Please note: Older issues of the newsletter are likely to contain
broken links -- the newsletter is presented here "as published."
Table of Contents
Researcher's Notes by Heather Boyd
- Using the World Wide Web to Conduct Surveys
- Record of American Democracy (ROAD)
- American Community Survey
To Sample or Not to Sample
Since our last issue of DPLS News, Congressional opposition to the Census Bureaus plans to augment the Census 2000 with statistical sampling methods remains, although the disaster relief bill was finally passed last spring without the rider forbidding the use of sampling. Instead, language was inserted requiring a full report from the Census Bureau detailing the proposed methodologies and error rates. The report, which is now available at http://www.census.gov/main/www/stat_activities.html (in PDF format), persuasively documents the reasons behind the Bureaus plan to use limited sampling methods to reduce costs and improve accuracy of the upcoming decennial census within the Constitutional requirements for apportioning U.S. representative seats.
The current appropriations language (passed by the Senate July 29) may still encumber the planning for Census 2000 by prohibiting the use of FY 1998 funds "to make irreversible plans for the use of sampling" in taking the 2000 decennial census (S1022, section 209).
Why is Congress resisting the tide of evidence in favor of statistical sampling? According to the COPAFS April-June Newsletter (Council of Professional Associations on Federal Statistics), "Some Republicans in the House fear that they would lose about 25 seats in the House if minorities (who tend to vote Democratic) are counted due to adjusting the census." Wisconsins own Senator Kohl--a Democrat--also opposes the use of sampling because Wisconsin would have a lost a Congressional seat to California if sampling was allowed to correct the 1990 Census.
According to the Census Bureau, "The 1990 Census failed to match the accuracy of the 1980 Census despite the Census Bureaus exhaustive attempt to make traditional methods work.... The undercount was 4.7 million people; the undercount rate of 1.8 percent in 1990 was 50 percent greater than the rate had been in 1980." The Milwaukee Complete Count Campaign, which boosted the mail response by 2% but still left out 2.9 percent of the citys population according to the Post Enumeration Survey, "demonstrated the limits of spending money on outreach."
Children make up 26% of the U.S. population, but 52% of the undercount. Others disproportionately affected are rural renters, African Americans, Hispanics, and especially American Indians living on reservations. The Bureau claims the cost of conducting the census using physical renumeration only will be $675 million more than if limited sampling is allowed.
Updates since press time:
- November 14, 1997
- November 10, 1997
- October 24, 1997
- October 3, 1997
- October 1, 1997
- September 29, 1997
- September 26, 1997
- September 23, 1997
- September 15, 1997
GSS Student Paper Contest
The National Opinion Research Center (NORC) at the University of Chicago announces the fourth annual General Social Survey (GSS) Student Paper Competition. The papers will be judged on the basis of their: a) contribution to expanding understanding of contemporary American society, b) development and testing of social science models and theories, c) statistical and methodological sophistication, and d) clarity of writing and organization. Papers should be less than 40 pages in length (including tables, references, appendices, etc.) and should be double spaced. Both undergraduates and graduate students may enter and college graduates are eligible for one year after receiving their degree.
Papers will be judged by the principal investigators of the GSS (James A. Davis and Tom W. Smith) with assistance from a group of leading scholars. Separate prizes will be awarded to the best undergraduate and best graduate-level entries. The winners will receive a cash prize of $250, a commemorative plaque, and the MicroCase Analysis System, including data from the 1972-1996 GSS.
Two copies of each paper must be received by February 15, 1998. The winner will be announced in late April, 1998. Send entries to: Tom W. Smith, General Social Survey, National Opinion Research Center, 1155 East 60th St., Chicago, Il 60637. For further information, phone: 773-256-6288, fax: 773-753-7886, email: email@example.com.
by Heather Boyd
What are the relationships among aspects of Christianity and environmentalism? The 1993 General Social Survey (GSS) includes items on religion as well as an extensive battery of questions about environmental attitudes and behaviors. Its national random sample makes results potentially more generalizable than many previous studies that used neither random samples nor national samples. Further, the presence of multiple indicators of environmentalism allowed me to construct indices (with Cronbach's alpha ranging from .59 to .85).
I have submitted a journal article that is now under review. Briefly, the most robust predictor of environmental thought and behavior was "fundamentalist tradition" (as defined by NORC's GSS Methodological Report no. 43). Bivariate correlations revealed that fundamentalist tradition (as opposed to "moderate" or "liberal" tradition) was inversely related to willingness to spend to protect the environment; perception of danger posed to the environment by pollutants; and frequency of environmental behaviors, such as recycling. Biblical literalism and belief in God were also inversely related to willingness to spend. However, having a religious tradition, as opposed to having no preference, was positively associated with willingness to spend to protect the environment. And, frequency of prayer was positively associated with frequency of environmental behaviors.
After demographic influences were controlled for using hierarchical linear regression, fundamentalism significantly accounted for variance in perceived danger and frequency of environmental behaviors. Also, frequency of prayer significantly accounted for variance in frequency of environmental behaviors. If you would like more information about this project, please contact me at firstname.lastname@example.org.
