When another scientist wrote that Linda Bartoshuk, PhD, had obtained effects that were not only wrong, but wrong in the wrong direction, Bartoshuk got mad. Then, she got motivated to prove her accuser mistaken. In doing so, she unearthed what she maintains is a serious yet virtually unrecognized methodological error that cuts across a broad swath of experimental psychology.

The problem that Bartoshuk identified arises from the fact that people cannot share one another's sensory experiences. Individuals learn to apply intensity adjectives such as "weak," "moderate" and "strong" to various domains, but the intensity indicated by those adjectives depends on the range of their experiences. To be accurate in drawing comparisons across individuals or groups, adjective-labeled scales must have universal meaning. But they don't.

"I don't know why this didn't occur to me years ago, and I don't know why it didn't occur to somebody else years ago," Bartoshuk said, speaking at APA's 2000 Annual Convention in Washington, D.C., Aug. 4­8. "This is a dreadful mistake."

Bartoshuk, a psychologist at the Yale University School of Medicine, conducts research on genetic variation in people's ability to taste a particular bitter chemical, called 6-n-propylthiouracil, or PROP, and how that variation shapes health, directly and indirectly.

Psychophysicists have known since the 1930s that although most people can taste PROP and related substances, others cannot. A third group of people, dubbed supertasters, can taste PROP in spades. For them, even a minute amount is almost intolerable. Bartoshuk and others who study taste perception have made a number of discoveries about the ability to taste PROP. They've learned that there are both gender and race differences in taste perception. Women--and particularly Asian women--are the most likely to be supertasters. For many white men, in contrast, a test dose of PROP is likely to elicit only a blank stare.

Research also has revealed that the ability to taste PROP is linked to a sensitivity to salty, spicy and sweet tastes, as well as liking for many fruits and vegetables that contain bitter compounds. This constellation of taste patterns affects the foods people choose, Bartoshuk said, and probably their health as well.

Bartoshuk continues to conduct research on a number of aspects of taste perception. But these days, she's turned increasing attention to a measurement problem that, she believes, has implications far beyond her own area of expertise.

An artifact nobody noticed

Bartoshuk's work unexpectedly led her to the realization that generations of researchers have consistently mismeasured taste sensation, and probably many other psychological constructs as well.

Other researchers had published critiques of her work, saying that her findings on the differences between nontasters, tasters and supertasters were wrong, that her effects were not in line with many others' results.

Then, one day last winter, she and a student, Derek Snyder, were poring over data that compared nontasters and supertasters, plotting PROP-tasting ability against salt-tasting ability. Usually, Bartoshuk's team uses a technique, called magnitude matching, that links taste ratings to ratings of sound or other sensations, skirting the problem of systematic variability in how people use rating scales. In this instance, though, they had used an adjective-labeled scale, which did not.

Unexplainably, their results showed that supertasters rated salt to be less salty than did nontasters. That couldn't be right, Bartoshuk knew, because supertasters have more taste buds than do nontasters.

"Both of us looked at it and the light bulb went off," she remembered.

Supertasters live in a much more intense taste world than do nontasters, Bartoshuk explained. That means that an adjective like "strong" applied to taste reflects a much stronger taste to a supertaster than it does to a nontaster.

"When we treat taste data as if 'strong' means the same to all, we in effect squeeze down the supertaster responses," Bartoshuk said. For substances like salt, which are only slightly stronger to supertasters than to nontasters, supertaster responses are squeezed down so far that it looks like supertasters taste salt as less salty than do nontasters.

"We started digging through our old papers and found more examples, including in published papers," Bartoshuk said. "All of the sudden, I realized the magnitude of this. I thought, 'This is incredible--we've stumbled across an artifact that nobody noticed.' I keep thinking that somebody somewhere has noticed it, and I just don't know about that paper or that literature. But what I do know is that if anyone has done it, it hasn't penetrated anywhere, because this mistake is being made all over the place."

The scaling error that Bartoshuk and Snyder observed is easy to overlook because it usually leads to null results. In the case of taste, it happened that the error produced a reversal of effects, making the problem more obvious.

So far, to Bartoshuk's surprise, her cautions have met with only lukewarm interest.

"When I mention it, people will say, 'Yeah, I've worried about that,' but it's as if the whole field has embraced the adjective-labeled scaling technique, so nobody has pursued their worries."

That's a mistake, Bartoshuk believes, because it's likely that the problem that has plagued taste research has broader ramifications. The same error, she said, can occur with any adjective-labeled scales--including category, visual analogue and Likert scales--that implicitly assume that the adjectives reflect the same intensities to all. She speculated that research on a wide range of topics, such as jealousy, depression, pain, addiction and eating disorders, may also be tainted.

"Think about the situations in which you're most likely to want to make across-group comparisons," Bartoshuk said. "It's probably when you suspect there's a difference between the groups. If there is, the chances are that the groups are using different scales [in making ratings], so the problem may be much worse than we thought."

Rosetta Stone

Bartoshuk believes there are at least two solutions to the scaling problem. One, familiar to psychophysicists, is magnitude matching. The other, called a Labeled Magnitude Scale (LMS), was developed by Yale psychologist Barry Green, PhD, in 1993 and was recently modified by Green and Bartoshuk.

The LMS circumvents the problem that besieges traditional adjective scales by not only providing an absolute lower bound, at which there is no sensation, but also an absolute upper bound, labeled "strongest imaginable sensation of any kind." Bartoshuk and Green believe that anchoring the scale in this way puts different people's responses on the same metric, creating a "universal ruler."

Unlike researchers in some other fields, taste researchers possess what Bartoshuk calls a Rosetta Stone for evaluating intensity scales: the density of mushroom-shaped structures on the tongue, called fungiform papillae, each of which contains conduits to several taste buds. The greater the density of fungiform papillae, the greater is PROP's perceived taste intensity.

Using fungiform papillae as a gauge, Bartoshuk's group has found that, unlike traditional adjective-labeled scales, both magnitude matching and Green's LMS scale accurately reflect disparities between supertasters, tasters and nontasters. That's good news, not only because it verifies Bartoshuk's suspicions about adjective-labeled scaling techniques, but also because it provides a solution for researchers in other areas of psychology, who may not have a Rosetta Stone.

Now, Bartoshuk hopes, others will take action.