A new test to detect monotonic and non-monotonic types of heteroscedasticity

Creative Commons License

Çelik R.

JOURNAL OF APPLIED STATISTICS, vol.44, pp.342-361, 2017 (Journal Indexed in SCI) identifier identifier

  • Publication Type: Article / Article
  • Volume: 44
  • Publication Date: 2017
  • Doi Number: 10.1080/02664763.2016.1169258
  • Title of Journal : JOURNAL OF APPLIED STATISTICS
  • Page Numbers: pp.342-361
  • Keywords: Monotonic heteroscedasticity, non-monotonic heteroscedasticity, heteroscedasticity tests, absolute transformed centered external variable, REGRESSION-MODELS, HETEROSKEDASTICITY

Abstract

A direct parametric test is proposed to detect monotonic and non-monotonic types of heteroscedasticity. After giving brief information about non-monotonic types of heteroscedasticity, the test algorithm is introduced. Proposed test and usual heteroscedasticity tests are compared on monotonic and non-monotonic types of heteroscedasticity in real and artificial data.