Title: Using the Current Population Survey CPS in Social Science Research
1Using the Current Population Survey (CPS) in
Social Science Research
Wei Sun College of Arts and Letters University of
Notre Dame
2Why Use the CPS?
- 1. Up to date annual data can be used between
two Decennial Censuses - 2. Diversity diverse information on the labor
force, demographics, and other specific topics - 3. Popularity Primary source of monthly
estimates of total employment, self-employed
persons, domestic and unpaid workers, wage
workers, and unemployment widely used by
researchers - 4. Reliability large and nationally
representative sample for generating reliable
statistical inference
3Major Social Science Data Sets Used by Journal
Articles
Source ICPSR
4CPS Overview
- The CPS is a monthly survey conducted by the
Bureau of the Census for the Bureau of Labor
Statistics. - The CPS is a probability sample designed to
produce national and state estimates of labor
force characteristics and demographic status of
the civilian, non-institutional population 16
years of age and older.
5Overview (continued)
- The CPS samples housing units from lists of
addresses obtained from the Decennial Census of
population and housing. - A housing unit is interviewed for 4 consecutive
months, then dropped from the sample for the next
8 months, then interviewed for 4 more months. - Each month has eight rotation groups. This method
is a compromise between a permanent sample and a
completely new sample.
6Locating Information on Survey Design
- Details about survey design can be found in CPS
Technical Paper 63RV, 2002. Please visit the
Census Bureaus web site to locate the report at
-
- http//www.census.gov/prod/2002pubs/tp63rv.pdf
7Delivery Format
- Monthly Current Population Survey
- Basic monthly data including demographic and
labor force data - CPS Supplements
- March CPS- Annual Social and Economic Supplement
Includes annual data on income, work experience,
non-cash benefits, and migration
8Delivery Format (continued)
- Other Supplements
- January displaced workers, tenure, mobility
- February contingent workers
- April child support
- May tobacco use, work schedules
- June fertility
- August job training
- September volunteers
- October school enrollment
- November voting and registration
- December food security
9Accessing CPS Data Programming Skills Required
- 1. http//www.census.gov/cps/
- Download raw data and extract data with
statistical software - 2. http//www.nber.org/
- From 1992-2006
- Data sets, document files, and extraction
programs - 3. IPUMS-CPS
- http//cps.ipums.org/cps/index.shtml
10Accessing CPS Data Less Programming Skills
Required
- Data Ferrett http//dataferrett.census.gov/
-
- A data search engine created by the Census Bureau
and the BLS - Includes March CPS back to 1992 and other
supplements - Dont need to create extraction program, data can
be downloaded in your preferred format - Can tabulate several types of data
- Easy for undergraduate students to use
11The Sample Size
- March 2005 CPS Supplement
- Over 600 Variables
- Total Records 396,461
- 98,664 Household Records
- 87,147 Family Records
- 210,648 Person Records
12The Raw Data Structure
- The CPS is delivered as a hierarchical raw data
file with three levels - Household
- Family
- Person
13Illustration of Record Sequence
- Household 2 (2 families)
- Family 1 variables
- Person 1 variables
- Person 2 variables
- Family 2 variables
- Person 1 variables
- Person 2 variables
- Person 3 variables
- Household 1 (1 family)
- Family 1 variables
- Person 1 variables
- Person 2 variables
- Person 3 variables
14Importing Raw Data
- The CPS March file has three record types.
- The value in column 1 indicates what
- record type the line is.
- Importing raw data is a process of
- converting hierarchical data to
- rectangular data.
- SAS, SPSS, and Stata programs
- are available at www.nber.org
-
15Output in SAS Hierarchical Nature of Data File
Reflected in Rectangular Format
16Sample Weight Variables
- Household weight
- Family weight
- Final weight
- Earnings weight
- March supplement weight
17Major Household Variables
- Household type
- Economic (Income and non-cash benefits)
- Housing characteristics
- Geographic variables
18Major Family Variables
- Family type
- Family structure
- Family income (poverty level, source of income,
non-cash benefits)
19Major Person Variables
- Family Interrelationship
- Core Demographic
- Race and Ethnicity
- Education
- Labor Force
- Income
- Migration
- Health Insurance
20CPS Compared to the American Community Survey
(ACS) PUMS
- CPS has smaller sample Unable to produce
reliable single-year estimates for some states or
sub-state areas - CPS is better source of official national
estimates of poverty - Some questions are similar and overlap
- CPS is the primary source of information about
national trends in the family - CPS collects more than 50 sources of income
including non-cash benefits while ACS only
includes 8 sources
21CPS Compared to the Survey of Income and
Program Participation (SIPP)
- Both collect similar information on income
source, labor force, and program participation.
SIPP has more information on assets and
liabilities - CPS panel duration is shorter than SIPP, SIPP
follows people across waves while CPS does not - March CPS Sample is larger than SIPP
- SIPP over-samples low income households, CPS is
less biased - Detailed comparisons are based on research
questions.
