PrepTalks: Dr. Dennis Mileti “Modernizing Public Warning Messaging”


I’m going to speak today on modernizing public
warning messaging. I started studying how the public responds
to disaster warnings 45 years ago, and I’m still doing it today. I decided that the purpose of my talk would
be, if I had to address the men and women in our nation who will issue public alerts and
warnings in the future, what would I tell them? And that’s what I’m going to say today. I have five things to say. First, focus on alerts and warnings for imminent
rapid onset events with a small amount of time between when an event is detected to
impact. Just to give you an example – if they just
detected Sarin gas in the Washington subway, and you’re on the escalator going down and
you get a wireless emergency alert on the cellphone, focus on that message or other
things comparable to it. If we’re talking about warnings that in three
days a Tsunami is going to impact one of our cities, we don’t need to prepare for warnings. People will figure it out. When public response time is short and warning
delays have large public health and safety consequences, the wireless emergency alert
technology of the future in our nation can provide the largest public good to remove
delays from the warning system. Delays are anything that prolongs the time
between when a threat is detected and public protective action is initiated. So just to be clear, when you detect a threat,
it doesn’t mean anybody’s are going to protect themselves. That comes way down a sequence of stuff you
have to be a social psychologist to get excited about. Delay types, so there are three major points
of delay in warning systems in our country the first is warning issuance delay. The amount of time those who would press the
alert button delay pressing it for a variety of different reasons, and in imminent events,
that’s not a good thing. Second is audience dissemination delay. Many people think because if you issue a warning
over whatever technology is available being used that it will be received by the people
at risk. The data suggest it’s not that straightforward. And third public protective action initiation
delay. Once people get the warning message that means
what delays them from stepping over the threshold of their home and getting in the station wagon,
sorry, SUV and peeling rubber out of town Delays are additive and this diagram illustrates
that a hazard notification is received, then we have issuance delay, a warning is issued,
and then we have dissemination delay, a first warning is received, and then we have protective
action delay. And if you were on the escalator getting your
first Alert about Sarin gas at the bottom, you could see there’s a problem here if the
delays are too long. Planning can reduce issuance delay. Let me say that again, emergency planning
works. Not planning doesn’t work quite as well. Gee who told me that? Oh, it was FEMA, no it was President, Eisenhower,
I think said that actually. Warning plans and procedures is the essence
of planning with threat conditions, warning triggers, public protective actions catalogued. Here’s an example this is a planning matrix
for dams and levees Why is it for dams and levees? I was asked to prepare one for the Army Corps
of Engineers, so I did it for dams and levees, but it pretty much looks like the emergency
planning matrix for emergency planning at our nation’s operating nuclear power plants
because it’s virtually identical to it. And anyone who knows red, go, six, five, four,
rev, one, that’s going way back in time, will recognize this, there are physical observations
that lead to different classifications of different threat levels that lead to different
levels, in this case flood threat, that lead to different possible public protective actions. Planning suggests this can be thought through
before an event happens. I’m sad to say our nation doesn’t participate
or require or even help local emergency managers plan for giving warnings, unless it’s in the
emergency planning zone for a nuclear power plant, and I’m out to change that. Planning also includes primary factors like
having written plans, rules and procedures, threat classes. Secondary factors like identifying responsibilities
for who has the decision responsibility to make a decision, who they need to get approval
from. Tertiary factors like threat verification
procedures, inter-agency contact ad nauseum. For my fourth thing I want to say is disseminate
alert and warning messages wisely. Every dissemination channel for disseminating
warnings has pros and cons. They include factors about the audience who
would receive the message that either help them get it or keep them from receiving it,
technological factors about the ability of different technologies to reach the audience. For example there’s a matrix I don’t expect
you to see; just trust me that some of us in the country have gone through every warning
dissemination mechanism we have at our disposal and have judged it in terms of the percentage
of at-risk audience that it can reach whether it’s a quick dissemination or slow dissemination. For example wireless emergency alerts are
a very quick route notification. When cops drive up and down the street with
