We define and investigate the unifying concept of Information Quality (InfoQ) as the potential of a dataset to address a specific (scientific or practical) goal using empirical analysis. Although this concept is at the core of empirical analysis, it has only been partially addressed in both the statistics and the data mining literature. In this work we develop InfoQ, map out its various dimensions, and discuss how it can be assessed. We demonstrate the generality of the concept by considering various aspects of experimental and survey design, thereby showing its unifying importance to the wide context of empirical analysis.