The Junk Science Problem

You have likely seen the term "911 call analysis" in discovery packets or police reports. It feels intuitive, but it is scientifically hollow.

This technique claims to detect guilt based on word choice and cadence. Yet, researchers who attempted to corroborate these claims failed completely. A ProPublica investigation documented its use in over 100 cases across 26 states.

The methodology relies on the proprietary COPS Scale. This scale flags innocuous behaviors as indicators of guilt. For example, saying "I don't know" or being too polite triggers a flag. These are normal human responses, not evidence of deception.

The scientific community rejects this approach for a reason. The creators have never released their raw data or sample methodology. This lack of transparency violates standard research practices. Consequently, courts have repeatedly excluded this evidence under Daubert and Frye standards.

Despite these rejections, the technique persists in investigations. Law enforcement often uses it to generate investigative leads. This creates a confirmation bias that diverts attention from reliable physical evidence. The University of Michigan’s Ford School notes that this practice poses an ongoing risk to impartial legal outcomes.

You need to understand how this junk science enters your case. It often appears disguised as lay witness testimony. Prosecutors may guide juries through the audio without calling an expert. This workaround bypasses strict expert testimony requirements. You must be prepared to object under Federal Rule of Evidence 701 and 702.

Consider the case of Riley Spitler. His murder conviction was overturned because this analysis should not have been admitted. This outcome highlights the danger of relying on unvalidated psychological profiling. When you review the evidence, look for these specific flags in the transcript.

Identify where the police report cites "guilt indicators." Cross-reference these claims with the actual audio. If the report mentions "awkward phrases," listen to the context. Is the caller simply confused? Or is there genuine evasive behavior?

This review process requires precise timestamping. You need to isolate specific moments in the audio to challenge the narrative. Tools like 911 call transcript analysis help you create a clear, timestamped record of these interactions.

Similarly, reviewing related communications is essential. If the investigation relied on post-arrest statements, you must compare those to the initial emergency call. Use jail call transcription to ensure every word is captured accurately.

The goal is to expose the gap between scientific validity and investigative narrative. You are not just listening to a phone call. You are dismantling a flawed forensic theory.

Here is what to watch for in the audio:

• Innocuous phrases flagged as guilt indicators, such as "Huh?" or "I don't know."
• Absence of an immediate plea for help, which is often misinterpreted as deception.
• Polite language or awkward phrasing that the COPS Scale categorizes as suspicious.
• Discrepancies between the police report's interpretation and the actual caller's intent.

Remember that the Illinois Supreme Court has agreed to revisit convictions based on this method. The institutional backlash is growing. Your role is to hold the prosecution to a higher standard. Demand empirical validity before any jury hears these claims.

Ground your arguments in the specific words spoken. Do not accept generalizations about "guilt markers." Force the state to prove the scientific basis for their interpretation. This is how you protect your client from the junk pile of fake scientific theories.

Legal Challenges and Workarounds

The technique keeps getting into courtrooms through the side door. When Daubert or Frye challenges succeed, prosecutors don't always drop the analysis; they repackage it as lay testimony and let a detective walk the jury through the call anyway.

That workaround is documented, not hypothetical. Because courts have repeatedly rejected 911 call analysis as expert testimony, Harpster's own training materials advise prosecutors to guide juries through the audio and transcript during lay testimony, folding the analysis into in-court statements to sidestep the expert-witness requirements. An Orange County lead detective wrote to Harpster that his consultation supplied "suggestions on how to introduce the 911 calls to the jury during trial," according to emails ProPublica obtained. That's the play to watch for.

Your objection lives in the FRE 701/702 distinction. Lay opinion must be rationally based on perception, helpful to the trier of fact, and not based on scientific, technical, or other specialized knowledge. A detective reciting COPS Scale indicators is applying a proprietary 38-indicator methodology built from a 100-call sample. That's specialized knowledge wearing a lay witness's badge.

When you object, have these ready on the record:

• FRE 701(c): the testimony rests on specialized knowledge, so it's expert testimony in disguise and must clear Daubert or Frye first.
• Daubert factors: the method fails testability, peer review, known error rate, standards controlling operation, and general acceptance. Harpster and Adams have never released their sample methodology or data.
• Foundation: the witness has no linguistics, psychology, or speech-analysis credential. Harpster himself has never testified as an expert in a criminal trial.
• Prejudice under FRE 403: the "indicators" invite the jury to read guilt into "Huh?" or "I need help," which swamps any probative value.
• A limiting instruction request if the court lets any of it in, plus a preservation colloquy for appeal.

Build the absurdity exhibit before the hearing. The record already contains a Michigan caller flagged as a suspect for saying "I need help" and a Washington mother whose first word, "hi," was treated as suspicious. When you have the call audio and a timestamped transcript in front of you, you can pin each "indicator" to the exact second it occurs and show the court what's actually being criminalized. That's the kind of line-by-line 911 call transcript analysis that turns an abstract reliability argument into something a judge can see.

