Data Preparation

Researchers will routinely be faced with the challenge of ensuring that the data to be analysed are in the best form to provide evidence to assess the research questions. Contrary to expectations, this is not merely a time to transform skewed distributions and delete possible outliers. Data preparation involves checking that the data reflect the expected processes in the populations of interest. It is important that the research keeps careful audit trails of all such procedures and actions. The initial phase involves descriptive statistics and figures such as scatter plots and may lead to a re-evaluation of the original analysis plan. Next, after models have been fit, a diagnostics phase involves widely used procedures to assess the appropriateness of the model for the actual data. These procedures, in slightly modified forms, can also be applied to research that uses existing data. Suggestions for further reading are provided for those who seek more advanced material.