A record of all the values for all the cases for all the properties in a research project constitutes a dataset, which will usually be laid out in the form of a data matrix, a framework of rows and columns for storing numeric data in a table. This chapter explains that all data analyses begin with such a data matrix, but variable-based analyses focus on the columns, looking at the distributions of property values across the cases, whereas case-based analyses focus on the rows, reviewing the combinations of values across the properties. The analysis of any kind of data, whether qualitative or quantitative, whether variable-based or case-based, at a minimum includes data preparation and a description of the properties (or of the cases) in a dataset, but is also likely to include elements of interpreting, pattern seeking, evaluating, explaining, applying and presenting results to an audience. The analysis of any dataset needs to be approached holistically, that is as a complete, self-contained entity, set in the context of the objectives for which the research that generated the dataset was designed to achieve and with a well-rounded view of what all the evidence is saying. Data analysis means getting the most out of a dataset, approaching it in several different ways so that the data tell the complete story. These processes raise a number of ethical issues, for example of privacy, confidentiality, deception, imposition, integrity or misrepresentation.