In workflow management, the data flow perspective specifies how data are produced and consumed by activities in a workflow. Data flow analysis can detect data flow anomalies occurring in a workflow while its control flow can be syntactically error-free. Currently, most commercial workflow management systems do not provide the tools for data flow analysis at design time. We have previously proposed a data flow analysis approach and developed the basic concepts and the essential algorithms. As another step forward, this paper examines the issues of data flow anomalies and their verification from a theoretical point of view and validates the correctness of the proposed approach.