As an understandable reaction to the crude, dogmatic empiricism of the German Historical School, Ludwig von Mises became an uncompromising spokesman for pure, a priori economic theory—untainted by sensory observation :
What we know about the fundamental categories of action is not derived from experience. We conceive all this from within, just as we conceive logical and mathematical truths, a priori, without reference to any experience.… As an a priori category the principle of action is on a par with the principle of causality…no kind of experience can ever force us to discard or modify a priori theorems. They are not derived from experience; they are logically prior to it and cannot be either proved by corroborative experience or disproved by experience to the contrary.
Note that the typically modest assertion that Misesist theorems are precisely analogous to math and causality is nowhere supported by any argument. More to the point, contrast Mises' remarks with those of Objectivist philosopher Leonard Peikoff :
Since truth is the identification of a fact of reality, a "non-empirical truth" would be an identification of a fact of reality which is validated independently of observation of reality. This would imply a theory of innate ideas, or some equally mystical construct.…Without reference to the facts of experience, man has no basis for his "logical" propositions, which become mere arbitrary products of his own invention.
Mises, though, does claim to possess innate ideas leading to "non-empirical truth." Since this mystical notion conflicts so violently with Objectivist epistemology, we should examine its foundation. For simplicity, let us first consider a concise defense of Misesian method by Israel Kirzner :
…empirical confirmation of the theorems obtained by abstract knowledge is neither possible nor necessary. It is not possible, because there are no constants in the realm of human actions; it is therefore impossible to investigate the consequences of changes in one variable with assurance that no disturbance is at the same time being caused by changes in other variables. On the other hand confirmation of economic theorems is not necessary, because the theorems themselves describe relationships logically implied by hypothesized conditions.
First of all, the lack of constants, or of controlled experiments, in human actions needn't prohibit statistical predictions of useful accuracy. Probability statements, expressing relative frequencies of repeated events, are essential to any intelligent predictions of future experiences. We must act on the best available information about what is likely to happen in response to certain changes in prices, quantities, money units, etc. If the "impossibility of rationally comprehending life in its fullness and diversity" kept us from measuring relations between economic variables, it would likewise prohibit empirical work in the insurance business—which also deals with "unpredictable" human action .
While perfect prediction in such matters might imply determinism, a tolerable range of accuracy simply follows from the empirical fact that human action is, as Mises paradoxically over-emphasizes, almost "axiomatic" in its regularity. If "knowledge of the essence of human action" were as simple as Mises implies, how could the same human action be so random that we couldn't measure it ?
INFORMATIVE OR SELF-EXPLANATORY TRUTH?
A statement may be constructed in a formally valid manner and yet be empirically false and vice-versa. Consistent relationships between statements do not insure that the elements of the syllogism are not nonsense or that the conclusion will not contradict reality .
The convention of separating analytic (self-consistent) from synthetic (empirically falsifiable) statements is a sometimes useful, though not exhaustive, categorization . Much of the controversy surrounding the use of this analytic-synthetic dichotomy concerns the refusal of some of its users to admit that ethical reasoning is "meaningful," in the sense of being potentially capable of factual refutation.
There surely exist, however, at least in principle, some nonarbitrary limits on what can be considered good or bad. If, for example, that which is designated "good" must be conducive to survival of man as we understand him, then the needs of man and the relation of moral principles to those needs are subject to logical analysis and factual confirmation. There is, therefore, no reason why moral reasoning need conflict with the scientific method . To say, as does Brand Blanshard, that the analytic-synthetic dichotomy necessarily excludes value statements, is to concede that value statements have no basis in logic or fact. Far from being a devastating criticism of positivism, this objection to the analytic-synthetic dichotomy instead involves a tacit acceptance of crude positivism's most glaring inconsistency.
