What 911 Call Analysis Is and Why Prosecutors Keep Using It

If your client's 911 call has been described in a police report as showing a "guilty demeanor" or "defendant mentality," there's a good chance the officer is applying a method called 911 call analysis, and it's worth understanding where that method came from. The technique, most closely associated with former Detective Tracy Harpster, rests on a scoring framework known as the COPS Scale, which assigns meaning to how callers speak during emergency calls in homicide investigations.

The scale identifies 38 indicators organized into six categories in its 2017 version, up from 16 indicators in 2006 and 18 in 2009. Those indicators are used to infer whether a caller is genuinely a victim or is actually the offender. The claimed payoff for prosecutors is intuitive: the theory holds that 80 percent of callers who say they need help are innocent, meaning the analysis is aimed at the remaining 20 percent, and your client may be in that bucket.

The adoption numbers explain why this keeps showing up in your cases. Prosecutors and investigators in at least 26 states have used 911 call analysis in investigations and prosecutions, and Harpster has consulted on at least 2,500 homicide investigations as of March 2024. His training course runs 8 hours and costs departments thousands of dollars, which is a modest investment for a technique that can help convert a caller's panic into evidence of guilt.

Here's the problem for the defense. The COPS Scale is built on a sample of just 200 911 homicide calls, down from earlier studies of 100 calls each. According to an analysis in the University of Cincinnati Law Review, the method has not been subjected to peer review, and Harpster and his collaborators have never made their underlying data publicly available. No one has been able to replicate the findings, and Harpster himself has never testified as an expert witness in a criminal trial despite the thousands of consultations. That combination, small sample, no peer review, no public data, no replication, fails both Daubert and Frye standards, which is why the National Association of Criminal Defense Lawyers has called it another example of forensic junk.

The harder problem is how prosecutors get this material in front of a jury anyway. Rather than qualifying an analyst as an expert and facing an admissibility hearing, prosecutors have been coached to weave the analysis into lay witness testimony. Harpster has provided prosecutors with "ten tips" to guide juries through the audio and transcripts and create lasting impressions, without ever formally claiming expertise. A detective who listened to the call for 20 minutes can testify about what "stood out" to him, and the COPS Scale does the work invisibly.

That's the core threat: a method built on 200 sample calls is being used to interpret your client's words as evidence of guilt, and it often enters the record without ever being tested as science. When discovery includes 911 audio alongside bodycam and police reports, a grounded review workspace like Evidence Copilot can produce timestamped, diarized 911 call transcript analysis with word-level timing, so you can compare the caller's exact words against the officer's later characterization of them. The same workflow applies to jail call transcription, where diarized transcripts with citations back to the exact second let you verify what was actually said before the state's interpretation hardens into fact.

Challenging the Method: Daubert, Frye, and the Lay-Testimony End Run

The most dangerous move the prosecution makes isn't putting an expert on the stand. It’s keeping the expert off it.

When a prosecutor introduces 911 call analysis, they often bypass Daubert entirely by having a detective or dispatcher offer "lay" opinions. This tactic lets the jury hear linguistic conclusions without the rigor of expert qualification.

Detective Tracy Harpster, the method’s creator, has consulted on over 2,500 homicide investigations but has never testified as an expert witness in a criminal trial, according to University of Cincinnati Law Review analysis. Instead, he provides prosecutors with "ten tips" to guide juries through the audio.

This allows the state to present specialized linguistic theories as common-sense observations. The method claims that 80 percent of callers who say they need help are innocent, a statistic derived from small, non-public samples, as noted by Miami defense attorney Julia Kefalinos.

You must attack this infiltration directly. Under Federal Rule of Evidence 701, lay opinion is limited to the witness’s direct perception. If the detective is applying a scale of 38 indicators, that is specialized knowledge, not a lay observation.

To catch this, you need to know exactly what the witness heard versus what they interpreted. A transcript alone often misses tone and hesitation. That’s why you should review both the text and the original audio to spot where the witness’s "perception" diverges from the record. Tools for 911 call transcript analysis help you pinpoint these gaps before you walk into the courtroom.

Here is how to structure your cross-examination of the responding officer:

• Ask about training: "Did you receive specific instruction on how to interpret 'polite' language in 911 calls?"
• Pin down the source: "You said the caller sounded 'defensive.' Which specific word in the transcript led you to that conclusion?"
• Challenge the baseline: "You’re using a scale to judge this call. Is that scale part of your department’s standard operating procedure?"