22How CPS Data is Used in Research By Method
- 1. Cross sectional analysis (single cross
section) - Mostly commonly used research method
- 2. Time series aggregate analysis for long-term
trends - Example Sources of Health Insurance and
Characteristics of the Uninsured. - http//www.ebri.org/pdf/briefspdf/EBRI_IB_10a-2006
1.pdf
23How CPS Data is Used in Research Pooling CPS
Data
- 3. Pooling the March CPS Across Years
- Why pool CPS data?
- Increases sample size, especially for studying
minorities, immigrants, single families, and
other population subgroups. - Research example
- Farley, Reynolds and Richard Alba. The New
Second Generation in the United States.
International Migration Review. Vol. 36. No. 3
(Fall 2002).
24How CPS Data is Used in Research Pooling CPS
Data (continued)
- Warnings
- 1. When pooling two consecutive years, more than
50 percent of the sample is overlapped. - 2. Some variable names and value intervals change
from year to year (see next slide) - Get pooled CPS data from IPUMS
- http//cps.ipums.org/cps/index.shtml
25Example of Changed Variables Across Years
26How CPS Data is Used in Research Matching CPS
Data
- 4. Matching the March CPS Supplement
(Longitudinal Aspect) - Why match CPS data?
- To study short-term dynamics, e.g. the length of
an unemployment spell - To utilize CPS address based sample design
- Example Foreign Born Emigration A New Approach
and Estimates Based on Matched CPS Files.
Jennifer van Hook, Weiwei Zhang, Frank Bean, and
Jeffery Passel. Demography. Vol. 43. No. 2 (May
2006).
27How CPS Data is Used in Research Matching CPS
Data (continued)
- The CPS is not good for matching because
- 1. 50 percent of the sample is lost when matching
two consecutive years of data - 2. Duration of the sample is short, limiting
longitudinal nature - 3. Demographic changes make matching more
complicated and decrease the sample size - 4. There are tradeoffs in merging merge
individuals who should merge with merging
individuals who shouldnt.
28How CPS Data is Used in Research Matching CPS
Data (continued)
- Basic (simple) approach
- Household ID, Household line number, Person line
number - Madrian and Lefgan approach
- Basic merge first
- Delete observations that sex, race, or age do not
match after basic merge - Source An Approach to Longitudinally Matching
CPS Respondents. Madrian, Brigitte and Lars John
Lefgan. Journal of Economic and Social
Measurement. Vol. 26 (2000).
29How CPS Data is Used in Research By Research
Area
- 1. Policy Research
- The CPS provides the government with
- Monthly estimates of labor force characteristics
and the general demographic composition of the
population. - Important indicators for planning and evaluating
government programs - Related publications
- Bureau of Labor Statistics, Monthly Labor Review
- Bureau of Census, Current Population Report
30How CPS Data is Used in Research By Research
Area (continued)
- 2. Community Research
- Study immigrants (immigration policy, education,
political participation), local state policy on
minimum wage, health insurance. - Conduct research with community partners to
promot justice and equity in a diverse,
multiethnic, multicultural world.
31How CPS Data is Used in Research Academic
Research
- 3. Selected topics using CPS data
- Demographics Living arrangement, marriage, child
support, population change, race and ethnicity,
gender, and undocumented population - Education Grade distribution, economic returns
to education, school enrollment, income gap
between high and low education levels - Health insurance and employee benefits
Characteristics of uninsured people, pensions,
health insurance for children -
32Academic Research Selected Topics Using CPS
Data (continued)
- Income and poverty Earnings mobility, earnings
inequality, income distribution, minimum wage,
poverty - Labor force Employment instability, job
mobility, part-time work force, at-home work
force, youth employment - Health Insurance, fertility, tobacco use,
smoking policy, food security - Voting/election can be used to study voting
participation among immigrants and across
immigration generations
33Academic Research Selected Topics Using CPS
Data (continued)
- Immigration The CPS is a major data source for
studying immigration - CPS surveys have questions on
- Nativity
- Parental Nativity
- Citizenship
- Year of entry
- Main reason for moving
- Topics include
- Estimation of undocumented immigration
- The effect of immigrants on the economy
- Characteristics of the foreign born, etc.
34Accessing Literature that Utilizes CPS Data
- http//www.icpsr.umich.edu/ICPSR/citations/index.h
tml - Type CPS or Current Population Survey in the
search field under Search the Bibliography of
Data-Related Literature - This is a tool to train students how to do
research - http//bibliography.ipums.org/
35Journals that Published Papers Using CPS Data
36Journals that Published Papers Using CPS Data
(continued)
37Want to Learn More?
- Please contact
- Wei Sun
- Office 945 Flanner Hall
- Telephone 631-6166
- Email wei.sun.7_at_nd.edu