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

You open the discovery folder and find a file labeled "911 Call Analysis." It’s not a standard police report or bodycam footage. It’s a forensic technique that claims to identify guilt based on how a caller speaks.

Retired Detective Tracy Harpster developed this method. It purports to detect deception through word choice, cadence, and grammar. The technique has been documented in more than 100 cases across 26 states.

You will encounter it in trial transcripts, police reports, and discovery dumps. Prosecutors often use it to build a narrative before physical evidence arrives. This creates a specific risk for your defense strategy.

The method relies on the COPS Scale. This proprietary framework identifies 38 specific "guilt indicators." These include being too polite, using awkward phrases, or failing to show high emotion. The underlying data for this scale is not public. Harpster and his co-author have not released the raw calls or methodology for peer review.

Courts have repeatedly rejected this technique as expert testimony. In Alabama, Nevada, and Michigan, judges ruled it fails the Daubert and Frye standards. The Michigan conviction of Riley Spitler was overturned specifically because this analysis should not have been admitted.

Despite these rulings, the technique persists. Prosecutors often bypass expert testimony rules by using lay witnesses. A detective or dispatcher might testify about "suspicious behavior" without formal qualification. This allows the jury to hear the implications without the scrutiny of scientific validation.

This approach creates a confirmation bias loop. The initial suspicion from the call shapes the interrogation. It influences forensic interpretation and narrows the investigative focus. The University of Michigan Ford School notes this diverts attention from more reliable evidence.

You need to identify these signals early. When reviewing the 911 audio, look for specific patterns that match the COPS Scale indicators. This is where precise transcription matters. You must catch every hesitation, filler word, and tonal shift.

Tools like 911 call transcript analysis help you isolate these moments. You get a timestamped, diarized transcript. You can flag specific phrases and cross-reference them with the police report. If the report cites "emotional instability" but the audio shows a calm, logical caller, you have a discrepancy.

Similarly, jail call transcription can reveal inconsistencies in subsequent statements. If the caller’s tone changes significantly after detention, that context matters. You need the full audio context, not just a summary.

The National Association of Criminal Defense Lawyers has called this method "dangerous and insidious." They urge it be added to the "junk pile of fake scientific theories." You should treat every 911 call analysis reference in your case file with that same skepticism.

Your goal is to challenge the admissibility of any testimony based on this framework. You also need to expose the confirmation bias in the investigation. The data is in the audio. Your job is to ensure the jury hears the truth, not a forensic guess.

Why Courts and Defense Organizations Call It Junk Science

If a prosecutor offers a detective's opinion about a 911 caller's "guilt indicators," the scientific record is already against them. Courts, defense organizations, and even reform-minded prosecutors have converged on the same conclusion: 911 call analysis is unvalidated junk science dressed up as forensic expertise.

The admissibility failures are well documented. The technique, built on retired Detective Tracy Harpster's work, has been rejected as expert testimony in Alabama (2009), Nevada, and Michigan, and the Michigan conviction of Riley Spitler was overturned because the analysis should never have been admitted. Under the Daubert five-factor test (testability, peer review, error rate, standards, general acceptance) and the Frye "general acceptance" standard, the method fails at every step. Harpster and his co-author never released the underlying data from the COPS Scale, never explained how they obtained the calls used to build it, and never published an error rate. At least eight independent replication studies have tried to validate the guilt indicators and found they don't hold up under scrutiny.

That track record explains the institutional pushback:

• NACDL president Nellie King called the business of 911 call analysis "dangerous and insidious," urging that it be added to "the junk pile of fake scientific theories."
• Fair and Just Prosecution has called on elected prosecutors to review past cases through post-conviction processes.
• North Carolina's Office of Indigent Defense Services issued a statewide warning telling attorneys to watch for the technique in their cases.
• An Ohio judge wrote to ProPublica: "I never anticipated that prosecutors, officers of the court, would engage in systematic organized frauds."

The stakes are not abstract. According to the Innocence Project and ProPublica reporting, faulty science is the second most common factor in wrongful convictions, and the technique has appeared in more than 100 cases across 26 states. Because the analysis fails scientific scrutiny, prosecutors increasingly present it through lay testimony from trained detectives rather than qualified experts, evading Daubert challenges entirely. ProPublica documented that some prosecutors knew the method "would not be recognized as scientific evidence but still disguised it in trial against unwitting defendants." Harpster even distributes "ten tips" for getting the analysis in front of juries without expert qualification.

For your practice, the practical takeaway is this: when 911 call audio shows up in discovery, pull the transcript early and check whether any detective involved received the 8-hour, up-to-$3,500 training or consulted with Harpster. If the analysis influenced the interrogation or investigative focus, that confirmation bias contaminates everything downstream, which is exactly the argument the University of Michigan Ford School brief makes when it says the technique "reinforces investigative confirmation bias and diverts attention from more reliable evidence."

Tools that keep the audio, transcript, and your annotations tied to exact timestamps make that challenge easier to build. When you're preparing a Daubert motion or an FRE 701 objection, a grounded review of the 911 call transcript analysis in the case, alongside any jail call transcription, keeps every quote you cite anchored to the source moment, with human review on every AI-assisted output.

The Lay-Witness Workaround (and How to Spot It)

Prosecutors know the COPS Scale won't survive a Daubert hearing, so they stopped offering it as expert testimony. Instead, they route it through people the court already trusts: detectives and dispatchers who testify as lay witnesses about what struck them as "off" in a call.