Another problem with the separation of analytic from synthetic statements is that most significant questions require a blend of logical and empirical techniques. Peikoff is misleading, however, when he says the dichotomy necessarily implies that logically necessary propositions "do not represent knowledge of reality" but only "arbitrary whim." Analytically valid statements do represent reality—if their premises and elements conform to sensory experience—though logically certain propositions appear possible for only a small range of relatively unimportant matters. Experience is a test for rationally deduced knowledge of reality.
The basic issue here is: "What counts as a good reason for accepting an assertion?" The Misesist answers that only a priori reasons count, and the naive positivist supposedly admits only a posteriori reasons. A more fruitful approach traces the growth of knowledge by a process of "conjectures and refutations." A relation between economic magnitudes is deduced, and this hypothesis is then tested against rival notions for its comparative ability to predict the effects of changes in the relevant variables. The hypothesis which is least-often refuted by observation is acted upon as true and certain (within the context of continually changing information and circumstances) .
Against this approach to discovery, Blanshard raises the dilemma of universals and particulars: "To falsify 'some swans are white' you must verify that no swans are white. Now you may have observed any number of swans and found no white swan among them, but that will never entitle you to say there are no white swans." 
This argument loses some power when you change colors: "You may have observed any number of swans and found no purple swan among them, but that will never entitle you to say that there are no purple swans." If any number of observations will never entitle me to say there are no purple swans, am I not likewise prohibited from denying the existence of unicorns or angels or of anything which anyone wishes to assert? Absence of contradictory evidence does not affirm the possible existence of purple swans .
POSITIVISM OR DEDUCTIVE TESTING?
In a letter to Tibor Machan, a prominent libertarian philosopher wrote: "Reynolds…knows nothing about epistemology, and hence seems to swallow the Friedmanite positivism, which holds, among other things, that sufficient research into economic statistics will give them a theory in economics, that they are verifying economic theory with their graphs and the like."
After conceding ignorance, I probably should define "positivism." Positivists stress the importance of sense data, generalize from particular observations (induction), and sometimes too-rigorously confine their definition of "meaningful" statements. Milton Friedman's hypothetico-deductive method, on the other hand, says "we cannot perceive 'facts' without a theory" and "the evidence for a hypothesis always consists of its repeated failure to be contradicted." "Factual evidence can never 'prove' a hypothesis," writes Friedman, "it can only fail to disprove it.…Observed facts are necessarily finite in number, possible hypotheses, infinite." 
Clearly, this does not say that research will yield a theory or even that graphs will verify a theory. It says, rather, that measurements presuppose theories, and facts can only contradict theories. Since the "positive" aspects of economics deal with many measurable quantities (prices and quantities, rather than "value" or "social cost"), this would seem to be an especially mild and acceptable variation of "positivism."
Most objections to Dr. Friedman's 1953 essay on method concern his treatment of the "realism" of assumptions. In order to be descriptively precise, Friedman said, a theorem would have to be swallowed-up in endless minute detail. But the usefulness of theories depends on separating the essential from the trivial, on explaining much by little. For this reason, Friedman suggests "the relevant question to ask about the 'assumptions' of a theory is not whether they are descriptively realistic, for they never are, but whether they are sufficiently good approximations for the purpose in hand. And this question can be answered only by seeing whether the theory works, which means whether it yields sufficiently accurate predictions." 
Many detractors have wrongly interpreted this Chicago methodology to imply that logical analysis of a theory is never appropriate. G.C. Archibald, for example, criticizing Friedman and Stigler's critique of "monopolistic competition," said "they have concentrated upon a priori discussion of his assumptions, instead of on discovering what facts were needed to give the theory content, and endeavoring to obtain them so they might test it." George Stigler correctly replied that "the possibility that a theory will yield useful predictions is reduced by logical weaknesses in its construction."  If the implications of a theory are ambiguous and/or self-contradictory, the theory cannot be disproven and is therefore empty. "Sensation is the test of truth," said Will Durant, "but reason is its discoverer."
