Variation in Service Quality Scores - Some General Patterns

Customer evaluations of service quality may exhibit variation at the individual customer level, or in the aggregate level, over time. This study investigated three types of aggregate-level variation in customer evaluations of service quality. Using multiple data sets, it examined (1) the relationship between the mean score for an aspect of service quality and its variance; (2) the amount of variation in service quality scores given to an organization over time; (3) the degree to which scores for aspects of service quality are stable relative to scores for other aspects of service for the same organization and (4) the extent to which a mean score is related to the proportion of responses that fall below the neutral point on the scale (e.g. below 5/10). The study finds the following: (1) Service quality aspects that receive high mean scores have low variance, and those that receive lower scores have higher variance. (2) Scores for aspects of service quality are very stable over time (for the three organizations with repeated cross-sectional data). (3) There is also considerable stability in the relative ranking of service quality aspects over time. Aspects of service quality that have higher average scores than other aspects of service quality tend to retain this relative position over time. (4) There is a strong association between mean score for a service aspect and the proportion of responses that fall at or below 5/10. The relationship can be expressed simply as Proportion of responses at or below 5/10 = 122 - 13 x Mean Score. (R2 = 0.90). The results suggest that service quality surveys may not need to be conducted very often. They also suggest that aspects of service quality that receive lower scores are due to more heterogeneity in responses, rather than a large number of customers perceiving that aspect of service to be poorer. This result implies that when managers notice that some aspects of service score less highly than others, that the challenge for remedial action is to identify and remedy causes of variability in service quality rather than trying to lift performance ‘across the board'. Lastly, the results help contextualize mean scores - ‘7 out of 10' means that (in these datasets, at least) approximately (122 - 13 x 7 = 31) 31% of respondents gave at best a neutral score