Abstract
Several power transformations proposed in the past are examined to find out the type of distributions that they can normalize, and a general family of transformations, “the Generalized Modulus Power Transformation” (GEMPT), is proposed. The GEMPT will remove skewness and kurtosis and induce normality from a broad class of distributions, which we investigate, implying certain limitations for all power transformations. The use of GEMPT is illustrated and shown to lead to a better approximation to a normal distribution in an example in which the response is expected to follow a rectangular hyperbola.
Original language | English (US) |
---|---|
Pages (from-to) | 2933-2952 |
Number of pages | 20 |
Journal | Communications in Statistics - Theory and Methods |
Volume | 17 |
Issue number | 9 |
DOIs | |
State | Published - Jan 1 1988 |
Externally published | Yes |
Keywords
- Michaelis-Menten equation
- bimodality
- kinky distributions
- kurtosis
- normality
- skewness
ASJC Scopus subject areas
- Statistics and Probability