the blue lights doing this with a bull speaker, it is slow notification, very effective, but
slow. At any rate that potpourri exists in different
levels in different cities and counties in our country and selecting a good mix of both
old fashioned technologies and wise technologies is probably very good. We’re just beginning to collect data, in fact
I own it. I just finished a survey with a colleague
John Sorensen of evacuation at the Oroville Dam in February of
2017, and we measured these things up one side and down the other, and I can tell you
in that one event what was the most effective dissemination technology, how long each different
technology took to reach different segments of the audience, or percentage of the population
evacuated, what percentage didn’t. I haven’t shared the data with the core yet,
so I’ll just give you a snippet of a sneak preview. Wireless and modern emergency warning technology
is fast. Mama is faster. What that means is the quickest way people
got their first alert was because their mother or a friend or a relative got in touch with
them. Well, that’s not really a surprise your mother
has more interest in you than other devices. Here’s some wire diffusion data from the Boulder
flood. That’s how the WEA message reached the population
in that flood a few years ago. As you can see it got up to about 18% of the
population and eventually peaked at about 30% and the initial speed with which it got
out was very, very rapid. We were just beginning to collect data on
wireless emergency alerts. Here’s historical diffusion data what that
suggests to you is somebody took a statistically representative sample of people who received
warnings in a particular American disaster event and asked, “What was your first alert? What time did you get it?” and then plotted
graphs of the different disasters that are on this slide are there varied diffusion rates. In this slide you can see informal notification,
which is under the red line, which is the cumulative sum. The blue line is at the top. That’s Mama. Diversity reduces diffusion delay. If there were one thing I wish I could tattoo
inside your foreheads, it’s this: multiple dissemination channels for public disaster
warnings yield quicker and more comprehensive audience penetration. There never has been; there never will be
a silver bullet for distributing warnings. One technology is insufficient because you
need multiple technologies to reach different subpopulations in an at-risk audience. So I’d recommend doing two or three modern
technologies and two or three old-fashioned technologies. We use SMS tested methods like radio route
notifications, special ways to reach unique subpopulations, and if you’re into wireless
emergency alerts, particularly, IPAWS’s WEA system, and I’m a IPAWS and WEA systems zealot,
I know that’s the warning system of the future in this country. I’m a hundred percent behind it, and I want
to make it as good as it possibly can be. It has to be one flower in the bouquet of
dissemination channels that are used. And it has its advantages, as well as, disadvantages. And five: issue messages that reduced public
action delay. Now here’s the social psychology part. I’m grinning because now I’m feeling like
I’m back at the University of Colorado in a lecture hall. This is the problem you’re up against when
it comes to giving warnings. It’s human beings. The math is people immediately take protective
actions when they receive a warning message. Here’s my response to that, a social psychologist
that studied this phenomenon for 45 years, if you believe that you’re nuts. Let me say it in real simple language. When all the forest animals are running away
from the flames people who get the warning delay taking protected action and instead
waste time searching the net, watching television and talking with neighbors trying to decide
what, if anything, to do about the fire. It’s called millet. It’s fundamental to how human beings invent
new realities in their minds. A warning tells people they need to go from
perceiving that they’re safe to perceiving that they might die. That doesn’t happen quickly. It’s not automatic people need to, let me
just say it in a way you’ll get, clues before they come to that change of mind. And we see data, for example in the Joplin
tornado, where people got a five-minute warning of an impending tornado and were found dead
in rubble with their hands on their cell phones as they were milling trying to interact with
other people. If they didn’t have a cell phone, and they
were in their backyard, they’d be talking over the backyard fence. Milling is basic human nature. Its basic human nature to search for more
information, confirm warnings with others, check out what others are doing, and personalize
threat perceptions. It takes time. Human beings don’t respond to warnings. When people get warnings, what they do their
natural hardwires say, “Let’s go hang out with other people and talk about it.” Now the prime objectives therefore in an American
warning system of the future, the one I’m hoping to bring forth, is to minimize issuing
alerts and warning messages in America that motivate milling and delay, and instead to
