Likelihood-Ratio Test and F-test for Two Exponential Means Equality: a Monte Carlo Power Exploration

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F62156489%3A43110%2F20%3A43918268" target="_blank" >RIV/62156489:43110/20:43918268 - isvavai.cz</a>

Výsledek na webu

<a href="https://mme2020.mendelu.cz/wcd/w-rek-mme/mme2020_conference_proceedings_final.pdf" target="_blank" >https://mme2020.mendelu.cz/wcd/w-rek-mme/mme2020_conference_proceedings_final.pdf</a>

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Likelihood-Ratio Test and F-test for Two Exponential Means Equality: a Monte Carlo Power Exploration

Popis výsledku v původním jazyce

In statistical practice, the exponential distribution is frequently evoked to explore problems of survival or utility times or in general, time to event scenarios. This study investigates power of statistical tests to verify equality of two exponential distributions: a traditional F-test and a likelihood-ratio test (LRT) in common and approximate variants. The tests were examined for power in both exact and bootstrapped forms. Two Monte Carlo simulations were set up to research the test power in response to the combined sample size, size of samples generated from the exponential distributions and mean ratio as a measure of the population means inequality. 15,000 MC runs were generated with additional 1,000 resampled data for the bootstrapped alternatives. The rejection rates generally increased with sample size, balanced samples and inequal means. Power was found inadequate for n 30 in all examined combinations. A similar power was found, when samples were balanced, although it turned divergent with more unbalanced data. The bootstrapped tests generally showed increased empirical test size relative to nominal = 0:05 and superior power over the exact tests for simulated combinations with greater means occurring in large samples.

Název v anglickém jazyce

Likelihood-Ratio Test and F-test for Two Exponential Means Equality: a Monte Carlo Power Exploration

Popis výsledku anglicky

In statistical practice, the exponential distribution is frequently evoked to explore problems of survival or utility times or in general, time to event scenarios. This study investigates power of statistical tests to verify equality of two exponential distributions: a traditional F-test and a likelihood-ratio test (LRT) in common and approximate variants. The tests were examined for power in both exact and bootstrapped forms. Two Monte Carlo simulations were set up to research the test power in response to the combined sample size, size of samples generated from the exponential distributions and mean ratio as a measure of the population means inequality. 15,000 MC runs were generated with additional 1,000 resampled data for the bootstrapped alternatives. The rejection rates generally increased with sample size, balanced samples and inequal means. Power was found inadequate for n 30 in all examined combinations. A similar power was found, when samples were balanced, although it turned divergent with more unbalanced data. The bootstrapped tests generally showed increased empirical test size relative to nominal = 0:05 and superior power over the exact tests for simulated combinations with greater means occurring in large samples.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

50202 - Applied Economics, Econometrics

Projekt

—

Návaznosti

I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Rok uplatnění

2020

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Mathematical Methods in Economics 2020: Conference Proceedings

ISBN

978-80-7509-734-7

ISSN

—

e-ISSN

—

Počet stran výsledku

7

Strana od-do

11-17

Název nakladatele

Mendelova univerzita v Brně

Místo vydání

Brno

Místo konání akce

Brno

Datum konání akce

9. 9. 2020

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

000668460800001