Data in the Wild: A KM Approach to Collecting Census Data Without Surveying the Population and the Issue of Data Privacy

Research Motivation

Why do we need a new approach to census data collection? First, the census is expensive. The reported cost of the 2010 census was approximately $13 Billion (United States Government Accountability Office,.2011). The 2020 Census, if administered the same way, will cost approximately $17.5 Billion (House Oversight Committee, 2017; Scherer and Bahrampour, 2017). Along with the census costs, the American Community Survey (ACS) (a newer development discussed later) costs as much as $204 million per year to administer (Griffin, 2011). The proposed strategy would drastically cut costs by using existing open source data sets as well as existing government raw data to reduce or eliminate collecting similar or identical information through Census surveys or the ACS. The US Census Bureau would also begin to utilize state information databases to cut down on its data collection efforts. California and Hawaii have robust websites with information on their specific states already in use, making some of the data the census and ACS currently collects redundant for those states.

Second, the census is not accurate. Accuracy is crucial for decision making. Unfortunately, the census may be inaccurate as most of the questions on the ACS have the potential to produce inaccurate information because of the design of the questions and the way answers are provided. Specifically, most questions about annual income and monthly and annual expenses are answered by providing write in totals. This has the potential of producing data that is not accurate as people have no incentive to be exact when answering the questions nor do they know exactly how much they spend. In addition, the data provided is not checked against any other systems to determine the data’s accuracy. The length of the survey (consisting of a minimum of 11 pages and 48 questions per member of a household with multiple parts to majority of the questions) introduces fatigue. Thus, if options exist for automated data collection, self-reporting and long surveys should not be the first choice. A more accurate alternative to survey data is open source or other data sources. For example, utility companies have a vested interest in keeping accurate information for billing purposes, and banks have a vested interest in keeping accurate track of how much money their clients have on account.