An Investigation of the Effects of Carrier Groups on Airline Quality Rating Components Using a Two-way Analysis of Variance

The Airline Quality Rating is a quantitative determination of the quality of U.S. domestic air carriers based on parameters published by the U.S. Department of Transportation. The rating is unique in that it is of interval scale and is comparable across carriers and time periods (Bowen & Headley, 2012). In order to gain insight into the relationship of the AQR metric to the carriers to which it is applied, it is helpful to group the carriers and examine the effects of those groupings on the four individual factors that comprise the AQR. Such a methodology allows one to better understand the relationship of the AQR metric to each of the carrier groups and ultimately improve the predictability of the metric. The authors employ a two-way analysis of variance to determine differences between carrier group means for each of the four AQR factors while examining longitudinal effects, along with post-hoc difference testing. While the post-hoc test results indicate significant differences between some groupings, suggesting that separate econometric models for those groupings might be created with the goal of more accurately forecasting the metric, some of the assumptions upon which the ANOVA is predicated are violated. This article will examine those violations and suggest that further research using nontraditional methods (e.g., Bayesian analysis) is indicated.