by Alan Journet, Conservation Chair, Trail of Tears Group
Every decision we make is fraught with the possibility of error. When we take the car for service because the engine warning light is on, thinking the warning light indicates danger rather than thinking it is just a gauge malfunction, we could be wrong. If we conclude there is no danger and continue driving, we could also be wrong. Such mistakes have been codified in Decision Theory. If we take the car in when we really did not need to, we have made a Type I Error. On the other hand, if we keep driving when we should have heeded the warning, we have made a Type II Error. Every time we make a decision, the possibility of error looms before us.
The Judicial Burden:
We are all familiar with the legal principle that an individual charged by the state with a crime is presumed innocent until proven guilty. Again, a decision has to be made. But unlike the example above, here a persuasive case has to be made before a decision is reached. That the burden of persuasion should fall on the prosecution to convince a judge or jury that the accused is guilty beyond a reasonable doubt seems entirely consistent with our individual freedoms, liberties, and societal mores. Most of us agree that innocent individuals should not be convicted of crimes. Maybe we feel this because we would like such a requirement to be imposed on the state in the event we were charged with a crime we did not commit. Maybe we just do not think the state should risk imposing sentences (the death sentence for example) on innocent citizens.
As a result of these reasonable concerns, we generally agree that the burden of proof should fall on the prosecution to make the case for conviction rather than falling on the defense to make the case for acquittal. However, since both juries and judges are human, we must acknowledge again that, when decisions are finally rendered, mistakes can be made. Convicting an innocent person is a Type I Error while acquitting a guilty person is a Type II error.
By placing the burden of proof on the prosecution the judicial system strives to avoid making the Type I Error of convicting an innocent person. In so doing it favors the Type II Error of allowing a guilty person freedom. Unfortunately, as the likelihood of making one error decreases, the likelihood of making the other inevitably increases. Every time we make decisions we are forced to choose between these errors; we cannot simply check in our chips and get out of the game.
The Scientific Burden:
In science, the burden of proof follows that of the legal system. In this discipline, we collect and analyze data to test hypotheses. The null hypothesis is the judgment of no pattern, regularity or relationship, while the alternative hypothesis is that a pattern, regularity or relationship exists. When we are evaluating, for example, the relationship between nicotine and cancer, or the occurrence of global climate change we establish a null hypothesis for data analysis that would state either no relationship between nicotine and cancer or no pattern of climate change. Against this, we would then establish the alternative hypothesis that there is a nicotine-cancer relationship or that there is global warming. In science, we desperately wish to avoid making the Type I Error of stating there is a connection between nicotine and cancer, or that there is evidence of global warming when such does not exist. Scientists are desperately concerned that their credibility will be lost if they incorrectly conclude patterns, regularities and relationships exist when they do not. As a result, they we will only reject the null hypothesis if the chance is less than 5 percent that the null hypothesis is correct.
As a consequence, before the scientific community suggests patterns, regularities or relationships exist, there is a greater than 95 percent probability that they do exist. Sometimes this requirement is even more stringent, being set at the 1 percent or 0.1 percent and 99 percent or 99.9 percent level respectively. Again, the consequence of reducing the probability of making the Type I Error of concluding patterns, regularities or relationships exist (when they do not) is that we increase the probability of making the Type II Error of concluding no pattern, regularity or relationship exists (when such actually exists). Whenever we make decisions, in life, the courts, or science, we cannot get out of the game.
Should The Burden of Proof Always Be The Same?
The question then arises as to whether all human decisions should follow the judicial and scientific principle of favoring a Type II Error over a Type I Error. Let us consider medical health questions for a moment. Suppose that a physician has suggested that you may be suffering from a serious disease, and thus proposes a painful and costly treatment that has a good probability of curing the condition—if it exists. You now have a choice: you could reject the medical advice and the treatment and risk assuming you are not ill when actually you are (i.e. make the Type II error of accepting the false null hypothesis that you are ok and have no medical condition when you actually do). Alternatively, you could accept the medical advice and the treatment and, therefore, infer you are indeed ill and could benefit from treatment when actually you are quite well (i.e. make the Type I error of rejecting the true null hypothesis thereby accepting treatment when actually you do not need it). Given such a choice, and after seeking a second, third, and maybe specialist opinion, you are likely to conclude that it is better to make a Type I Error than a Type II Error: it is probably better to accept treatment (unless the treatment itself is hazardous), even when you actually may not need it, than risk rejecting treatment when you actually do need it. This approach is consistent with the view that ‘it’s better to be safe than sorry.’ Those familiar with the Precautionary Principle (which states that in the absence of certainty, it is better to be prudent and take the course of action that will lead to reducing risk or hazard than follow the course that increases the risk or hazard) will note that the decision to favor a Type I over Type II error is entirely consistent with that principle.
In this case we have not placed the burden of proof on the prosecution (the physician) to demonstrate beyond any doubt that we are sick. Rather we are likely to accept the expert’s professional judgment, acknowledge the potential cost of inaction in relation to the potential benefits of action, and submit to treatment even though it may be uncomfortable or expensive.
Deciding where the Burden of Proof should lie in decision-making has significant implications for the outcome of those decisions. Who should be assigned the burden of proof depends upon the nature of the issues and our philosophical and ethical approach to them—specifically our weighing the relative costs of Type I and Type II errors.
