Econometric Theory/Statistical Inference/Hypothesis Testing

From Wikibooks, open books for an open world
< Econometric Theory‎ | Statistical Inference
Jump to: navigation, search

Basic Concepts[edit]

To conduct a successful hypothesis test, the following are required:

  • Testable Hypothesis

We need to have a null (H_0) and alternate (H_1) hypothesis.

  • Feasible test statistic

A test statistic is a random variable whose value for given sample data determines whether the null is rejected or retained. It is feasible when:

    • Its probability distribution is known when the null hypothesis (H_0) is true.
    • Its value can be calculated from the given sample data
  • Decision rule

A decision rule clearly delineates the:

    • Rejection region - the set of values of the test statistic for which H_0 is to be rejected.
    • Non-rejection region - the set of values of the test statistic for which the H_0 is to be retained.

Procedure for testing a hypothesis[edit]

  1. Formulate H_0 and H_1.
  2. Specify the test statistic and its distribution.
  3. Calculate the sample value of the test statistic under H_0 for the given sample data.
  4. Select a significance level (α) and determine the corresponding critical values (for the particular distribution).
  5. Apply the decision rule and state the conclusion (or inference) implied by the sample value of the test statistic.

Notes[edit]

  • The null hypothesis will always be the position where there is an equality (either strong or weak), and the alternate hypothesis will have the inequality.