THE VALUE OF EVIDENCE IN LAW
Vol. 39 No. 2 Summer 1988
During the last fifteen years in the United States there has been much talk about uncertainty by two very different groups of scholars. First, legal and non-legal scholars in America and elsewhere have been using formal theories of probability and inference to study uncertainty in factfinding. Second, the collage known as Critical Legal Studies has devoted much of its time and energy to the proposition that legal reasoning and legal doctrines are "indeterminate" and uncertain.
These two groups of scholars have largely ignored each other. This is not as it should be. Although the two groups seem to be talking about very different kinds of uncertainty, the puzzles that interest these two groups are related. I have suggested elsewhere that the vast and still-burgeoning literature about the interpretation of texts, symbols, and meaning is important to theorizing about the nature of inference from evidence.1This is because many problems of factual inference involve the job of making guesses about human aims and meanings. Today I want to try building a bridge from the other side of the river. I will argue that theories of factual inference have much to say about theories of legal inference; I will argue that insight into the logic of inference from evidence offers important insights into the logic of legal reasoning; and I will argue that reasoning about facts and reasoning about law are more alike than different.
My thesis is unconventional, and possibly even startling. It amounts to the claim that law is a fact and that all sound law rests on evidence. This claim is not as bizarre as it may seem. I am not arguing that it is possible to be certain of the meaning of law. I am making the much more modest claim that there are better and worse ways of guessing about legal questions; that legal reasoning is a way of guessing about legal questions; and that guessing about legal questions is in many ways like guessing about factual questions.
There was a time—so we are told—when not only facts were facts. There was a time when law was also a fact; there was a time when the legislator, if he existed at all, thought that his job was to discover the law rather than invent it. However, this vision of law as discovered rather than legislated has passed away; we now generally believe that law is legislated rather than merely discovered.
Our vision of the nature of facts has also undergone a profound transformation. There was a time when facts, like the law, simply awaited discovery. If we can believe what some legal historians tell us,2 the rules of evidence and proof were not always designed to facilitate the weighing of factual probabilities. Instead, the law once took the view that evidence, when appropriately packaged and processed, established the truth beyond any doubt.
This sort of theory of evidence and proof is dead. The theories of evidence and inference that now dominate the law are firmly planted in the so-called rationalist tradition of evidence scholarship, a tradition that has been masterfully described by William Twining.3 In that tradition, it is axiomatic that all knowledge of facts is merely probable and always uncertain. We believe that the law of evidence may reduce the frequency of errors in factfinding, but we also believe that factual certainty is unattainable and that no matter how much evidence we have or how careful we are, we can always make mistakes about the facts.
I believe that certainty about any thing is unattainable. Therefore, by saying that law is a fact I am not saying that there is a technique that will restore our faith in our ability to resolve difficult legal questions in a positively correct way that is free of all doubt. All hard cases may have a right answer but they do not have a clear right answer. I am only arguing that the logic of our reasoning about legal uncertainty is much like the logic of our reasoning about factual uncertainty.
The prima facie case for an affinity between legal and factual inference is not hard to make out. Legal interpretation may be seen as an effort to draw inferences about the meaning of laws from reports of the meaning of those laws. There are various models of the structure of inference that seem to portray the types of guessing that this vision of legal interpretation generates.
In terms of one style of inferential theorizing, the meaning of a law is the fact whose probability is to be assessed, and the reports of the meaning of the law—eg, the language in a book, committee reports, and the like—are treated as evidence. This style of theorizing asserts that the probability that a law has a particular meaning depends on such evidence. It asserts that different pieces of evidence—different reports about that law—generate different estimates or guesses that the law in question has a particular meaning and it tells us that the project of making a final guess about the meaning of that law consists of trying to combine and put together those separate estimates in a coherent fashion.
In terms of another style of inferential theorizing, legal interpretation involves trying to grade (on an ordinal scale) the relative plausibility of various hypotheses about the meaning of the law and we examine evidence and its range to see how well it supports those alternative hypotheses about the meaning of the law.
In terms of yet another style of theorizing about uncertainty, legal interpretation involves the effort to express as precisely as possible the fuzziness of the words that are used to express legal meaning.
