Mature scientific research results in the area of schedulability analysis have had a very limited impact on real industrial applications. This, we believe, is that current models are not able to accurately capture the complex temporal behavior of actual systems. In this paper we address a common assumption made in schedulability analysis methods that tasks can experience their worst case execution time independently from each other. This assumption is not very realistic for real systems since tasks collaboratively often perform certain functionality and thus depend on each other. Our aim, in this paper, is to capture execution time dependencies between tasks and to take advantage of this information when performing response time analysis. We introduce the concept of execution modes to the task model with offsets (transactional task model). The execution modes are used as a generic way to specify temporal dependencies between tasks that execute within a transaction. We then present extensions to the Response-Time Analysis (RTA) theory to analyze transactions with execution modes.