Premium
King Kong, Storks and Birth Rates
Author(s) -
Wirth Steffen
Publication year - 2003
Publication title -
teaching statistics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.425
H-Index - 13
eISSN - 1467-9639
pISSN - 0141-982X
DOI - 10.1111/1467-9639.01141
Subject(s) - bootstrapping (finance) , statistics , ordinary least squares , linear regression , standard error , econometrics , regression , regression analysis , correlation coefficient , mathematics , correlation , geometry
Summary This article discusses the influence extreme observations can exert on linear correlation between variables. It shows that standard p ‐values obtained according to ordinary least squares and median regression can be misleading. Bootstrapping can prevent underestimation of the standard error of the regression coefficient and therefore unmask the deceiving correlation.