Maximum-likelihood estimation of receiver operating characteristic curve parameters

Autor:	John C. Ogilvie, C. Douglas Creelman
Rok vydání:	1968
Předmět:	Normal distribution Variables Standard error Best fitting Receiver operating characteristic Applied Mathematics media_common.quotation_subject Maximum likelihood Statistics Curve fitting General Psychology Mathematics media_common
Zdroj:	Journal of Mathematical Psychology. 5:377-391
ISSN:	0022-2496
DOI:	10.1016/0022-2496(68)90083-7
Popis:	Conventional least-squares curve fitting procedures are inappropriate for Receiver Operating Characteristic (ROC) plots because both axes represent dependent variables and are both subject to error. A brief review of the problem of interpreting ROC plots is followed by a solution which uses maximum-likelihood estimation of the best fitting straight line derived from a theoretical model which assumes underlying logistic distributions. The logistic is used simply because it is more convenient mathematically than the normal distribution. The normal and logistic models are compared empirically. Two advantages of this estimation procedure are that standard errors for the estimates of the parameters are obtained and there is also a test for goodness-of-fit of the theoretical model. Solutions are given for both yes-no and rating procedures.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::908bc8cbc78f36ebb1e8d6f679e1262b https://doi.org/10.1016/0022-2496(68)90083-7 Zobrazit plný text záznamu