Public Release: 

People are fairly good at judging health risks--but only when they stick to personal experience

Psychologists find that two different thought processes can explain how we gauge our risk of disease

American Psychological Association

WASHINGTON -- Psychologists have gained insight into how people judge their personal health risks. The findings suggest that people aren't horribly off the mark as long as they do not rely on media reports and stick to what's happened to people they know. The study appears in the current issue of JEP (Journal of Experimental Psychology): Learning, Memory and Cognition, which is published by the American Psychological Association (APA).

The findings challenge the assumption, says Ralph Hertwig, PhD, of the University of Basel in Switzerland, that people make huge blunders when inferring the likelihood of, say, dying of a heart attack or in a car accident. He says, "People can arrive at relatively accurate estimates as long as they rely on their personal experiences of the frequencies of such events ... by thinking of how many of their relatives, friends and acquaintances died from these causes."

He continues, "However, when they start sampling from the virtual world as created by the mass media, they are more likely to arrive at distorted estimates of likelihood." For instance, if people sample from the virtual world, they might readily conclude that many more people die due to more rare but dramatic causes such as mad-cow disease or airplane crashes, than due to more typical causes such as asthma.

The authors are concerned that as "factors such as overpopulation, poverty, and global climate change pave the way for new health risks, it becomes even more important to better understand how the public perceives and judges risks."

Hertwig and other cognitive psychologists first extended theories of how people think about event frequencies to health risks in particular. Then, at the Max Planck Institute for Human Development in Berlin, they tested these models by asking participants to assess health risks in various ways.

The researchers presented one group of 45 participants with pairs of causes of death and asked them to choose the cause that took more lives per year. They presented two other groups, of 30 and 35 participants, with pairs of types of cancer and pairs of infectious diseases, respectively, and asked them to choose the disease with the higher incidence rate (the number of new cases appearing in a population in a given time). The Federal Statistical Office of Germany and the Robert Koch Institute provided the disease data.

On average, participants were 71.2 % correct in the causes-of-death set; average accuracy was slightly lower for the cancer set (68.2 %) and markedly higher for the infection set (80.6 %).

The scores not only demonstrated reasonably good accuracy, but also made sense in terms of two of the four proposed models: "availability by recall" and "regressed frequency mechanisms."

In a second study, the psychologists directly tested how well these two models predicted risk assessments for each OF 276 disease pairs and each of 80 participants, based on what a post-experimental test showed about their knowledge. Hertwig was surprised that "two quite different models, based on different underlying assumptions, explain the data equally well."

According to the availability by recall model, people assess the odds of an event by the frequency of experienced episodes within their social network. They might, for instance, figure their odds of having a heart attack by thinking about the people they know who have had heart attacks.

The other thought process, a "regressed-frequency mechanism," assumes that people base their health risks on automatically encoded frequency information arising from a goulash of various exposures -- including obituaries and news reports, doctors' warnings, public-awareness campaigns and so forth. Because it's hard to reliably process all that information, however, people's estimates shift toward the average value in a category, a statistical phenomenon called "regression toward the mean." As a result, small frequencies (such as dying from vitamin overdose) are overestimated and large frequencies (such as dying from rectal cancer) are underestimated.

The researchers ruled out the use of a supposed "fluency mechanism," in which media coverage shapes people's risk perceptions (individuals would have felt at high risk of getting West Nile virus last summer), and another mechanism according to which people have only a sense of the frequencies of high-level categories of risks such as natural hazards.

The authors speculate that people switch their risk-assessment strategies depending on available information from memory, not using the same mechanism for each single inference. They write, "For instance, if a person cannot retrieve any episode within his or her social circle, he or she may attempt to rely on a sense of fluency or frequency."


Article: "Judgments of Risk Frequencies: Tests of Possible Cognitive Mechanisms;" Ralph Hertwig, PhD, University of Basel; and Thorsten Pachur and Stephanie Kurzenhäuser, Max Planck Institute for Human Development; Journal of Experimental Psychology: Learning, Memory and Cognition; Vol. 31, No. 4.

(Full text of the article is available from the APA Public Affairs Office and at

Ralph Hertwig can be reached by E-mail at or by phone at (+41) 61 2670611. The American Psychological Association (APA), in Washington, DC, is the largest scientific and professional organization representing psychology in the United States and is the world's largest association of psychologists. APA's membership includes more than 150,000 researchers, educators, clinicians, consultants and students. Through its divisions in 53 subfields of psychology and affiliations with 60 state, territorial and Canadian provincial associations, APA works to advance psychology as a science, as a profession and as a means of promoting human welfare.

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.