论文部分内容阅读
多数心理测验只能达到顺序量表水平,不适合算术运算和参数统计检验,测验分数是心理属性的一个区间的代表,而不是一个点的精确表示。但是,在实际运用中,我们假定心理测验是等距量表,即,条目内、条目之间、被试之间的分数具有相同的单位。我们认为条目和对条目的反应具有质性和量性的区别,因此,提出了条目质性的概念;我们还认为,有的心理属性的行为样本不具备难度性质,或者条目的难度没有实际意义,或者有理由假定条目难度是相等的,我们提出条目集群的概念来处理这类条目的分数分配方法。经典测验理论的标准分和条目反应理论的logit是在等距假设前提下对测验分数的数学表示,虽然它们本身都是等距量表,但并不代表测量内容也是等距的。明确等距假设的风险有利于更好地运用测验。有多种途径可以降低等距假设的风险。
Most psychological tests can only achieve the order of the scale, not suitable for arithmetic and parameter test, the test scores is a representative of a range of psychological attributes, rather than a precise representation of the point. However, in practical application, we assume that psychological tests are equidistant scales, that is, the scores between subjects within an item and between subjects have the same units. We think the entry and the response to the entry have a qualitative and quantitative difference. Therefore, we propose the concept of the item’s quality. We also think that the behavior samples with some psychological properties do not have the difficulty property or the entry difficulty is not practical , Or it is reasonable to assume that the entries are equally difficult, we propose the concept of entry clusters to handle the fractional allocation of such entries. The standard scores and entries of the classical test theory The logit of the reaction theory is a mathematical representation of the test scores on the assumption of equidistance. Although they are themselves isometric scales, it does not mean that the measurement content is also equidistant. Defining the risk of equidistant assumptions helps make better use of tests. There are several ways to reduce the risk of equating hypotheses.