That's the workaround, and it's documented. According to Innocence Project reporting, prosecutors present 911 call analysis indirectly through lay testimony from trained detectives and dispatchers rather than as expert testimony. Dr. Jessica Salerno's advice for defenders is direct: identify and challenge those attempts. When an officer describes a caller's "guilt indicators" or "deceptive cadence" without FRE 702 qualification, object under FRE 701. That's specialized knowledge wearing a badge.

The indicators themselves are the problem. Harpster's 2017 book expanded his sample to 200 calls and organized 38 indicators into the proprietary COPS Scale, per the University of Cincinnati Law Review. Several ensnare ordinary, innocent callers:

• Being "too polite," including a simple "hi" as a greeting (flagged in a 2018 Washington case)
• "Defendant Mentality," meaning word choices the analyst associates with guilt
• No immediate plea for help
• Saying "Huh?" or "I don't know"
• Unintelligible comments and awkward phrasing

Salerno's research found police and civilians are significantly more suspicious of callers who show low emotion. For officers, even moderate emotion isn't enough; a caller has to sound highly emotional to be read as non-guilty. Think about what that means for a caller in shock, or one who's dissociating, or one who just watched someone die. If the public learns cops use behavior in the worst moments of their life against them, Salerno warns, people will stop calling 911 to help others.

Your job is to catch the workaround early, then pull the training records that prove it. Discovery requests for detective and dispatcher training histories, Harpster consultation emails, and any COPS Scale materials in the file do double duty: they support your FRE 701 objection and they build the confirmation-bias argument, since initial suspicion based on a call shapes interrogations, forensic interpretation, and which leads get pursued. Running 911 call transcript analysis against the rest of the discovery, report against recording, statement against call audio, surfaces those contradictions before the hearing does.

Challenging the Investigation: Confirmation Bias, Location Data, and Brady

When a detective walks into an interrogation room already believing the caller is guilty, everything that follows bends around that belief. The University of Michigan Ford School policy brief found that 911 call analysis "reinforces investigative confirmation bias and diverts attention from more reliable evidence." That contamination is your leverage.

The mechanism is straightforward. Initial suspicion from a caller's word choice or tone shapes subsequent interrogations, influences forensic interpretation, narrows investigative focus, and determines which leads get pursued or ignored, according to the Innocence Project. You see this concretely in the Fremont, Ohio case where Detective Jason Kiddey used a 911 call analysis handout to get a man to "crack." The man pled to manslaughter and received an 11-year sentence.

So your discovery requests should target the training trail:

• Whether any investigator or detective attended 911 call analysis training, which runs 8 hours and costs participants up to $3,500
• Any emails between investigators and Tracy Harpster or John Adams consulting on the case
• Any reference to the COPS Scale or "guilt indicators" in reports, notes, or case files
• The investigator's training records and qualifications, which bear directly on Giglio material if their credentials are overstated

Under Brady and Giglio, this material cuts two ways. If call analysis shaped the investigation, you can argue every downstream conclusion, from the interrogation strategy to the forensic interpretation, flowed from an unvalidated premise. And if the state plans to put a trained detective on the stand to walk the jury through the call, training records become the foundation for your FRE 701 objection.

Location evidence deserves the same skepticism. An estimated 240 million calls hit 911 each year, and in many areas 80% or more come from wireless devices, per NENA statistics. The critical distinction: Phase I wireless service delivers only the location of the cell tower handling the call, not the caller's location. Phase II delivers caller location. Nearly all PSAPs have some capability (99.4% have some Phase I, 99.2% have some Phase II), but "some" is doing heavy lifting in that sentence.

When a report says the caller was "located" at a specific address, subpoena the PSAP records. Determine whether the jurisdiction had Phase II at the time of the call and whether the recorded location reflects the tower or the caller. A tower a mile from the scene is not an alibi breaker.

When you're working through hundreds of pages of discovery to find that one training email or one "guilt indicator" notation, tools like Evidence Copilot let you search transcripts and OCR'd documents with citations back to the exact page, so nothing slips past. The work is still yours; the retrieval just gets faster.

Your Defense Playbook: From Motion to Suppress to Post-Conviction Review

If 911 call analysis touched your case at any stage, your job is the same whether you're preparing a motion to suppress, defending a trial, or reopening a conviction: build the record that shows the technique isn't science, then attack every place it leaked into the investigation. Here's the sequence.

Step 1: Move to exclude under Daubert/Frye. If a prosecutor offers a detective as an expert on caller deception, file a motion in limine or Daubert challenge. The record is on your side: courts in Alabama (2009), Nevada, and Michigan have rejected the technique as expert testimony, and Michigan's Riley Spitler conviction was overturned because the analysis should never have been admitted. Build your motion around the Daubert factors the method plainly fails: no peer review, no published error rate, and underlying data Harpster and Adams have never released. At least eight replication studies found the guilt indicators don't hold up.

Step 2: Object to the lay-witness workaround. This is where most 911 call analysis sneaks in now. ProPublica documented that prosecutors who knew the technique wouldn't survive Daubert "got creative," presenting it through detectives testifying as lay witnesses. Harpster even distributes "ten tips" for getting the analysis in front of juries without expert qualification. When a witness starts opining about "guilt indicators" or "deceptive cadence," object under FRE 701: that's specialized knowledge reserved for FRE 702, and the witness isn't qualified.

Step 3: Subpoena the PSAP records. Location evidence deserves its own challenge. Phase I wireless 911 delivers only the cell tower's location, not the caller's, and "Enhanced 9-1-1" is not synonymous with wireless 911, per NENA statistics. Demand records showing whether your jurisdiction had Phase II capability at the time of the call and what the recording actually captured.