Public Safety Communication Research and Best Practices

Start with the decision the reader needs to make.

A useful warning helps someone decide whether it applies to them and what they should do. A notice that says only “police activity nearby” can attract attention while leaving its audience unsure of the affected place, timing or next action. The responsible authority supplies the incident facts and appropriate instruction; this guide examines how that information is communicated.

FEMA training identifies source, hazard, location, timing and protective behavior among the content a warning needs. The original diagram below translates those components into an editorial review aid. It is not a message template approved for every hazard or a substitute for an agency’s warning policy.

Source: Alert and Warning Message Content ↗

Use familiar places and concrete instructions.

NIST’s imminent-threat guidance recommends understandable source, hazard, area, timing and protective-action information. Its language guidance favors familiar descriptions and reduced jargon. Here, the practical question is whether a reader can identify the actual boundary and action without knowing the department’s internal abbreviations.

Give a reader the same context they would have on a phone, rather than explaining the scenario first. Ask them to point out the affected area and say the requested action in their own words. A map can reinforce a boundary, but a reader should not have to infer an instruction from a blue shape alone.

Source: Alerting under Imminent Threat: Guidance on alerts issued by outdoor siren and short message alerting systems ↗

Keep accessibility and language in the authoring process.

FEMA’s IS-247.C summary recommends plain language, equivalent audio and text, an explanation for images or maps, and attention to the population’s primary languages. It also recommends checking automated translations with a speaker of the language. These are review tasks, not properties guaranteed by adding a translation button.

Check the complete instruction with the formats people will actually use. Listen to spoken output, enlarge the text, read the message without its image and have the planned language versions checked. Confirm that a translated boundary, time or action still means the same thing. Include residents who use the relevant formats in an appropriately planned exercise.

Source: IS-247.C Course Summary: Integrated Public Alert and Warning System (IPAWS) for Alert Originators ↗

Character count is a constraint, not a result.

A 2024 study by Carlson and colleagues randomized 481 U.S. adults to mock warning messages in five 90/360-character pairs. Longer complete messages did not increase intended compliance overall in that experiment; both versions contained core warning elements. That finding does not establish that length never matters.

The exposure was hypothetical and the outcome was self-reported intention, not observed behavior in an emergency. The practical implication is to test a message’s content and interpretation within the limits of its channel. Do not remove a critical location or action simply to make a message shorter, and do not treat a longer message as proof it works better.

Source: Do 360-Character Wireless Emergency Alert (WEA) Messages Work Better than 90-Character Messages? Testing the Risk Communication Consensus ↗

Choose channels by reach, task and authority.

CISA’s alerting guidance recommends multiple channels for reach and corroboration and warns about irrelevant messages consuming public attention. A dedicated app, a local opt-in system, a public webpage and an authorized warning broadcast can have complementary roles. Their recipient models and operating permissions need to be checked separately.

For a specific event, identify who is affected, the urgency, the approved warning authority and the available pathways. Keep the core instruction consistent while adapting the format. Preserve alternatives for people without a particular app, account, connection or compatible device. A police activity archive answers a historical question; it should not be presented as a current warning.

Source: Ten Keys to Improving Emergency Alerts, Warnings & Notifications ↗

State uncertainty without abandoning the next step.

CDC’s crisis-communication guidance discusses the difficulty of processing complex information under stress and recommends clear communication about what is known and unknown. That supports a useful editorial separation: confirmed facts, the instruction currently in force and what is still being checked.

Do not fill a gap with an explanation that has not been verified. Put an update time and responsible source beside the current instruction. The fictional correction example below changes a boundary without pretending that every reader saw the first notice. The issuer remains responsible for release decisions and for protecting information that cannot be made public.

Source: CERC: Psychology of a Crisis ↗

Original editorial examples

Make the missing information visible.

Plan the message after the first alert.

A USGS publication record describes a framework for post-alert messaging developed for ShakeAlert earthquake early warning. We reviewed its abstract, not the full study, and do not treat its findings as an all-hazard communication standard. Its narrower context still prompts a useful planning question: what information will people need after the initial attention signal?

