Internal consistency is one of the four basic types of reliability. It examines if a list of statements are well enough related to each other to create a reliable construct.
If you want to be sure people are telling the truth about a topic, they should give coherent and consistent information about all aspects. All answers should point in the same direction. And between all aspects there should be a positive correlation. If not, the information becomes unreliable.
Suppose you want to know if people are satisfied with the house they live in. In a questionnaire only one question can be fine to make the respondent’s opinion clear. It can also be questioned more in details. People who are more satisfied with the total, should also be more satisfied with the details. Those details can be satisfaction with the living room, the bathroom, the entrance, the kitchen and so on. The list can become quite long. Between all aspects correlation coefficients can be computed and all should be positive.
Okay, but that is a score between the aspects, we need a score for the total, that is for all aspects together. Is there any score available? Yes, there is. This is the formula:
This score is known as Cronbachs alfa. It is a score for internal consistency. As you can see, the minimal number of items is 2, otherwise k – 1 will be zero and dividing by zero is not allowed. Also notice that there is no maximum numbers of items.
You can find more information about building and testing instruments for measurements in our paper Factor analysis and Cronbach´s alfa.
Learn how to compute Cronbach's alpha in SPSS with our SPSS-tutorials.
Related topics to internal consistency
- Factor analysis