One of the reasons why requirements engineering (RE) is so difficult is that requirements change ‘on the fly. ’ To investigate the sources of requirements change, 18 managerial supervisors of a logistic warehouse management system filled out a structured requirements-engineering questionnaire, the REquest, which assessed the level of agreement to the current system, the future system, and the stakeholders ’ needs. The results confirmed the assumption in goal-oriented RE that requirements are tightly connected to goals. More importantly, however, we discovered a mechanism that rules the level of agreement to requirements, which we coin the goal-torequirements chiasm or the χ-effect: Variance in what the system won’t have is for 70 % explained by goals stakeholders want to achieve with the system. Variance in what the system must have is for 90 % explained by goal states that stakeholders want to avoid. Moreover, we found evidence for an emotional component (i.e. valence) in the requirements evaluation that has a moderating effect on agreement to requirements. The χ-effect emphasizes that won’t requirements and goals to avoid are as important to requirements change as must requirements and goals to achieve with the system. In this light, structured questionnaire design is a systematic and controllable addition to common requirements-validation methods. Categories & Subject Descriptors H.1.2 [Models and Principles]: User/Machine Systems–Human information processing; K.6.3 [Management of Computing and Information Systems]: Software Management–Software development.

Current goal-oriented requirements engineering methods focus on the definition of optimal requirements that an information system needs to support in order to help its stakeholders to achieve their goals. But, the lack of systemic reasoning and disregard for questions of interpretation lead to insufficient attention given to activities and implicit policies affecting the definition of these goals. This results in the optimization of the requirements for potentially inadequate goals. Our framework relates stakeholders' goals to the their activities, their policies and their interpreted constraints and capabilities. It enables requirements engineers to better understand the rationale for goals found through requirements elicitation techniques and shows that conflicting goals can be reconciled by understanding how they fit in a higher-level activity. This results in the formulation of a more adequate set of goals that the information system should support in order for the organization and stakeholders to perform their activities.