The Lay Testimony Trap: Why Courts Are Letting It In

The prosecution knows 911 call analysis won't survive a Daubert hearing. So they stopped trying to qualify it as expert testimony and started sneaking it in another way.

Here's the playbook. A detective or dispatcher takes the stand and walks the jury through the audio and transcript of the call. They point out the caller's word choices, the pauses, the "missing" grief. They never mention the COPS Scale, never mention Harpster, never mention the 38 guilt indicators. The jury hears an experienced investigator explaining what they heard on the call. It sounds like common sense. It isn't.

The method's own creator trains prosecutors to do exactly this. As the UC Law Review's analysis details, Harpster advises trained prosecutors to "guide juries through both the audio and transcript of the calls" without seeking expert qualification. The goal is to get the conclusions in front of the jury while keeping the methodology out of reach of the gatekeeping standard.

It works more often than it should. The technique has surfaced in more than 100 cases across 26 states, and it is almost always introduced through this lay-testimony side door. Police departments pay up to $3,500 for training programs that promise competency after just eight hours of instruction.

The admissibility record is mixed but the trend line is running against the prosecution. Recent appellate courts have started to push back hard.

• In Michigan, 16-year-old Riley Spitler's murder conviction was overturned on the ground that 911 call analysis should never have been admitted.
• In Illinois, the state supreme court agreed to review the Jessica Logan case after ProPublica's reporting on how the technique shaped the investigation.
• In Alabama, Harpster himself was rejected as an expert in 2009.
• In Nevada, a judge excluded a detective trained in the method.

Your motion practice has to attack the workaround directly. Under FRE 701 and 702, testimony that rests on specialized training in a proprietary analytical method is expert opinion in lay clothing, and dual-role law enforcement witnesses add an unfairly prejudicial aura of reliability that the rules are designed to prevent. Object before the detective opens their mouth, and demand disclosure of the training, the underlying data, and any prior disqualifications.

The cross-examination theme writes itself once you have the record. Dr. Jessica Salerno's Cornell research shows that low emotion triggers suspicion in police officers regardless of whether the caller is guilty or innocent. Robert Roberson's flat affect was read as guilt; he was later diagnosed with autism spectrum disorder after more than two decades on Texas death row. When the state's case leans on how your client sounded on the call, that's the argument.

Getting the transcript right is half the fight. Precise 911 call transcript analysis with word-level timestamps and speaker diarization lets you pin the detective to the exact second they claim "guilt" appeared on the tape. The same discipline applies to jail call transcription when the state tries to read coded meaning into a conversation your client had while sitting in a cell.

Deconstructing the Science: The COPS Scale and Failed Replications

Every Daubert challenge to 911 call analysis eventually lands on the same uncomfortable question: where's the science? The answer, documented across multiple investigations, is that it traces back to a master's thesis with a sample size of 100 calls.

That thesis, written by Detective Tracy Harpster in 2006, was expressly exploratory research. One co-author stated the findings weren't ready for real-world application, yet the method went from 16 indicators in the original study to 18 guilt indicators in a 2009 paper to 38 indicators in the proprietary COPS Scale by 2017, built on an expanded sample of just 200 calls, according to a UC Law Review analysis.

The replication record is worse. At least eight subsequent studies failed to replicate or validate the method, per the Innocence Project. Researchers who tried to corroborate Harpster's claims simply couldn't.

Then there's the data problem. Harpster and co-author Dr. Susan Adams have never released the underlying calls used to build the COPS Scale, how they obtained them, or their research methods. Harpster has also never testified as an expert at a criminal trial, which means the technique's creator has never once defended it under oath as scientific evidence.

When you map these gaps against the Daubert factors, the challenge writes itself:

• Testability and error rate: no released data means no independent testing and no known error rate.
• Peer review and publication: the foundational "research" is a proprietary scale built from 200 calls, contradicted by at least eight failed replications.
• General acceptance: ProPublica canvassed roughly 50 departments and associations; only 11 responded, and those that did distanced themselves or minimized the technique's role.
• Contrast with admitted behavioral evidence: in Boyd v. City & County of San Francisco, suicide-by-cop testimony survived Daubert because the expert tied conclusions to roughly 10 peer-reviewed articles. 911 call analysis has no comparable literature.

Harpster's counter-narrative, laid out in a letter to the editor, is that he has reviewed over 2,500 equivocal death cases and is unaware of any wrongful conviction resulting from the analysis. Anticipate it. But case volume isn't validation, and a method whose data has never seen daylight can't meet the burden of establishing admissibility that sits with the proponent.

Before you draft the motion, get the call itself under control: a clean, timestamped 911 call transcript with accurate speaker diarization is the foundation for both the Daubert briefing and any cross of the detective who "walked the jury through" the audio. Tools like Evidence Copilot produce diarized, word-level transcripts of 911 audio and jail calls, so every claimed "guilt indicator" can be checked against the actual recording rather than the analyst's characterization.

