Hosmer-Lemeshow testean erabilitako talde-kopuruaren azterketa simulazioen bidez

• Autores: Ane Moreno Oya, Irantzu Barrio Beraza
• Localización: Ekaia: Euskal Herriko Unibertsitateko zientzi eta teknologi aldizkaria, ISSN 0214-9001, Nº. 45, 2024, págs. 327-344
• Idioma: euskera
• Títulos paralelos:
  • Analysis of the number of groups used in the Hosmer-Lemeshow test using simulations
• Enlaces
  • Texto completo
• Resumen
  • euskara

    Egun, Hosmer-Lemeshow (HL) testa erregresio logistikoko ereduen doikuntza-egokitasuna neurtzeko maiz erabiltzen den hipotesi-kontrastea da. Ordea, lagin tamainarekin lotuta dauden hainbat eragozpen ditu eta, hori dela eta, azken urteetan eraldaketa ugari jasan ditu. Lan honetan, g talde kopurua aldatuta, testaren erabakien egonkortasuna aztertu dugu. Aukeratutako eredua egokia denean, HL testaren errendimendua ona dela lortu dugu. Gainera, ez da lagin tamainaren araberakoa. Bestalde, egoera honetan, proposatutako talde kopuru gomendatuaren erabilerak ez du eragin nabarmenik. Aldiz, doitutako eredua desegokia denean, HL testa lagin tamainarekiko sentikorra da eta, batez ere lagin txikietan, errendimendua eskasa da. Honetaz gain, lagin handietan bai gaixotasunaren prebalentziak bai ereduaren konplexutasunak eragina dute.

  • English

    Nowadays, the Hosmer-Lemeshow (HL) test is a tool often used to measure the goodness of fit of logistic regression models. However, it has several inconveniences associated with sample size. As a result, it has undergone many modifications in recent years. In this work, by changing the number of groups, we have analyzed the stability of the test’s decisions. When the model is correct, the performance of the HL test is good. What’s more, it doesn’t depend on the sample size. On the other hand, in this case, the use of the proposed number of groups has no significant effect on the results. When the chosen model is misspecified, the performance of the HL test is poor and it is sensitive to sample size. Above all, its performance is poor in small samples. Apart from that, in big samples, both prevalence and the model’s complexity have an effect on the decisions of the test.