The second major reliability statistic is the standard error of measurement. The Standard error of measurement is a function of the coefficient of reliability and the degree to which the scores on a test tend to spread out around the average score (O’Connor and Kleyner, 2012).
Correlation and Dependence
Dependence defines to the statistical relationship between 2 data sets or random variables whereas correlation defines wide number of statistical relationships by managing dependence. Dependence also defines to any situation in which random variables do not satisfy the probabilistic dependence of mathematical condition. Whereas correlation can also be referred to the more than 2 random variables departure from independence, but technically defined several specialized kinds of relationship between mean values Correlation is an association measure whereas dependence is the absence or presence of a relationship between variables measure. Correlation is the appropriate measurement of dependence only for a specific set of joint distributions. If the dependence framework is not described by one of the distributions the estimated or calculated correlation relies not only on the dependence nature but also on the marginal distribution behavior. It follows the stochastic correlation nature which is not driven by the stochastic dependence nature but by the alterations in the marginal behavior. To decide whether to introduce stochastic correlations and other members it will be useful to know whether stable non correlation dependence with differing marginal is dealing or with a genuinely varying framework of dependence (Mari and Kotz, 2001; Eydeland and Wolynesic, 2003, p 234).
第二个主要的可靠性统计测量的标准误差。测量的标准误差的系数可靠性的功能和在何种程度上倾向于在考试成绩平均分摊（奥康纳和kleyner，2012）。
相关
依赖定义之间的统计关系的2个数据集或随机变量，而相关管理依赖定义广泛的统计关系的数量。的依赖也定义在任何情况下，随机变量不满足的数学条件概率的依赖性。而相关性也可以被称为超过2独立随机变量的偏离，但技术上定义一些特殊种类的平均值的相关关系是一个协会的措施而依赖变量之间的关系的存在或不存在的测量。相关性是依赖不仅对联合分布的一组特定的适当的测量。如果依赖框架是不一的估计或计算关联不仅依赖于依赖性也对边际分布行为的分布。它遵循随机相关性质不是由随机依赖自然而在边际行为改变驱动。决定是否引入随机的相关性和其他成员就会知道不同的边际稳定的无相关关系的处理或与一个真正不同的框架依赖有用（Mari和科茨，2001；艾德兰和wolynesic，2003，p 234）。
