Question 6: In terms of test - retest reliability, CHQ-C demonstrated intraclass correlation coefficient (ICC) of 0.75. What is meant by 'test-retest reliability', and how does the intraclass correlation coefficient measure it?
‘Test-retest reliability’ is the agreement between measures taken at two different occasions. We cannot usually measure the repeatability of a questionnaire scale by making two measurements in quick succession, because respondents will remember their first answers and may try reproduce them. We must allow sufficient time to elapse between administrations of a questionnaire for respondents to forget their replies. At the same time, we do not want so much time to elapse that the subject’s true value of the construct (i.e. their quality of life) changes. Two weeks is a typical time interval between administrations of a questionnaire for test-rest reliability.
The intraclass correlation coefficient is a version of the correlation coefficient between repeated observations which ignores the order in which observations were made. Like all correlation methods for measuring repeatability, it depends on the way the sample was selected and the variability of quantity being measured. It estimates the correlation between repeated measurements in the population from which the subjects form a representative sample.
The ICC is the proportion of the variance of the measurements which is variation in the true values of the quantity being measured.
Back to question.
This page maintained by Martin Bland.
Last updated: 5 March, 2007