Three Approaches to Technical Risk Assessment

The Actuarial Approach

Characteristics

Risk is measured as expected value (i.e., arithmetic average) based upon previous occurrences of undesirable events. Undesirable events are defined as physical harm to humans or other ecosystems, wherein these events can be observed with sense data (e.g., excluding events such as subjective perceptions). The underlying concept of reality is positivist: that undesirable events are easily recognized, agree upon by all, and limited to what is observed.

Assumptions

This approach assumes that sufficient data exists to make meaningful predictions about future events and that the causal mechanism that underlies the occurrence of previous undesirable events will remain stable over the prediction period.

Strengths and Limitations

The actuarial approach provides quantification of undesirable events and an indication of what frequency to expect for future occurrences of these events. It assigns risk without prejudice because it does not provide an explanation of why undesirable events occur. Because it does not attempt to identify causal mechanism, however, it gives little guidance on how to prevent or predict future occurrences of undesirable events. Also, because the actuarial approach attempts to quantify hazard, it is subject to the types of observation and measurement errors described in the next sections on critiques of risk assessment.

Examples of Use

The assignment of automobile insurance rates for different segments of the population provides a good example of the actuarial approach to risk assessment. Younger drivers pay higher insurance rates than do older drivers because historical evidence shows that younger drivers have more accidents than do older drivers. This assignment of a higher insurance rate to John, Jr. is not an expression of prejudice against him; it is simply an assertion that persons in his age group are more likely to be involved in an automobile accident than persons in the age group of John, Sr. John, Jr. might be an excellent driver, perhaps even a better one than is John, Sr. But Jr. pays a higher rate for automobile insurance due to the expected value for his age group.

The Toxicological/Epidemiological Approach

Characteristics

The toxicological/epidemiological approach attempts to identify causal mechanisms in occurrences of undesirable events. This focus on explaining why negative events occur requires the application of scientific theory to analysis of previous events. Hence, this approach represents a considerable advancement over the actuarial approach in its attempt to explain the occurrence of undesirable events. As with the actuarial approach, undesirable events are considered to be observable and reality is thought of as positivist in nature.

Assumptions

This approach assumes that the correct theoretical explanation has been applied to the data on previous undesirable events. It assumes that events can be explained and that future events will, under conditions specified by a theory, occur in accordance with theoretical predictions.

Strengths and Limitations

The toxicological/epidemiological approach provides quantification of undesirable events, an explanation of these events, and therefore a rationale for predicting the occurrence of undesirable events in the future if theoretical conditions exist. But it depends upon correct specification of theory. If the theory is free from misspecified spurious or suppressor relationships, then the risk assessment will be relatively correct subject to the errors of observation and measurement as described below and in the critique of risk assessment.

Examples of Use

Modeling events is a common practice in science and modeling undesirable events is one of the keystones of risk assessment. In quantitative risk assessment, technical experts attempt to derive expected frequencies of undesirable events based upon experience with previous events and theoretical expectations of conditions occurring that would lead to an undesirable event occurring in the future. Thus, a technical expert might conclude theoretically that the use of Pesticide A will result in one additional person in ten million contracting cancer than would be the case if Pesticide A were not to be used.

The Probabilistic Approach

Characteristics

Probabilistic risk assessment is the use of modeling applied to technology systems rather than to a single event. This approach relies upon the application of logic systems such as fault-tree or event-tree analyses to arrive at a quantitative assessment of overall system failure as some function of the probability of the failure for each of the components of the system. Undesirable events are considered as observable and positivist in nature.

Assumptions

The approach assumes that theories for each individual risk assessment are correct and that the procedure for combining the probabilities of individual failures to arrive at an overall assessment of failure is correct.

Strengths and Limitations

The probabilistic approach is useful for quantifying the probability of system failure for complex technologies. It is difficult to model, however, the probability of common mode failure, (the simultaneous breakdown of more than one system component) and human-machine interactions. The approach is limited by all the possible errors of observation and measurement outlined in the critiques of risk assessment. But note that such failures in precision can occur for each component of the overall system and that each probability for error is multiplied by the probability for error in subsequent components. Thus, the overall assessment of failure can be highly inaccurate.

Examples of Use

Many technologies consist of a system of individual technologies. In fact, the technologies people tend to fear the most--nuclear power plants, petrochemical refineries, and food safety nets--consist of many individual components that might fail, which would result in total system failure. The risk management approach taken in such cases is to institute backup systems in the event that one system fails and early warning systems to detect an impending component or system failure.