PREFERENCE AND DESCRIPTION IN ECONOMIC DISCOURSE
Tibor Machan recently revived a favorite complaint of both Austrian and Chicago economists and turned it against them. Murray Rothbard's version of Machan's lament is that "economics…is filled with implicit moralizing."  Chicago patriarch Frank Knight completed the thesis (in 1940) when he wrote: "In so far as any treatment of economics makes explicit references to the merits of any social policy, some theory of value, beyond factual preference, is necessarily involved."  In line with this tradition, Machan begins by arguing that Ludwig von Mises must consider socialism ineffective because it doesn't satisfy human wants, and that this presumes an ethical position which should be made explicit, rather than denied.
Mises' reasoning could, however, easily be interpreted to mean only that socialism cannot achieve the results claimed for it by its advocates. This position would not require making any value judgments but only factual predictions: To the extent that socialists practice those distortions of the market which they advocate, they will confuse the measurements of scarcity and desire (prices and incomes) required to attain any goal without considerable waste. Trial and error price-administration, proposed by Lange as a solution to this dilemma, could achieve optimal trading efficiency only if the "socialist" state didn't influence the initial positions of the trading units or the terms of trade (i.e., no income redistribution, price control, taxes, or unmarketed capital and land) .
Machan next generalizes from this allegedly ethical basis of Mises' rejection of socialism and implies that no aspect of any economic question can be answered apart from ethical considerations (that is, no "value free" economics is conceptually possible). This is a strong suggestion and one which does not at all follow from the observation that moral judgments are often implicit in what are alleged to be purely factual or logical descriptions of reality.
Milton Friedman proposed clear separation between "positive economics," concerning what is, and "normative economics," concerning what ought to be. Whether or not the normative issues are scientifically defensible, as Machan and I believe, positive economics can and does provide "a system of generalizations that can be used to make correct predictions about the consequences of any change in circumstances." 
"Let us admit," says Machan, "that value theory as an enterprise is not in good shape. Neither, however is any theory (least of all logical positivism, the philosophical foundation of positive economics) which holds that judgments of value ('X is valuable') cannot be true or false."  But positive economics is only a method of acquiring predictive knowledge and is compatible with any and all schools of moral philosophy.
This independence of positive and normative issues seems tacitly acknowledged by Machan's remark that Mises' ethical subjectivism "has nothing to do with economics." Indeed, positive economics is the very enactment of Machan's prescription that social scientists should "stop acting as if they could incorporate into their own analyses ethical and political problems." 
Perhaps the best reconciliation we can derive from this detour is that economics has factual and ethical aspects and that both should be explored with all promising techniques of analysis and research.
SUMMARY AND CONCLUSION
I have tried to show that 1) the epistemologies of von Mises and Peikoff are polar opposites; 2) empirical confirmation of economic theorems is generally possible and desirable; 3) the scientific method is not embarrassed by objective ethics or the formal non-falsifiability of particular possibility assertions; 4) Friedman denies positivism's antitheoretical position and refutes inductions; and that 5) Chicago economists expressly acknowledge the need for reasoned ethical views but for views which are carefully distinguished from factual assertions.
An economist venturing out on the dry and dusty road of philosophy encounters many unusual characters. At one point he finds someone telling him not to think; later he comes upon another telling him that thought is everything, so he should close his eyes and stuff his ears. One group may tell him that whatever he wants is right, while still another claims that whatever they want him to do constitutes his moral obligation.
Without getting too deeply mired in this slush, I have hopefully provided enough words to show that the "Chicago" methodology is consistent with Objectivist ethics and epistemology, while von Mises' subjectivism and a priorism definitely is not .
Copyright 1972 by Alan Reynolds.
NOTES AND REFERENCES
 von Mises, Ludwig EPISTEMOLOGICAL PROBLEMS OF ECONOMICS (Van Nostrand, 1960), pp. 13, 14, and 27.