maximize issuing alert and warning messages that reduce delay and are actionable. That is motivate public protective action. Two things I need to say, the second is the
slide, the first thing is the Department of Homeland Security funded a lot of social psychologists
and sociologists and communication people to study warnings a few years ago, and there
were many breakthroughs learned. One of the things that was learned that that
90-character WEA messages accomplished nothing. What people do is get that message and say,
“what the blank was that?” However a message that provides them with
the information that they would go in search of can actually minimize delay and maximize
an appropriate, more timely response. Hallelujah! We discovered what we need to put in a warning
to communicate to the person at risk to motivate them to take a protective action. I think that’s the essence of what a warning
system should do and The FCC believed us, and they said we’re going to take the 90 character
limit off WEA messages, and we’re going to give you three hundred and sixty characters
for a WEA message because that’s the size of a screen. Let me tell you what, such a message would
look like. For your information, here’s some historical
data. These are curves about people getting the
warning and initiating a protective action. The access is when they got the warning. Do you see any straight lines going up? That would suggest people don’t immediately
respond to warnings. If you look at how they bend and lazily move
over across time to initiate a protective action, there it is. I’m rubbing your nose in it. I mean to be rubbing your nose in it. People don’t immediately respond to disaster
warnings. We know, by the way, all the correlates that
dozens of people have discovered the relative weight of the correlates of what affects public
response to warnings. I could send you these, and they’re like chocolate
chip cookies I could serve on a plate. But what impacts protective action initiation
behavior in Americans the most? It’s the message contents It’s the message
contents. It’s the message contents. Any questions about what the most important
factor is? It’s the message contents
It’s what the message says, and how it says it. And there also are contextual factors like
whether there are personalization visualizations. It’s really hard to issue warnings for flooding
on a sunny day for example. And it’s about message repetition. People need to get warnings many times, because
human beings, even me, have thick skulls. Any of you encounter anybody like that? You know what I’m talking about. So what does the message need to be? It needs to be specific. For example if you are in between the river
and First Street, move north of Main Street that tells people who’s at risk. If you don’t tell them who’s at risk and who
isn’t, people prefer believing they’re not at risk. Never say evacuate if you’re near the river
that’ll mean different things to different people. Some will leave, some won’t. You need to be clear. A wave of water 20 feet high, moving faster
than a person can run is clear to the person being warned. Not 10,000 cubic foot per second flow moving
at a 20 foot per second speed. People don’t know what that means. Fluvial geomorphologist do. Dam operators do. They love these words. It communicates not out of the public. Here are the things that need to be in a warning
message based on the new DHS research: Source hazard location, personalization consequences,
protective action, protective action time, how actions reduce consequences, and the expiration
time. What is he nuts? He wants a query mold that in a warning. Yes. Here’s the warning three hundred and forty
nine characters that will fit in the next generation WEA message. Elm County Sheriff: the source needs to come
first. People want to know who they’re listening
to. What do we do in the country now? We put the source last
The source has to come first. Elm County Sheriff: floodwaters are approaching
Wood City and will hit two blocks on both sides of Elm Creek from highway 110 to Maple
Road, people outside will be washed downstream. The water will be above rooftops. Move two blocks plus from the creek now and
be there no later than 6:00 p.m. to avoid the flood. This message expires at 11 p.m. May 15 2017. Three hundred and forty nine characters inside
of our three hundred and sixty character WEA long message. Granted, I picked a simple one. I didn’t want to make it rough on me. So what conclusion can we encourage comprehensive
local planning for alerts and warnings in the thousands of communities where alert originators
reside and practice their craft? We have a really big country with lots of
diversity in our emergency response people and can we upgrade practices by alert and
warning message providers? Can we educate them? Can we design a course for them to give them
this information? Can we give guidance to them? Can we write a guidebook for them? Can we encourage them to engage in training
drills and exercises? Well Shazam that all sounds like it’s right
up FEMA’s alley, so I’d like to thank you very much and I want to point out you can’t
see the second clock. I ended precisely on time. Thank you very much
[Applause]

Leave a Reply

Your email address will not be published. Required fields are marked *