This paper presents guidelines and naming conventions to assist in managing datasets, programs and printouts for small to medium sized research projects. First, the limits and nature of the type of project are considered.
A number of definitions are included to help further specify the problem. Four main areas are addressed. First, a consistent set of rules for naming programs and datasets is presented. Next, techniques for storing and archiving the programs and datasets are presented, using the naming conventions just developed. Third, rules for printout management are given, based on the naming rules. Finally, the documentation of all components of the project is discussed. Although trivial to implement, adherence to the naming rules provides a sound basis for a high quality, easy-to-use documentation system.
1. CONTEXT OF THE PROBLEM
Datasets, programs and printouts grow like well watered vegetables and weeds around any research project involving computerized data collection or analysis. This paper will present rules to assist the researcher or scientist in the complex task of nurturing such a garden. The rules have evolved over the last several years and stem from the experiences of the authors and their colleagues on small to medium size research projects. A medium size research project can be loosely defined as one involving 1000-10000 observations, l-10 persons involved in data processing and analysis, at a yearly cost of $lOOOO-$100000.
The projects in this middle range face a unique problem. Small studies are so small that typically one person has complete control over all of the data management and analysis associated with the study. The very large study has the advantage of economy of scale. Explicit allocation of money
This paper is based upon a presentation given at the February,