One or more of these models may well capture and describe the types of uncertainty that we see in the law, and the various models I have alluded to do seem to portray the properties or structure of some of the uncertainty we see in legal problems. However, if in fact one or more of these models accurately portray the types of uncertainty found in the law, the logical structure of legal and factual uncertainty is to some extent identical. The models I have mentioned also portray the structure of our guessing about facts.
I have just made out a prima facie case for the factual character of questions of law. I made out this case with a fair degree of ease and without great intellectual labour. There is a good reason for this. Formal theories of uncertainty have a curious and interesting property. By reason of their formal character they can be used to portray any set-up that involves uncertainty of a particular sort. Hence, once the concession is made that legal interpretation involves uncertainty, legal problems can be portrayed by one or more formal theories of uncertainty. The project now at hand, therefore, is to determine what if anything is wrong with trying to portray legal uncertainty as uncertainty that is a function of evidence or information.
The objections against treating law as fact fall into two broad categories. On the one hand, there may be normative objections. It may be said that our values and our sense of justice make it inappropriate to treat law as fact. On the other hand, there may be epistemological objections. It may be said that the realities of law and legal reasoning make the portrait of law as fact uninstructive and uninformative.
In this lecture I want to focus on the epistemological issues because my interest and expertise run in that direction but please bear with me while I first make some brief comments about the value of evidence in law.
Normative considerations, taken by themselves, speak strongly in favour of treating legal inference as an inquiry into a question of fact whenever possible. The role that evidence plays in adjudication suggests why this so. Evidence is important in adjudication primarily because we think it is very important to draw reliable inferences about material facts. The reliability of factfinding is in turn intimately connected with the ideal of the rule of law.
I submit that reliability is also important for inferences about legal norms. In matters of law, just as in matters of fact, it is very important to us that we get the right answer and, if we cannot do that, it is important that we make guesses about law that are more likely right than wrong. So I believe that a commitment to the ideal of the rule of law implies that reliable lawfinding is desirable.
In my mind, then, the real difficulty with my thesis about law as fact is not with the desirability of treating law as fact. I take for granted that it is good to have law that rests on good grounds and good information. The real difficulty with my thesis is the question of the possibility of this ideal of law as fact. The question is whether lawfinding and lawmaking in our world have properties that make it meaningless to talk about inferring laws from evidence. I turn to this question next.
Surely the central objection to my analogy between legal and factual inference rests on the sense that legal interpretation is a far more subjective process than factual inference. Moreover, my thesis of law as fact embraces legislation and it seems odd or vacuous to talk about "inferring" law from "evidence" when what we have is law that was created by a legislator.
Consider first the argument based on the alleged subjectivity of legal interpretation. The claim that legal interpretation is a subjective process cannot sustain a distinction between legal interpretation and factual inference. The business of drawing factual conclusions from evidence is a very, very subjective one. Much of the work of the new evidence scholarship has centred on the use of formal theories of probability as models of factual inference. From the beginning, this new evidence scholarship stressed the subjective nature of inference. The early leaders of this new wave of scholarship made much use of subjective Bayesianism. This is a theory that purports to describe how a rational trier logically and coherently combines his subjective probability estimates.
In more recent years, the emphasis on the subjectivity of inference has deepened. Many scholars now acknowledge that (1) there is not just one formal theory of probability or inference, but there are at least several different theories, each of them equally coherent: (2) the choice among these different formal theories as models of rational inference cannot be made on purely logical grounds, and (3) the choice among these equally coherent theories also cannot be made on empirical grounds.
The corollary of this thesis of the subjectivity of inference is the proposition that there is no objective basis for preferring one model of inference over another and that intuition or something very much like intuition is the final basis for the choice of a particular formal theory as a model of inference. To be sure, the choice of a particular formal theory as an appropriate portrait of the shape of rational inference involves some thinking and reasoning—indeed, much of both—but that choice is nonetheless ultimately a subjective one.
The emphasis in formal theorizing on the subjectivity of inference has a close parallel in informal theorizing about the nature of inference. The conclusion that we must trust our intuition and common sense when evaluating evidence is not peculiar to those who debate formal theory. It is a widely-held sentiment. This suggests that the emphasis in the new evidence scholarship on the subjectivity of inference is not mere happenstance.