Children's Stats Evaluated
Amidst the masses of data collected by federal agencies for various purposes, many facts about the nations children can be found. Whether these scattered facts are enough to show what is happening to children and how they are faring in society has been called into question. A recent Executive Order by President Clinton to the Interagency Forum on Child and Family Statistics culminated in the 1997 publication: Americas Children: Key National Indicators of Well-Being, published by the forum. The publication (available in PDF format on the Web at http://www.cdc.gov/nchswww/about/otheract/children/child.htm) describes 25 vital indicators with graphs and highlights by population subgroup. The key indicators fall under the main topics of population and family characteristics, economic security, health, behavior and social environment, and education. A yearly theme is also identified--child abuse and neglect, drawing on the periodically collected Third National Incidence Study of Child Abuse and Neglect, (1993-1994).
Perhaps more interesting than the indicator trends are the gaps identified by a lack of consistent data collection in specific areas: "homelessness, long-term poverty, mental health, violent crime and other behavior problems, early childhood development, and children with special needs" (p. 2). The Forum, a collaboration among key federal statistic-gathering agencies such as the Bureaus of the Census, Labor, and Justice, and National Centers for Education Statistics and Health Statistics, promises to address those gaps in future efforts.
An earlier effort to examine gaps in national statistics on children resulted in the publication Integrating Federal Statistics on Children: Report of a Workshop, Washington, D.C.: National Academy Press, 1995. This volume includes an in-depth evaluation of 12 major continuing surveys and suggestions for improvement on their value for research on children. Participants included UW Sociology Professors Robert M. Hauser.and Larry Bumpass.
Online Archival Repository:
Wis. Migrant Farmworkers
The 1978 survey was based on a 10 percent statewide sample of migrant workers, age 16 years or older, using lists of names randomly selected from employers' payroll files during July and August. The 1989 survey used the same sampling method and a slightly revised version of the 1978 interview instrument. Since severe budget restrictions precluded a statewide survey in 1989, the follow-up study was limited to three counties--Columbia, Dane and Waushara--containing about one third of the migrants expected in the season. These three counties are an approximate representation of the state as a whole. The focus of the surveys was on the health care needs of migrant farm workers, and their opinions on obtaining health care in the local area. Extensive information was collected on chronic conditions, illnesses, and medical utilization patterns.
Association of Public Data Users Conference: Preserving Public Data, Washington, D.C.: Oct. 20-22, 1997. Fee: $350 or $500, Email: email@example.com, phone: 504-280-3154.
Analysis of Complex Data Sets Seminar, Univ. of Cologne, Germany: March 2-20, 1998. Fee: 300 DM. http://www.za.uni-koeln.de/events
Using the World Wide Web to Conduct Surveys
Have you ever wondered about using a Web site to conduct surveys and collect information from site visitors? If you are interested in such experiments, you can head to VLAB (http://elsa.berkeley.edu/vlab). Some researchers at the Dept. of Economics at the University of California-Berkeley have created this site with funding from the National Science Foundation. Their goal is to explore the use of the Web as a medium for economic surveys, experiments and instruction. Currently there are four different surveys, which require 5-10 minutes to complete. You are invited to have a test-drive and your suggestions for improvements are welcome.
Record Of American Democracy (ROAD)
The ROAD project provides United States electoral outcomes from 1984 through 1990 in various geographic levels. "The NSF-supported Record of American Democracy (ROAD) project includes a wide variety of election returns, socioeconomic summaries, and demographic details about the American public at unusually low levels of geographic aggregation."
Approximately 30-40 electoral variables for all of the roughly 170,000 precincts nationwide are included for state and national elections. Added to the political variables for each year in every state, 3,725 variables have been merged from the 1990 U.S. Census. The collection also includes geographic boundary files so users can easily draw maps with these data. Its URL is http://gking.harvard.edu/roaddoc/road.html.
American Community Survey
The Census Bureau has begun to release data from the first stages of the experimental American Community Survey. As part of the Continuous Measurement System, the new survey is designed to capture small area socio-economic data that becomes quickly out of date with each decennial census. Anticipating the "devolution" of federal programs to state and local block grants, the Census Bureau wishes to provide local planners, as well as federal statistics users, with accurate population and housing figures each year based on the "continuous" surveys. The goal of providing yearly estimates for every community in the United States will take 7 years according to the Bureaus plan:
* Demonstration period 1996-1998
* Comparison sites 1999-2001
* National comparison sample 2000-2002
* Full implementation 2003 and beyond
The 1996 data released on the website are preliminary and for only four communities from the demonstration period. Also on the website are variable definitions and explanations of their compatibility with the 1990 Census of Population and Housing.
Another fruit of the Federal Interagency Council on Statistical Policy (see previous article "Childrens Statistics Evaluated") is a new website designed for "one-stop shopping for federal statistics." FedStats organizes online data from 70 agencies into an alphabetical index of 275 terms. In addition, users can search for their own keywords across 14 statistical agencies (or by check-boxing selected agencies). A regional statistics page from several agencies and a Fast Facts section help make this tool user-friendly for data users of all levels of expertise.
End of October 1997 Newsletter.