1. Why are there so many similarities between analogical arguments and inductive generalizations, despite the big differences between them?
2. Why do these methods for testing inductive arguments look so completely different from the methods for testing deductive arguments?

1. Why are there so many similarities between analogical arguments and inductive generalizations, despite the big differences between them?

The question is how much the differences amount to. Certain concepts that we use in describing inductive arguments put the wrong spin on what is, from a theoretical point of view, a matter of emphasis or approach. You use the same processes of collecting and evaluating evidence in both cases.

The misleading concepts are "general" and "specific." At most, the distinction between specific and general lets us divide inductive arguments into two convenient categories: those that use observations to reach a conclusion about one item (or a few) and those that use them to establish a claim about a large number of items. These are, of course, analogical arguments and inductive generalizations, respectively.

The difference in the types of conclusions is the main feature that differentiates analogical arguments from inductive generalizations, so we should not be surprised to find the two kinds of inductive argument so similar to each other. Moreover, you may notice how easy it is to describe any analogical argument as a special case of an inductive generalization. Suppose that a number of things have properties a, b, and c--call them a+b+c things--and that all the a+b+c things you look at also have property d. You conclude (with more or less confidence, depending on sample size and so on) that "All a+b+c things are d." Now say you're looking at one specific a+b+c thing, X. You can further conclude, "X is d." This conclusion to an analogical argument also can be read as a consequence of the conclusion to your inductive generalization.

2. Why do these methods for testing inductive arguments look so completely different from the methods for testing deductive arguments?

If you had to distinguish quickly between deductive and inductive arguments, it would be hard to improve on David Hume's observation about the repetition of relevant experiences. His observation is this: If you apply logical principles to a deductive argument even once and see its validity, you will not become more certain of your result by repeating the test. Maybe repeat once to check your work; then you have convinced yourself of the argument's validity. Repetition is irrelevant to logical truth and is positively rejected by logical method.

But inductive arguments become more and more convincing the more observations you make in support of them. This is why, although we draw the premises of deductive arguments from experience as well, we more often speak of inductive arguments as having their basis in experience and observation. What we perceive, and how we perceive it, matters more directly to their effectiveness.

The methods outlined in Chapter 10 therefore focus on the quantity of the information being used in a way that no deductive principles had to. Far more than anything else, this quantitative element in inductive argumentation produces its characteristic look.

Even the stress on similarity between sample and target class often turns out to mean something about sheer quantity. This is obviously true in inductive generalizations, where we can assume a large random sample to represent the target class--which means, we can assume it to resemble the target class.

Analogical arguments also use size to help ensure similarity. All things considered, a large enough random sample will probably be diverse with respect to any given property. Suppose you just bought your first toaster; you wonder if it might electrocute you under normal circumstances (no knives jabbed in, no baths during which you also make toast). You begin to ask people on the bus, in class, and at work whether they have ever been electrocuted by their toasters. While asking you realize that you've forgotten what brand of toaster you own. This is not necessarily a problem: Ask enough people, randomly enough, and the chances increase that some will own the same brand of toaster as yours, while others do not. If they all reply that no toaster of theirs ever shocked them, you will trust the answer, just because your large and random sample achieved diversity with respect to brand.

If sample size commonly helps to make the sample resemble its target, as well as being desirable in itself, it makes sense to call the sheer numbers of observations collected the single most important factor in establishing an inductive argument's conclusion. Here, again, lies the difference between deductive arguments and inductive arguments: The former pay no attention to numbers of claims, whereas the latter hardly pay attention to anything else.