Public Use Microdata Samples (PUMS)
Public Use Microdata Samples v Tabulated Data
The Bureau of the Census publishes statistics in 2 format, tabulated statistics and public use microdata samples (PUMS).
- Tabulated data are preconfigured tables, such as those in books.
- Microdata samples reflect the individual answers of each respondent, excluding personal identifying information. Data are not compiled into tables. Users create their own tables with statistical packages, such as SPSS or SAS. This sometimes requires a good amount of statistical sophistication.
The following illustrates differences. Suppose 4 people complete a census questionnaire as illustrated in figure 1.
Figure 1
Response to Fictitious Census Questionnaire
Name of Householder Address Spouse's Name Children, Their Sex, and Their Age Population in Housing Unit Highest Grade of School the Householder Completed Income in 1999 Jones, Sam 126 Kenmore Ave Mary Sam M 10 5 12 45,000 Joan F 12 Kathy F 9 Thomas, Steven 127 Kenmore Ave Karen 2 12 50,000 Kelley, George 128 Kenmore Ave Jill Erin F 15 5 16 70,000 Jason M 11 Emily F 9 Wu, Gang 129 Kenmore Ave Hong M 2 2 14 60,000 Figure 2 represents PUMS data illustrating how each of the 4 households responded. It is no longer possible to identify who completed each questionnaire.
Figure 2
PUMS Data Based Upon Fictitious Census Questionnaire
Number of People in Housing Unit Sex of Householder Presence of Spouse Sex and Age of Child 1 Sex and Age of Child 2 Sex and Age of Child 3 Highest Grade of School the Householder Completed Income in 1999 Case 1 5 M Yes Male-10 Female-12 Female-9 12 $45,000 Case 2 2 M Yes 12 $50,000 Case 3 5 M Yes Female-15 Male-11 Female-9 16 $70,000 Case 4 2 M No Male-2 14 $60,000 Figure 3 displays the fictitious tabulated data.
Figure 3
Tabulated Data Based Upon Fictitious Census Response
Population 14
Presence of children 18 and under by sex Male
3
Female
4
Highest grade of school householder completed Mean 13
Median 13.5
Income Mean
$56,250
Median
$55,000
Per capita
$16,071
Why bother using PUMS?
Some questions can only be answered using only PUMS data. Suppose you want to know the median income of widowed polish women who live in owner occupied homes in the U.S. Figure 4 displays the best one can do using tabulated data. Explanations show how this does not answer the question.
Figure 4
Limitations of Tabulated DataSource: 2000 Census of Population and Housing, Summary File 4
Polish widows in the U.S. 381,049 Median income of Polish females. (This figure includes all Polish women regardless of marital status.) $19,230 Polish females living in owner occupied housing units with no husband present. (This includes women all Polish women who are unmarried.) 6,188,750 PUMS data enables you to 1) isolate Polish females living in owner occupied housing units with no husband present who also happen to be widows, and 2) calculate median income within this subset.
Sources of PUMS Data
1990 Census of Population and Housing. Public Use Microdata Samples. A (5%) Sample (U.S. Bureau of the Census)
US CD-ROM 67: C 3.285: CD 90-PUMSACensus 2000 Public Use Microdata Sample (PUMS) (U.S. Bureau of the Census)
Capen Multimedia Center US DVD 27: C 3.285: P 96/ 2000
Capen Multimedia Center US DVD 28: C 3.285: P 96/ 2000Integrated Public Use Microdata Series (IPUMS) (University of Minnesota. Minnesota Population Center)
http://www.ipums.umn.edu/
Created PUMS data for the censuses of 1850-2000, plus the American Community Surveys of 2000-2004. Extract data for use with statistical packages, or analyze statistics online. Users must completethe free registration process. These data are comparable to those in the MicroAnalyst series.Inter-university Consortium for Political and Social Research (ICPSR) (University of Michigan. Institute for Social Research (ISR))
http://www.icpsr.umich.edu/
ICPSR is the largest data archive in the world that makes available to its members data files and related documentation. Contact Leslie McCain, UB's ICPSR liaison, for additional information. ( 645-3102; mccain@buffalo.edu )ICPSR maintains a Census 2000 Web site at http://www.icpsr.umich.edu/CENSUS2000/ . Search for other census data on the ICPSR home page.
MicroAnalyst (QueryLogic)
Lockwood Reference Desk CD-ROM HA181 .M53 1999
Four CD-ROMs provide PUMS statistics covering 1850-1990. These data are comparable to those in the IPUMS series.Public-Use Microdata Samples (PUMS) (U.S. Bureau of the Census)
http://www.census.gov/main/www/pums.html
Covers decennial census data from 1980-2000, and the American Community Survey.
Home