Efficiency and performance drive most public sector modernization efforts. Government mandates also may require this. For U.S. agencies, the Government Results and Performance Act (GRPA) and the Government Paperwork Elimination Act (GPEA) both require them to improve processes and automate activities, when appropriate. GPEA calls for all U.S. federal agencies, including the Census Bureau, to allow the public to submit information and engage in transactions electronically.
Against this backdrop, the Census Bureau retained Fenestra Technologies Corp., to plan and create a data capture and exchange system known as GIDS (General Instrument Design System) for the 2002 Economic Census initiative. GIDS streamlines the development of Optical Character Recognition-ready paper and Web services-ready electronic forms, facilitates sharing and reuse of metadata using eXtensible Mark-up Language (XML), and dramatically lowers respondent burden and manual efforts of Census personnel.
The United States Census Bureau, part of the U.S. Department of Commerce, conducts an economic census to millions of businesses across the country every five years. “The Economic Census is indispensable to understanding America’s economy,” according to Alan Greenspan, Chairman of the Federal Reserve Board of Governors. Current economic statistics all have their origins in the data that are collected in this census.
In this effort, more than 650 individual survey forms (10-12 pages each) are sent out. Because the forms measure current economic activity, their questions change every five years to reflect the changes in the dynamic U.S. economy. They are customized for specific industries, according to the North American Industry Classification System (NAICS). Therefore, the portfolio comprises over 6,000 pages of forms – a massive body of questions.
Realizing that its process for authoring these surveys was unwieldy and overly time-consuming, the Bureau set requirements for a new process and system. The requirements were to be able to:
Key considerations in developing this advanced system included:
During the development of GIDS, Fenestra focused on gaining a deep understanding of Census’ data and its uses: identifying the data to be captured (both numerical and text data), creating models for structuring and storing survey metadata, and generating policies and methodologies to refactor captured data and metadata for sharing and re-use. (Metadata in this application define both the data model and characteristics of the forms or questionnaires.)
After thoroughly understanding the business context and employing rigorous R&D methodologies, our team designed and built the Generalized Instrument Design System (GIDS). Now nonprogrammers can create complex survey questionnaires that can be deployed and responded to in paper or electronic formats for massive data collection and capture.
The use of a metadata repository that differentiates between layout and content allows the automatic generation of about 80 percent of the 650 forms needed (both hard copy and online), and also allows response content to be reused across paper and electronic variations. Since it includes a sophisticated paper and electronic typography system, GIDS can also be used to produce both survey form layouts and highly complex data dissemination reports.
As deployed by the Census Bureau, GIDS facilitates:
GIDS uses XML to enable capturing systems to communicate seamlessly with the Census Bureau’s storage systems. Using XML as the basis for data interchange helps to insure that the information remains accessible and transportable, as the systems at the Bureau change over time. Data quality and access are enhanced with this “capture once and re-use” approach to data gathering, editing and publishing.
GIDS and the use of Web Services help Census:
Commercial-off-the-shelf (COTS) products aren’t always the answer. Therefore, it is crucial to understand the dynamics of research and development as they affect your agency.
Bear in mind that technical solutions exist only to support fundamental business solutions. These solutions are most effective when based on careful research both to understand the data and its purpose, as well as to address work process changes and evolving staff responsibilities.
Among the guidelines to increase the likelihood of success:
Identify/respond to project stage: Learn to recognize if a project is at a research, enhancement, addition, or upgrade phase. Each phase requires a different strategy and approach. Combining the wrong approach with the wrong phase is a near certain recipe for project disaster. For example, a research project includes many unknowns that can affect the ability to schedule with a high degree of accuracy. On the other hand, an upgrade usually involves known issues, and release planning can be done quite easily.
Find Commonalities: Look for interfaces and linkages that can unify programs and processes. For the Census Bureau, XML provided a uniform basis for interchange between data gathering/publishing systems and data storage systems.
Support Cultural Change: Changing technology without adjusting the surrounding business processes guarantees minimal success. It is vital that you build understanding between stakeholders groups of the importance of business process change to the success of the overall project. Individuals not immediately involved in the core project rollout must be apprised regularly so that they can become productive team members when needed.
Watch for Systemic Errors: With automated processes, a single error or defect will often appear to occur many times rather than just once. This is normal. However, individuals used to a one-at-a-time approach may have some difficulty adjusting to this concept. Seeing hundreds of errors appear can be unnerving, yet those are the easiest to resolve.
Don’t be Afraid to Start Over: After the learning phase of a program, effective approaches that worked may begin to exhibit shortcomings that seem correctable. But often, the wisest course is to take the valuable lessons learned and start at the beginning—building a more elegant solution that more often than not represents a huge improvement over the first iteration.
Perform Due Diligence at the Outset: Moving to an automated process moves most of the costs and effort to the creation of the first successful deliverable. The payoff comes after that.
Our champion is the U.S. Census Bureau. Through their visionary thinking and willingness to transform core processes, they are realizing the true benefits of reusable metadata – and XML.
Over time, the agency plans to build on this reservoir of knowledge, practical experience and reusable software components to enable data collection and processing to proceed more accurately and less expensively in the future. Hopefully businesses and other agencies will adopt and use these capabilities, in order to accelerate complex information collecting, capturing, sharing, analyzing, publishing and editing processes for timely decision-making and proactive leadership.
John Elrick is a software designer and developer with nearly two decades of experience in architecting software solutions for businesses and government. He is currently the Projects Manager at Fenestra Technologies.
Fenestra researches, designs and deploys practical and turnkey solutions for highly complex content problems. Using a "triage" approach, the firm deepens feature sets and accelerates faster implementation cycles.
Core competencies include forms creation and processing, data collection and validation, complex typography formatting, and the creation of dynamic, scalable and flexible data management solutions for both electronic and paper purposes.
The U.S. Census Bureau continues to partner with Fenestra (since the early ‘90’s), and companies including IBM, ExxonMobil, and Advanced Productivity Software have also retained the firm for specialty expertise.
Further information about this case study and GIDS is available at http://www.fenestra.com.