Data warehousing and decision support at the national agricultural statistics service

Easy access to large collections of historical survey and census data and t he associated metadata that describes them has long been the goal of resear chers and analysts. Solutions to problems such as understanding the behavio r of current survey data, respondent burden, improved statistical technique s, and data quality are often found in the careful analysis of historical d ata. Many questions have gone unanswered because the data were not readily available, access was limited, metadata were not well defined, or query per formance was intolerably slow. This article describes the database modeling techniques that permit end users fast and easy access to large amounts of microlevel data contained in different data systems and from different time frames. Also, techniques for tracking metadata changes and standardization are discussed. A generalized dimensional model is presented that can be us ed for any census or survey to track the full history of the data series.