For a local communication exercise, identify who posts the next update, where it appears and how an ended or corrected condition is distinguished from an older message. A request for camera footage also needs a clear response route and expectations. It should not lead a resident to assume that a submission is monitored immediately.

Source: Developing post-alert messaging for ShakeAlert, the earthquake early warning system for the West Coast of the United States of America ↗

Measure receipt, understanding and action separately.

A system’s acceptance of a send, a notification appearing on a device, a reader understanding the instruction, a stated intention and an observed action are different observations. None alone establishes a reduced injury rate, a solved case or a faster emergency response. The measurement diagram below shows why these stages should not be collapsed into one success percentage.

Record the denominator, missing observations and conditions for each proposed measure. A small comprehension exercise can identify confusing wording. It cannot estimate the behavior of everyone in a jurisdiction or establish a product’s effectiveness during a real incident.

Use a repeatable, voluntary comprehension exercise.

NIST’s building-communication guidance distinguishes language, readability and response testing and describes paraphrase and usability methods. Our proposed exercise adapts that testing approach outside its building-specific context. Use clearly labeled fictional material and informed participants; do not issue a real public alert to test copy.

Write the intended interpretation before the session. Show the message in its planned format without coaching, ask open-ended questions, record the participant’s words and then revise what is unclear. Include relevant languages and access needs in the plan. Keep identifiable participant or case information out of this website and use an agency-approved records process for any real evaluation.

Source: General Guidance on Emergency Communication Strategies for Buildings, 2nd Edition ↗

A blank protocol you can adapt.

1. Record the fictional scenario, intended interpretation, message version, format, language, planned access needs and session conditions.
2. Show the message without a presenter’s explanation. Participation should be voluntary and appropriate to the exercise.
3. Ask the open-ended questions below. Record the reader’s words before explaining or revising the draft.
4. Compare the observations with the intended interpretation, identify ambiguity, revise and repeat under recorded conditions.

Keep a blank cell blank when there is no observation. Do not convert one person’s interpretation or a convenience sample into a jurisdiction-wide compliance percentage. This protocol has not been run for Lucas or Sentinel; no results are reported.

Apply the findings to Lucas and Sentinel with the same limits.

Lucas is the resident app; Sentinel is the department workspace. Their intended connection is a department’s message, a resident’s relevant observation and an accountable agency review. An evaluation can examine whether that connection is understandable and workable in the current release.

The material here is guidance and original editorial application, not published Lucas pilot findings. No participating department, delivery benchmark, security certification or operational outcome is established by this review. When verified pilot evidence exists, it should be published with a protocol, version, sample, missing observations and limitations rather than added as an unattributed success claim.

How this resource was developed

Bounded narrative evidence review, not a systematic review or meta-analysis. Consulted content/source-claims.json first and reused three unchanged guidance records with their existing verification dates and retrieval limits. Searched for primary official guidance and original research covering message content, dissemination, accessible language, uncertainty, follow-up and comprehension testing. Selected eight sources with different practical roles. Reviewed relevant full-text sections when retrievable and recorded abstract-only access explicitly. Kept operational guidance, evidence syntheses, original experiments and this publisher's proposed applications distinct. No numerical findings were pooled.

New research was reviewed November 1, 2026. Reused guidance retains its original October 26 or 27 review date in the source notes. Selection focused on message content, audience access, uncertainty, channel fit and testing. No quantitative results were pooled.

• This is a selective U.S.-focused reading list. It does not establish that every relevant study was located or that every community will respond alike.
• Building and earthquake guidance has a narrower setting than all public safety communication. Applications outside those settings are editorial proposals.
• No comprehension sessions, live warnings, delivery tests, product comparisons or outcome studies were performed for this resource.
• Authority, local policy, hazard expertise and current incident facts determine actual warning thresholds, release decisions, protective actions and approved channels.