing 911 audio:

• The technique is used to initiate investigations, not just prove guilt at trial.
Detectives may use it to elicit confessions during interrogations.Undetermined deaths have been reclassified as homicide based solely on this analysis.The COPS Scale indicators are subjective and lack peer-reviewed validation.Prosecutors may "guide juries" through the audio without expert qualification.

When you receive 911 audio, you need a precise record of what was said. Timestamped transcripts help you pinpoint the exact moments where the prosecution might argue for "guilt indicators." This precision is essential for building a Daubert challenge.

Evidence Copilot’s 911 call transcript analysis provides word-level timestamps and speaker diarization. This allows you to isolate specific phrases and compare them against the COPS Scale criteria. You can then cite the exact second of the audio in your motion.

This level of detail supports your cross-exam preparation. You can draft questions that target the subjective nature of the indicators. The goal is to show the jury that the analysis is opinion, not fact.

Understanding the baseline facts of this technique is your first line of defense. You must recognize it before it shapes the narrative of your case.

How It Sneaks Past Daubert: The Lay-Testimony Workaround

A technique that can't survive a Daubert hearing doesn't disappear. It changes costumes. That's the core finding in the University of Cincinnati Law Review's takedown of 911 call analysis: the method is "mixed into in-court statements and lay testimonies to sneak it past the standards for expert testimony." Cornell Law Review calls this "shadow forensics," and it explains why the challenge you're prepared to make isn't always the one you face.

The workaround works like this. The technique's founders are never proffered as experts, so there's no expert to cross. The underlying methods are never disclosed, so there's no methodology to attack. Instead, proponents encourage prosecutors to "guide juries through both the audio and transcript of the calls," as the Cincinnati Law Review documents. The analysis arrives as a detective's casual impression, not expert opinion, which means your Daubert motion may never get off the ground.

The challenge still works, though. A Michigan 16-year-old's murder conviction was overturned because 911 call analysis was improperly admitted. Harpster was rejected as an expert witness in an Alabama trial in 2009, and a Nevada judge rejected a detective trained in the method. Build your motions to target both routes:

• The expert route: no published, peer-reviewed foundation, a proprietary 38-indicator COPS Scale resting on roughly 200 calls, half from callers already convicted or confessed
• The FRE 701 route: a detective's "impression" of a caller's grammar is a disguised expert opinion, not a permissible lay observation
• The interrogation route: the method is used to elicit confessions and reclassify deaths, so check whether it contaminated the interview itself

That last point matters more than most defenders realize. The technique doesn't just show up at trial. It shapes investigations, and in one Ohio case a suspect pled to manslaughter after a detective used call analysis to draw out a confession. Pull the interrogation footage and any jail call transcription into the same review as the 911 audio, and compare what the client said before and after contact with the detective.

Accurate, timestamped 911 call transcript analysis is where this starts. You can't attack what you can't quote, and every indicator on that scale traces back to a specific word or pause in the recording.

Book a 20-minute walkthrough at evidencecopilot.com/book, or start on the Solo plan at evidencecopilot.com/signup. Every AI output is marked for human review, and every citation points to the exact moment it came from.

Spotting It in Discovery: A Screening Workflow for 911 and Jail Calls

Discovery dumps hide 911 call analysis in plain sight: no one labels a report "Harpster-trained," so contamination shows up as tone, sequencing, and word choice in the paperwork. A screening pass catches it before the interrogation narrative hardens.

Start with the transcript itself. You want a timestamped, diarized 911 transcript with word-level timing, because the technique's claims live at the word and pause level ("Huh?", "I don't know," awkward phrasing). A Cornell Law Review analysis shows the methodology is never disclosed and the founders never proffered as experts, so your transcript is often the only place the analysis becomes visible. Evidence Copilot's 911 call transcript analysis gives you that diarized, word-timed view with citations into the exact second.

Then run the contamination checks in order:

• Did call analysis shape the interrogation? Pull the interrogation transcript and look for the caller's own words quoted back at them as proof of guilt. The technique is used to initiate investigations and elicit confessions; in a 2012 Ohio case a suspect pled to manslaughter after a detective used call analysis to draw out a confession (University of Cincinnati Law Review).
• Did it reclassify the death? Check the medical examiner's original manner of death against the final one. One baby's death moved from undetermined to homicide on a detective's 911 call analysis alone (same review).
• Does the police report track the call's phrasing? Mirrored wording suggests the report was written from the analysis, not the event.
• Do jail calls match the statements? Compare what your client says on a recorded jail call against the 911 audio and the report. Statement-vs-call-audio inconsistencies surface on ingest as unreviewed signals; Evidence Copilot's jail call transcription flags them before anyone asks.
• Was a COPS Scale worksheet or training certificate in the file? The 38-indicator proprietary scale rests on 200 calls, half from already-convicted or confessed callers, with the underlying data never made public.

