Am I having a heart attack? My self-diagnosis concludes I must be, because the symptoms match what I found on Google. However, a more objective reflection that also takes into account the risk of having a particular condition, might lead someone else, like a doctor, to suggest I have the hiccups.

This somewhat exaggerated example, highlights the findings of a new study, published recently in the Journal of Consumer Research, that propose using the internet to self-diagnose can be unwise because we tend to focus on symptoms rather than the risk of having the illness.

Authors Dengfeng Yan and Jaideep Sengupta, from Hong Kong University of Science and Technology, write in their introduction:

"In today's wired world, self-diagnosis via internet search is very common. Such symptom-matching exercises may lead consumers to overestimate the likelihood of getting a serious disease because they focus on their symptoms while ignoring the very low likelihood that their symptoms are related to any serious illness."

Psychological Distance

For their study, the researchers looked at two pieces of information that influence people's decision as to whether they have a disease or not: the base rate (the rate of the disease in the general population), and the case information (eg the description of the symptoms).

They had a theory that how much reliance a person places on base rate and case information depends on the "psychological distance" to them of the person who is ill (self being the closest of all, strangers being very distant).

Their theory was that when assessing themselves (psychologically very close), people would place more importance on case information, and the influence of base rate would be weak. But when assessing others, especially strangers, then the influence of symptoms would be weak and base rate would be strong.

Self-Positivity and Self-Negativity

Conversely, if these theories are right, then they should also work the other way around: self-positivity (underestimating risk to self) would occur when base rate is high, but case information doesn't provide a good symptom match. And self-negativity (over-estimating risk to self) would occur when base rate is low, and case information does provide a good symptom match.

An example of self-negativity would be assessing a set of symptoms as indigestion when considering them happening to a stranger, and perceiving them as heart attack when happening to oneself.

An example of self-positivity would be underestimating the risk of becoming infected with HIV ("it won't happen to me").

Experiments Show Psychological Distance Matters

The researchers examined these self-positivity and self-negativity biase in a series of experiments with hundreds of undergraduates.

They explored many disease scenarios including flu, hepatitis C, breast cancer and osteoporosis. In each scenario, the participants had information on base risk (the prevalence in the general population), and case risk (a person's profile of symptoms and behavior). In some experiments the participants were asked to consider themselves as having the symptoms, in others they were asked to consider strangers as having the symptoms.

When they analyzed the results, the researchers found their theories were confirmed: psychological distance matters.

The less a participant knew the person they were being asked to consider, the more they relied on base risk, whereas the closer they were to the subject, the more they relied on case risk such as symptom matching.

Yan told NBC News:

"We found the effect to be quite strong, as evidenced by the fact that we replicated our findings using different manipulations of psychological distance, and across five different types of health risks."

See a Real Doctor for an Objective Opinion

The researchers said this study and others like it are important because, if consumers are more likely to misdiagnose themselves, then this could lead to them taking up treatments and buying drugs that are not appropriate, which has a wider impact on public health.

The easiest answer, they conclude is to get rid of the bias by seeing a real doctor instead of "Dr Google".

Real doctors will take the prevalence of the disease into account, because they are viewing the patient from a distance, they say.

"This will prevent symptoms from exerting a disproportionate influence on the diagnosis," they conclude.

Written by Catharine Paddock PhD