Internal consistency refers to an important requirement for composite scales, in which the individual items or tests making up the scale are statistically associated with each other. If so, then it can be assumed that they are all measuring the same construct or attribute. It is important to note that measures of internal consistency do not constitute tests of the unidimensionality of items in a scale. For example, if the first half of a scale consists of cognitive items (correlating highly among themselves), and the second half items concerning achievement motivation (also correlating highly among themselves), the scale would have a high internal consistency value, despite the fact that there were two distinct dimensions. Moreover, referring to internal consistency as ‘internal consistency reliabilit’, as is often done, can be a source of confusion as this conflates two distinct concepts (viz., internal consistency and reliability), which do not necessarily go together. There are a variety of estimates of internal consistency, the most widely used being Cronbach’s alpha (others being, for example, the Kuder-Richardson formulas and the Spearman-Brown prophecy formula).
See Composite (or sum) scale, Construct, Cronbach’s alpha, Multitrait-multimethod matrix, Reliability