DEVELOPING A SUSTAINABLE INFRASTRUCTURE FOR COMMUNITY POPULATION ESTIMATION IN HAWAII

• Sponsoring Institution: National Institute of Food and Agriculture
• Project Status: TERMINATED
• Funding Source: HATCH
• Reporting Frequency: Annual
• Accession No.: 0216712
• Project No.: HAW00392-H
• Project Start Date: Nov 1, 2008
• Project End Date: Sep 30, 2011

Project Director
Min, H.

Recipient Organization
UNIV OF HAWAII
3190 MAILE WAY
HONOLULU,HI 96822

Performing Department
CENTER ON THE FAMILY

Non Technical Summary
There has long been substantial interest in timely updated and accurate sub-county level population figures and trends in both the public and private sectors. Policymakers, program funders, and businesses need such data to make wise decisions regarding the allocation of critical resources. Government agencies, social service providers, and business entrepreneurs rely on up-to-date and reliable community demographics to assess and justify community needs and to evaluate program outcomes and performance. In the past decade, the demand for such data has increased exponentially due to factors such as political decentralization, community consciousness and mobilization, an increased level of data literacy among people, the emergence of community epidemiology, and the availability of innovative methodologies and techniques. However, most of the demographic data relating to Hawaii's communities are from decennial censuses or private data vendors. The former usually takes two to three years from data collection to data release. When community demographics for the post-censusal years are needed, the common practice is to either directly use or extrapolate from the latest available census data. Although private data vendors can provide community population data in a timely manner, data reliability is often uncertain and the cost of the data is usually beyond the reach of most community organizations. Although block-level estimates are the best way to capture local changes, neither census nor private vendor data can be used directly as community estimates. These data usually require some level of data processing before they can be used as community estimates. In addition to the lack of community population data, the definition of "community" differs by contexts. For example, it could be defined as a high school complex, administrative district, neighborhood board boundary, group of census tracts or postal zip code areas, and other ways. Moreover, the boundary of a given definition is subject to transformation over time due to changes in political affiliation, population dynamics, or landscape. Further complexity stems from the fact that different stakeholders may have specific needs requiring estimates at different levels of detail and precision. For example, some stakeholders may need population estimates by 5-year age categories every 3 years, while others may need annual estimates by single age and sex. To create reliable community demographics that meet the needs of various stakeholders would be cost prohibitive for any single group. On the other hand, without the availability of up-to-date community level demographics, planning, evaluation, and resource allocation decisions would be based on faulty data. The challenge is to research a feasible and cost-effective way to develop a community population estimation system that provides the valid and timely data needed by a wide range of individuals and communities.

Animal Health Component

(N/A)

Research Effort Categories

Basic

30%

Applied

60%

Developmental

10%

Classification

Knowledge Area (KA)	Subject of Investigation (SOI)	Field of Science (FOS)	Percent
802	6099	3080	10%
803	6020	3080	60%
903	6099	3080	30%

Knowledge Area
903 - Communication, Education, and Information Delivery; 803 - Sociological and Technological Change Affecting Individuals, Families, and Communities; 802 - Human Development and Family Well-Being;

Subject Of Investigation
6099 - People and communities, general/other; 6020 - The family and its members;

Field Of Science
3080 - Sociology;

Keywords

small area population estimation,

community demographics, hawaii,

component method ii method, administrative record method,

housing unit method,

remote sensing method

Goals / Objectives
To develop a sustainable and cost-efficient infrastructure for generating on-going and timely community population estimates in Hawaii is the primary goal of this project. These estimates will be precise enough to validly describe community characteristics, adaptable to changes in community boundaries, and useful to stakeholders. The objectives to accomplish the goal are: 1) build a coalition with community organizations, government agencies, and demographic experts to increase support for community population estimation and design procedures for exchanging data and sharing the resulting population estimates; 2) set up a secure infrastructure for data exchange, warehousing, and processing; and acquire reliable data needed for community demographics; 3) use censual year data to understand the fundamental population dynamics and structure in Hawaii at the state and community levels, as well as cross-community population differences which will help to inform the selection of the best estimation technique for Hawaii; 4) assess appropriate techniques and apply the most desired technique for generating intercensual and post-censual community population figures and trends, with consideration of the precision of the estimates, length of production time, resources required, and the needs of key stakeholders; 5) disseminate project results and gather community feedback on the results; 6) develop an enhancement and sustainability plan. The expected outputs are: 1) report on community population change between 1990 and 2000; 2) data warehousing and processing system; 3) 1991-2000 annual community population estimates by using component II method, administrative record method, housing unit method, and remote sensing method; 4) estimation technique evaluation report; 5) documentation of proposed infrastructure and procedures; 6) documentation of requirements and data sources; 7) final community population estimation data and user guide; 8) project evaluation report; 9) enhancement and sustainability plan.

