▲ 127 points
back
144 comments
This is reminiscent of an experience we had at ClearHealth with managed hospital aesthetics. Not a scientific study in any sense, but as part of rebranding a hospital group, the color scheme was changed from a neutral green to a much brighter orange. The change coincided with a small but meaningful increase in mortality for procedures done on the associated floor. Enough that we reverted to the older aesthetic. The mortality change reverted as well. Chalked it up to coincidence. This process then repeated at two other separate campuses. I can't rule out this was some sort of reverse "Hawthorne Effect" where our communications about the change were the source of the problem but it bothered me enough to mostly put an end to our appetite for large scale aesthetic redesigns.
Color has a measurable effect on humans. There's an extensive body of research on the psychological effects of color; every designer worth their salt knows this. The fields of branding and marketing use color purposefully every day as well.
Green has a soothing, calming effect on people because it's most common in our natural surroundings. For example, TV shows use "green rooms" filled with plants (and sometimes even painted green) to calm down guests before they go on stage for this very reason.
Orange, on the other hand, excites people and can elevate their heartbeat. It's a cross between red (hot-blooded, passion, rage) and yellow (energy, happiness, attention). You could say it encourages your emotions to interfere with rational thought.
So you can keep your appetite for large-scale redesigns! Just add a step to incorporate color psychology.
A factory here, a few decades ago spend 150 k to select a color scheme for maximum productivity. No one believed in it but one investor insisted because he used the service before and the consultant had a lot of numbers to show.
50 cm above the floor the wall had 6 washed out grayish colored bars 20-25 cm width with green and orange slightly brighter. I didn't notice a productivity gain but it did allow excellent peripheral vision.
>but it bothered me enough to mostly put an end to our appetite for large scale aesthetic redesigns.
But how do we know a different color might not save lives? Only half joking.
Perhaps wall color makes the doctor thinks you're sicker than you actually are.
The sicker you look, the stronger medicine they want to give you. Always made me wonder how many people would have recovered naturally. And adding a strong medicine with potential dangerous side effects into the mix, must be killing a substantial number of people that would have recovered on their own without it.
Like how many people put on the ventilators would have been better off without them, for example.
Bright orange is a warning color in nature, neutral green suggests an hospitable environment.
It could be that the warning color added subconscious stress to doctors and patients alike.
In one car I've driven, whenever you turn it on, a pop-up appears that says (paraphrasing): "Taking your eyes off the road to interact with this screen can cause an accident. Press OK to continue."
Irony...
Car manufacturers get bonus points when the screen obscures the backup camera feed.
My car's crash safety rating was significantly dinged because lower trim lines don't have low beams that follow the steering.
I wish they'd deduct a star for "there is a circumstance in which the car is on, and the main console is displaying a non-time-critical-popup, or refusing to accept input".
On a related note, why do car manufacturers make you come to a complete stop on the freeway so the person in the passenger seat can type an address into the navigation system?
Yeah, my car shows similar on startup.
"Our infotainment system is a danger for anyone to use because of the terrible design. Please press "I agree" to confirm it is all your fault and responsibility somehow." Mercifully but even more ridiculous, it just goes away without being pressed when you start driving.
Thats a huge PITA on newer cars. On both, mine (Skoda) and my wife's (Nissan) I have to accept their data protection policy every time I start the car. The dialogue usually pops up a few seconds after I start driving.
Keyboard error or not present, press F1 to resume.
I love google maps's insistence on trying to convince you to use driving mode when you start navigating by car. Of which literally every time I've done so I was you know starting to pull out of my driveway, airport parking lot, etc. They seem to have solved this recently by not giving you an option and just forcing you into driving mode.
My Chrysler does this. Thankfully the screen goes away on its own after some amount of time (I've never bothered to time it, but it's definitely seconds rather than minutes) if you don't click the Okay button.
Oh mazda...
Do you _have_ to press ok or does it go away after a while?
Things I think should be disallowed:
- traffic messages that are displayed by rotating through three or more screens; two is fine if necessary, as you can ingest all of it in short order; but to keep staring at it waiting for the third screen is just way too dangerous
- digital advertising billboards, for the same reason; these are even worse because the ads are generally poorly designed with tiny fonts that are hard to read at a glance
I have a strong preference for an analog speedometer for the same reasons. In an instant your eyes can discern whether you are accelerating or decelerating. On a digital display, with a measly 1 MPH resolution, you have to keep looking down over and over to determine if you are maintaining your speed.
Aircraft with digital displays usually use a scrolling column of numbers analogous to a tape measure for critical information like airspeed and altitude. This provides information at a glance with digital precision and a more compact display because the full range doesn't need to be shown all the time.
The digital speedometer on my vehicle is surrounded by an "analog" speedometer. I'm not sure why anyone would need it to determine whether they are accelerating or decelerating, but it's there.