One trap runs the other direction. In State v. Outland, the New Jersey Appellate Division held that once a defendant introduces his own 911 call, the State may impeach his credibility with sanitized priors under N.J.R.E. 806, even if he never testifies; the court called the recording "in effect, equivalent to making an assertion of innocence from the witness box." If you're weighing whether to play your client's call, brief that risk first and prepare a limiting-instruction request.

The same discipline applies to the rest of the call record. If your client's follow-up statements came over jail call transcription, you need those words timestamped and quotable too, because the State will use every syllable.

Strategic Risks of Introducing 911 Calls

The best evidence you can offer for a client is sometimes the client's own 911 call. It can establish panic, consistency, an immediate report. But in State v. Outland, the New Jersey Appellate Division showed exactly how that same recording can become the vehicle for a client's worst history to reach the jury.

Outland reported a theft of roughly $35,000 and played his 911 call at trial. The court held that introducing his own call was "in effect, equivalent to making an assertion of innocence from the witness box," even though Outland never testified. Because a 911 call is hearsay, the court applied N.J.R.E. 806, which allows a hearsay declarant's credibility to be attacked "by any evidence which would be admissible for those purposes if the declarant had testified as a witness." The State then introduced Outland's sanitized criminal history: four prior indictable offenses and state prison time. He was sentenced to a sixteen-year extended term as a persistent offender subject to the No Early Release Act.

The mechanism travels. Federal Rule of Evidence 806 is nearly identical, and the Outland court noted that other federal and state courts have used similar rules for the same purpose. If you play your client's call, you may have vouched for the declarant's credibility, and the State may respond on rebuttal with priors the jury would never otherwise hear.

Before any decision to play a client's call, weigh these questions:

• Does your client have prior convictions that would be admissible to impeach if he testified? If yes, the call may deliver them without testimony.
• Is the call truly exculpatory on its face, or does it merely sound sympathetic? The Outland court treated the act of playing it as an assertion of innocence.
• Can you request a limiting instruction? The Outland court found admission proper only when paired with one.
• Would a transcript excerpt or a witness's account of the call achieve the same goal without opening the 806 door?

This is where careful 911 call analysis pays for itself. Working from a timestamped, diarized transcript lets you test what the call actually proves before you commit to playing it, which is the same discipline we apply when reviewing 911 call transcript analysis requests and jail call transcription for defense teams: know what the audio says, and what it costs, before the jury hears either. The trial court in Outland admitted the recording as a present sense impression and excited utterance, reasoning the caller sounded "excited and unreflective." That same characterization is what gave the impeachment its force. Verify the rule against the primary opinion in your jurisdiction, but treat Outland as the cautionary template it is.

Practical Workflow for Defense Review

If the State's case leans on a 911 call, your first job is to get the call out of the summary and into your own hands. The raw audio, a verbatim transcript, and the responding officers' report rarely tell the same story, and the gaps between them are where defenses live.

Start with transcription. Request the audio in its native format and produce a word-level transcript with timestamps, because you'll need to quote the call precisely in a motion or cross-examination. Speaker diarization matters here: you need to know which words belong to the caller and which belong to the call-taker. Call-takers follow a structured protocol, asking for location, callback number, the nature of the emergency, and specific details, and they may deliver real-time instructions like CPR guidance while responders are en route, so separating the two voices prevents you from attributing dispatcher language to your client.

Next, map the call against the police report. Note what the caller said, when they said it, and what the report claims happened. Auto-flagging tools can help here: in Evidence Copilot, every new file is compared against earlier evidence in the case, so a mismatch between a statement and call audio surfaces as a review signal before you have to ask. Whatever you use, the task is the same: build a side-by-side of call content, report content, and bodycam or interrogation footage.

Then check the transcript against the analysis the State intends to use. If a detective is applying 911 call analysis, understand what that means. The technique, built on the proprietary "COPS" Scale and its 38 guilt indicators, has appeared in more than 100 cases across 26 states, yet researchers who tried to corroborate it failed, and Harpster and Adams have never released their call sample, methodology, or data. The flags are often absurd on their face: callers have been treated as suspects for saying "I need help," and a Washington mother's first word, "hi," was deemed suspicious before her baby's death was reclassified as a homicide.

Your transcript review should capture:

• Exact wording of any phrase the State may characterize as a "guilt indicator," with timestamps
• Statements in the call that contradict the police report's account of timing, sequence, or detail
• Call-taker questions that may have shaped the caller's answers, useful against suggestion arguments
• Any place the detective's written analysis paraphrases the call inaccurately

Finally, anticipate the courtroom strategy. Because courts have repeatedly rejected call analysis under Daubert and Frye, the documented workaround is lay testimony that "guides juries through both the audio and transcript," smuggling expert-style opinion past FRE 701/702 limits. A precise transcript lets you quote the call back, line by line, and object when the witness's "observations" become conclusions.

Thorough 911 call transcript analysis is tedious work, and many teams pair it with jail call transcription so every recorded statement in the case sits in one citable workspace. The goal throughout: every quote you use in a motion or cross-exam traces to the exact second of audio, and nothing in your file rests on the State's paraphrase.