CogKnoSys

Knowledge you can walk. Read the idea below, then step up to the general, across to peers or down to the specific.

Mental

Inductive Reasoning

Inductive reasoning draws conclusions that the evidence makes likely, but not certain.

Learn, discuss, create, buy or invest

The idea

Inductive reasoning is any way of reasoning in which the evidence makes a conclusion likely, but not certain. In deductive reasoning, true premises guarantee the conclusion. In inductive reasoning, the conclusion is at best probable. Say every swan you have ever seen is white. You might conclude that all swans are white, yet one black swan would prove you wrong.

Its main types are generalization, prediction, statistical syllogism, argument from analogy and causal inference. A generalization moves from a sample to a whole group. For example, a poll of some voters can be used to estimate what all voters think. A statistical syllogism goes the other way, from a fact about a group to a likely conclusion about one member. An argument from analogy sees that two things share several traits and infers that they share one more.

A generalization is stronger when the sample is bigger and more like the whole group. Two classic mistakes are the hasty generalization, which draws a rule from too few cases, and the biased sample. Induction also assumes that nature stays the same, so the future will be like the past. No one can prove that idea from the data alone.

Check yourself

Question 1 of 2

Which kind of inductive argument moves from a fact about a group to a likely conclusion about one member?

Pick one. The first pick counts.

Composition

what it's made of

  • Observations — The specific cases or data a person has actually seen.
  • Sample — The group of cases studied; larger and more representative samples give stronger conclusions.
  • Pattern — A regularity noticed across the observations.
  • Probable Conclusion — A claim the evidence makes likely but does not guarantee.
  • Background Assumption — The unproven idea that nature is uniform, so the future will resemble the past.

Connection

how it relates

Under Logic and Reasoning

Related

  1. Evaluating Evidencestrength 70 of 100

    Inductive conclusions are only as strong as the evidence behind them, so judging that evidence comes first.

  2. Identifying Assumptionsstrength 60 of 100

    Induction rests on the hidden assumption that the future will resemble the past.

  3. Recognizing Biasstrength 50 of 100

    A biased sample is a classic way for an inductive generalization to go wrong.

Change

how and why it evolves

Past
Arguments that assume the future will mirror the past are sometimes called Humean, after the philosopher who was first to examine them closely.
Now
Today scientists and pollsters use statistical tools such as Bayesian inference to measure how strongly a sample supports a conclusion.
Next
As computers learn patterns from ever larger data sets, questions about when inductive conclusions can be trusted will likely matter even more.

Knowledge you can walk. Read the idea, then step up, across or down.

Inductive Reasoning

Inductive reasoning draws conclusions that the evidence makes likely, but not certain.

Learn, discuss, create, buy or invest

The idea

Inductive reasoning is any way of reasoning in which the evidence makes a conclusion likely, but not certain. In deductive reasoning, true premises guarantee the conclusion. In inductive reasoning, the conclusion is at best probable. Say every swan you have ever seen is white. You might conclude that all swans are white, yet one black swan would prove you wrong.

Its main types are generalization, prediction, statistical syllogism, argument from analogy and causal inference. A generalization moves from a sample to a whole group. For example, a poll of some voters can be used to estimate what all voters think. A statistical syllogism goes the other way, from a fact about a group to a likely conclusion about one member. An argument from analogy sees that two things share several traits and infers that they share one more.

A generalization is stronger when the sample is bigger and more like the whole group. Two classic mistakes are the hasty generalization, which draws a rule from too few cases, and the biased sample. Induction also assumes that nature stays the same, so the future will be like the past. No one can prove that idea from the data alone.

Check yourself

2 quick questions, instant feedback.