Professor Barber – Physics Dept.

I have been brought up to think that theories are invented so that one can foretell what will happen. One invents them and reinvents them until they work successfully.

For example I find that when I stir a teacup or a bucket of water, and leave it, the leaves in the tea, or sand grains in the water, are left in a little heap in the middle on the bottom of the vessel.

My first theory says this is natural; it will always happen when the water rotates. So I make the situation a little different and hang the bucket on a string and make it spin. The sand grains move to the outside edge now. The same thing happens with tea leaves. This is not what my first theory predicted. The theory must be wrong.

My second theory is that the sand moves to the outside edge because it is denser than water; the bucket acts like a "centrifuge". But theories are of little use unless they predict successfully in new situations. I therefore reverse the argument and predict that if the particles were less dense than water they would move to the middle. So I sprinkle cork fragments on the water (they float of course, being less dense) and I spin the bucket. Nothing happens; the floating particles begin to go round as the water picks up the speed of the bucket but they show no inclination to move to the middle in preference to the outside.

Things are getting mysterious and I seem to be losing sight of my first question 'why did the sand gather in the middle?' But in research it does not pay, I think, to insist on answering the question one first thought of. The point of research, to my mind, is to succeed in making sense of something. It doesn't matter quite where one begins. The point is to succeed in making sense of something, then with luck the area of enlightenment will spread.

So back to the problem. The sand gathers in the middle only when the water is spinning but the bucket is not. This is a third theory. Let's invert it and predict that if the bucket is spinning and the water is not, then the sand should move to the outside. This is a good prediction because we have already seen it happen. The sand moves to the outside as soon as one rotates the bucket, long before the water has had time to pick up the spin.

So let's push the theory a little further. One should never leave a theory alone. One should always keep on pushing a theory until it breaks down (or until one gets tired of pushing); because a theory that finally breaks down points the way to an even better theory.

So think of some further twist. Perhaps if the sand were lighter than water it would move in the reverse way. But we have tried cork dust and nothing happened. Perhaps this was because the water surface became curved. When the water spins it climbs up the side of the bucket. Incidentally this might page 5 explain the sand motion at the bottom, for if the water moved outward as it began to pick up the spin it might carry the sand with it, and then as it lost its spin and the surface became flat again it could carry the sand back.

So let's stop the water surface from becoming curved by filling the bucket to the top and putting a glass lid on it. I prefer to do this with a closed jam jar on a gramophone turntable. Now predict what will happen to the sand in the jam jar. I think one should always predict what will happen before one does an experiment. This helps one to realise whether the theory is working or not. It keeps one's mind on the job.

The last theory suggests that if the water surface is kept flat, the sand should not move. It's wrong. The sand moves to the outside when the jar starts turning, and then if one stops the jar and lets the water spin inside it, the sand gathers in the middle. The curve on the surface is not the cause of the motion.

So let's go back to the theory about the relative motion of the jar and the water. This says nothing about the density of the sand. Try cork fragments. There is a glass top to the jar now and the cork fragments float against it. Toss a coin and say perhaps that they will behave like the heavy sand. This proves true when you make the experiment. They move out at first, then gather in the middle after you have stopped the jar turning.

It seems that the relative density of the fragments and the water is unimportant. I won't go further, but merely say that I think that the relative rotation of the jar and the water inside it, sets up circulating currents that waft the fragments about in the way we see.

My point in saying all this is that I think that theories are made to allow us to predict what will happen. We alter them until they do. But if one cannot use a theory to predict a behaviour, there seems little use in the theory, and until it can be shown to give successful predictions in new situations it is not to be taken seriously.

One last point is that, in physics at least, one can usually invent a variety of theories and that these must be assessed by seeing whether or not their predictions are true. Occasionally one has different theories that are equally successful in prediction and so cannot be distinguished by experiment. In that case one always choses the one that works out most simply. I will give an illustration.

My friend and I live three miles apart and I set out one day at 9.00 by my clock at home and reach his house when his clock says 9 30. I leave when his clock says 10.30 and reach home at 12.00 by my clock. This seems wrong, since the first journey takes half an hour and the other takes one and a half hours. It might happen day after day till it occurred to me that things would make more sense if my friend put his clock forward half an hour. If he did so I should arrive at 10.00 and leave at 11.00, having taken an hour each way. This would seem better.

But if my friend were obstinate he could maintain that both clocks were perfectly good clocks and showed the right time. If I carried my own clock with me and showed him that it read 10.00 on arrival he might, if sufficiently obstinate, point out to me a new law of nature, that clocks always gain half an hour when carried from my house to his, and conversely lose half an hour when carried the other way. This might strike me as a wrong headed view, but I could live with it if he insisted, and could make all the necessary additions or subtractions to keep appointments with him. But sensible people would arrange to alter one of the clocks by half an hour. If my clock is right, then the right way to adjust his clock is the way that makes for plain simple sense. It doesn't mean (to me) that it is the Right way with a capital R but it is right in the sense that it makes for the simplest arguments. So in physics we choose the simplest way to look at things and call it "right".