Two numbers set your posture. ProPublica counts more than 100 cases across 26 states where law enforcement used Harpster's methods, and North Carolina's indigent defense office issued a statewide warning telling attorneys to watch for it. Treat every 911 call in a homicide or unattended-death file as high-risk until proven clean.

The screening pass is cheap. The remediation, if you miss it, is a habeas petition.

Building the Challenge: Motions, Cross-Exam, and the COPS Scale's Weak Foundation

A 911 call lands in discovery, and the prosecutor never calls an expert. Instead, a detective narrates to the jury what the caller's "awkward phrases" really mean. That is the move you have to be ready to dismantle.

The technique travels under two names. When a detective or analyst takes the stand, the State will frame their testimony as "lay opinion" under FRE 701, or as an investigator explaining what they heard. Either framing sidesteps Daubert because the methodology stays undisclosed and the founder never testifies. As a 2026 Cornell Law Review article frames it, the field operates as "shadow forensics," sitting alongside bitemark evidence, hair microscopy, and arson investigation in a category that has contributed to "hundreds of exonerations, and likely thousands more wrongful convictions." Faulty forensic disciplines remain the second most common factor in wrongful convictions nationwide.

Your motion practice has to attack both routes at once. A motion in limine should target the expert path (no published peer review, no known error rate, no general acceptance) and the lay path (the witness is not a percipient observer of the call, they are applying a proprietary framework). The Michigan Court of Appeals overturned a 16-year-old's murder conviction because 911 call analysis was improperly admitted. That ruling is your template.

Then dismantle the COPS Scale on its own terms. The proprietary "Considering Offender Probability in Statements" instrument rests on a sample of 200 911 homicide calls, half drawn from callers already convicted or who had confessed, with the underlying data never made public. The indicator list grew from 16 in the original 2006 thesis to 18 in a 2009 paper and now 38 indicators of guilt or innocence. Indicators include things like "having a 'Defendant Mentality,'" saying "Huh?" or "I don't know," and using "awkward phrases." Each indicator is a cross-exam handle.

Build the questions from the source material, not from memory. For each indicator the State leans on, pin the detective to the underlying call quote and the indicator definition, then walk the jury through the circular logic: the indicators were derived from a sample where guilt was already presumed.

• "You scored this caller's 'I don't know' as a guilt indicator. How many of the 200 calls in your training sample came from people later proven innocent?"
• "Where is the underlying data? Has it been peer reviewed?"
• "Has the methodology been tested against a sample of known-innocent callers?"
• "What is your published error rate?"

The point is to make the jury hear the gap between confident in-court narration and the actual evidentiary foundation. The same workflow that produces a clean, timestamped 911 call transcript analysis is what lets you build those questions from the exact words on the page, with every quote tied to a source line. If the State also leans on post-arrest statements, pull the corresponding jail call transcription and surface any inconsistency between what the caller said at the scene and what detectives later claimed they heard.

The 2026 Climate: Post-ProPublica Leverage and What to Cite

The institutional ground has shifted, and 2026 briefs should say so plainly. Prosecutors, judges, and defense organizations have moved from skepticism to open condemnation of 911 call analysis, which gives you citable momentum no defender had two years ago.

Start with the scale of adoption. Law enforcement has employed Harpster's methods in more than 100 cases across 26 states, according to ProPublica's investigation. That same reporting documents the institutional response now unfolding:

• North Carolina's Office of Indigent Defense Services issued a statewide warning telling attorneys to watch for 911 call analysis in their cases.
• Fair and Just Prosecution urged member prosecutors to review past convictions through post-conviction review.
• The Illinois Supreme Court agreed to review Jessica Logan's case after the ProPublica reporting.
• NACDL published "911 Call Analysis: Another Example of Forensic Junk," and its president, Nellie King, said the system can "add 911 call analysis to the junk pile of fake scientific theories."

Pair those with the academic frame. Christopher Lau's Cornell Law Review article calls the technique "shadow forensics": methods never disclosed, founders never proffered as experts, so the technique largely escapes Daubert or Frye review. Faulty forensics and misleading testimony rank as the second most common factor in wrongful convictions, which is why this framing lands.

Present Harpster's rebuttal fairly, or the brief looks selective. He disputes ProPublica's framing, notes the outlet "never reported that 911 call analysis caused the conviction of innocent persons," says the analysis has helped exonerate people, and points to alternative evidence in the Logan and Helmick cases, as documented in Governing's coverage. Even supporters concede the limits: the Colorado Bureau of Investigation's Susan Medina called it "nothing more than a potential investigative lead."

That concession is your wedge. Cite the institutional warnings, the shadow-forensics framework, and the technique's own proponents agreeing it isn't decisive. Then show where the call sits in your discovery: transcribed with word-level timestamps, quoted, and cited, the way Evidence Copilot's 911 call transcript analysis supports review.