Responsible
Thinking:
Principles

Back

Back to main page

Back to home page


Printer friendly

Testability

An important property of any statement or claim is whether it is testable. Are there means by which we could decide whether this statement is true? What would we expect to observe if the statement is true? What would we expect not to observe? Most everyday statements are easily tested. If someone says it is raining, this can be tested by looking out the window. We would expect to see droplets falling through the air and puddles on the ground with ripples from the falling drops. We would not expect to see sunshine and dry pavement. Other statements are much harder to test, like "The universe is expanding." Scientists do have ways of testing this, however, mainly by observing a color shift in the light from stars that occurs if they are moving away from us at a high rate of speed.

If a statement is completely untestable, it implies that the statement has no consequences. There is no difference between the statement being true and being false. It is reasonable to consider such a statement to be meaningless. If I claimed I was Julius Caesar in a previous life, but I couldn't remember anything that Caesar knew, nor in any significant way did I have characteristics the same as his, my claim would be completely untestable. It would also be safe to regard it as meaningless.

There are two problems with untestable claims. One is that there is no reason to care about the truth of the claim if there are, in fact, no consequences. Who cares if I was Julius Caesar in a previous life if nobody is affected in any way by that fact? The second problem is that the person making the claim could have no way of knowing it was true if it is untestable. How exactly did I determine that I was Caesar in a previous life if in fact that doesn't have any detectable consequences?

The main danger from untestable statements is that we may mistakenly judge the originator as having some deep insight. Fortune tellers and others who claim to have mystical knowledge will often make claims that are either completely untestable or are extremely difficult to test. They can do this without having to worry about being shown to be wrong, since it is nearly impossible to gather any evidence to refute their claims.

Often the reason something is untestable is vagueness. When a statement is sufficiently vague, it is effectively untestable. Suppose you are told you will have bad luck. Does that mean the next time you get to a traffic signal it will be red instead of green, or might it mean you will die a horrible death within the next week? Will this bad luck occur within the next five minutes or the next twenty years? The chances are that unless a person's luck is consistently very bad or very good, we would be hard pressed to say whether this statement was confirmed or not. It is easy to imagine that someone could find excuses to believe they had bad luck even though their luck on the average was no different from anyone else's.

Sometimes the situation is even more severe than mere vagueness - even in the most extreme situations we can neither confirm nor deny the truth of some statements. Consider "We are one with the universe" or "People have a natural right to own property." These statements lack any observable consequences. I do not know of any way I could present evidence to either confirm or refute either of these statements. Although both of these statements carry with them a sense of importance, they are totally untestable. Rather than worrying about whether they are true or false, we might do best to regard them as meaningless until such time as someone clarifies them in such a way as to imply some consequences.

One common category of assertion that is effectively untestable is those that depend on definitions. We frequently hear the claim that "abortion is murder." The truth of this assertion depends on the definition of "murder". Often we will judge definitions by what is in a dictionary or what is common usage. This does not seem adequate in this case, since the implication of the statement is that abortion is morally wrong because murder is morally wrong. Most people would agree that what is morally right or wrong should not depend on either the whims of dictionary writers or the results of public opinion polls about word usage. "Abortion is murder" could legitimately be part of a larger argument in which the author provides a definition of murder and perhaps relates it to a basis for moral beliefs, but as it stands, it cannot be shown to be true or false. The same problem occurs in other cases when the truth of statements is essentially dependent on the definition of a word that is not universally agreed on, such as "alcoholism is a disease" or "prostitution is a form of slavery". People who agree on what alcoholism is and what prostitution is would not necessarily agree on whether they are included in the definition of "disease" or "slavery" nor is there a practical test to determine who is right. We could get farther by asking what policies are promoted by those making such statements and whether we think these policies would be beneficial. Definitions are based on human usage, not on any independently testable property of the universe.

An aspect of testability is falsifiability. A reputed prophet Nostradamus made many vague claims, and with the right interpretations of some of his metaphors, some of the claims could be said to have come true. Over time we might note that Nostradamus has had several successes and no failures, so we might feel he has a good track record and should be taken seriously. Suppose one of his claims was that "New City" will be destroyed in an earthquake. Even though we assume that "New City" is not the real name of the city, and we don't know when the earthquake was supposed to occur, we might agree that this hasn't happened. On the other hand, we can't count this as a failure, since it might occur tomorrow or next year or some other time in the future. In fact it is impossible for this statement to fail - not because Nostradamus is so gifted, but because the prediction is stated in such a way that failure is impossible. Similarly, if all a prophet's statements are failure-proof, the prophet is safe from contradiction, whether or not he or she knows anything at all.

A common criterion for whether a claim has scientific value is whether it is falsifiable. If the statement is false it ought to be possible for some evidence to show it is false. If, no matter what happens, the statement would not be shown false, then, even if it is true, anything could happen.  Nothing is ruled out.  Therefore the statement is not useful for science.

Another example of a statement that isn't falsifiable would be a claim that a tall man will have a positive influence on my life. If a tall man hires me for a job, I might be impressed that the prediction was true, but if not, how do I know that a tall man might not have influenced me in some other way, for example by previously turning down the job I was just offered? The truth of a statement means it will not be found false, but if it could never be found false anyway, such "truth" doesn't count for anything. Looking for falsifiability helps prevent us from being fooled by statements that really don't tell us anything worthwhile.

We have to be careful in situations where something is very difficult to test or falsify. Some scientific claims can be very difficult to test, but have definite implications if we could make the right kind of observations. The temperature inside of stars, for example, cannot be measured directly but there are certain implications such as what atomic reactions take place or how dense the star is. In science as well as prophesy we should not credit a person with being right until their idea has been successfully tested. If we can see the claim could never be tested or falsified, we should not attach importance to it.