Application in Context What is the technical risk assessment of food irradiation? Quantitative Risk Assessments Food irradiation is perhaps the most studied food processing technology, with over 50 years of research on its effects. In theory, the dosages of gamma rays applied in food processing should allow for virtually no chance of survival for microorganisms living in or on the food. Studies that seem to show adverse health effects from eating irradiated food have been discounted because of serious flaws in methodology. The U.S. Food and Drug Administration has determined that irradiated food is safe to eat. Critical Thinking Qualified epidemiologists have critiqued the studies used by the FDA in approving food irradiation as having serious flaws in methodology. These persons nevertheless support the technology. Therefore, note that technical assessments are critiqued as being flawed by both proponents and opponents of food irradiation. Is it possible to conduct scientific inquiry that is not flawed? If not, then how should the consumer interpret conflicting accounts of scientific credibility?

Critiques of Technical Risk Assessment

Ortwin Renn

Understanding that risk is multifaceted explains how societies can be in conflict over technology adoption and why technology adoption sometimes can take a long time to achieve. Renn makes these observations about approaches to risk assessment:

1. All approaches have benefits and drawbacks.
   
2. All approaches are necessary for a complete understanding of risk.

Renn offers these critiques of technical risk assessment:

• People have different values and preferences that affect their perceptions of risk.
  
• Human interaction with technological systems is difficult to model quantitatively. Outside the domain of all forms of technical assessment, for example, is the probability of technology failure due to human mismanagement or irresponsible behavior.
  
• The institutional structure designed to manage risk itself might be inadequately designed or managed to do so.
  
• Technical approaches imply risk management practices in proportion to quantitative risk assessment. People, however, also desire risk management policies that include objectives such as fairness, equity, and sense of morality/ethics.

John Adams

Adams, (Risk, 1995) points out that risk is not easily measured, agreed upon by diverse audiences, or managed. In asking, "Can we assess risk better?," he is not so much posing a problem that has a one best solution as challenging us to become more involved in understanding and evaluating relationships among science, technology, and society.

Adams notes that sometimes the public and scientific experts differ in their evaluations of technology risk. This disagreement occurs, in part, because the public uses a wide variety of criteria, including some nonscientific criteria, in its evaluations of risk. Adams distinguishes between formal and informal approaches to risk evaluation, wherein formal approaches emphasize technical assessments of health and safety hazards and informal approaches address social, political, economic, and ethical issues. He observes that the typical response of technical risk evaluators to nontechnical evaluations is a patronizing effort to further educate the public about the real risks associated with a technology.

Adams rejects as a false dichotomy the notion that technical experts know actual risk and the public harbors uninformed, misinformed, and even irrational perceptions of risk. He asserts that individual and group risk-taking involve instead a balancing act between social, political, economic, and ethical costs and benefits. He argues that adherence to the false dichotomy has led to many misguided attempts to educate the public into thinking correctly about a new technology. Given that such educational efforts are necessary but not sufficient motivators of attitudinal and behavioral change (even when scientists do have a good knowledge of actual hazards), scientific experts experience inevitable failures and subsequent frustration in their attempts at risk communication.

Adams points out the actual versus perceived dichotomy is false in two respects:

1. Technical risk assessments are neither entirely objective nor necessarily very precise.
   
   
   • Sometimes no data exists upon which to make a risk assessment. For new technologies, this problem is common. For complex technological systems, the problem increases geometrically. That is, there might be no data for a particular component of the system and there might be no data for the combination of various components with one another.
     
   • Inadequate data, improper recording of data, and data that are difficult to disaggregate also can create problems in technical risk assessments.
     
   • When technical risk assessments are demanded, and the data are inadequate for such assessments, guesswork must be made, which elicits problems with values and opinions entering into presumably objective indicators of risk.
     
   • Technical risk assessment is further hampered by accident migration (the tendency for ignored areas of accident occurrence to experience increased accidents) and regression towards the mean (the natural ebb and flow of accidents associated with a certain range of events).
     
   • Cultural filtering determines which types of risk will be assessed and the outcome of the risk assessment. Noise (measurement error in collecting data), "near misses" (ambiguous data), and bias (misrepresentation of data) also affect quantitative risk assessment.
     
   • Deriving cost/benefit analysis for a technology not yet in use can be especially difficult. First, to assess expected utility, the user of the technology must be fully informed of the risks associated with it. The educational requirements for a complex technology, however, can be extensive. Second, users must be able to incorporate subjective evaluations into their expectations of utility. As noted, these evaluations depend upon the social construction of risk, which not only include many subjectively defined shared values, but require some lag time to fully develop.
   
