Quantitative measurement of quality attribute preferences using conjoint analysis

Abstract

Conjoint analysis has received considerable attention as a technique for measuring customer preferences through utility tradeoffs among products and services. This method can be applied to the area of software architecture, in particular, to analyze architectural tradeoffs among quality attributes. By eliciting customer utilities through conjoint analysis, software engineers can focus on the useful quality attributes identified, which will increase the chance of delivering satisfactory software products to customers. This thesis aims to propose a quantitative method of measuring quality attribute preferences using conjoint analysis and demonstrates exploratory results from the Project Management Center (PMCenter) project. In addition, it shows this method can be extended for exploratory and confirmatory purposes. The proposed method is complementary to the Architecture Trade-off Analysis Method (ATAM) in that ATAM relies on customer feedback to elicit important quality attributes, whereas this method can be used to measure the utilities of quality attributes in a quantitative manner. The method can provide a new framework for utility measurement of quality attributes in shrink-wrap product development tough an aggregate-level conjoint model, which will be a great advantage over ATAM because the process of conjoint analysis can remove repetitive and frequent feedback steps when eliciting quality attribute utilities from the customer.