 Peikoff, L., quoted in A. Ellis IS OBJECTIVISM A RELIGION? (Stuart, 1968), pp. 236-37. Recall also Ayn Rand's famous "A is A" passage, where she says "abstractions are man's method of integrating his sensory material" (FOR THE NEW INTELLECTUAL, Signet, 1961, p. 22).
 Kirzner, I. "Divergent Approaches in Libertarian Economic Thought," INTERCOLLEGIATE REVIEW (January-February 1967). See also I. Kirzner THE ECONOMIC POINT OF VIEW (Van Nostrand, 1960), pp. 177-85.
 Knight, F. RISK, UNCERTAINTY AND PROFIT (Harper, 1965), p. 215: "It would be as ridiculous to suggest calculating from a priori principles the portion of buildings to be accidentally destroyed by fire in a given region and time as it would be to take statistics of the throws of dice." On probability, see H. Reichenbach, THE RISE OF SCIENTIFIC PHILOSOPHY (University of California, 1951), Ch. 14.
 von Mises, L. HUMAN ACTION (Van Nostrand, 1966), p. 64. On determinism, see J.W. Krutch THE MEASURE OF MAN (Grosset, 1953).
 Wilson, J. "The Relation Between Symbol and Content in Logic and Mathematics," INDIVIDUALIST (March 1971).
 Compare B. Blanshard REASON AND ANALYSIS (Open Court, 1962), pp. 124-25; J. Benardette "Sense Perception and The A Priori," MIND (April 1969); C.H. Whitely "Empirical, Analytic, Evaluative," ANALYSIS (March 1971).
 Olson, R. THE MORALITY OF SELF-INTEREST (Harbinger, 1965), Ch. 3; Rapoport, A. OPERATIONAL PHILOSOPHY (Science Edition, 1965), Ch. 10.
 Pollock, J. "The Structure of Epistemic Justification" in STUDIES IN THE THEORY OF KNOWLEDGE (Basil Blackwell, 1970), p. 78.
 Popper, K. THE LOGIC OF SCIENTIFIC DISCOVERY (Science Edition, 1961), pp. 15-56; Brunner, K. MONOGRAPH ON MONETARY THEORY (Chandler, forthcoming).
 Blanshard, op. cit., p. 228.
 Peikoff, L. "Certainty Without Omniscience" (lecture notes of J. Wollstein 1970), Sec. II B.
 Friedman, M. ESSAYS IN POSITIVE ECONOMICS (University of Chicago, 1966), pp. 9, 23, and 34. Also G. Stigler ESSAYS IN THE HISTORY OF ECONOMICS (University of Chicago, 1965), p. 28: "a good theory is at least as reliable as a report of the facts."
 Friedman, op. cit., p. 15.
 Archibald, G. "Chamberlain versus Chicago," REVIEW OF ECONOMIC STUDIES (October 1961). See Stigler and Friedman's replies in ibid. (February 1963).
 Rothbard, M. POWER AND MARKET (Institute for Humane Studies, 1970), p. 191; Machan, T. "Values and Economic Science," REASON (August 1971).
 Knight, F. ON THE HISTORY AND METHOD OF ECONOMICS (University of Chicago, 1963), pp. 169-70.
 Lange, O. ON THE ECONOMIC THEORY OF SOCIALISM (McGraw-Hill, 1964). For contrasting examples of intervention in practice, see A. Nove THE SOVIET ECONOMY (Praeger, 1961), Chs. 3, 4, 5, and 8; and A. Reynolds, "The Case Against Wage and Price Controls," NATIONAL REVIEW (September 1971).
 Friedman, op. cit. p. 4.
 Machan, T. "Has the Sceptic Won?" INDIVIDUALIST (May 1971), p. 17.
 For further comparisons of these "Divergent Approaches in Libertarian Economic Thought," see Kirzner, op. cit., and A. Reynolds, "The Purge of Chicago Economists," REASON (July 1971).
This article originally appeared in print under the headline "Economics, Ethics, and Epistemology".