Paper Type
Complete Research Paper
Description
Organizational and technical changes challenge standards of data warehouse design and initiate a redesign of contemporary Business Intelligence and Analytics environments. As a result, the use of multidimensional models for performance oriented reasons is not necessarily taken for granted. Simple data models or operational structures emerge as a basis for complex analyses. The paper therefore conducts a laboratory experiment to examine from a non-technical perspective the influnce of different data modeling types on the representational information quality of end users. A comparison is made between the multidimensional model and the transactional model respectively the flat file model. The experiment involves 78 participants and aims to compare perceived and observed representational information quality aspects of ad hoc analyses regarding the data modeling type. The results indicate a higher observed quality for multidimensional modeled data, while different types of data models do not influnce the end user perception of the representational information quality.
IS THERE STILL A NEED FOR MULTIDIMENSIONAL DATA MODELS?
Organizational and technical changes challenge standards of data warehouse design and initiate a redesign of contemporary Business Intelligence and Analytics environments. As a result, the use of multidimensional models for performance oriented reasons is not necessarily taken for granted. Simple data models or operational structures emerge as a basis for complex analyses. The paper therefore conducts a laboratory experiment to examine from a non-technical perspective the influnce of different data modeling types on the representational information quality of end users. A comparison is made between the multidimensional model and the transactional model respectively the flat file model. The experiment involves 78 participants and aims to compare perceived and observed representational information quality aspects of ad hoc analyses regarding the data modeling type. The results indicate a higher observed quality for multidimensional modeled data, while different types of data models do not influnce the end user perception of the representational information quality.