The Structure of Customer Satisfaction: Effects of Survey Measurement

INTRODUCTION Measurement errors and questionnaire effects in survey data can have substantial impact on estimates of statistical relationships. Although the customer satisfaction literature contains some discussion of scaling issues (Westbrook and Oliver 1981), statistical models of customer satisfaction have not explicitly recognized measurement issues. This paper enhances the traditional model of customer satisfaction and attitude formation by explicitly incorporating these effects, thus linking the three separate research streams whose bases are in customer satisfaction, structural equation modeling and in survey research. A structural equations/measurement error model is developed that distinguishes between observed variables from a satisfaction survey and underlying constructs. The model explicitly incorporates measurement error and questionnaire effects where response correlations exist among proximate survey items or those with similar wordings or response categories. The model is estimated with a telephone service survey in which identical overall quality items are included at both the beginning and end of the questionnaire, with a series of attribute rating questions in the middle.