Average inter-item correlation is a way of analyzing internal consistency reliability. It is a measure of if individual questions on a test or questionnaire give consistent, appropriate results; different items that are meant to measure the same general construct or idea are checked to see if they give similar scores.
For example, it might be used to see how consistently a set of ten questions intended to test reading comprehension seem to be at honing in on the test subject’s reading abilities; or how well a questionnaire recorded true opinions by presenting three questions, each worded slightly different, on the subject.
Calculating Average Inter-item Correlation
Calculating average inter-item correlation is a three step process:
- Identify the questions or items meant to test the same thing or construct,
- Calculate the correlation between all pairs. For example, if you had three questions (a, b, c) meant to identify a test subject’s ice cream preference, you would calculate the correlations ab, bc, and ac. If you had six questions, a, b, c, d, e, and f, you would need to calculate correlations: ab, ac, ad, ae, af, bc, bd, be, bf, cd, ce, cf, de, df, and ef.
- Find the mean of all those correlations.
The ideal range of average inter-item correlation is 0.15 to 0.50; less than this, and the items are not well correlated and don’t measuring the same construct or idea very well (if at all). More than 0.50, and the items are so close as to be almost repetitive.
- Trochim, William. Web Center for Social Research Methods: Types of Reliability.
Retrieved from http://www.socialresearchmethods.net/kb/reltypes.php on April 24, 2018
- Phelan & Wren. Exploring Reliability in Academic Assessment. Retrieved from https://chfasoa.uni.edu/reliabilityandvalidity.htm on April 24, 2018
- FSSE, Faculty Survey of Student Engagement. Internal Consistency Reliability. Retrieved from http://fsse.indiana.edu/pdf/pp/2013/FSSE13_Internal_Consistency_Reliability.pdf on April 24, 2018