Doing a Double Take
研究利用神秘顾客数据，发现隐藏场合（观察时间）的方差会显著增加服务评估所需数据量，对基准测试和评估者细分决策影响巨大。
Service managers require precise enough measures of service performance to make particular decisions. Generalizability theory (G-theory) has begun to be used to design service assessment studies for decision making. However, initial applications address a limited range of decisions and may underestimate the amount of data needed for decision making if the variance due to the hidden-occasions (time-of-observation) facet is substantial. This study uses test-retest mystery shopping data to investigate the variance due to the main and interaction effects of test occasions and the consequences of ignoring them for different managerial decisions. Accounting for the hidden-occasions facet reveals a need to collect more than twice as many data when benchmarking services and over 10 times as many data to segment service assessors on the basis of their evaluative responses. Thus, accounting for variation due to occasions is crucial for G-theory applications to deliver assessment data of the required quality.