Arguments are built on premises that are believed, or assumed, to be true. Some premises we establish by deductive arguments that have preceded, but very many of the premises on which we must rely cannot be established by deduction. Our reasoning process usually begins with the accepted truth of some matters of fact, in David Humes phrase. To establish matters of fact we must rely on reasoning that is inductive. Induction thus provides the starting pointsthe foundationfor the rea- soning that concerns us most. We reason to establish truths in our everyday lives, to learn facts about our society, to understand the natural world. Deduction is certainly powerful in enabling us to move from known (or assumed) propositions to other propositions that those premises entail, but in the search for truths with which our reasoning must begin, it is insufficient. The inductive arguments with which we establish matters of fact differ fundamentally from the deductive arguments that were the concern in Part II of this book. One essential contrast between the two families of argument (noted much earlier in our discussion of basic logical concepts, Section 1.5) lies in the relation of the premises to the conclusion in the arguments of the two great families. In deductive arguments, the claim is made that conclusions follow with certainty from their premises. That claim is appropriate because any deductive argument, if it is good, brings to light in its conclusion what was already buried in its premises. The relation between premises and conclusion, in deduction, is one of logical necessity. In every deductive argument, if it is valid and if its premises are true, its conclusion must be true. In inductive argumentsthe concerns of this chapter and those that follow the relations between premises and conclusion are not those of logical neces- sity. The claim of certainty is not made. The terms valid and invalid simply do not apply. This does not mean that inductive arguments are always weak; sometimes they are very strong indeed, and fully deserve our confidence. Anal ogi cal Reasoni ng 11.1 Induction and Deduction Revisited 11.2 Argument by Analogy 11.3 Appraising Analogical Arguments 11.4 Refutation by Logical Analogy 11 M11_COPI1396_13_SE_C11.QXD 10/22/07 9:21 AM Page 482 Scientists now assert without reservation (for example) that smoking is a cause of cancer. This is true, but it is a truth that cannot be known with the demon- strative certainty of a valid syllogism. If p or q is true, and not p is true, we may conclude that q must be the case, beyond all doubt. It is a truth we establish as an inescapable consequence of the relations of the concepts involved. Empirical truthsabout the consequences of smoking, or the causes of cancer, and all oth- ers of that sortcannot satisfy the standard of deductive certainty. By that stan- dard, as one distinguished medical investigator observes, no one will ever be able to prove that smoking causes cancer, or that anything causes anything. 1 In the realm of induction, as we seek new knowledge of facts about the world, nothing is beyond all doubt. We must rely on arguments that support their conclusions only as probable, or probably true. Some such arguments are of only moderate worth; others are very powerful, as we shall see. The strengths and weaknesses of inductive arguments, and the techniques for the evaluation of such arguments, are the focus of Part III of this book. Arguments grounded on analogies, aiming to establish particular conclu- sions, are examined first, in this chapter. Arguments that go beyond particulars, aiming to establish generally applicable causal laws, are examined in the fol- lowing chapter. The uses of hypotheses and their confirmation in developing scientific theories follows in Chapter 13; and we conclude, in Chapter 14, with an analysis of the concept of probability itself, the conceptual instrument with which inductive conclusions are commonly expressed. 11.2 Argument by Analogy The most common type of inductive argument relies on analogy. If I report that I got very good service from a computer of a certain make and model, you may infer that a new computer of the same make and model will serve you well. That conclusion has some degree of probability, but the argument is far from compelling. When a new book is called to my attention and I infer that I will enjoy reading it because I have read and enjoyed other books by the same author, I may have my confidence in that author strengthened when I read the bookor I may be disappointed. Analogy is the common ground of our every day inferences from past experience to what the future will hold. Here follow two more carefully formulated analogical arguments. The first concludes, on the basis of what we commonly think to be prudent and fair, that it would be prudent and fair to adopt now a major change in public policy: Some people look on preemployment testing of teachers as unfaira kind of double jeopardy. Teachers are already college graduates, they say. Why should they be tested? Thats easy. Lawyers are college graduates and gradu- ates of professional school, too, but they have to take a bar exam. And a number 11.2 Argument by Analogy 483 M11_COPI1396_13_SE_C11.QXD 10/22/07 9:21 AM Page 483