In point of fact, the belief that inference is subjective is not happenstance. It is the inevitable result of the fact that modern Western man by and large lacks an ontology, a theory of fundamental reality. The lack of a theory that establishes a hierarchy of reality among phenomena in the world removes any basis for claiming that there are privileged sources of knowledge. The absence of this sort of ontology necessarily produces an associationist theory of knowledge with a subjective flavour.
An empiricist theory of empirical knowledge asserts that that knowledge depends on the regularity with which phenomena are associated with each other. This empiricist theory is intrinsically subjective because the only thing that vouches for knowledge is the experience of regularity in the cosmos. However, if individuals differ, either physiologically or conceptually, their experience of regularities in nature also differs; they either havedifferent experiences or they process their experiences in different ways. In short, like it or not, we are stuck with a subjective theory of inference. The extent of that subjectivity is deep because it is now also apparent—which it was not before—that we also have no logic of inference that can legitimately claim priority over other logics, and we have no empirical data that can show one mode of inference is right and that another is wrong. Hence, the argument that legal reasoning is more subjective than factual inference is very hard to make out. Factual inference is a very subjective business.
Unfortunately I do not have the time to discuss various permutations of the notion that legal interpretation is more subjective than factual inference. However, allow me to note that many of the distinctions we might try to make between legal inference and factual inference are illegitimate or overdrawn. First, legal reasoning is not distinctive merely because it involves the interpretation of symbolic meaning and texts. Inferences about facts often require this too. Second, legal interpretation is not distinctive because it involves imagination: inference from evidence often involves the same thing. Third, legal reasoning is not distinctive merely because the choice of an interpretative matrix is partially or largely left open: the same is true for the interpretation of evidence to determine facts. Fourth, it should be clear now that legal reasoning does not differ from inferential reasoning merely because the an of legal interpretation cannot be reduced to a formula and requires the use of intuition. Finally, it is not even clear that the normative dimensions of legal reasoning differentiate factual and legal reasoning. There are normative elements in reasoning about evidence and facts. I have sketched out some of these points in a recent paper.4
In sum, then, we have not yet discovered any sense in which the subjectivity of legal interpretation is essentially different from the subjectivity of factual inference. Nonetheless, there remains a possibility that legal interpretation has a special kind of subjectivity that is not found in inference from evidence. There is the possibility that legal interpretation is uniquely subjective because it involves pure legislation that is not grounded in any kind of evidence or information. Let us now examine this objection to law as fact—this objection to the notion that evidence is important in legal interpretation.
The final and most powerful objection is that it is vacuous to talk about inferring law from evidence. The basis of this objection is the legislative character of the activities of legal decision-makers. If legislators and judges make law rather than find or discover it, no amount of evidence can show that a law is more probably "true" or more probably "false". When a legal decision-maker acts in the fashion of a legislator, the basis for the law chosen apparently lies in the decision-maker's values and ideological preferences rather than in any evidence.
This is a powerful attack. My response to it is that law can be treated as a fact even in a world where new laws and norms are undoubtedly being generated. My response is by way of confession and avoidance. I agree that legislators and some judges do occasionally import personal values and personal ideological preferences into the law; but I assert that even in these cases there is an important sense in which law can remain a fact. Moreover, I assert that the extent of entirely free and unconstrained legislation is easily exaggerated.
Consider the activity of a judge or legislator who has decided that it may be necessary to create an entirely new legal rule. This person's decision to create a new legal rule is not necessarily unconnected with a factual inference, his decision may depend on factual inferences. In reality, of course, even the legislator who feels free to create a new legal rule may well have a great deal of interest in evidence. He may accumulate large masses of evidence through committee hearings, parliamentary commissions, and the like. This quite rightly suggests that the legislator is inferring what the law should be even though he also feels free to create new legal rules for reasons that he personally thinks are sufficient.
This phenomenon is not difficult to explain. The legislator may have the freedom to decide what conditions would warrant new legislation. But he may of course very much want to know whether or not those conditions actually exist, and it is plain that he must acquire and assess evidence and information to determine if they do exist. For example, the legislator may believe that a social spending scheme would be required by a certain level of unemployment but, without any change in ideological preference, he may decide against the scheme if he believes that in fact very few people are unemployed. Hence, the bare fact that the legislator is free to conclude for entirely personal reasons that corrective legislation of a certain type would be advisable under certain conditions does not convert him into a pure creator of legal norms.
