The Census Web page is your one stop source for Census information and data: http://www.census.gov

Key Products:

• Census of Population & Housing
• Economic Census
• County Business Patterns
• Census of Housing

• Integrating information
• Measuring & Describing Change over Time
• Measuring & Describing differences in location (space)

All are factors in how you might approach using census data as an analytical tool

A.  Political representation

Article 1, Section 2 of the Constitution requires decennial census--primary purpose to assign seats in the House of Representatives to the state. Also used to draw districts at the state & local level.

B.  Revenue Implications

Since WWII, federal aid programs and revenue sharing funds have been allocated by formulas that rely on census data--not just pop, but per capita income, unemployment rates, age of housing, etc. For example: in 1989, a total of $58.7 billion, $236 per capita was transferred from federal to state and local governments for formula grants.

C.  Difficulties in Counting/Data collection

Becoming increasingly difficult. more heterogenous - non-English speaking, homeless, etc. Nonresponse increasing -- 63 % in 1990, 75% in 1980 on first mailing

D.  Undercounting

Concerns increasing--particularly in undercount of minorities. Census bureau conducts a "post mortem" after each census. A series of studies of what worked well and what didn't. Reduction in undercount is a primary goal.

E.  Political Aspects of the Census

Questionnaire Content. Content determination through three stages: considered needs of users outside of federal gov, Bureau formed 10 interagency working groups, chaired by staff and organized along content lines, and OMB organized a federal agency council to advise OMB on census data needs.

Reporting the Data. This is less well documented and is based on user input in regional and local meetings.

Adequacy of Coverage: basically a political dispute over the accuracy of the data

Budget: Determines how well bureau gets to do its job. 1990 cost $2.58 billion

Basically four universes: population, workers, housing, households.

Basic definitional units:

• housing: housing unit (room or group of rooms intended as separate living quarters)
• household: occupied housing unit -- households=#occupied housing units; one or more person sharing a housing unit - household size
• Families - persons living together and related by blood, marriage or adoption
• Population: persons - counted at place of usual residence
• Economic status: income/labor force participation; individual/family-per capita/hh income; data on persons at work

Housing sample universe

housing unit sample frame - mailed to addresses - strives for 100 percent sampling of housing units.

Household universe

defined by occupied housing units--actually consists of three intersecting universes:

1. occupied housing units

2. household groups living in those quarters

3. persons in the households

each has a separate set of characteristics, ie HU, # rooms, hh # persons, persons -- age

Population Universe

ultimate goal to obtain a complete population count - contains info on individuals abstracted from their household and housing contexts. variables include age, race, and sex Two key variables contain information on individual relationships with others: marital status and household relationship

Labor Force/Employment Universe

Defined on basis of population: persons who seek and hold jobs--key rational for defining as a separate universe is that jobs are held at a different location than place of residence.

Labor force: included both employed and unemployed persons who are available for work.

Separation of employment from the housing unit and place of residence makes local analysis difficult for this universe

Content of Questionnaire

Data collected by two different questionnaires: sample and 100% (hand out, forms if possible)

short form questionnaire - to most households

long form - to sample, includes short form questions and other questions answered by a sample of the pop. It is subject to sampling error

Household Linkage

Average household size=occupied housing units x persons per unit

Occupied housing units = noninstitutional population/persons per unit

Data are reported for 39000 jurisdictions and 7 million blocks nationwide.

The quality and quantity of information increases as the geographic scope increases. (hierarchy principle)

Pitfalls

1.  the same name may be applied to more than one spatial unit. For example: LA CMSA, LA-LB PMSA, LA urbanized area, LA county, LA city

2.  geographic areas at lower levels may be subdivided by higher level geographic units. A common example is split tract by city limit.

TIGER (topologically integrated geographic encoding and referencing system). based on topology, a branch of mathematics that describes the spatial relationships among points, lines, and areas in a two-dimensional plane. Each line segment and point is coded with a series of geographic identifiers (name of street, address ranges, l & l, potilitcal/statistical, etc). national file is 25 gigs/average county 5.4 megs

Comparability of Boundaries over time: Tracts are most stable, however, are changed occasionally. Goal to maintain tracts at around 4000, but range from 1500-8000. Published books show changes

1. Summary Tabulations

aggregated by geographic area. published in standard printed format and also available on CD-ROM (you'll get a chance to work with this)

2. Microdata (PUMS)

Public Use microdata samples. Randomly selected records from the sample questionnaires and contain the full set of data (raw numbers). Three sample rates--1:1000, 1:100, 5:100 -- smaller rate used for national sample, higher rates for local areas. Only available down to county or MSA level. 100000 min pop

100% v Sample

Sample data are weighted up to reflect #'s in entire pop.

Rate: 1 in 6, higher rates for smaller areas (1 in2 for juris less than 2500)

is subject to sampling variability and error. Can calculate statistical si

Editing and Imputation of Basic Records

Census exercises editing procedures to improve data quality. Surveys are screened for missing or erroneous entries. Efforts made to catch before processing. CB uses two procedures: allocation and substitution.

Allocation:

Made in central processing to fill missing entries or to replace unacceptable entries.

Procedure is to replace blanks with info borrowed from a different household. Replacement is drawn from last complete questionnaire that had chars similar to the present case. Believed to yeild better results that missing values.

Substitution:

In some cases, entire record is missing for a housing unit or person. Can occur because of nonresponse or mechanical error. If CB believes the person or unit exists, a full set of characteristics is substituted for that record. Only applied to 100% data items.

Accounting for imputed info:

publishes tables of imputed cases

Undercount:

No adjustment made in 1990 or 2000. overall undercount is not important, is by geographic area. Effectively askes CB to generate substitution records for persons never directly observed or known to live at an address

Major Census Data Groupings and Linkages

Source: Analysis with Local Census Data, Myers, 1993

Source: Analysis with Local Census Data, Myers, 1993

[Home | Syllabus | Schedule | Assignments | Project |  Links]

This page maintained by Bob Parker, ©2002
October 07, 2003