That reminds me of the Citroën CX speedometer that was like a horizontal ribbon (probably a cylinder though):
Hmm, idk. I don't like their modeling approach.
Their dependent variable is the number of crashes per hour, scaled by the "population average" for segments of the same length. I guess this is fine so long as you look at the same areas. They give two tests to try and control. One is checking upstream crashes but I kind of suspect the signage is positioned where you're supposed to be extra careful due to increase risk, so that may not be relevant. The other is looking at the same areas before and after. This does seem like a fair argument.
They also include specific effects for exact date-hour combos, but I feel like this is a bad approach as you would want to ideally capture higher level weekday + hour seasonality traits; not specific date hours.
They say on the multi variate: > In particular, we show that clustering by segment-year-month reduces the standard error in half, clustering by just geography produces slightly smaller standard errors, controlling for rain more flexibly does not affect our results, not controlling at all for rain doubles our estimated treatment effect, not controlling for holidays increases our estimate slightly, and dropping hours immediately before and after campaign weeks (i.e., hours outside of campaign weeks that sometimes display fatality messages, see fig. S6) further increases our estimate
Which implies a lot of it may be random sample bias of the test set all being on worse weather days, even if they try to control for it here.
Regardless. I think it's a plausible analysis. The effect size is just 1-2% though.
How do they control for situations where the sign is displaying a traffic alert that preempts the informational display?
Presumably, if the sign is displaying "congestion ahead" warnings, then traffic has often already slowed to a crawl, which could lead to a few percentage point decrease in crashes.
Also, unless the signs were placed poorly, they're likely in high-congestion areas, and frequently toggling between normal and congested display modes.
This should be replicated outside of Texas. When I lived there my coworkers and I used to get a good laugh at the traffic signs like "do not drive into water". Who could be so dumb?
Actually of course texans are no dumber or smarter than anyone else, but there were very many opportunities for a laugh, parallel to the "known to the state of California to be harmful" signs. Still, it would be interesting to see if there are cultural differences before drawing too broad a conclusion.
People drive into water all the time. They see water across the road, and using past experience of the road conclude that the road surface is only a few inches below the water. Better wet than late so they decide to drive through it. Except unknown to them, the water washed out the road and they fall into a trench deep enough to wreck their car or worse.
It's so easy to find videos of this online. People should know better but hubris is a hell of a drug. Anyway, I doubt the signs help much.
I have had a long standing suspicion that the proposed attentional mechanism is the same reason why nagging someone almost never actually helps get something done, because it shifts their attention away from the actual activity and toward a negative social interaction.
Far better to restructure the environment so that the person can get the work done. In the case of traffic deaths we have really good data that says that you should change the structure of streets and roads so that people instinctively drive safely.
1] never nag some one to do something, while also expecting them to abandon doing something that was previously nagged about.
2] when some one shifts focus to do something, and is nagged to do exactly that thing, the chance of it being done, with quality, or at all, will approach zero.
Sounds sort of like the near-irresistible impulse to achieve the high score in the "Your Speed Is" sign game.
I was bummed when they capped the reading at 70 on the signs in the I-5 mountain passes in Oregon, but that's Oregon for you, I guess. At least they let you pump your own gas now.
Near my parents house they painted dots on a dangerous road with signs "keep 2 dots apart". The dots didn't make much a different until someone painted pacman about to eat one of the dots on the road. Now the DOT repaints that pacman and thee average following distance is longer.
Reminds me of the "Werther Effect," a phenomenon where the publicity of suicides leads to a spike in suicide rates, particularly when reported in a sensationalist manner. - If the same principle applies to traffic crashes, then it's vital that to consider how and if this needs to be presented to the driver.
Edit: Changed the last paragraph to apply it to traffic.
I think that displaying this information on those DMS is a bad idea unless you give it some context.
In Texas they throw up the message similar to "1669 deaths this year on Texas roads" as in the article. Drivers who read the sign have no way of knowing whether this is lower than, higher than, or about the same as the number of deaths for that date in previous years. As the year goes by the number gets larger but there is still no context.
A better way to convey this information if you are going to do it at all would be use a reward/penalty message format that is readily understandable at highway speeds.
The message should be simple. For example - "Texas Traffic Fatalities lower this year than last. Good Job! Drive Friendly" or "Texas Traffic Fatalities higher this year than last. Drive Friendly"
That tells anyone reading exactly what they need to know and the message highlighting an improvement contains a simple reward message. When drivers become accustomed to seeing the message in this format they will immediately recognize that fatalities must be higher because the reward is missing from the message. That could trigger an effort to drive more safely.
Personally I hate the DMS signs and billboards. In a world where I made the rules there would be no billboards at all as the immortal Ogden Nash said:
I think that I shall never see a billboard lovely as a tree.
