Summary

Overall, the Federal Aviation Administration (FAA) has done a credible job analyzing and defining the requirements of the Safety Performance Analysis System, an automated decision support system that FAA is acquiring to help it target its limited inspection and certification resources. However, FAA's current estimates for system software are subjective, not supported by verifiable analysis, and therefore may be unreliable. To identify aviation safety risk precursors, the system depends on information from many databases, including those compromising FAA's Aviation Safety Analysis System. As previously reported by GAO, FAA, and others, these Aviation Safety Analysis System databases contain incomplete, inconsistent, and inaccurate data. If the data quality is poor, the system's input into safety decisions will not be reliable and will not effectively support FAA's inspection and certification system.

GAO found that: (1) FAA has generally implemented good development and acquisition procedures for SPAS; (2) FAA has maximized user involvement and system prototyping in developing and evaluating SPAS; (3) FAA has reduced SPAS development risks by using an independent verification and validation agent; (4) FAA is exploring the potential of its proposed corporate-wide area network to accommodate SPAS in order to avoid the acquisition of duplicate communication networks; (5) FAA cost estimates for SPAS software may not be reliable, since they are subjective; (6) FAA lacks a strategy for improving SPAS data sources, particularly ASAS, which jeopardizes the system's utility; (7) ASAS databases contain incomplete, inaccurate, and inconsistent data on airline inspections; (8) FAA has not yet defined its long-term data quality goals; and (9) if FAA fails to improve ASAS data, it could improperly target its limited inspection and certification resources on less important problems.