We say that the earth moves round the sun because this theory allows us to invent simple universal rules in physics. A person might maintain otherwise if he did not mind his physics becoming extremely complicated. But it would be "wrong" because it was more complicated than it need be.

Mr Mulgan – Political Science Dept.

One way of finding out what the scientific method is might be to examine what the scientist actually does. But though this approach would work reasonably well in the case if the natural sciences it would not be suitable for the social sciences, especially political science, because it begs what is usually considered to be the vital question, i.e. to what extent is the method of political 'science' really 'scientific'? In other words, it is the natural scientist who seems to set the pace while the social scientist is then judged on how successfully he keeps up (or back?) with him. We must therefore start with some motion of what science is before we can assess the role of scientific method in political science. It would be disingenuous to try to give an account of the attitude of political scientists in general to 'science' and the 'scientific method' for the simple reason that there is no such general attitude. What follows will therefore be very much a personal view.

Scientific knowledge must first of all, I assume, be 'objective'. It must be based on evidence which can be tested according to generally accepted standards. 'Mr Holyoake is the Prime Minister of New Zealand' is an objective statement because there are generally accepted ways of proving or disproving it. 'Mr. Holyoake is a political wizard' is more suspect because there may not be clear agreed criteria for deciding the truth of this statement in which case it becomes 'unscientific'. Science obviously involves more than such propositions about isolated particulars. It aims to establish generalised knowledge about classes of event or object and, where possible, to construct scientific theories which will explain why particular phenomena occur by fitting them into a general pattern. Ideally, a scientific theory gives us a set of necessary and sufficient conditions for the occurrence of a particular type of result (if and only if What will happen (given conditions a, b, c, … then result x will follow). It is worth noting that obtaining predictions need not be the main purpose of such a theory. We may want to use a scientific theory to explain past events which we already know to have happened. The point is that if we have given such a scientific explanation of such a past occurrence, it follows that, with the aid of the knowledge we now have, we could have then predicted the result.

Not all scientific theories, however, need be of this type. There are, we are told, some areas of the natural sciences where it has proved so far impossible to construct theories which, even with the help of an 'other things being equal' clause, will fit all observed instances. The best that can be done is a statement of the probability with which a particular event is likely to occur. (Given conditions a, b, c. … there is a particular probability that x will occur.) Though such propositions may seem less satisfactory than the 'pure' form of scientific theory, if they are constructed reliably and objectively there is no reason other than sheer dogmatism, for saying that they are 'unscientific'.

The 'scientific method', then, is the method of arriving at scientific knowledge, involving such procedures as the construction of possible theories (hypotheses) and testing them objectively. What part does this method play in the study of politics?

We will begin with the question of evidence. Difficulties arise even with the initial classification of the subject matter. Categories such as Labour party member, voter, MP and so on cause no bother. They are clearly objective because there are precise ways of finding out who belongs in any of these categories. But we may want to use categories like 'Communist sympathiser' or 'authoritarian' where more caution will obviously be needed. If anyone uses such terms in the analysis of politics we will expect him to give explicit criteria which do not depend merely on the whims or intuitions of the writer so that the terms can be applied in the same way by other people and the results of the observations based upon them objectively tested. Such criteria might be membership of certain organisations or the performance of certain types of action. But often we will want to go further than this and take into account attitudes or opinions which can only be discovered by questioning the individuals concerned. Here the difficulties of being objective are greatly increased.

In fact, the political scientist in his desire to be objective and scientific is commonly faced with a dilemma. If he wishes to examine, say the extent of people's interest in politics, one way of doing this might be to stick to clearly observable characteristics such as participation in elections, subscribing membership of political parties, etc. The trouble is that he will then be accused of using crude and misleading indicators of what we really mean by 'political interest'. If, on the other hand, he starts to examine opinions and attitudes, the objectivity of his results will be questioned.

But we should not overemphasise these difficulties. Human beings are it is true, particularly complex objects to classify and describe—they are capable, for example, of deliberately misleading the observer. But this only makes scientific observation especially difficult; it does not necessarily make it impossible. Social scientists are becoming increasingly aware of the pitfalls in their subject and have evolved techniques for minimising them. Unless we adopt the implausible and philosophically dubious scepticism of those who say that it is impossible ever to know the attitudes or opinions of others, there seems to be no reason why such problems cannot be overcome. (That the difficulties are merely minimised and not removed altogether should not be taken as an objection to the claims of the social scientist to be scientific. All observation, of non-human as well as human phenomena necessarily involves a certain amount of unreliability and uncertainty.)

Scepticism is perhaps more justifiable when political scientists go beyond mere scientific classification and description and attempt to construct scientific theories about politics. Some people hold that it is in fact impossible to provide a scientific explanation of human behaviour because human being possess free will which means that their actions are outside the realm of cause and effect. What 'free will' means, whether or not we have it, and if we do have it, whether or not it is incompatible with 'determinism' are highly intricate and hotly disputed philosophical problems. At any rate, whether or not it is in principle possible to explain human action in terms of a scientific theory, there is no doubt that no-one has yet managed to do this with political action. Not that there is any lack of suggested theories. Quite the reverse. There are a large number of them but no theory has yet been found which is capable of explaining even a small segment of political life.

