Reliability and validity in binary ratings: areas of common misunderstanding in diagnosis and symptom ratings.
Reliability and validity in binary ratings: areas of common misunderstanding in diagnosis and symptom ratings.
复制标题
二元评级的可靠性和有效性：诊断和症状评级中常见的误解领域。
DOI：
--
复制
发表时间：
1978
影响因子：
--
通讯作者：
I. Gottesman
中科院分区：
文献类型：
--
作者：
G. Carey;G. Carey;I. Gottesman
二元评级（例如，精神分裂症与非精神分裂症）的可靠性及其对有效性的影响之间可能存在混淆。高可靠性并不能保证有效性，但矛盾的是，低可靠性在所有情况下并不意味着有效性差。基础比率或实验设计的改变可能表明有效性较高，即使可靠性被认为较低。仅仅通过提高可靠性来改进精神疾病命名法存在“衰减悖论”的风险，即可靠性的进一步提高会使评级的有效性降低。最后，在做出诊断时随机误差的假设并不总是成立，因此必须相应地调整统计分析。需要新的统计方法来仅对假阳性或假阴性率进行索引，以便量化会降低一些有效性系数的误差。
Confusion may exist between the reliability of a binary rating (for example, schizophrenia versus not-schizophrenia) and its implications for validity. High reliability does not guarantee validity, but paradoxically, low reliability does not imply poor validity in all contexts. Changes in the base rate or in experimental design may indicate high validity even when the reliability was thought to be low. Attempts to improve the psychiatric nomenclature by increasing only reliability run the risk of the "attenuation paradox" where further increases in reliability will make the ratings less valid. Finally, the assumption of random error in making diagnoses does not always hold, so that statistical analyses must be adjusted accordingly. New statistical methods are needed to index only false-positive or false-negative rates in order to quantify the error that will reduce some validity coefficients.