What Do Heteroskedasticity Tests Detect?

No 1990022, CORE Discussion Papers from Université catholique de Louvain, Center for Operations Research and Econometrics (CORE)

Original Title: What Do Tests for Heteroskedasticity Detect, published as CORE DP 1990-022 listed above.

It was revised to have new title: Implicit Nulls and Alternatives for Hypothesis Tests, and issued as discussion paper in a number of other series -- U of Columbia, Bonn, etc. The paper was never published, though it has a lot of new ideas, and interesting approaches to assessment of hypothesis tests.

TITLE:

Implicit Nulls and Alternatives for Hypothesis Tests

ABSTRACT:

A test can be said to ‘detect’ an alternative hypothesis only if its power against this alternative exceeds its size. We use this principle to define the implicit null and alternative for a test. We analyze the performance of several tests for location and scale parameters.

REFERENCE:

Zaman, Asad; Jayasri Dutta, "Implicit Nulls and Alternatives for Hypothesis Tests", Dec 1992