A goodness of fit statistic for the geometric distribution
José A. Ferreira
Abstract
Open-access reader
José A. Ferreira
Abstract
Open-access reader
We propose a goodness of fit statistic for the geometric distribution and compare it in terms of power, via simulation, with the chi-square statistic. The statistic is based on the Lau-Rao theorem and can be seen as a discrete analogue of the total time on test statistic. The results suggest that the test based on the new statistic is generally superior to the chi-square test.
A significance statement is not available in the OpenAlex record.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
We propose a goodness of fit statistic for the geometric distribution and compare it in terms of power, via simulation, with the chi-square statistic. The statistic is based on the Lau-Rao theorem and can be seen as a discrete analogue of the total time on test statistic. The results suggest that the test based on the new statistic is generally superior to the chi-square test.
Key concepts: Ancillary statistic, Statistic, PRESS statistic, Goodness of fit, Pearson's chi-squared test, Mathematics, Chi-square test, Statistics