Burden of Proof and Business
When we adopt the current legal principle of ‘innocent until proven guilty’ in evaluating industrial activities or development proposals that have potential environmental costs, the consequence is a tremendous advantage to industry and developers. This is because the burden of proof falls upon opponents of industry or development to demonstrate that the environmental or health costs outweigh the benefits before an activity can be halted. By adopting the philosophical position that we would rather assume an activity is harmless and allow it to go forward when it actually is harmful (Type I Error) than halt an activity because of potential harm when it actually is benign (Type II Error), we promote potentially hazardous activities, inadvertent though this consequence may be.
An alternative to this approach is to adopt the Precautionary Principle. In this case, the burden of proof is switched 180 degrees and falls upon project developers to demonstrate the safety of their project before permits are granted.
Burden of Proof and Health Issues:
Unfortunately, as with the environmental example above, when standard cost-benefit analysis is applied to health, decisions, it usually favors short-term economic gain and the risky choice rather than the more conservative or prudent choice promoting health. When industries propose releasing toxic chemicals into our air and water, the burden of proof falls upon health agencies to demonstrate the human health costs of potential pollution. It seems that we would rather make the Type II Error of assuming safety when there is danger than make the Type I error of suspecting hazard when none exists. In such analyses, each human sickness and life is necessarily assigned a value. This value falls generally in the $1 million to $10 million range per life, though it focuses in the $2—$3 million area. Of course, this value may fall well below what many of us think human lives are worth, especially when it is our own life. When the probability that a toxin may cause death or sickness is calculated to be very small for each exposed individual, even though, statistically, sickness and death at a low frequency may be inevitable, the overall cost-benefit calculations tend to favor pollution rather than protection of human health.
On the other hand, if we were to adopt the view that industries should demonstrate they pose no threat and meet a stringent safe minimum standard before being permitted to conduct business, with the burden of proof falling upon them to demonstrate their activities are safe, the balance would favor Type I Errors and a healthy environment rather than Type II Errors and a polluted environment (see Figure 1).
Burden of Proof and Conservation
When the question concerns protecting our natural resources and biodiversity, deciding where the burden of proof should be assigned becomes critical. One of the essential problems that conservationists have is in demonstrating the value of conserving species since biodiversity is generally valued only in terms of its anthropocentric or utilitarian value, i.e. its direct economic or recreational value to humans. In conventional cost-benefit analyses, the economic benefits of projects are weighed against the economic costs or potential costs of damage to other species. The burden of proof thus falls upon the conservationist to demonstrate the costs of loss of species. Regrettably, for many species this is a very difficult task indeed.
There are, however, two alternative views that shift the burden of proof:
- One approach is to consider the value of the life support system upon which we depend and estimate the value of each of its component species. While this is a difficult task, one measure is to consider the cost of the Biosphere II ‘experiment’ conducted in Arizona from 1987–1989. In that adventure, a closed, supposedly self-sustaining, microcosm was established to support eight intrepid human biosphere inhabitants for two years. Interestingly, it failed because the system was not self-sustaining; the residents had to be supported by input from the outside. This adventure was called Biosphere II to distinguish it from the rest of the planet which supports us (Biosphere I). The project cost for construction and maintenance was $200 million for eight individuals over two years. Thus, the cost per person per year was $12.5 million. Now, if we extrapolate this to the (conservative estimate of) six billion humans on the planet, we find the planetary ecosystem to be valued at $7.5x1015 (in 1987 currency).
Now, we do not know exactly how many species there are on the planet, but if we take the largest estimate of 10 million, we can compute the average value to our ecosystem of each species as $750 thousand per annum. Now, it is clear that some species are more important than others, but the problem is that we do not understand our planetary ecosystem well enough to know which is which. As population biologist Paul Ehrlich once noted, we sit on the planet as a passenger might sit on an aircraft watching mechanics taking rivets out of the wings and fuselage. We know there is much redundancy in the system, but we do not know exactly which rivet will be removed and cause the collapse of the entire system. Prudence, therefore, demands that we value all species equally, and that we value them very highly. Some might argue that this would be the conservative approach, and might further note that the term conservation is derived from the same root as the root of conservation. - According to the second approach, rather than considering only the anthropocentric value of other species, we might adopt a biocentric world view. According to this world view, we extend our sense of the intrinsic value of life beyond the limits of ourselves, our family, our community, our nation and even our species, to encompass all other life forms on the planet. If we do this, then all species have intrinsic value.
The interesting consequence of both these approaches is that the value of biodiversity is seen as vast; it may not be infinite, but it is certainly incalculably large. With such a perspective, the cost-benefit analyses discussed above change dramatically as the value of conserving species reaches a level somewhere between vast and infinite. Similarly, while each specific industrial activity or development project might not in itself pose the threat of species extinction, the potential contribution that each might have towards this result must be assessed (see Figure 2).
Conclusion:
While it is inevitable that every decision we make confronts us with Type I and Type II errors, and many decisions we make confront us with health and survival risks, when it comes to making political decisions regarding public health, environmental protection, or conservation, we should also be aware that where the burden of proof lies will influence the frequency at which Type I and Type II errors are made. Consequently, where we as a society decide the burden of proof should fall will influence the extent to which we and our environment are protected from the threatening activities of industries, developers, and those who would exploit our natural resource for short term benefits.
Figure 1 Cost-benefit Analysis and the Burden of Proof
Figure 2 Burden of Proof and the Value of Biodiversity