measurement and scale

reliability

The extent to which the measurement instrument gives the same results when the measurement is repeated.

当重复试验的时候，依然会得到相同的结果, 即测量结果的可靠性。没有random error. 

如何测量？how to assess scale reliability?

方法一：分两半，分别计算score，再算二者的correlation。

好处：一次处理。解决了test-retest 方法中的问题。（什么问题？）

问题：如何分两半

方法二：Cronbach's a 

a can be viewed as the expected correlation of two tests that measure the same construct.

a 可以看做是两次对相同变量检验的相关程度。mean correlation over all possible split-half reliabilities.

它含蓄地假定了：the average correlation of a set of items is an accurate estimate of the average correlation of all items that pertain（属于） to a certain construct

由公式推导出：a depends on 1)number of items ;2) mean correlation among items

a大于0.7 is acceptable

Validity

the extent to which the measurement instrument indeed measures the construct that we intend to measure.

确实是我们想测量的那个东西

content validity

content （items）是否覆盖全面（the complex construct）

测量方法：定义construct- 产生items-专家看看是不是都fit- 做item analysis（factor analysis & cronbach）

assume uni-dimensional, confirmatory factor analysis.(上一章) 啥意思？

dimensionality of the scale matches construct 

criterion validity

和其他变量的关系，是否预测成功（scale predicts the behavior it is supposed to predict）

如何做？

--这个新的scale应该要predict （在此例中 tendency to purchase grooming products）检验什么（scale score）和什么(predicted behavior)的显著相关性(significantly correlate with)

final scale scores = average 

SPSS-correlation （以前章节）

construct validity

是根据理论说的要测量的那玩意吗

·convergent - 这个scale和其他measures of the same construct 正相关吗

·discriminant - 这个scale和 measures of different constructs 不相关吗

·nomological - 这个scale 和throry一致吗

如何做？

建立一个network（是什么鬼）of caual relationships

检验hypothesis

if yes, validity is established.

if no, there are two possibilities 

  1 scale is not valid (前两个) 

  2 theory is wrong