Implications:
Obtaining dependable measures of student performance may require greater task sampling than is usually assumed
Generalizability may be increased by distributing those task samples across different occasions of testing
Obtaining an accurate picture of the interchangeability of assessment methods may require administering tasks on different occasions