Perhaps, unless the billboards fall, I'll never see a tree at all. -- Ogden Nash (after Joyce Kilmer's poem about trees)
I would not allow any form of dynamic advertising along roads or highways. That means that if billboards are allowed there would be no LED billboards flashing a new message periodically and distracting drivers. All allowed billboards would be totally static and unless they advertised critical services for travelers they would be unlit at night. That means that no lighted billboards unless they are advertising travel-related infrastructure such as restaurants, clean restroom facilities available for travelers, auto towing, maintenance, or repair shops, etc. All advertised infrastructure can only be lighted during the hours when it is open and available for use by travelers. If the business is closed then the lights must be off.
There are already too many distractions for drivers with dash-mounted tablets, cell phone holders, navigation screens, and drivers do not need flashing shit along the road to distract them from the important task of maintaining a safe following distance from other traffic and monitoring other drivers on the road while scanning the highway for hazards along their route like traffic light status, objects in the roadway, unsecured loads, etc.
The rest of your comment, I agree with but:
> That tells anyone reading exactly what they need to know and the message highlighting an improvement contains a simple reward message
Do they need to know it though? Is someone going to see that, and adjust what they're doing because numbers are trending up? It seems unlikely.
If you have to show something, how about a standard caution sign that says "elevated accident zone"? It conveys everything that /might/ be helpful, is highly relevant to the specific area the driver is in, and does not require critical thinking (numbers up? numbers down? how does this affect me? Is it for this spot, or all of Texas?). Caution signs are also taught as part of driver's ed as warning of road hazards - deer crossing, falling rocks, etc.
Ohio has a good implementation of safety messages in my opinion: http://zerodeaths.ohio.gov/create-your-message.aspx
Whoever they have picking messages is doing a really good job keeping things consistently corny.
Around here, the signs flip between time and distance to other highways or cities and a safety message. If there is an accident closing lanes, that's displayed instead of the safety message. The signs seem okay at night. Nothing like blinding billboards.
Texans know that they need more "masculine" slogans for road signs: https://en.wikipedia.org/wiki/Don%27t_Mess_with_Texas
Highly recommend this "If Books Could Kill" episode on the concept of "nudges": https://podcasts.apple.com/us/podcast/nudge-part-1-a-simple-... (critical of the concept but I think quite fair and well-researched)
I don't even think it's a cognitive overload thing, it's a "oh shit, you're right, this is hard, skirrrrt" sort of thing.
Like telling someone they will do badly on a test and comparing them with a control group that isn't told that. Reverse placebo effect.
Haven't owned a car for years, but when I did, those signs just never... hit.. enough to warrant a significant behavior change. Obviously it is a lot, but that's just not how it feels while on the road. Can't imagine I'm the only one.
That is the problem - it is really hard to figure out how anyone would make a change. either you know how to drive safe and you do, or you don't and think you do. I've never seen anyone back off to the recommended 2 second following distance for example. (that is something I know about and do - but I have no idea what other safety advice now exists that I don't even know of)
TLDR the strategy - "Let's add attention grabbing signs on poorly designed roads that are know to be dangerous instead of improving the road themselves" does not work.
Having signs on the side showing complex information (how is that metric defined, exactly?) unrelated to the task at hand (driving safely) won’t help much.
Showing relevant information with a clear impact on the driver has proven to work effectively: have color-coded feedback telling you that your insurance premium just went up when you drive too fast, erratically, too close to the car in front don’t look ahead, or turn without checking behind the mirrors and behind you.
how about you don't "nudge" i.e. attempt mass mind control in the first place. it's the same thing as dark patterns. involuntary experimentation on people, driven by analytics. if lying to people helps reduce car crashes by 1% and dog bites by 5%, let's roll out our lying operation nationwide
Do you think that there are no "nudges" if nobody comes along and tries to "nudge"? The situation as-is is not some kind of default that's nudge-free. Whenever someone makes a decision, it probably results in some kind of "nudge".
Accepting this reality, and working with it, is completely different from "mass mind control". Any time a city decides which streets to repair, it's a nudge. Any time a city decides which streets to renovate at the same time, it's another nudge. Any time a city decides not to repair streets it's a nudge. How do you get out of this "nudge-less"?
Anyone have any idea what the deal is with California's new:
The 55 MPH Speed Limit Will Be Enforced
If I get a speeding ticket, can I just argue that the sign's use of the future tense clearly indicates the lowered speed limit doesn't apply yet?
The word "will" isn't strictly future tense. It also conveys purpose or intent. e.g. "If you suggest a silly reading of a traffic sign, I will explain why you're being ridiculous."
I’m pretty sure we do driver harrowing on purpose, here in Portugal. This is not to excuse rash driving, which is rampant, but rather to indicate the presence of sadistic individuals in our roadway infrastructure safety and design areas.
Adding any sort of complexity / unnecessary information will do that. Not quite following why this requires a study.
Editorialized title on this submission. The year of this article is 2022, which isn’t reflected in the HN title.
Actual title: “Can behavioral interventions be too salient? Evidence from traffic safety messages”