In Sir Karl Popper's work on the methodology of social sciences, he tries to provide a clear criterion for demarcation between science and non-science, and thus, came up with falsificationism.
Popper believes that his falsificationism provides a clear methodological and philosophical distinction between theories of a scientific nature and those which are not. However, since its conception, falsificationism had come under a lot of criticism.In this essay, I will demonstrate that falsificationism in its strict Popperian sense is an invalid and inappropriate methodology for science, and hence Economics, and examine its problems, yet, the more relaxed general notion of falsification is one that is reconcilable with Economics. Firstly, we must examine falsificationism. If we were to be a hard-core Popperian, and take falsificationism to its utmost seriousness, we arrive at logical falsification.
This can be explained as scientific theories consisting of basic statements with easily observable empirical properties.These basic statements are falsifiable if and only if it is inconsistent with some set of finite basic statements. Thus, A is only falsifiable if, for instance, it is inconsistent in a set {A, B, C, D, E}. However, if there are no sets of basic statement where A is inconsistent in, then A is not falsifiable. Thus, considering only this logical property, we can now define theories as theories which contain basic statements which are falsifiable. So why then, is falsifiability a valid criterion of science? The whole concept of falsificationism stems from Hume's problem of induction as means of achieving knowledge.
Hume thought that the way in which we obtain knowledge is by way of inducing empirical experiences into universal laws and generalisation, such as when one sees swans, and observes that they are all white, and then from the basic statements that all swans observed are white induce a universal statement that all swans are white. This according to Hume is fallacious since just by observing a finite set of instances, it is wrong to induce from that set something that is greater from it, namely the universal generalisation.Thus, to solve this plaguing problem in scientific method and epistemology, Popper resorts to falsificationism and presents the asymmetry between falsifying statements and verifying them. According to Popper, one can never verify a theory as it faces the problem of induction, but it is possible to falsify them by finding a set of basic statements where the theory in test is not consistent with.
Notice the difficulty or - in fact, impossibility - of forming a valid deductive argument supporting basic statements, but it is possible to form deductive arguments against a theory:There exists theory A Theory A is not consistent with the set {A, B, C} Therefore, theory A is false This seems to be a valid deductive argument However, in a logical standpoint, this is also problematic. Probabilistic claims cannot be falsify, since although an incidence has a 0. 0000001% probability of occurring, it can occur, and if it does not occur, under given finite observations, it is still not satisfied, for example. Much of science, and especially economics, is subjected to this kind of conjecture. Furthermore, it is as hard to falsify a basic statement by itself as it is to verify it.Consider the basic statement that if A exerts a force on B, B will exert an equal and opposite force on A.
For if we were to falsify the statement by itself it is next to impossible because we have to have some other non-basic statements, such as various 'background knowledge'. If this is true, then, it offers no solution to problem induction since there would be no asymmetry between verification and falsification, since the two are equally hard. Thus, Popper acknowledges this and proposes that theories be tested as a system of theories consisting of various auxiliary assumptions and background knowledge and different sets of theories.Granted, logical falsifiability seems to be a valid criterion, since if a basic statement in a scientific theory is ambiguous as to its truth and falsity, it will yield no knowledge at all, however, in this sense, logical falsifiability is in itself rather trivial, since all statements that are meaningful and scientific should be one way or another falsifiable. So, we relax falsificationism from its pure logical framework and examine whether systems of science are falsifiable or not in practice, and this more pragmatic and methodological usage of falsificationism is perhaps intended by Popper.Popper believes that one should first consider only theories that are testable.
Testable theories are considered as fit for further critical examination, not as 'true' theories to be later falsified. The more testable and more falsifiable theories are ones which are more preferred. Then, scientists should seek to try to falsify the theories, not to support them. Theories that are falsified should be dropped, or form further assumptions that can be further tested, not ad hoc assumptions that do not increase their empirical power.Lastly, theories that are not yet falsified should be accepted as conjectural truths, not certain truths, as it is possible that it may be some day falsified.
