.

The Psychology of Warnings

Marc Green


Warning signs and labels are everywhere. Many factors are conspiring to fill the world with more injunctions against our desired behavior. Daily life is increasing in complexity while technology continually invents more powerful machines, drugs and chemicals. Moreover, the cost of injury and insurance as well as accompanying litigation is spiraling upward rapidly. As a result, failures of warning compliance cause costly social and economic impact.

Unfortunately, warnings often fail to change people's behavior. Either the warning goes unnoticed, or, as increasingly happens, the warning is seen but ignored. For many years, designers focused their concern on sensory aspects of warnings: color, shape, location, pictures vs. text, size and so on. However, recent research suggests that effective warning design depends as much on the contents of the viewer's head as on the contents of the warning's message.

People who see a warning must decide whether or not to comply. However, "warning viewers" (people for whom the warning is intended) are not blanks slates but rather start with a mental model containing three components. First, the viewer has general knowledge about the world and how it works. Second, s/he has a set of beliefs and expectations based on experience with the same or similar environment, product or technology. Lastly, the viewer enters the situation with a goal and strategy for achieving that goal. The goal can be specific ("I want to arrive at my destination as soon as possible") or more diffuse ("I want to feel good about myself"). Understanding what the viewer "brings to the table" is critical for creating effective warnings.

The Cost-Benefit Analysis

If a warning tells the viewer to refrain from behavior which will enable him/her to easily and directly achieve a goal, then the viewer makes a cost-benefit analysis. In some cases, the viewer might lose the goal altogether ("No Smoking" signs), or exert more effort ("Detour" signs). Without knowing the viewer's goal (fast travel, least effort, keeping shoes clean, etc) the complete cost cannot be precisely specified.

The benefits are practical, injury avoidance, and psychological. A driver deciding whether to comply with a speed limit sign might calculate the costs of arriving later. However, the benefit of compliance is the sense of safety and lowered anxiety ("I won't be as likely to get in an accident," "I won't have to spend time and effort looking for a radar trap," etc.) There may be other psychological benefits, such as the sense of being "a good citizen" or "a team player."

The size of the benefits depends on another psychological factor, perception of danger, which has two components, hazard and risk. Hazard is the "badness" of the possible outcome. For example, being in a major auto accident is a greater hazard than getting a speeding ticket. Risk is the probability of the outcome. Being killed by a falling meteorite is certainly a severe hazard, but it is also a very small danger because the risk probability is tiny. (Note: much of the research literature confuses the terms risk and danger, using the term risk to cover both. However, the difference is important for understanding viewer perception.)

The viewer then must take both hazard and risk into account when making the cost-benefit analysis. If the person believes that there is great danger, then s/he will see a larger benefit in compliance. Conversely, perception of small danger means low benefit and compliance will decrease.

Finally, decision-making factors affect compliance. Two people could make the same cost-benefit analysis, but one might ignore the warning and the other doesn't. People accept different danger levels, have different attitudes about their ability to control danger and are differentially affected by social and cultural norms.

In sum, people who view a warnings use a mental model to perform a cost-benefit analysis. The three main process components are 1) cost of compliance, 2) perception of danger level and 3) personal and social and cultural decision-making factors.

All elements of the cost-benefit analysis are psychological. It is perceived risk, perceived hazard, perceived control and perceived norms that matter, not actual ones. This may be obvious, but is worth saying explicitly since many of the people who design warnings are engineers. They are not so likely to consider mental models or psychological concepts. That's one reason that much of the warning research is so concerned with physical factors such as warning color, shape, etc and less with goals and motivation.

What's Ahead

Another page Warnings and Warning Labels discusses warnings from an information processing viewpoint, this article takes a higher-level look at how warnings affect behavior. The remainder of this article discusses each of the three components. In the next section, I discuss factors affecting cost of compliance while the subsequent section examines perceived danger levels. The final section outlines the role of person/social variables.