Project Methods
To obtain optimal techniques and procedures for the small area population estimation, the following techniques will be employed: 1) Component II method: It is widely used because it provides detailed information on an area's future population, births, deaths, and migrants by age, sex, and race. This information is useful for many areas of planning and public administration. This method follows each cohort of people of the same age throughout its lifetime according to its exposure to mortality, fertility, and migration (net-migration). The whole procedure is repeated for each year of the projection period, resulting in the projected population. 2) Administrative records: It will be used to gauge the small area population more accurately. It keeps by federal, state, and local government agencies for purposes of registration licensing, program administration. Data on births, deaths, marriages, and divorces are called vital statistics. Other administrative records are school enrollments, building permits, driver's licenses, Medicare enrollees, voter registration lists, and property tax records. These records provide information on variety of demographic events and population characteristics. 3) Housing unit method: It is regarded as one the most reliable methods for making local estimates and one of the easiest to apply, because the necessary data for applying the housing unit method are readily available at the local level and are often the only data readily available at the local analysts. These estimates are typically based on building permits, electric customers, and similar indicators of population and household changes. The general design of the housing unit method is to update the number of occupied housing units from the previous census using data on building permits and demolitions, certificates of occupancy, utility connections, or real and personal property tax information on residential units, and then to convert this current estimate of households into a population estimate by use of a population-to-household ratio. 4) Remote sensing method: It is the measurement or acquisition of information of some property of an object or phenomena by a recording device that is not in physical or intimate contact with the object or phenomena under study. It includes aircraft, spacecraft and satellite based systems. Its products can be analog or digital images. Remotely sensed images need to be interpreted to yield thematic information. It is particularly very useful for population studies and the following are a few of them. Jensen and Cowen, for example, identify three ways in which population estimates can be attained through Remote Sensing. These are individual dwelling units count, measuring urban extent and, land use/cover classification. It also can provide intercensal population estimates. For instance, night-time lights have been used as a proxy to estimate population.

Progress 11/01/08 to 09/30/11

Outputs
OUTPUTS: This integrated research/extension project aims to develop a cost-efficient and sustainable infrastructure to produce sub-county level population estimates by utilizing state-of-art applied demography, information technology, and geographic information system (GIS) technologies. In the first year, we compared the basic indicators containing individual, family, and household information and other social and economic characteristics by using summary file 1 and 3 for both 1990 and 2000 census data that provided us the base population information to be compared with the our population. Due to different definitions and categories between two census questions, however, we had to select only comparable variables from the summary files and compiled the series of documentation that is important for future reference; e.g., the document that explains the procedure of variable selection, the SAS programming files, the codebook and the data dictionary for the selected variables, and the original and the selected variable files. In the second year, our main task was to compile the large datasets from various sources to conduct small area population estimation by using GIS techniques in the third year. The datasets are compiled from various agencies such as from the U.S. Census Bureau, the U.S. Department of Labor, the National Center for Health Statistics, the Hawaii State Department of Education, the Hawaii State Business, Economic Development and Tourism, Department of Health, and the Hawaii State Department of Health. The information included in the dataset is population size by age and sex, migration status, and military population size, births and deaths, life tables, and the enrollment size for private and public schools. To conduct GIS techniques, we finished geocoding and census data compilation. Due to many unstandardized geo addresses in Hawaii, the geocoding was the most important and difficult task in the project and often required manual work to recode unstandardized geo addresses. In the third year, by incoorporating the three main popular methods, called school enrollement methods, housing unit methods, cohort-survival methods, and GIS techniques, we estimated population estiamtes 2000 and 2010 by 43 schoool complex areas. The GIS procedure "intersect" was performed on the map of 43 school districts with the map of census blocks. This kind of overlay of one map on top of another created crisscross of areas. The share of these areas of housing units and population were determined and then aggregated for school districts. One of the main tasks of this project is to create a strong partnership with stakeholders on the estimation method, products, and further development. We have continued to contact and share the information about this project with the state departments such as DOE and DOH to share the ideas and information on this project. In addition, we have met with the state geographers to share this idea and reach an agreement of sharing and exchanging the most update GIS information with COF. PARTICIPANTS: Individuals (me from the Center on the Family served as principal participant): Hosik Min, Shijen He, Shriram Bhutada, Sarah Yuan, Carol Lin, Saiful Momen Institutions (from the State Department) Alvin Onaka, PhD, State Registrar and Chief of the Office of Health Status Monitoring, Department of Health Keith Yamamoto, Chief, Alcohol and Drug Abuse Division, Hawaii Department of Health Joan M. Delos Santos, Data Processing Systems Analyst, State of Hawaii Office of Planning Craig T. Tasaka, Planning Program Manager, State of Hawaii Office of Planning Royce Jones, Hawaii-Pacific Manager, Environmental Systems Research Institute Inc. (ESRI) TARGET AUDIENCES: Administrators of state agencies and local nonprofit community agencies relating to health, human services, education, and economic development Academic researchers PROJECT MODIFICATIONS: Nothing significant to report during this reporting period.

