Variance as a Method for Objectively Assessing Pilot Performance

Abstract

The advent of high-precision GPS navigation and digital attitude/heading reference instruments creates opportunities for researchers to obtain objective aircraft attitude data now in actual flight as well as simulation. With this wealth of data, appropriate analyses and interpretation are critical to efficiently and accurately assessing pilot performance. A combined (constant plus variable) error measure applied to attitude is a common approach to determining pilot performance. However, a variable error measure has previously proven effective in various contexts. This paper will report results obtained from twenty novice participants performing a simulated flight training and test task. Attitude data were analyzed using both common root mean square error (combined error) and a variable error measure. Results indicated that the variable error approach was sensitive, precise, and efficient when measuring pilot performance. Characteristics of flight operations are discussed that render this approach particularly useful in an aviation context.