Introduction

Your client called 911 the night her husband died. She was calm on the phone, said "hi" to the dispatcher, and didn't beg for help in the first ten seconds. A detective who attended an eight-hour training course is about to tell the jury that those details mean she's guilty. That's 911 call analysis, and in 2026 you need to know exactly how it works and how to break it.

The technique traces back to Detective Tracy Harpster's 2006 master's thesis, which analyzed 100 homicide calls and produced 16 supposed guilt and innocence indicators. It has since grown into a proprietary framework with 38 indicators, marketed to police departments for up to $3,500 per training, and Harpster has consulted on at least 2,500 homicide investigations. According to ProPublica's investigation, law enforcement has used the method in more than 100 documented cases across 26 states.

The problem is that the science collapsed. The FBI's Behavioral Analysis Unit tested the scale in 2025 and found officers misclassified calls over 40% of the time, and that some of the most influential "guilt indicators" actually appeared more often among innocent callers. At least eight replication studies failed to validate the method at all. Yet the technique keeps reaching juries, usually through a side door: instead of proffering Harpster as an expert, prosecutors put trained officers on the stand to offer lay opinions about the caller's tone and word choice. A Cornell Law Review article calls this evasion "shadow forensics," and it has largely escaped Daubert and Frye scrutiny because the underlying methodology is never disclosed in court.

This guide covers what defenders actually need in 2026:

• The empirical record you can cite in an admissibility fight, including the FBI's misclassification findings
• How to spot the lay-testimony workaround before trial, from COPS Scale language in reports to officer training records in discovery
• Cross-examination angles built on the failed replications and researcher allegiance effects
• Post-conviction review momentum, from North Carolina's statewide warning to the Illinois Supreme Court's reexamination of a affected conviction

The practical work starts with the evidence itself. Reviewing the call audio against the police report, bodycam footage, and your client's statements is where inconsistencies surface, and tools like Evidence Copilot's 911 call transcript analysis and jail call transcription can help you flag those gaps with citations to the exact moment in the recording. The State's narrative hardens early; your job is to find where it breaks before it does.

Key Concepts

If a detective tells your jury that a caller "didn't ask for help fast enough" or was "too polite," you're probably looking at 911 call analysis. Understanding where it came from, what the science actually says, and how it slips into trials is the foundation for every challenge you'll build.

The method traces to Detective Tracy Harpster of Moraine, Ohio, whose 2006 master's thesis analyzed 100 homicide calls and produced 16 guilt/innocence indicators. It grew through a 2009 paper with 18 indicators into a 2017 book built on 200 calls and 38 indicators organized into the proprietary "COPS" Scale. Harpster has consulted on at least 2,500 homicide investigations, and the method's creators claim it helped solve more than 1,500 homicides, according to the Innocence Project.

The validation record tells a different story. At least eight replication studies failed to validate the method's guilt indicators. An FBI Behavioral Analysis Unit evaluation found sworn officers misclassified calls over 40% of the time, and some of the most influential "guilt indicators" were actually more common among innocent callers. The FBI team concluded the scale "should not be used in actual investigations to evaluate callers' potential innocence or guilt," per the APA's summary of the research. An independent Villanova analysis found the scale's total score failed to differentiate deceptive from truthful callers, and documented a researcher allegiance effect: the scale's own authors reported far stronger results than independent replications did.

So how does an unvalidated technique reach juries? Through what Cornell Law Review scholarship calls "shadow forensics." Prosecutors rarely qualify Harpster or his methods as expert testimony subject to Daubert or Frye scrutiny. Instead, officers and dispatchers who attended call-analysis training testify as lay witnesses about a caller's culpability. Harpster, who has never testified as an expert in a criminal trial, advises prosecutors to use the technique in closing arguments and context, guiding juries through call audio to "create lasting impressions."

The scale of use is real. ProPublica documented more than 100 cases across 26 states where law enforcement applied these methods. After that investigation, North Carolina's Office of Indigent Defense Services issued a statewide warning to attorneys, and Fair and Just Prosecution called on members to review past cases through post-conviction processes.