Weaponizing Emotional Bias: The Cross-Examination Angle

The flat, quiet 911 call from your client is doing damage in ways most defenders never map. Dr. Jessica Salerno's research at Cornell gives you a concrete, science-backed cross-examination theme: suspicion based on emotional display has nothing to do with actual guilt.

Salerno's team found that both civilians and police officers become significantly more suspicious when callers express low emotion. For officers, the effect was harsher. Even moderate emotion wasn't enough to change their perceptions; only highly emotional callers did. And in follow-up studies using real 911 calls with known outcomes, low emotion was associated with guilt whether the caller was later convicted or exonerated. That last point is your hammer: the display predicts suspicion, not truth.

This matters because 911 call analysis trains detectives and dispatchers to treat exactly these displays as "guilt indicators." The COPS Scale's 38 indicators include emotional presentation, and prosecutors now introduce the method through lay testimony from trained detectives rather than qualified experts, so jurors hear about "guilty" behavior without learning the science is disputed. When your client's flat affect gets characterized as suspicious, you can cross on the mechanism itself:

• Ask the detective whether they were trained that low emotion indicates guilt, and whether they know that at least eight studies have failed to replicate or validate the method.
• Establish that the training came from programs promising competency after just eight hours of instruction, costing departments up to $3,500.
• Walk them through Salerno's finding that low emotion triggers suspicion regardless of whether the caller is innocent, using real calls with known outcomes.
• Connect it to the cascade: mistaken suspicion from a call interpretation shaped the interrogation, the forensic interpretation, and the investigative focus that followed.

Robert Roberson spent more than two decades on Texas' death row, narrowly avoiding execution three times, after his flat affect was read as evidence of guilt. He was later diagnosed with autism spectrum disorder. Trauma, culture, shock, and neurodivergence all shape how people respond in crisis, yet the legal system keeps relying on subjective assumptions about what grief and innocence should look like.

To run this cross well, you need the call itself diarized and timestamped, so every "low emotion" moment the detective flags can be played against the transcript and the known science. A 911 call transcript analysis workflow in Evidence Copilot lets you cite the exact second of audio behind every flagged display, and jail call transcription covers the downstream statements the suspicion produced. Every AI answer carries citations and stays pending human review; the calls and judgments remain yours.

Salerno recommends that defense attorneys learn to identify and challenge the technique when it arrives through lay testimony. Her warning is worth quoting at trial: "If the public learns that cops are using their behavior in the worst moments of their life against them, people are going to stop calling 911 to help others."

Building the Challenge File: Practical Steps for Motion Practice

Your challenge file wins or loses before the hearing starts. Prosecutors have used Harpster's method in more than 100 documented cases across 26 states, and because courts have rejected it as expert evidence, they increasingly present it through trained detectives or dispatchers instead. Your motion practice has to attack both the technique and the workaround.

Start with disclosure demands. The COPS Scale's underlying calls, research methods, and data have never been publicly released, and its creator has never testified as an expert at a criminal trial, as documented in a UC Law Review analysis. So serve discovery aimed at:

• The specific training the testifying detective received, including the fact that departments pay up to $3,500 for eight hours of instruction
• Any written protocol, indicator list, or worksheet the analyst used on your client's call
• The analyst's qualifications, prior disqualifications, and any error-rate publications
• The underlying data supporting the 38 "guilt indicators," with exclusion or continuance if disclosure is withheld

Next, map each Daubert factor against the documented gaps: the 2006 origin study examined only 100 calls as exploratory research, at least eight subsequent studies failed to replicate the method, and no peer-reviewed validation exists. The proponent bears the burden of establishing admissibility, and sample defense motions argue dual-role law enforcement testimony casts "an unfairly prejudicial aura of reliability and trustworthiness" on lay opinion, a framing you can borrow from the federal defender motions bank.

Prepare for the counter-narrative now. Harpster claims review of more than 2,500 equivocal death cases with no known wrongful convictions, and prosecutors have acknowledged the technique "will not hold up to scientific scrutiny," which is precisely why they get creative with lay testimony. Expect the state to argue the detective is merely describing "what the call sounded like." Your response: any testimony that walks the jury through indicators of guilt is expert opinion in lay clothing under FRE 701/702, and it should not dodge gatekeeping.

Finally, build the factual record that undercuts the narrative. A grounded evidence review of the 911 audio against the police report and bodycam frequently surfaces inconsistencies: a report describing a "calm, calculated" caller when the audio shows panic, or a detective's timeline that doesn't match the call's timestamps. This is where a tool like Evidence Copilot earns its place in the workflow; upload the discovery dump and the AI evidence agent returns timestamped, diarized transcripts with clickable citations into the exact second of the 911 call or page of the report, with candidate inconsistencies surfacing automatically before you ask. Every output is marked AI-generated and pending human review, so the work product stays yours. Whether you use 911 call transcript analysis support or jail call transcription for related communications, the goal is the same: pin every claimed "indicator" to the actual audio, then impeach with the record.