Analyzing Work Sample Task Performance Using Three Data Sets
Abstract
The Armed Services Vocational Aptitude Battery (ASVAB) has historically been validated against technical school grades. Job performance measurement (JPM) data were collected in the 1980s to link ASVAB cut scores to more realistic job performance measures for eight Air Force Specialties (AFSs). Due to the expense of these work samples, a major challenge has been to find reliable and valid surrogates. For 261 Aerospace Ground Equipment Mechanics (423x5) first termers, data on performing 16 hands on tasks were collected. These tasks, selected by subject matter experts (SMEs) at special JPM workshops, ranged from having seven to 26 steps and from low to high difficulty. Three data sets were generated for each task members getting all, some, or none of the task's steps correct. For all tasks, the largest numbers of personnel were associated with getting some steps correct. Generally, less difficult tasks were associated with larger numbers of personnel getting all steps correct. Getting all steps correct and some steps correct tended to be close in average task performance time, as well as having smaller average times, than did getting no steps correct. A correlation of .79 (p < .001) was found between the average time for all steps correct and independent SME estimates of the average time required by first termers for task completion. Members with all steps correct tended to have more recent task experience, more overall task experience, more average task performance per month, higher task experience ratings, higher Armed Forces Qualification Test (AFQT) scores, and higher grades in technical school. Future research needs are discussed.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1999
- Accession Number
- ADA363981
Entities
People
- Donald L. Harville
Organizations
- Armstrong Laboratory