Person calling 911

(© MargJohnsonVA - stock.adobe.com)

Research Shows How Investigators Misjudge Callers’ Voices, Emotions

In A Nutshell

  • Listeners focus first on emotion: 77–85% mentioned it, and more emotion and urgency for the victim predicted less suspicion.
  • Perceived information or impression management, plus cognitive load, tended to raise suspicion across studies.
  • Police, not laypeople, rated male callers as more suspicious than female callers when scripts were identical.
  • Popular “911 linguistic analysis” trainings lack solid replication; treating demeanor as evidence risks wrongful suspicion.

ITHACA, N.Y. — Calling 911 should be straightforward. Someone witnesses a crime, picks up the phone, and reports what happened. But new research reveals a troubling reality: the very people trying to help can find themselves targeted as suspects, simply because they don’t sound the “right” way during one of the most stressful moments of their lives.

A pre-print study involving nearly 2,000 participants across four separate experiments found that both laypeople and police officers judge 911 callers based on behavioral expectations that research suggests are not reliable indicators of guilt. When callers violate these expectations by seeming too calm, too emotional, or failing to provide information in the “correct” manner, they trigger suspicion that can snowball into full-blown criminal investigations.

“Citizens engaging in the well-intentioned act of calling 911 risk observers—including police—holding expectations for their behavior and targeting them as a suspect if they violate those expectations,” the researchers wrote.

Importantly, the results should not keep you from calling the police in an emergency situation. Among study limitations, the authors note that real 911 calls provided ecological validity, albeit with less control over variables, whereas simulated calls used in the study allowed for tighter control but potentially reduced realism.

The study, led by researchers from Cornell University and Arizona State University, analyzed real and simulated 911 calls to identify which behaviors sparked suspicion. Participants listened to calls reporting violent crimes and rated callers on measures including urgency, emotionality, cognitive load, and whether they seemed to be managing information or their impressions.

Calling the police: Dialing 911 on smartphone
Calling 911, unfortunately, carries risks for innocent bystanders. (Photo by gioele piccinini on Shutterstock)

The Emotion Paradox

Perhaps the most striking finding: between 77% and 85% of participants spontaneously mentioned the caller’s emotional display when asked for their first impressions. Yet humans experiencing trauma exhibit wildly different emotional responses, from hysteria to numbness, depending on factors including personal history, relationship to the victim, and neurological differences like autism.

Callers perceived as more urgent and emotional drew less suspicion. Those who seemed to be carefully managing information or trying to control how they came across raised red flags. Cognitive load, or appearing to struggle with clear thinking, also increased suspicion in some studies.

Police officers showed patterns similar to laypeople, with one notable exception: officers rated male callers as significantly more suspicious than female callers, even when both genders recited identical scripts describing the same situation.

These judgments matter because suspicion often precedes formal accusations. Once someone becomes a suspect, research shows that confirmation bias can create a chain reaction throughout an investigation, with each piece of evidence interpreted through a lens of assumed guilt.

When Behavior Becomes Evidence

The paper documents several cases where 911 call behavior led directly to wrongful accusations and convictions. Gary Gauger was labeled “flat” during his call about his parents’ deaths and later coerced into a false confession before being exonerated. Sandra Ortiz’s “calm tone” was used against her at trial, leading to a wrongful murder conviction. Thomas Perez falsely confessed to killing his father after a dispatcher felt “something was off” because he seemed to lack concern and was “rambling.” Perez’s father was found alive days later.

Amanda Knox was described as both overly emotional and insufficiently emotional at different moments following her roommate’s murder. Robert Roberson currently faces execution based partly on the perception that he didn’t show enough emotion after his daughter’s death. He is autistic and does not express emotion in expected ways, but this neurological difference was not initially recognized or considered.

Making matters worse, a controversial police training called “911 linguistic analysis” has spread through law enforcement agencies despite lacking scientific support. The course, developed by a retired deputy, claims to teach officers how to identify guilty callers based on behavioral indicators like opening a call by saying “hello” or being overly polite to the dispatcher.

