Debugging is time and energy intensive. Many tools have been developed to help solve the problems associated with debugging, but programmers still rely on editing their code using traditional, manual techniques. One reason behind this is that many techniques succumb to the Isolation Flaw, where they isolate suspicious code to the point that it loses necessary context. Additionally, traditional debugging relies on the ways in which humans rely on the creation, testing and modification of hypotheses. An ideal tool will both avoid the Isolation Flaw while assisting the developers in their hypothesis cycle. This thesis consists of an empirical study that evaluates how debugging changes between traditional debugging and debugging with visualization assistance. The visualization chosen is based on the Tarantula fault localization tool. Each participant is given one program with the visualization and one without it to debug. Results imply that debug times using the tool to debug the programs in this study sometimes resulted in faster debugging, but usually was not significantly different from traditional debugging. The visualization decreased the average time between locating the file that the bug was in and fixing the bug for all bugs, implying that participants who reached that file did so on a more accurate hypothesis about the cause of the bug. While the tool may not consistently improve the speed of debugging for programs of this size and bugs of this complexity, it offers promising results regarding the context-dependent learning that is missing from debugging tools that contain the Isolation Flaw.