Research data across the lifecycle

Nowadays, many things are referred to as "data" - people may say that when they mean "statistics," or "some factual/numeric information," or even just "something in digital format" - so it will be helpful to draw some boundaries and clarify some definitions first.  

The purpose of this guide, then, is to help you with making the data that you generate in the course of your research discoverable and resuable by others, publish or share it, protect it as may necessary or required, and preserve it for posterity.

What we mean by "research data" warrants some deliberation, to delineate it from other types of information, as mentioned above.  As the Australian National Data Service puts it: "Providing an authoritative definition of research data is challenging, as any definition is likely to depend on the context in which the question is asked." That is true!  

• In the USA, one "official" definition comes from Circular A-110 by the Office of Management and Budget (emphasis added): Research data is defined as the recorded factual material commonly accepted in the scientific community as necessary to validate research findings, but not any of the following: preliminary analyses, drafts of scientific papers, plans for future research, peer reviews, or communications with colleagues. 
   
• Research data can take many formats – for example, sound recordings of interviews from a population sample can be research data; text documents containing poems can be research data for an analysis of word patterns; images of the universe can be research data for an analysis of the movements of celestial bodies; a spreadsheet or database containing the recorded results of a telephone survey can be research data.  Based on its author's background, an emphasis in this guide will be on quantitative data in the social and allied sciences.

So, what is not research data?  Well, we would typically not call a journal article that presents scientific findings, including results from data analyses, "research data."  Nor would we refer to a statistical table that summarizes and presents data analyses as "research data."  Usually, we think of research data as a bunch of information - numerical datasets, images, sounds, etc. - that is so large and/or complex that it requires computer-aided analysis to make sense of - as opposed to a poster, article, presentation, or podcast that is intended to be comprehended by viewing or hearing it.

And yet, the water gets muddy again ... a whole lot of journal articles, or a whole lot of (or large enough) statistical tables, can themselves become input for (meta-)analysis.  So the output of research, in different formats and at different levels of information aggregation, can become the input of other data analysis ... we can therefore think of a lifecycle of research data.