Table 1.

Specific evidence to support validity arguments.

	Scoring	Generalization	Extrapolation	Implication
Definition of Kane (26)	Rule is appropriate Rule is applied as specified Scoring is free of bias Data fit the scaling todel	Sample is representative of universe of possible observations Sample is large enough to control for random error	Observed score is related to the target score No systematic errors likely to undermine the extrapolation	Implications (interpretations) are appropriate Properties of scores support the implications (interpretations) associated with the label
Operational sources of evidence	Development of scoring dimensions/development of selection items Evidence of independence of scoring dimensions Distinguishability of scored items Evidence of scoring reliability Quality control of scoring	Internal consistency reliability across projects Assessment of sources of measurement error Sampling of observations (number of items or sites, breadth of content) Sample size	Relationships with other variables/measures (correlation with other scores) Development of items to reflect the full breadth of real-life tasks Retesting performance	Impact on the physically exercise person (i.e., viewing the act of assessment as an intervention) Impact on the project Accurate classification of individuals Standard setting process