Notice the emphasis on falsification and critical acceptance of theories, considering them as testworthy, not as truth. When undergone tests, theories are not supported or confirmed, but just more testworthy. They can, however, gain more empirical power by being able to give more accurate predictions, as Lakatos - one who follows the Popper tradition - have pointed out with theories gaining 'novel facts' after being able to withstand tests.In a methodological sense, this idea of theories not being supported, but always being taken as conjectural truths, is perhaps an overly cynical stance, as Hausman suggests in An Appraisal of Popperian Methodology.
To say that one cannot seek support for a theory and anyone who occupies himself seeking support is being dogmatic rather than critical (as a scientist should be), is not valid, since although it is true that bad support can be more easily found, it cannot be said that good support is of no merit whatsoever.For example, the moon landing is a good support of many theories such as there are no aliens on the moon, the gravity on the moon is less than that on earth, and while these observations do not increase empirical content of a hypothesis by way of falsificationism, it certainly supports the hypothesis, although not from a purely logical point of view, but from a pragmatic one.However, the more serious problem arises in Durheim-Quine thesis. This states that testing hypotheses in a complete scientific system, including all assumptions, theories, background assumptions etc.
is impossible, since it is not possible to state which hypothesis is actually being tested as testing that hypothesis also involves all of its other assumptions and theories. Popperian and his supporters have tried to give a solution to this, by examining various instances of the experiment E under other assumptions A: {E, A}, and if at least one of {E', A}, {E'', A} ...
{En, A} is falsified while {E, A}, {E, A''} ... {E, An} is not falsified, then, it can be said that it is E that is false, not assumptions A.However, this clearly under the same problem of induction. This is especially troublesome in economics since economic phenomenons are much more complex than in the natural sciences, with much more causal factors at work.
Even the simplest consumer choice model when applied to real life can face major problems, for example, counteracting causes resulting in agent's intransitive preferences. Another problem is how Popper arbitrarily takes for granted some background knowledge, that although not considered to be true.Because only the whole system is testable, it is important to consider which and which part of the system is more true relative to the more falsifiable part of the theoretical system, and take them as truths and carrying out tests on the theory in question. But this sort of easy decision seems very subjective and non-characteristic for scientific method. Even more enlightening is that if we in allow background assumptions to be truths, we can actually verify theories by fitting this into the covering law or the deductive nomological model (D-N).
By stating general laws that we take for granted, and initial conditions, we can derive a conclusion without the problem of induction! This, again, proves Popper's asymmetry thesis invalid since in this case, both validation and falsification is possible. Hence, economic theories can be better put into this covering law framework without being falsifiable, and still remain science, for example, neoclassical economics being based on the rationality assumption. If we are to practically apply falsificationism to economic, there are many problems.First, it is impossible for much, if not all, of economic theories to be tested in a controlled situation.
This also makes prediction a hard thing to do, and social scientists end up explaining past phenomena. Doing this is clearly not falsifiable, since theories will have endless 'immunising stratagems', or assumptions that are made up to protect the theory against falsification. Such is the same with making predictions, since economic generally makes wrong predictions, but those are blamed on other factors, be it the background knowledge or other external causes.Yet, if we were to fit the theories into the D-N model, we do not have to face this problem, especially if we make the outcome probabilistic, as many of economic phenomenons are, such as econometrics which is a very formalised branch of economic. Also, if we look at economic advances individually, as Blaug (1994) did, we can see that advancees in economic theories have in fact lead to greater predictive ability and greater empirical content after successive tests.
This should at least give some qualification to the scientific status of economics, for it can be tested, albeit with some difficulty, and after under going tested, has gained greater empirical content. In conclusion, economic is indeed science, although it has some problems with the strict falsificationism methodology, but as a more trivial and pragmatic falsificationism as described above, economics, like other forms of science, is able to reconcile itself with falsificationism, which in the end equates to forming clear testable hypothesis and testing them with rigour and critical approach.