Impacts
This project achieved four main outcomes and impacts. (1) We obtained the base population information by comparing the census data between 1990 and 2000 from the SF1 and SF3 data. We did the rigorous crosswalk between them to match the variables those have different definitions and values and used the distribution methods to align the SF3 with the SF1 because only the SF1 contains the 100% sample, while the SF3 contains the partial. (2) One of the main tasks was to obtain population estimates for 43 school complexes by incorporating various data sources. Using GIS techniques as well as other three major popular methods, school enrollement methods, housing unit methods, cohort survival methods, we estimated population estiamtes 2000 and 2010 by 43 schoool complex areas. The outputs can be provided in both the spreadsheet table format and the maps. Each tables and maps contained the basic population information such as population size by age and sex. However, those estimates for few areas do have some errors because of incomplete information. (3) Another task is to build a sustainable data incorporation for further population estimation. To create the GIS usable datasets was the one of the most challenging parts in this project. To enhance the data accuracy, the team used the private sector information such as zipcode and digital maps. Also, the property tax data including tax map, dwelling records, commercial tax records, permit records, and timeshare construction helped the project enormously. The team set up the procedure to replicate the population estimation for the additional year(s). However, each additional year requires all the necessary sets of data to conduct the estimation. (4) Because this project requires various kinds of administrative data, the center has a successful relationship with DOE and DOH in sharing the data and the idea of this project, and continued to update the current status of the project and exchange the data. The team had the official meeting with the state GIS experts and continued to contact them to solve the geocoding issue.