Multiple studies have failed to replicate the training’s central findings. One analysis currently in press found the method “failed to differentiate between deceptive ‘false allegation’ and truthful ‘true report’ callers.” Yet officers continue taking the training and citing it in investigations.

In the case of Jessica Logan, a mother who slept through her alarm to give her sick infant medicine and woke to find him not breathing, the investigating detective used the training’s checklist to claim her 911 call proved involvement in her son’s death.

Police officer interrogation with witness or suspect
Witnesses’ emotional responses to officer can make them appear more suspicious. (Photo by Ground Picture on Shutterstock)

The Mistaken Assumption

The Center for Substance Abuse Treatment notes that witnesses to violence often experience just as much trauma as direct victims. Initial reactions can include intense emotions manifesting physically through shaking or rapid breathing, or the opposite: numbness, dissociation, and withdrawal.

Personal history with trauma, cultural background regarding emotional expression, relationship to the victim, and proximity to the event all influence how someone reacts. Autistic individuals and others who are neurodivergent may process and express emotion differently than neurotypical people. As the researchers note, there is no “one size fits all” reaction to a traumatic event.

Yet the study found that people hold firm expectations anyway. Participants rated callers who violated these expectations as more suspicious and indicated police should investigate them further.

Psychologists have long documented the fundamental attribution error: people’s tendency to attribute others’ behavior to internal characteristics while underestimating external situational factors. Someone struggling to provide a coherent narrative on a 911 call might be doing so because they’re in cognitive overload from witnessing violence, not because they’re lying and trying to keep their story straight.

Observers, however, tend to interpret these behaviors as intentional deception rather than trauma responses or personality differences.

The study’s fourth experiment tested 88 real 911 calls with a diverse participant group. Even across this wider range of calls and situations, the patterns held. Urgency for the victim and emotionality predicted less suspicion. Perceived information management and impression management predicted more. Cognitive load also increased suspicion, particularly for male callers.

What Comes Next

Lead author Samantha Bean and her colleagues argue their findings reveal the starting point of a dangerous process. Once suspicion takes hold, investigators may approach the case with assumptions that warp how they interpret evidence, conduct interviews, and pursue leads.

“It is critical to examine why someone is initially suspected because research and anecdotal case evidence suggest that, once someone is a suspect, a chain reaction of confirmation bias can lead to a wrongful conviction,” the researchers wrote.

Study authors stopped short of recommending specific policy changes but emphasized that law enforcement and the public need better education about trauma responses and the limitations of behavioral analysis. Officers who understand that trauma has no script may be less likely to target innocent witnesses. Dispatcher training that focuses on gathering information rather than assessing caller credibility could help separate the act of reporting from the question of involvement.

For now, anyone who calls 911 faces an uncomfortable reality: the way they sound during a crisis could determine whether they’re treated as a helpful witness or a potential suspect. Someone who spent their life learning to stay calm under pressure might be suspected of not caring enough. Someone whose trauma response includes emotional flooding might be dismissed as overdramatic. Someone who struggles to think clearly while witnessing violence might appear to be hiding something.

Until law enforcement agencies and the broader public absorb that lesson, Good Samaritans who pick up the phone to report a crime will continue rolling the dice on whether their particular trauma response happens to match what observers expect.


Paper Summary

Methodology

The research team conducted four studies involving participants who evaluated 911 calls. Study 1 recruited 943 participants through Amazon Mechanical Turk who listened to one of six real 911 calls. Study 2 included 624 participants from CloudResearch who heard simulated calls performed by professional voice actors using controlled scripts. Study 3 involved 298 law enforcement officers recruited through police department newsletters in a large metropolitan area, also using the simulated calls. Study 4 used 250 CloudResearch participants who each listened to eight calls from a pool of 88 real recordings obtained from police departments.