   
2. Technical risk assessments exclude considerations of political, social, and ethical goals.
   
   Even to the extent that technical risk assessments accurately reflect hazards, because they ostensibly exclude consideration of political, social, and ethical goals, they provide only a limited appraisal of the value of a technology. A technology might contain few hazards but engender much outrage. The abortion of a human fetus, for example, is a fairly safe technology (for the mother) but raises strong emotional feelings in American society. On a different note, some argue that risk assessments intentionally include political and economic considerations. See: Risky Business: How Scientific Are Science-Based Risk Assessments?

Michael Bell and Diane Mayerfeld

Bell and Mayerfeld (The Rationalization of Risk, 1999: full text article) , like Renn and Adams, note important limitations to technical approaches to risk assessment:

• Quantitative risk estimates are precise, but often are not accurate because they rely upon a whole series of assumptions, guesses, and extrapolations that limit their accuracy.
  
• Estimated risks often do not account for multiple hazards that occur in conjunction with one another in complex technological systems. For example, we might estimate the risk of pesticides A and B, but often we do not estimate the risk of pesticide A in combination with pesticide B.
  
• Numbers often carry disproportionate effect in technological assessments of risk.
  
• Technical risk assessments often falsely homogenize populations. That is, the risk for a child might be different than the risk for an adult.

Judith Bradbury

Bradbury (Science, Technology, and Human Values 14: 380-389, 1989) notes that all forms of risk assessment have limitations. Therefore, risk communication strategies that rely too closely upon a single risk assessment framework will not be as effective as they could or should be.

Technical approaches define risk as the product of the probability and consequences of an adverse event. Assessments of probability and consequences are made by technical experts as part of quantitative risk assessments. From the perspective of the technical approach, risk can be evaluated independently of political, economic, or social conditions. Thus, risk resides primarily in the technology and its relationship to foreseeable consequences.

The technical approach implies communication strategies that educate the public about technical risk assessments. Technical assessments are considered to represent actual risks. When these risks are deemed by regulatory agencies to be minimal, then, if possible, they are not conveyed to the public to avoid unnecessary concern. When risk assessments become public and consumer perceptions do not coincide with actual risk, then, from the technical perspective, acceptance of a new technology can be unnecessarily delayed or implementation can become more expensive than necessary. Thus, public rejection of the logic of technical risk assessments is considered to be irrational. Risk communication strategies thereby focuses upon educating an ignorant and sometimes irrational public about actual risk. Strategies seek to reduce outrage based upon inaccurate perceptions so as to retain a focus on actual risk.

Bradbury asserts that this approach ignores the economic, political, and cultural dimensions of risk assessment and management. The public is not necessarily ignorant of technical risk assessments nor are they being irrational in expressing skepticism about a technology when technical assessments show it to be a minimal risk. Rather, the public evaluates technology on a broader set of criteria than is considered in technical risk assessments. Bradbury asks, for example, "Who bears the burden of responsibility for defined risk?" and " Who decides how risk will be evaluated?" She argues that because technical assessments ignore economic, political, and cultural issues, risk communication strategies that focus upon education about technical facts are necessary but insufficient to sway public opinion.

Risk Perceptions and Risk Management

Managing risk, then, is a key motivator of much professional practice. But the objective of managing risk must be as free from misplaced concreteness as possible to avoid polemics. Myths about nature, or cultural outlooks, affect all risk evaluations. These myths--also called paradigms, ideologies, belief systems--are the set of assumptions about reality formed through shared experience, supported by interactions with others, and routinely go unquestioned. They are culturally constructed and maintained.

What happens when observations about reality do not correspond with our assumptions about it? Certainly, we should avoid being too hasty to revise paradigms; they must have enjoyed a great deal of support at some point in time to have gained their standing. Yet, to cling too long to paradigms with many anomalies is to engage in the fallacy of misplaced concreteness: to believe in the paradigm in spite of overwhelming evidence that refutes it.

But paradigms carry much emotional baggage. Revising them or exchanging them for radically different ones requires not only much scientific debate, but much soul-searching as well. Thus, evaluations of risk, if they place pressure on paradigms, which they sometimes do, will instigate debates about paradigms that reflect cultural outlooks. Hence, debates about high risk technology often entail emotionally charged debate that reflects cultural outlook.

Can we manage risk better? Adams suggests keeping in mind the following observations on the evaluation of risk by technical experts:

1. Remember, everyone else is seeking to manage risk, too.
   
2. They are all guessing; if they knew for certain, they would not be dealing with risk.
   
3. Their guesses are strongly influenced by their beliefs.
   
4. Their behavior is strongly influenced by their guesses and tends to reinforce their beliefs.
   
5. It is the behavior of others, and the behavior of nature, that constitute the actual risk environment.
   
6. It will never be possible to capture "objective risk."