Publications

• No publications reported this period

Progress 10/01/09 to 09/30/10

Outputs
OUTPUTS: The key outputs in the second year were a compilation of the two large datasets from various sources to conduct small area population estimation in the next year. One dataset is for the demographic techniques, called the housing method, and the other one is for the Geographic Information System (GIS) techniques. The dataset for the demographic method is compiled from various agencies such as from the U.S. Census Bureau, the U.S. Department of Labor, the National Center for Health Statistics, the Hawaii State Department of Education, the Hawaii State Business, Economic Development and Tourism, Department of Health, and the Hawaii State Department of Health. The information included in the dataset is population size by age and sex, migration status, and military population size, births and deaths, life tables, and the enrollment size for private and public schools. All these information is provided at the county level and since 2000 to present. This level of information meets the minimum requirement for the housing method. Another dataset for the GIS technique comprises of two parts; geocoding and census data compilation. The geocoding is the most important and difficult task in this project because Hawaii has a lot of unstandardized geo addresses. As these addresses are not easy to be coded to analyze, the manual work has to be done. These difficulties are found at the neighbor islands, Big Island in particular, and now is wait for the test run. As a strong partnership with stakeholders is critical, we have continued to contact and share the information about this project with the state departments such as DOE and DOH and the state geographers to share the ideas and information on this project. The final results will be reported to the stakeholders in the workshop or conference after we have accomplished this project, and will be published group's aimed refereed journal. PARTICIPANTS: Individuals (me from the Center on the Family served as principal participant): Hosik Min, Shi-Jen He, Sarah Yuan, Carol Lin, Shriram Bhutada Institutions (from the State Department) Alvin Onaka, PhD, State Registrar and Chief of the Office of Health Status Monitoring, Department of Health Keith Yamamoto, Chief, Alcohol and Drug Abuse Division, Hawaii Department of Health Joan M. Delos Santos, Data Processing Systems Analyst, State of Hawaii Office of Planning Craig T. Tasaka, Planning Program Manager, State of Hawaii Office of Planning Royce Jones, Hawaii-Pacific Manager, Environmental Systems Research Institute Inc. (ESRI) TARGET AUDIENCES: Administrators of state agencies and local nonprofit community agencies relating to health, human services, education, and economic development Hawaii State legislators Service providers and advocates who work with at-risk populations Academic researchers PROJECT MODIFICATIONS: Nothing significant to report during this reporting period.

Impacts
As this project is still in the process, real outcomes and impacts are yet to be reported. However, the followings are the progress report from the current work. The main outcomes of the second year are to prepare the datasets for the demographic and GIS analysis. The required datasets for the population estimation methods are from various Federal and State agencies and the data from those agencies are 2000 U.S. Census, 2001-2009 American Community Survey, 2001-2009 Current Population Survey, 2000-2008 Vital Statistics (2008 is the most recent one), 2000-2006 Life Table Report (2006 is the most recent one), 2000-2009 Military Population Report, and 2000-2009 School Enrollments Report from both public and private schools. From those data, we obtain the following demographic information that is essential to conduct the small area estimation: total population size by age and sex, migration status, military population size, vital statistics such as births and deaths, life expectancies from life tables, and the enrollment size for both private and public schools by each grade. All the information we obtained from these data could provide at the county level and are collected from 2000 to present. This level of information meets the minimum requirement for the housing method. We have compiled a series of documentation for future reference: the complete list for the original data sources, the STATA programming document to obtain the selected variable data from the original data, the data dictionary and the codebook for the selected variables, and the elucidation of the procedure we have taken. Another major dataset is for the GIS analysis. Although the GIS techniques require the accurate GIS information, specifically geocoding, the geocoding task involves a rigorous amount of time and effort because of the difficult technical issue such as unstandardized addresses. A graduate assistant who is majoring in GIS and has deeper knowledge and understanding of the Hawaii GIS system has continued to work on these issues and now almost completed. More specifically, we only need to do the test run to confirm the geocoding accuracy. After the test run, we will be able to analyze the small area estimation, which is the main procedure of this project. The data this GIS technique requires are also the same as the demographic methods: various kinds of administrative data such as school enrollment, vital statics data, and population data. These data can be shared by applying GIS information.

