Errors in a spreadsheet constitute a serious reason for concern among organizations as well as academics. There are ongoing efforts toward finding ways to reduce errors, designing and developing visualization tools to support error correction activities being one of them. In this paper, we propose a framework for classifying activities associated with spreadsheet error correction. The purpose of this framework is to help in understanding the activities that are important for correcting different types of spreadsheet errors and how different visualization tools can help in error correction by effectively supporting these activities. An experiment is designed to test the effectiveness of a visualization tool that supports one of the most important activities from the framework – chaining activity. Two groups of subjects, with and without the visualization tool, are required to correct two types of errors. Our hypotheses are derived based on the notion of cognitive fit between problem representation and task, and the results of the experiment support most of the hypotheses. Thus, this study demonstrates the usefulness of the activity-based framework for spreadsheet error correction, and also provides guidelines for designing and developing tools for spreadsheet audit. It also provides empirical evidence to the cognitive fit theory by showing that performance is significantly better when visual support tools result in a match between problem representation and the task in hand, as in the case of correcting link errors with the tool used in this study. Theoretical and practical implications of the findings are discussed.