In this weak sense, then, even a legislator who makes law treats a question of law as a question of fact.
There is a more subtle and more basic sense in which legislators may be trying to discover law rather than invent it. It should not be assumed too readily that legislators simply decide for themselves whether or not there is good reason to enact a new law to deal with a problem such as unemployment or police brutality. Legislators may well believe that the question of the appropriateness of new legislation for particular types of conditions depends on the existence and character of various mushy, but actual, states of affairs such as the sentiments of the electorate, the cultural values of their society, the general direction of the law in civilized societies, and the like. I believe that legislative and judicial decision-making frequently depends on assessments of these sorts of matters. If I am right about this, there is a strong sense in which even some of the most radical legislative decision-making amounts to factual decision-making. If assessments about matters such as the character of civilized customs play an important part in legislative decision-making, the creation of new law depends on evidence and on guesses about actual states of affairs in the world—in short, on factual inferences.
The claim that matters such as enlightened custom can be treated as evidence is likely to provoke an attack that resembles one we have seen before. My claim is likely to provoke the argument that this sort of putative evidence is no evidence at all, because standards such as "enlightened custom" are little more than an expression of the political, ideological, moral, or personal preferences of the legal decision-maker. The argument is that even if we assume that the judge or the legislator should rely on some evidence in deciding what law to declare, it makes no sense to speak of "inferring" the meaning of law from this sort of "evidence" because the matters that putatively amount to "evidence" are very much in doubt and therefore subject to arbitrary choice. The question of what should count as "evidence" has no obvious answer and different legislators and legal interpreters will take different things as appropriate indicia of the proper design of legal rules.
I think this argument is partly right but largely wrong. I believe that it is meaningful to speak of evidence of the proper design of a law even if, as an abstract matter, the choice of what things to count as evidence is left open. I assert that matters that abstractly appear to be open to choice frequently are in fact not open to choice and that therefore certain values taken as evidence by the lawmaker are in fact evidence merely because they are taken as such. Let me explain this mouthful of words.
Consider my argument that reports of law—eg, language in a lawbook—may be taken as "evidence" of the meaning of law such that, for example, we may express the legal problem in the form of p(M|E). The objection to this is that many evaluative assessments are involved in the decision to treat something as evidence of the meaning of a law and that these are so numerous that we are not entitled to regard the legal question being addressed as being in the nature of a factual question. More broadly, the objection is that the decision to take certain matters as "evidence" of law involves political, ideological or moral decisions that are arbitrary rather than objective or neutral.
For example, arguably the expositor of the law will arbitrarily choose to emphasize the aims and intentions of a parliamentary commission rather than the understanding of the electorate when he tries to decipher the probable meaning of a law. Similarly, in American constitutional theory a wide range of choice among different sources of "evidence" seems available. The interpreter may variously take as evidence of the meaning of a constitutional guarantee the views of Madison and Hamilton in the Federalist Papers, the views of the educated public, the debates or understandings of the members of the constitutional convention in Philadelphia, the evolving views of "civilized" nations, and so on. The argument is that it is vain to think of the lawfinder's inquiry as a factual inquiry because he remains free to decide whose understandings are to be taken as evidence of the meaning of a constitutional provision.
This critique of the theory of law as fact says less than meets the eye. The challenge wrongly assumes that the trier makes an arbitrary choice about what evidence to count merely because he seems to have the abilityto make that choice arbitrarily. Consider, for example, the real possibility that a lawfinder may already believe that certain reports of the meaning of a constitutional guarantee are significant reports that must be considered with care. This lawfinder already has a tacit belief about the authoritative sources of law. It of course remains possible to assert that the lawfinder has nonetheless acted arbitrarily in tacitly choosing to treat certain reports of meaning as significant evidence of the meaning of law. The possibility of saying this, however, does not itself warrant saying it. This conclusion produces a very interesting conclusion about the relationship between facts and values.