Participants heard calls reporting violent crimes including shootings. After listening, they completed open-ended responses describing their first impressions, then rated callers on scales measuring perceived urgency (for the victim and for themselves), emotionality, cognitive load, impression management, and information management. Participants also indicated how suspicious they found each caller and whether police should investigate the caller’s potential involvement. Trained coders analyzed the open-ended responses to identify which behaviors participants spontaneously mentioned. Statistical analyses examined relationships between perceived behaviors and suspicion ratings, controlling for other factors.

Results

Across all studies, emotionality emerged as the most salient factor, with 77% to 85% of participants spontaneously mentioning the caller’s emotional display. Callers perceived as more urgent (particularly regarding the victim) and more emotional received lower suspicion ratings. Callers who seemed to manage information or impressions drew higher suspicion. Cognitive load showed mixed results but trended toward increasing suspicion when present. A mini meta-analysis combining Studies 1 through 3 found medium to large bivariate correlations for several predictors, though the unique effects of individual predictors (controlling for others) were mostly small to medium. Urgency for the victim showed a large negative correlation with suspicion. Emotionality demonstrated a medium negative correlation. Information management and impression management both showed positive correlations with suspicion. Police officers generally showed patterns similar to laypeople but with one key difference: officers rated male callers as significantly more suspicious than female callers even when scripts were identical. Study 4’s conceptual replication with 88 diverse real calls confirmed that urgency for the victim, emotionality, information management, and impression management reliably predicted suspicion across varied situations and call types.

Limitations

The research design was correlational rather than experimental for most measures, meaning the team could not establish definitive causal relationships between specific behaviors and suspicion. Real 911 calls provided ecological validity but less control over variables, while simulated calls allowed tighter control but potentially reduced realism. Some behavioral predictors correlated substantially with each other, particularly urgency and emotionality, raising questions about whether participants perceived these as distinct qualities or as overlapping indicators of the same underlying state. Not all behavioral predictors remained significant when entered simultaneously in regression models, reflecting this potential overlap. The study focused on calls reporting violent crimes, so findings may not generalize to other emergency types. Participant samples were predominantly White, and the research did not examine how caller race might interact with behavioral perceptions. The study measured suspicion and expectation violation but did not track whether participants’ suspicions would translate to actual investigative decisions or case outcomes in real-world contexts.

Funding and Disclosures

Funding came from a National Science Foundation grant awarded to three of the authors (NSF Award #2146834), an Arizona State University Institute for Social Science Research Seed Grant, and a New College Scholarship, Research, and Creative Activities Seed Grant. The authors reported no conflicts of interest. Study materials, data, analysis code, and preregistrations are publicly available through the Open Science Framework.

Publication Details

Bean, S. R., Wulff, A. N., Lawrence, M. L., Reeder, I., Duran, N. D., Kassin, S. M., & Salerno, J. M. “From caller to suspect: Identifying behaviors that trigger suspicion in 911 calls,” pre-print published on PsyArXive, accepted for publication in Law and Human Behavior.

About StudyFinds Analysis

Called "brilliant," "fantastic," and "spot on" by scientists and researchers, our acclaimed StudyFinds Analysis articles are created using an exclusive AI-based model with complete human oversight by the StudyFinds Editorial Team. For these articles, we use an unparalleled LLM process across multiple systems to analyze entire journal papers, extract data, and create accurate, accessible content. Our writing and editing team proofreads and polishes each and every article before publishing. With recent studies showing that artificial intelligence can interpret scientific research as well as (or even better) than field experts and specialists, StudyFinds was among the earliest to adopt and test this technology before approving its widespread use on our site. We stand by our practice and continuously update our processes to ensure the very highest level of accuracy. Read our AI Policy (link below) for more information.

Our Editorial Process

StudyFinds publishes digestible, agenda-free, transparent research summaries that are intended to inform the reader as well as stir civil, educated debate. We do not agree nor disagree with any of the studies we post, rather, we encourage our readers to debate the veracity of the findings themselves. All articles published on StudyFinds are vetted by our editors prior to publication and include links back to the source or corresponding journal article, if possible.

Our Editorial Team

Steve Fink

Editor-in-Chief

John Anderer

Associate Editor

Leave a Reply