Publications

• No publications reported this period

Progress 10/01/08 to 09/30/09

Outputs
OUTPUTS: This integrated research/extension project aims to develop a cost-efficient and sustainable infrastructure to produce sub-county level population estimates. The present action research project will integrate science, technology, and extension to develop a blueprint and lay the foundation for the data system infrastructure. By utilizing state-of-art applied demography, information technology, and geographic information system (GIS) technologies, the project will contain the capacity to produce timely estimates as needed by key stakeholders; a strong and active partnership with stakeholders on the estimation method, products, and further development; the techniques that can incorporate the innovative information with GIS technologies to streamline estimation, eliminate human errors, and reduce operational cost; the capability to enhance estimation precision and details iteratively; and multiple channels to disseminate products and promote stakeholders' buy-ins. Various internal and external resources will be used to ensure the project's success and outcome quality. As for the first year, we conducted a research that compared the basic indicators containing individual, family, and household information and other social and economic characteristics by using summary file 1 and 3 for both 1990 and 2000 census data. This comparison task is very important for the project, because the census data, the most reliable source for the population information, can provide the base population to be compared with the population estimates that we will create by using various estimation methods. Due to different definitions and categories between two census questions, however, we have to select only comparable variables from the summary files. During the process, we have complied the series of documentation that is important for future reference; e.g., the document that explains the procedure of variable selection, the SAS programming files, the codebook and the data dictionary for the selected variables, and the original and the selected variable files. We have created the county level comparison dataset, and are working on obtaining the sub-county level population estimates. Another major task of the first year was to build a strong partnership with stakeholders. As the project requires various sources of data in nature, building a strong partnership with stakeholders is critical. We have continued to contact and share the information about this project with the state departments such as DOE and DOH to share the ideas and information on this project. In addition, we have met with the state geographers to share this idea and reach an agreement of sharing and exchanging the most update GIS information with COF. Because incorporating GIS techniques with current demographic ones is one of the main objectives, this relationship would be a great help on succeeding the project. Following the goals of the project, the results will be reported to the stakeholders in the workshop or conference and published in group's aimed collaborative book(s) or other refereed journal. PARTICIPANTS: Individuals (me from the Center on the Family served as principal participant): Hosik Min, Shi-Jen He, Sarah Yuan, Carol Lin Institutions (from the State Department): Alvin Onaka, PhD, State Registrar and Chief of the Office of Health Status Monitoring, Department of Health; Keith Yamamoto, Chief, Alcohol and Drug Abuse Division, Hawaii Department of Health; Joan M. Delos Santos, Data Processing Systems Analyst, State of Hawaii Office of Planning; Craig T. Tasaka, Planning Program Manager, State of Hawaii Office of Planning; Royce Jones, Hawaii-Pacific Manager, Environmental Systems Research Institute Inc. (ESRI) TARGET AUDIENCES: Administrators of state agencies and local nonprofit community agencies relating to health, human services, education, and economic development Hawaii State legislators Service providers and advocates who work with at-risk populations Academic researchers PROJECT MODIFICATIONS: Nothing significant to report during this reporting period.

Impacts
It is important to note that because all current projects are in the process, there are no real outcomes and impacts yet to be reported. However, the followings are the progress report from the current work. (1) One of the main tasks in the first year is to compare the census data between 1990 and 2000, which is the most reliable data on population. The results of the comparison would serve as the base population to be compared with the estimation outcomes generated by various demographic and GIS methods we would employ. To link the data with geographic location, we used the Census Bureau's summary files (SF). We did the rigorous crosswalk between them to match the variables those have different definitions and values. In this process, we have acquired the series of documentations for the future reference: the complete list for the original and selected variables; the SAS programming document to obtain the selected variable data from the original SF; the data dictionary and the codebook for the selected variables; and the elucidation of the procedure we have taken. The use the SF, however, has a limitation: Only the SF1 contains the 100% sample, while the SF3 contains the partial. Therefore, we are in the process of applying the distribution methods, considering the land size, the population size, and the population composition, to align the SF3 with the SF1. (2) Another major task is to build a strong partnership with stakeholders. Because this project requires various kinds of administrative data such as school enrollment, vital statics data, and GIS information, without the help from those government agencies it would be extremely difficult to achieve the project goal. In addition, considering that the project outputs would be used by those agencies for the decision-making, the use and contents of the final product would be decided based on their needs, and the product would be updated in continual basis, forming a strong and close partnership is crucial. The center has a successful relationship with DOE and DOH in sharing the data and the idea of this project, and continued to update the current status of the project and exchange the data. GIS is one of the most challenging parts in this project. In order to succeed this project, the center has to incorporate GIS techniques into small area population estimation, and the GIS techniques require the accurate GIS information, specifically geocoding. The geocoding task, however, involves a rigorous amount of time and effort because of the difficult technical issue such as unstandardized addresses. To solve this issue, the team had the official meeting with the state GIS experts. And the meeting led us to exchange the mutual interests on this project and agree to share the information through the project. In addition, to increase our capacity of GIS data handling, the center hired a graduate assistant who is majoring in GIS and has deeper knowledge and understanding of the Hawaii GIS system. This collaboration would bring us not only to save tremendous time and effort for our GIS experts, but also to increase the accuracy of the population estimates that we create.

Publications

• No publications reported this period