We should, however, be careful not to exaggerate, as some social scientists do, the accuracy of scientific explanation and prediction as found in the natural sciences, Scientific explanations cannot take into account all the different factors which may vary from one situation to another and they will always contain, explicitly or implicitly, an 'other things being equal' clause. The difference between the natural and the social sciences seems to be one of degree. In the former a considerable degree of accuracy can usually (but not always—vide meteorology) be achieved by ignoring all but a few variables, whereas in the latter 'other things' are usually so unequal, the excluded variables so significant, that the particular explanatory theory is very likely to be highly inaccurate. Similarly, when political scientists try to establish 'probability theories of political behaviour, their results are comparatively less accurate because of the complexities of the subject matter.

Most political scientists take this relative lack of success simply as a spur to try harder in applying scientific method to politics. There is still much that can be done in the often boring but useful work of testing hypotheses gathering accurate and reliable information. The difficulty in achieving any adequate body of explanatory theory may, some argue, be due to trying to answer the wrong sorts of questions about politics. Questions such as 'what are the conditions of political stability?' or 'why did Hitler come to power?', which have been the traditional concern of political scientists are too complicated. Though they may one day be explicable in scientific terms, scientific progress is more likely to be made by looking at simpler and more manageable problems, For example, some political scientists are now concentrating on the politics of small groups, local communities, small towns, etc., where it is thought, one is most likely to find those conditions, a combination of a small number of significant variables and a large number of particular examples, which have proved so helpful in the advance of the natural sciences.

But not everyone who is engaged in the academic study of politics would agree that the questions should be tailored to fit the method. For them, the method must suit the questions. Because science has not yet provided answers to the traditional questions, it does not follow that new questions must be put. Rather, these questions are, it is felt, so interesting and important that they must be answered with the best means available even if this involves using other methods besides the strictly scientific. Not that this necessarily involves a complete rejection of scientific method. Scientific method is to be used as far as it will go but other methods will be employed to supplement it. For example, in the case of evidence, objective evidence is valuable when it can be acquired but where it is not available, such political scientists will not feel impeded from making an assessment of, say, the motives of politicians or the subjects discussed in an unreported meeting.

Similarly, in the case of explaining politics, the political scientist will see what 'insights' are provided by the various 'models' that have been suggested. These vogue words, 'model' and 'insight', reveal the limited use that is being made of the scientific method. A 'model' is like a scientific theory except that it is explicitly defined as an abstract construction, that is it does not pretend to make any claims about what phenomena are actually like or how they actually work. This is not to say that the social scientist does not build his models with an eye to applying them to the real world. The point is that by explicitly defining the model as abstract, he removes the possibility that it may be invalidated by the discovery of counter-examples. If it is to be a 'useful' model it will certainly accord with reality to a certain extent and the more it accords the more useful it becomes, but it is not disproved as a scientific theory proper would bo as soon as it is falsified. The cynic might say that the model is a device by which the social scientist may indulge his yearning for the trappings of science—tables, formulae, mathematical elegance—without having to worry about the truth of what he is saying. Building castles is such fun that it does not seem to matter whether they are in the air or on the ground.

But as long as the model is tested as rigorously as possible and its limitations, the points at which it does not square with the evidence, are fully recognised, it can by an invaluable aid to understanding politics by giving us certain 'insights' or clues towards understanding a particular phenomenon. But since, for example, no single model of federalism is likely to provide the whole answer to the question why the Central African Federation broke up, the political scientist will often find the most satisfactory explanation to be one which combines a number of different 'insights' drawn from different, incompatible models.

Once he does this, he stops using scientific method and starts to behave in the same way as a historian who, if asked for the explanation of a complex event, say, the causes of war, will produce a number of different factors—political, economic, social, psychological and so on. None of these factors individually provides an adequate explanation and an important part of the hostorian's craft involves assessing the different weight to be given to these different factors. The word 'craft' is important. This exercise of assessment is a matter of individual judgment and is unscientific in the sense that it is not necessarily susceptible of objective proof or disproof. Of course, some such explanations may be proved wrong if they are based on false evidence but not all disagreements will be of this type where there are agreed criteria for settling them. This does not mean, as is sometimes said, that such judgments are completely 'arbitrary' as if they involved a completely random decision such as could be made by tossing a coin or drawing lots. All of us, in our everyday lives, continually make such judgments or assessments of people or situations and we recognise that they involve skill, experience and perspicacity.

True, such intellectual virtues are not susceptible of the same rigorous tests as scientific statements are but they are not for that reason therefore worthless. In many areas of our lives they are all we have to go on. If they are the best means at present available for answering certain important and interesting questions about politics then they should be used. Many political scientist, including the present writer, would even make a virtue out of necessity and value their subject as one which combines rigour of a science with the judgment more characteristic of the arts. Meanwhile, the efforts of the political scientists proper, as they would, not unjustly, describe themselves, are watched with interest. That their researches will add to our knowledge is beyond dispute. But whether all important questions about the workings of politics will ever be able to be answered satisfactorily solely by means of the scientific method is still very much an open question.