If lawfinders do often address legal questions against a background of "entrenched" assumptions about the proper sources of legal meaning, and if those assumptions are in some sense normative ones, the normative character of those background assumptions serves to demonstrate that values can be and often are treated as facts. It is largely immaterial that we can also describe those background assumptions as "values". If such "values"—eg, the belief that certain reports or practices are authentic reports of legal meaning—seem inevitable and necessary, if values are so entrenched that we are not willing to question them, then there is every theoretical and practical reason to treat such assumptions as background facts as well as background values. If so, even while noting the various-ness of the different things that different people choose to count as authentic reports of legal meaning, we are fully entitled to say that lawfinders to a large extent are involved in an inquiry that involves inference from evidence.
Consider Justice Blackmun's opinion in Roe v Wade,5 which dealt with the question of the constitutionality of criminal laws against abortion. In examining the question whether states can criminalize the abortion of non-viable foetuses and the question of how the line is to be drawn between viable and non-viable foetuses. Justice Blackmun comprehensively surveyed the treatment of foetuses in Western civilisation, ranging from ancient Persian times, through the English common law of the middle ages, through American common law, to the positions taken in modern times by organisations such as the American Medical Association and the American Bar Association. This wide-ranging (and fascinating) survey is usually omitted from American constitutional law casebooks. It is apparently seen as window-dressing. In fact, however, it is entirely possible thatJustice Blackmun did not regard this historical evidence as "window-dressing" and that he seriously took it as evidence of the proper meaning or design of American constitutional law in the twentieth century. In short, I find it entirely plausible that this historical material counted in Justice Blackmun's eyes and that his conclusion about the proper outcome in Roe v Wade depended in part on what he found as a result of his historical research.
There is no point in exaggerating the empirical flavour of inquiries into the meaning of law. It is, after all, quite clear that investigations of legal problems do involve theoretical and normative commitments that are not dictated in any unique or straightforward way by things that are indubitably taken as authentic evidence, and it also seems clear that in many instances choices are being made among theoretical and normative perspectives that cannot easily be explained as the product of factual beliefs or evidence. My primary aim here is to remind everyone that the presence of such relatively free theoretical and normative choices does not in itself eliminate the factual character of an inquiry. I want to emphasize the simple but important truth that even legal interpretation involves in part an effort to make some reliable assessments about the existence of actual states of affairs in an actual world and that the indeterminacy of legal interpretation—which is very real and often very great—does not demonstrate otherwise.
Sociologically speaking, it is vain to deny that the disagreements about the meanings of legal rules and about the authority of different sources of legal meaning are often very wide-ranging in societies such as the United Slates. I only want to say that it may be a mistake to conclude that these disagreements indicate that legal interpretation and legal argument do not in a significant sense involve an effort to resolve questions about actual states of affairs in the world. The great variety in legal reasoning may reflect a shredding of a socially-shared web of belief. Even so, it remains true that legal problems force us to decide how we think things really stand in our world. We can never be completely confident of our answers and conjectures but we should try to make the best guesses we can. Formal theories of inference may aid us in understanding the logic of our guessing about legal problems a bit better.
* [Editorial Note] The text of a public lecture delivered on 10 March, 1987 in the Queen's University of Belfast.
 Tillers, "Mapping Inferential Domains" (1986) 66 Boston Univ L Rev 883, 907-932.
 See eg Shapiro, Probability and Certainty in Seventeenth Century England: A Study of the Relationships between Natural Science, Religion, History, Law and Literature (1983); Nelson, The Americanization of the Common Law: The Impact of Legal Change on Massachusetts Society 1760-1830 (1975), pp 25-26,114. Cf. Hacking, The Emergence of Probability (1975); Shapiro, "To a Moral Certainty: Theories of Knowledge and Anglo-American Juries 1600-1850" (1986) 38 Hastings LJ 153.
 Twining, The Rationalist Tradition of Evidence Scholarship, in Well and Truly Tried (ed Campbell and Waller, 1982). See also Chapter 1 in Twining, Theories of Evidence: Benthan and Wigmore (1985).
 Tillers, "Mapping Inferential Domains" supra, n 1 at pp 932-936.
 410 US 113 (1973).
* Professor of Law and Director, Program for Evidence in Litigation, Benjamin N. Cardozo School of Law, Yeshiva University.
Evidence marshaling software MarshalPlan