Description
Extracts performance data from many published studies, classifies them with task and equipment taxonomies, and converts them to error probabilities and times for predicting performance in new designs.
Meister treats the published human factors literature as the discipline's formal knowledge, yet notes that it is rarely combined into anything predictive. The American Institutes for Research Data Store did exactly that: the team screened over 2,000 journal articles, accepted 163 that met strict criteria, and reformatted their data into probabilities and times for reading displays and operating controls, indexed by a taxonomy of control and display types. The common metric is human error probability (operations with one or more errors divided by total operations). Combined values feed combinatorial models such as THERP to predict function- or mission-level performance. Meister links this to use validation: research is validated when its assembled data predict real performance. Unlike meta-analysis, which estimates an average effect for one question, the Data Store builds a reusable, taxonomy-indexed reference table for prediction.
Attributes
- Discipline
- Human Factors & Error
- Loss type
- Extinct
- Data
- Analytical / expert
- Effort
- Months
Steps
- Develop taxonomies of human function, equipment function and human-equipment interaction (for example starting from Fleishman and Quaintance's task taxonomy).
- Define a review window (for example ten years) and search all research reports in it.
- Apply acceptance criteria: adequate description of the measurement situation, no confounds, current technology, similarity to the target operational situation, and data convertible to a common metric.
- Extract and classify performance data with the taxonomies, converting to human error probability and performance times.
- Combine values in a combinatorial model (for example THERP) to predict performance of new designs.
- Keep updating the store as studies appear, and check its predictions against measured system performance.
Sources
- Conceptual Foundations of Human Factors Measurement (Meister, 2004)
Also known as
- AIR Data Store
- Human error-rate database from the literature
- Archival HF data combination for prediction
Categories
- Accessibility
- Human Factors & Error