When you screen a file, watch for these fingerprints:

• Officer testimony or closing arguments built on COPS Scale indicators: no immediate plea for help, being "too polite," saying "hi" to the dispatcher
• A detective or dispatcher who attended an 8-hour training course, which departments paid up to $3,500 for, often with taxpayer funds
• Consultation emails between prosecutors and Harpster about how to present the calls to a jury
• Investigation decisions that flowed from a caller-involvement judgment, which the FBI found shaped subsequent interview questions

Behavioral research adds another angle: Dr. Jessica Salerno's work found both civilians and police became significantly more suspicious of low-emotion callers, a pattern that held regardless of whether the caller was later convicted or exonerated. When you're reviewing call audio and transcripts against the police report and bodycam, grounded tools with timestamped citations (such as Evidence Copilot's 911 call transcript analysis and jail call transcription workflows) help you flag report-versus-call inconsistencies before the State's narrative hardens.

Best Practices

When a prosecutor's file contains a 911 call, your first move is to assume the State has already run it through Harpster-style analysis, whether the discovery says so or not. The technique has been documented in more than 100 cases across 26 states, so the burden is on you to find the footprint before it reaches the jury.

Start with the admissibility record, because it's lopsided in your favor. An FBI Behavioral Analysis Unit evaluation found sworn officers misclassified calls over 40% of the time and concluded the COPS Scale "should not be used in actual investigations to evaluate callers' potential innocence or guilt." At least eight replication studies have failed to validate the method's guilt indicators. Anchor your Daubert or Frye objection in those 2025 findings, not in argument.

Then attack the workaround. The reason this technique survives scrutiny is that prosecutors rarely qualify anyone as an expert. Officers testify about "guilt indicators" as lay witnesses after attending an 8-hour training course that departments paid up to $3,500 for. Cornell Law Review scholarship calls this "shadow forensics": the methods stay undisclosed and the founders never take the stand, so there's nothing to exclude. Your job is to make the methodology visible.

Practically, that means:

• Screen officer testimony and closing arguments for COPS Scale markers like "too polite," no immediate plea for help, or saying "hi" to the dispatcher, and object when disputed methodology reaches the jury without expert foundation.
• Audit discovery for training footprints: check whether any officer, dispatcher, or detective attended Harpster-style training, and demand consultation emails. ProPublica documented prosecutors asking for "suggestions on how to introduce the 911 calls to the jury during trial."
• Raise confirmation-bias arguments. The FBI found caller-involvement judgments shaped subsequent interview questions, and a University of Michigan Ford School brief found the technique reinforces investigative confirmation bias and diverts attention from more reliable evidence.
• Compare the call against everything else. Pull the dispatcher's CAD notes, the bodycam, and the police report, and look for inconsistencies between the call audio and the written narrative.

That last step is where most defenses are won or lost, and it's where the volume problem hits. A single case can mean hours of 911 audio, jail calls, and bodycam that need timestamped, diarized transcripts you can actually cite. This is the workflow 911 call transcript analysis at Evidence Copilot is built for: every AI answer cites the exact second of audio it came from, flagged for human review, with auto-signals that surface report-versus-call inconsistencies before the State's narrative hardens. The same applies to jail call transcription when the State tries to read motive into a client's recorded phone calls.

Finally, watch the post-conviction window. The Illinois Supreme Court is reexamining Jessica Logan's conviction, North Carolina's Office of Indigent Defense Services issued a statewide warning to attorneys, and Fair and Just Prosecution has called on members to review past cases. If your client's conviction involved call analysis, that momentum is your opening.

Implementation

The defense work doesn't end with knowing the science. It ends with what you do on Monday morning when the discovery dump arrives.

Start by pulling the 911 audio and transcript into the same workspace as the police report and bodycam. Side-by-side review is where the inconsistencies surface. If the caller is described as "calm and calculated" in the report, but the audio shows a panicked voice asking the operator to repeat the address three times, that's a discrepancy worth flagging before the State's narrative hardens. 911 call transcript analysis tools that link each transcript word to the exact audio timestamp let you cite the precise moment a witness's characterization diverges from the recording.

