Statistical process monitoring may be applied for two purposes: (i) the quality assurance of the survey administration and data collection process and (ii) the analysis of collected data. On principle, the potential benefits of SPM tools for the collection and analysis of survey data have been realised for long. Simple control charts such as Shewhart X , s, c or u chart were considered by Hayes [1] in the 1992 first edition of his monograph on customer satisfaction and loyalty measurement, now distributed in the 2008 third edition. Biemer and Caspar [2] were among the first to introduce an SPM perspective into the quality assurance of survey operations. On a technical level, they used simple time-indexed plots only. More elaborate charting techniques were suggested by Mudryk et al. [3] and Sun [4] to monitor interviewers' errors during computer-aided telephone interviews, in particular the demerits d -chart, Pareto chart and a specially designed p chart. Spisak [5] studied the monitoring of sampling frames and emphasised the aspect of accounting for autocorrelation in the monitored data.An early study on SPM methods for direct monitoring of customer satisfaction data over time was provided by Wardell and Candia [6]. The authors criticised a naive way of adopting control charts from manufacturing environments. Wardell and Candia [6] considered a five-level Likert scale ranging from 1 (poor) to 5 (excellent). The considered data from a continuous hospital patient satisfaction survey are strongly left skewed. It is shown that a classical X chart produces an excess of out-of-control signals. As a more appropriate alternative, the authors considered a 2 chart, which accounts for the categorical nature of the data.Despite some promising attempts, SPM does not seem to have become a standard part of survey methodology. Classical SPM tools stem from the manufacturing industry and cannot readily be adopted for survey methodology, as shown by Wardell and Candia [6]. The most important impediments are in the different data characteristics, mainly (i) the occurrence of categorical and ordinal data; (ii) distribution specificities such as negative skewness (see also [7] or [8]); (iii) varying sample sizes; and (iv) multivariate and compound data.The most obvious instance of the issue (iv) is data from the popular SERVQUAL surveys, originally suggested by Parasuraman et al. [9], [10]. A typical SERVQUAL questionnaire requires 22 Likert scale ratings grouped into five dimensions of service quality: tangibles (environmental factors), reliability, responsiveness, assurance and empathy. The statistical monitoring of SERVQUAL surveys was studied by Jensen and Markland [11]. Separate monitoring of the five