Of course, we must avoid postulating a new element for each new phenomenon. But an equally serious mistake is to admit into the theory only those elements which can now be observed. For the purpose of a theory is not only to correlate the results of observations that we already know how to make, but also to suggest the need for new kinds of observations and to predict their results. In fact, the better a theory is able to suggest the need for new kinds of observations and to predict their results correctly, the more confidence we have that this theory is likely to be good representation of the actual properties of matter and not simply an empirical system especially chosen in such a way as to correlate a group of already known facts.

But a theory is not like an airline or bus timetable. We are not interested simply in the accuracy of its predictions. A theory also serves as a base for thinking. It helps us to understand what is going on by enabling us to organize our thoughts. Faced with a choice between a theory which predicts well but gives us little insight into how the system works and one which gives us this insight but predicts badly, I would choose the latter, and I am inclined to think that most economists would do the same.

To say that prediction is the purpose of a scientific theory is to confuse means with ends. It is like saying that the purpose of a spaceship is to burn fuel. … Passing experimental tests is only one of many things a theory has to do to achieve the real purpose of science, which is to explain the world.

The hallmark of a successful theory is that it predicts correctly facts that were not known when the theory was presented, or, better still, which were then known incorrectly. A good theory should have at least two characteristics: it should be in sharp contrast to at least one alternative idea and it should make predictions which are testable.

A theory is a good theory if it satisfies two requirements. It must accurately describe a large class of observations on the basis of a model that contains only a few arbitrary elements, and it must make definite predictions about the results of future observations.

A theory, to scientists, means something rather different from its popular use, which suggests something speculative or untested. A scientific theory is a cohesive body of knowledge, an explanation that is consistent among a range of cases and can allow you to predict what might happen in unknown situations.

The purpose of theory is to clarify the world in which we live, how it works, why things happen as they do. The purpose of theory is understanding. Understanding is energizing. It energizes to action. When theory becomes an impediment to action, it is time to discard the theory and return naked, that is, without theory, to the world of reality. People become slaves to theory because people are used to meeting expectations they have not originated—to doing what they are told, to having everything mapped out, to having reality prepackaged. People can have an antiauthoritarian intention and yet function in a way totally consonant with the demands of authority. The deepest struggle is to root out of us and the institutions in which we participate the requirement that we slavishly conform. But an adherence to ideology, to any ideology, can give us the grand illusion of freedom when in fact we are being manipulated and used by those whom the theory serves. The struggle for freedom has to be a struggle toward integrity defined in every possible sphere of reality—sexual integrity, economic integrity, psychological integrity, integrity of expression, integrity of faith and loyalty and heart. Anything that shortcuts us away from viewing integrity as an essential goal or anything that diverts our attention from integrity as a revolutionary value serves only to reinforce the authoritarian values of the world in which we live.

The user of a theory has a simpler task: his not to reason why, his but to sigh and try. If the right element in the diverse situations has been isolated, the theory will work: it will yield predictions better than those which can be reached with any alternative theory.

When the theory performs well you also think, “Are there parallel results in naturally occurring field data?” You look for coherence across different data sets because theories are not specific to particular data sources. Such extensions are important because theories often make specific assumptions about information and institutions which can be controlled in the laboratory, but which may not accurately represent field data generating situations. Testing theories on the domain of their assumptions is sterile unless it is part of a research program concerned with extending the domain of applications of theory to field environments

No theory is good unless it permits, not rest, but the greatest work. No theory is good except on condition that one use it to go on beyond.

A theory is something nobody believes, except the person who made it. An experiment is something everybody believes, except the person who made it.

I'm an experimentalist. I like to look at hard data. And there — string theory is a distant second to because it doesn't make any observable predictions. ... As an experimentalist, I like when a theory makes a prediction. My job is to not prove theorists right — it's to disprove everybody else.

A scientific theory is a concise and coherent set of concepts, claims, and laws (frequently expressed mathematically) that can be used to precisely and accurately explain and predict natural phenomena.
A theory should include a mechanism that explains how its concepts, claims, and laws arise from lower-level theories.

It's hard to observe without imposing a theory to explain what we're seeing, but the trouble with theories, as Einstein said, is that they explain not only what is observed but what CAN BE observed. We start to build expectations based on our theories. And often those expectations get in the way.