Next, audit discovery for the training footprint. ProPublica documented that departments have paid up to $3,500 for 8-hour Harpster-style courses, and prosecutors have emailed consultants asking for "suggestions on how to introduce the 911 calls to the jury during trial." A targeted discovery request can surface:

• Training records or certificates for any officer, dispatcher, or detective involved in the case
• Emails between the prosecution and 911 call analysis consultants
• Internal memos or slide decks referencing the COPS Scale or its 38 guilt indicators
• Prior testimony where the same officer described a caller's demeanor using guilt-coded language

When the training footprint exists, the evidentiary picture changes. You're no longer arguing about whether a caller's tone was "too polite" or lacked an "immediate plea for help." You're arguing that a specific methodology, one the FBI Behavioral Analysis Unit found misclassifies calls over 40% of the time, reached the jury through lay testimony designed to bypass Daubert scrutiny.

Build the suppression or motion in limine around the 2025 empirical record. The FBI BAU study and the Villanova failure-to-differentiate analysis, both published in Psychology, Public Policy, and Law, give you peer-reviewed ammunition. Pair that with the eight failed replication studies and the Cornell Law Review's "shadow forensics" framework, which names the lay-testimony workaround by its legal term. That article is the first scholarly piece to situate 911 call analysis alongside bitemark evidence and arson investigation in the junk-forensics lineage, a framing judges recognize.

Finally, cross-reference the 911 timeline against the bodycam and any jail calls. A defendant in custody often describes the same incident differently than the 911 caller did, and the delta between those accounts is where Brady and Giglio material lives. Jail call transcription that produces searchable, diarized text lets you search for the same locations, names, and timestamps across every recording in the case. When you find a statement on a jail call that contradicts the prosecution's theory of who placed the 911 call, you have something to work with.

The technique has contributed to wrongful convictions. Faulty scientific disciplines are the second most common factor in exonerations nationwide. Every 911 call in every case deserves the same review you'd give a confession tape: timestamped, cited, and stress-tested against everything else in the file.

Conclusion

The empirical record against 911 call analysis is now strong enough that the question for defense teams isn't whether to challenge it, but where in the case to strike first. Start with the science. The FBI's Behavioral Analysis Unit found sworn officers misclassified calls over 40% of the time, and at least eight replication studies failed to validate the method's guilt indicators. Harpster himself has consulted on at least 2,500 homicide investigations, yet the COPS Scale still has no independent validation behind it.

Because prosecutors rarely proffer the technique through an expert, your challenge usually targets lay testimony. Screen officer and dispatcher testimony, plus closing argument, for COPS Scale indicators like "no immediate plea for help," "too polite," or saying "hi." When disputed methodology reaches the jury without expert foundation, object. Cornell Law Review calls this evasion "shadow forensics," and naming it on the record matters.

Then audit discovery for footprints:

• Check whether any officer, dispatcher, or detective attended Harpster-style training (departments paid up to $3,500 for 8-hour courses).
• Pull consultation emails, including any prosecutor requests for "suggestions on how to introduce the 911 calls to the jury."
• Compare the call audio and transcript against bodycam and the police report for report-vs-call inconsistencies.
• Document confirmation-bias contamination: the FBI found caller-involvement judgments shaped subsequent interview questions.

Post-conviction momentum is real. North Carolina's Office of Indigent Defense Services issued a statewide warning, Fair and Just Prosecution urged members to review past cases, and the Illinois Supreme Court is reexamining Jessica Logan's conviction. Even the Colorado Bureau of Investigation concedes the technique is "nothing more than a potential investigative lead."

For teams working at case scale, grounded review helps here: Evidence Copilot's 911 call transcript analysis and jail call transcription keep every citation tied to the exact second of audio or page of the report, so inconsistencies surface before the State's narrative hardens. Every AI output is marked for human review, and the call is always yours.