reliable change index =
= ((post-test value) – (pre-test value)) / (standard error of measurement)
standard error of the measurement =
= (standard deviation of pretest population) * SQRT(1 – (reliability of the measure))
where:
• Reliability of the measure indicates the test-retest reliability of the instrument.
• Standard deviation of the pretest population includes the control group, the normal population and the experimental group pretreatment.
Modification of Christensen and Mendoza: These authors recommended using the (standard error of difference between the 2 test scores) rather than the (standard error of measurement). The (standard error of difference) indicates the spread in the change score distribution which would be expected if no actual change occurred.
standard error of differences =
= SQRT (2 * ((standard error of the measurement)^2)) =
= (standard error of the measurement) * SQRT(2) =
= (standard error of the measurement) * 1.414
Interpretation:
• An index value more than +/- 1.96 indicates that an actual change has occurred (such a value is unlikely without an actual change having occurred with a p value < 0.05).