First, some disclaimers. The article will not discuss many important issues, such as warning conspicuity and comprehension. Obviously, people won't comply with a warning that they don't see and are less likely to comply with a warning that they don't fully understand. These issues are discussed in another article .  Instead, this article focuses on compliance failures for warnings which are seen and understood but ignored.

Factors Affecting Compliance

1. Cost of Compliance

Many studies have found that warning signs are more likely to be ineffective if the cost of compliance is high. Reducing compliance costs is a very effective way to increase safety, but it is necessary to understand where the viewer's costs arise. Recall that the viewer has a goal in mind when using a device or navigating an environment. The costs relate to the ability to achieve the goal as quickly and as easily as possible.

Compliance with a warning could cost viewer's goal attainment in several ways. First, the viewer's route to the goal is completely blocked. For example, the rate of diving accidents for in-ground pools is very high and "No Diving" signs are frequently ignored. One reason is that the viewer is blocked from the goal of diving. With in-ground pools, there is usually a deep-end for diving, so at worst, the viewer need only walk around to the other end of the pool to achieve the goal. If a sign says "No smoking" and there is no designated smoking area, then the viewer must decide whether to ignore the warning or to give up the smoking goal.

If the goal is important, then there will be little compliance. In this case, the warning strategy is to provide a substitute method for achieving the goal or a substitute goal. If someone with a peanut allergy wants to eat a candy bar with the warning that it contains nuts, then suggest another sweet that doesn't.

Second, the viewer may have to spend mental effort to find an alternate strategy for reaching the goal. A "No trespassing" sign requires the viewer to plan a new route. In this case, detour signs or maps of alternate routes will increase compliance by reducing mental time and effort required to find a new path. A designated smoking area will increase compliance with a "No Smoking" warning, but the area should be easily located by maps or signs.

Third, the new strategy may increase the effort needed to achieve the goal. Reducing this extra effort is undoubtedly a powerful method for gaining compliance. For example, one study found that viewers were far more likely to comply with a warning to use safety gloves when the gloves were attached to the product. The cost (finding a proper pair of gloves) was drastically reduced, so the cost-benefit analysis swung toward compliance. No smoking compliance will be reduced if the designated area is far away or means passing through locked or secure doors and will be even smaller if it is located outside in the snow.

Lastly, merely reading and interpreting the warning increases the cost. Too much information may create an overload, causing the viewer to decide that it is easier just to ignore the warning. In some cases, warnings are buried deep within instructions or other information that would require extensive reading.

Another growing trend, application of multiple warnings, also contributes to the overload. I've seen the drug Levaquin dispensed in a bottle that is only 2 1/2 inches high and 1 1/8 inches in diameter, yet it had five different warning labels in addition to the normal drug information. The viewer must turn the bottle completely around to read all the warnings, which are irregularly affixed at different angles, with different colors and written in different fonts. It is a major task to read all the warnings and difficult to be sure that one hasn't been missed any amid the clutter.

Small print, by itself, can significantly increase effort making reading more difficult and by causing eyestrain. Each of the bottle's warning labels is the same size, but one contains only five words while another fits 16 words into the same space. Moreover, the reader may not have his/her glasses or may need to fish reading glasses out of a pocket. This is especially annoying to baby boomers, who have suddenly found themselves forced to take glasses on and off for near and far vision. They often find it easier to simply ignore the warning label.

In Canada, where many products must have French as well as English warnings, the problem is worse than in the United States. The necessity for warnings in two languages forces smaller print, increases clutter and generally makes the finding and reading of warnings a more difficult task. I'm sure that I am not the only person who ever saw the French warning on one side of a large package but didn't want to bother turning the package around to see the English side. Use of pictures in place of bilingual text is one solution, but its effectiveness is highly questionable .

2. Danger Perception

Perceived danger is the other side of the cost-benefit balance sheet: the greater the perceived risk and hazard, the greater the likelihood of compliance. Studies have identified several factors which influence the level of perceived risk.

