This letter is essentially an addendum to Hinton-Bayre (2010), where a comparison of several models of reliable change (RC) was made. RC statistics have been developed to assess whether an individual has demonstrated statistically significant change over time or following an intervening event or treatment. Popular settings for this include sports-related concussion, brain surgery, and cardiac surgery. The more unique element of this methodology is the ability to compare pre- and post-measurements. There is a growing amount of test-retest normative data available to the clinician to assess whether an individual's change is more than would be expected due to chance. Hinton-Bayre (2010) demonstrated that all currently popular RC models can be conceptualized by a fundamental expression, RC = (Y-Y′)/SE, where Y is the actual retest score, Y′ the predicted retest score, and SE the associated standard error. Contemporary RC models estimate the Y′ and SE differently. There is ongoing evaluation of the most appropriate RC model(s) (Maassen, Bossema, & Brand, 2009). Hinton-Bayre (2010) went on to demonstrate that most RC models can be calculated with the possession of simple test-retest statistics taken from a relevant control group, particularly, the initial test and retest means and standard deviations, and the test-retest reliability. Most papers will report means and standard deviations. However, it is less common for the reliability coefficient to be presented. In a two-point situation (test and retest), with the possession of an inferential statistic (viz. repeated measures t- or F-test), and using some simple algebra, the following expression can be used to determine the test-retest reliability coefficient.  

formula

Please note in the setting of a test (X) and retest (Y) control group, rXY is the test-retest reliability coefficient, σX the estimated initial standard deviation (σ2 is a variance), σY the estimated retest standard deviation, MY the retest mean, MX the initial test mean, t the repeated-measures student's t-test, and n the sample size. If an ANOVA (F)-statistic is presented, the reader should recall that in the two-group example, t = √F. Researchers in the field of repeated or longitudinal data collection are encouraged to present basic test-retest statistics including reliability estimates, or at least exact inferential statistics. This simple and essential reporting will readily allow the use of RC models to assess individual change. Of course, the clinician should always be mindful that the control group adequately match the individual in terms of characteristics, initial test scores, and test-retest interval. The author is willing to supply a spreadsheet file to facilitate the calculation of RC models reported in Hinton-Bayre (2010) and the above expression on request.

References

Hinton-Bayre
A. D.
Deriving reliable change statistics from test-retest normative data: Comparison of models and mathematical expressions
Archives of Clinical Neuropsychology
 , 
2010
, vol. 
25
 (pg. 
244
-
256
)
Maassen
G. H.
Bossema
E. R.
Brand
N.
Reliable change and practice effects: Outcomes of various indices compared
Journal of Clinical and Experimental Neuropsychology
 , 
2009
, vol. 
31
 (pg. 
339
-
352
)