subject selection As is the case with all types of sub s a few reliability studies influences the non selection in the results can study; the be generalized only to the types of subjects sud rdized Therefore, the reliability of an instrument n should be determined using the individuals on ic for whom the instrument will be used practice. lf lways the measure is a clinical one, it is best to deter marks mine its reliability on patients who would o tment narily require this measurement as part of their ement care. Watkins and associates did just this in their study of the reliability of knee range-of-motion measurements. In fact, they even divided their patients into diagnostic categories to determine ther whether the measurements were more reliable rtially for patients with certain types of knee dysfunc xam- em- tion. inappropriate use normal subjects to establish the reliability of clinical measures has stan- the potential to inflate reliability estimates be f cer- cause normal subjects may be easier to measure y vi- patients. Pain, obliteration of landmarks be and cause of deformity, or difficulty following direc tions because of neurological impairment may ard- make it difficult to take measurements in pa ects tients If a researcher ultimately wishes to determine zed norms for certain characteristics, it is appropriate for to determine the reliability of measurements the bilit bility using sure plac ties Do thre side tics