Familiarization

Viewer history with the product or environment strongly affects danger perception. The greater the experience with no negative outcome, the lower the level of perceived danger. This is the "cry wolf" or "familiarization" effect, where people quit paying attention to uninformative input. The effect can be both specific to a particular product (a single drug) or can generalize to all similar products (same type of drugs) or to the warnings source ("those government do-gooders are at it again!" or "The company is just covering itself.")

One of the ironies of warnings is that the more experienced and skilled the viewer, the stronger the familiarization effect and the more likely that the warning will be ignored. For example, diving team members are the most likely people to ignore "no diving" signs. Studies have found similar results across a wide variety of products, including, automobiles, computers, power tools and other consumer devices. Not only are such people likely to less likely to heed warnings, they are less likely to even notice them - familiarization also reduces warning conspicuity and salience.

Viewers can also acquire familiarization vicariously. You observe other people speeding without incident, so you might as well, too. Conversely, if a viewer has personal, firsthand knowledge of people being injured in a situation, then compliance increases. For example, I only started wearing safety glasses around power tools after meeting someone who had lost an eye from kickback on a table saw.

Another form of familiarization is based on the viewer's mental model of the device/environment's operation. If the viewer believes that s/he understands how the device works, there is a greater chance that s/he will ignore the warning. For example, TV sets have a warning against opening the back because of shock danger. If the viewer believes that s/he understands electronics, especially TV electronics, then there may be the belief that the danger is easily avoided and that there is no need to comply. Here, the familiarization is not with the specific product itself but rather with the principles on which it operates.

Dilution

The increasing trend toward multiple warnings also has the subtle effect of "diluting" the strength of more important ones. For example, one warning might say, "Do Not Swallow or Take Internally" (important) while another on the same product might say "Loses Potency After April 2002" (not so important). It is not clear why dilution occurs. Perhaps viewers mentally average the warnings together somehow or maybe they are less likely to remember the one important warnings amidst all the lesser ones. Lastly, I have already noted that multiple warnings may turn viewers off by increased cost of reading.

Warning Appearance

Marshall McLuhan's famous aphorism, "the medium is the message," applies to warning labels. The physical appearance of the warning may inadvertently communicate hazard severity. Most people have unconsciously learned the general rule that signs and signals grow in size and vividness with their importance, presumably so that they will be more readily seen. Viewers will then likely interpret warnings that are small, faint, or located peripherally as signaling lower risk.

However, size is relative. Viewers are likely to understand that a small warning on a small package is the result of space limitation. If there is a small warning on a wall or other large surface, then it will be construed as a weak signal for risk.

Similarly, crude and sloppy signs also appear to signal lower risk. The viewer will believe that if the designers didn't bother putting much effort into the warning, then they must not think it significant. (There are exceptions. New US roadwork signs are rendered in a child's scrawl and say "Please slow down, my daddy/mommy works here." They are very effective in drawing attention, since the crudity contrasts with standard road signs. However, familiarization will eventually dampen their effectiveness).

3. Decision Making

Risk Taking

When the viewer ignores a warning, it could be because s/he viewed the cost of compliance as very high or because s/he underestimated the danger. However, studies show that some people engage in risky actions across a wide variety of circumstances. They likely have a high tolerance for risk rather than a tendency to underestimate danger in specific circumstances. Moreover, these risk takers were less likely to comply with warnings. In fact, the prospect of danger may decrease their compliance because the viewer's goal might involve courting danger.

There is a large amount of research, especially related to alcohol, tobacco and HIV warnings, that has investigated individual differences in danger tolerance. Typical results show that risk taking is lower for females, for young children and for seniors. Beyond these simple demographic factors, however, studies have failed to isolate significant variables.

