Objective. The purpose of this study was to compare the test–retest reliability of three administrative methods of the Work Box™: (a) the original instructions, (b) a revised version of the original instructions, and (c) another revised version that was based on suggestions made by authors of the first two versions of the instructions.
Method. Sixty subjects without disabilities were randomly grouped so that 20 subjects were tested per administrative method. The assessment was administered to each subject on two occasions, with a 7-day to 14-day period between tests. Scores were recorded as time in seconds, and intraclass correlation coefficients (ICCs) were used to calculate the reliability.
Results. The ICCs for assembly, disassembly, and total scores were .589, .604, and .654, respectively, for the original instructions; .424, .572, and .545 for the revised instructions; and .781, .579, and .717 for the second revised instructions. Reliability was found to be higher for men than for women and for subjects who claimed to have more rather than less experience in similar manual dexterity tasks.
Conclusions. On the basis of the reliability of each administrative method and comments made by subjects about their understanding of the instructions, the second revised version of the instructions is recommended as the standard method. The results also indicate that the assessment is most appropriate for a population of men with manual dexterity experience. With further standardization, the Work Box could be a valuable assessment tool for therapists working in industrial rehabilitation settings.