Psychometric Paradigm

The psychometric paradigm refers to a research approach used in explaining how laypeople (nonexperts) perceive various hazards. Results of these studies suggest that laypeople use qualitative information, such as perceptions of dreadfulness and newness, and not simply statistical information, such as probability, in their intuitive risk assessments. The “cognitive map” that resulted from studies utilizing the psychometric paradigm became the icon of risk perception research. Engineers, public policy people, and others who deal with public reactions to risks often rely on the psychometric paradigm for a better understanding of laypeople's risk perception. This understanding continues to underlie many approaches to risk communication practice.

In the 1970s, proponents of nuclear power, who emphasized the small probability of an accident, became puzzled by the reactions of laypeople, some of whom opposed nuclear power because they emphasized potentially catastrophic and long-lasting environmental effects in their thinking. The question of “How safe is safe enough?” became an important research topic. In his seminal work that influenced the development of the new paradigm, engineer Chauncey Starr used revealed preferences in answering this question. He assumed that society, through trial and error, had achieved an optimal balance between the risks and benefits associated with a technology or activity. According to this approach, a new technology should be accepted if it has a similar risk-benefit ratio as well-accepted technologies.

Starr's work has been criticized for a number of reasons: Accepted does not mean acceptable; there are problems in the measurement of benefits and risks, as well as concepts such as “voluntary”; and it does not take into account that the people who reap the benefits of a technology may be different from the people who bear the risks. Starr's approach was nevertheless original in that it took social preferences into account. Baruch Fischhoff, Paul Slovic, Sarah Lichtenstein, Stephen Read, and Barbara Combs were not in agreement with Starr, and they proposed an alternative research paradigm that became known as the psychometric paradigm. Instead of revealed preferences, this approach relies on expressed preferences.

In research following the psychometric paradigm, participants are asked to evaluate a set of hazards according to several different rating scales. Participants assess, for example, severity of consequences (how likely is it that the consequences will be fatal) and newness (are the risks novel or familiar) for each hazard. In most studies, between 8 and 15 rating scales are used, and participants assess a set of hazards, typically between 15 and 30. In the initial studies, a very heterogeneous set of hazards was used, ranging from smoking and caffeine consumption to nuclear power and possession of handguns. The psychometric paradigm has also been used to examine laypeople's risk perception of more homogeneous sets of hazards, such as food hazards, railroad hazards, or various nanotechnology applications.

In most of these studies, the data structure of rating scales versus hazards versus participants was simplified. Mean values were calculated across participants, and the rating scales by hazards matrix was analyzed using principal components analysis (PCA, a form of factor analysis). Hazards were treated as a unit of analysis, and the number of rating scales was reduced to two or three principal components (PCs). Most studies report the two PCs “dread risk” and “unknown risk.” The scales that have been termed lack of control, fatal consequences, and dread potential tend to be highly correlated with the PC “dread risk.” The scales delay of effect, perceived newness, and perceived scientific knowledge tend to be highly correlated [Page 601]with the PC “unknown risk.” Factor scores are computed for the hazards, and these values provide the basis for the “cognitive map.” This map depicts the values for the PCs “dread risk” and “unknown risk” for each hazard. For hazards located in the high “dread risk” and high “unknown risk” quadrant (for example, nuclear power), even minor incidents may have a large signal potential. A high signal potential means that the public perceives risk-related incidents as warning signals. High media attention and higher costs due to new regulations or change in consumer behavior may be the consequences.

...