Control Perception and "Risk Hedging"

In some cases, viewers might acknowledge that significant danger exists but attempt to control risk by behaving in a "safe" manner. In a sense, they are trying to perform a "risk hedging" or compromise, where their goal or goal path is modified as a tradeoff for more safety. For example, many swimming pool accidents have occurred when a diver ignored a "No Diving" sign, although s/he understood that the water was not deep. The diver believed that a shallow dive angle would minimize and control the danger of striking the pool bottom. (Untrained people are very poor at controlling dive angle, so serious injury often occurs.) In other cases, a viewer might ignore tobacco warnings by smoking only low tar cigarettes (trading taste for safety), or a worker might use reading glasses in his/her pocket rather than take the effort of finding the safety goggles required by a warning (trading time and exertion for safety.). Partial compliance is also increased when the viewer believes him/herself to have a good mental model of device operation. Risk hedging is also an example of a more more general mental short-cut, "satisficing," where a good-enough solution is accepted to avoid spending the effort to find an optimal solution.

More specific warnings reduce likelihood of partial compliance. The Levaquin bottle warns "May Cause Drowsiness or Dizziness" Most such warnings are more specific and instruct the viewer to avoid driving or using heavy machinery. The Levaquin bottle's vagueness allows mental wriggle-room to rationalize lack of complete compliance. Unfortunately, the Canadian requirement for bilingual warnings on many products means that specific, and therefore longer, warnings will result in small print which viewers may not bother reading. (Yes, longer warnings are less likely to be read. Creating specific, but brief warnings, is a fine art.)

Social Factors

We live in a social context, so other people affect individual notions of norms, standards and acceptable behavior. It is not surprising, that degree of viewer behavior may be affected by whether other people are complying with the warning. For example, almost everyone has had the experience of walking with a piece of trash in hand and then seeing a "No Littering" sign. The likelihood of compliance is high if there is no litter on the ground. If the ground in strewn with litter, however, then the typical thought is that "what's one more piece?" or "if every ones ignores the sign, why shouldn't I?" Research confirms the role of social factors. For example, one study found that people would be more likely to reduce volume on earphones if role models complied first while another in a chemistry lab showed that a model increased the use of safety gloves. Lastly, as already noted, seeing other people ignore warnings can also reduce perception of danger.

Such results naturally lead to the notion that compliance might be increased by modeling - showing examples of compliance by people in-person or on video. There is some positive evidence, but it is unclear whether the effect is large or long lasting. This is doubtless a complicated area because the results probably depend on the viewer's notion of peer group, authority and other sociological factors.

Who Gains By Compliance?

Currently, airline pilots instruct passengers to keep their seatbelts fastened even in calm air. They explain that it will prevent injury in case of unexpected turbulence. The probability of any one passenger being injured is very low because the event has a very small probability. However, the airline flies millions of passengers, so from their viewpoint, the chance of someone being injured from unexpected turbulence is reasonably high. In the airline's cost-benefit scheme, warning compliance has a large benefit and low cost, since they are not the ones who experience the seatbelt discomfort and restriction.

The viewer performs the cost-benefit analysis from his/her own viewpoint, so the calculations may differ from those of the airline, government or public policy. To gain compliance, the viewer must be convinced that there is a real personal benefit. Governments and industry have several methods for convincing the viewer that it is in his/her best interest to comply: overstate the risk ("using cell phones increases car accident risk by a factor of five"), institute penalties for noncompliance (speeding tickets), or appeal to lower economic cost ("it will save you tax money"), social good ("fewer people will be hurt"), or good citizenship. The effectiveness of these strategies varies.

Conclusion

There is much more to creating effective warnings than choosing the right color, size, location and font or even the right message. It is imperative to understand what the viewer is trying to achieve and how the warning affects attainment of his/her goal. Next, the designer must consider the cost-benefit calculations that the viewer is likely to perform. Finally, the designer must consider the viewer's experience and knowledge and how s/he fits into the social world.

It might seem that attempting to understand each individual psyche would be a daunting task. Fortunately, humans have many more similarities than differences. People in similar situations, driving cars, using power tools or taking medication, likely have similar goals and similar experiences, although there may be some differences between groups such as novices and experts. In most areas of design, it is possible to create usable artifacts, such as computer interfaces, that serve large groups well. However, there is much work spent in identifying typical users, defining their mental model, doing task analysis (defining goals), and testing with proposed designs. Those who would design effective warnings should use a similar methodology.

1A version of the article will be published in Occupational Health and Safety Canada.