11 ACS Public Use Microdata Samples of 2005 and 2006 – How to Use the Replicate Weights B. Dale Garrett and Michael Starsinic U.S. Census Bureau AAPOR.

Slides:



Advertisements
Similar presentations
Using American FactFinder John DeWitt Project Manager Social Science Data Analysis Network Lisa Neidert Data Services Population Studies Center.
Advertisements

The Microdata Analysis System (MAS): A Tool for Data Dissemination Disclaimer: The views expressed are those of the authors and not necessarily those of.
1 CDBG Income Survey Requirements For Grant Administrators.
Migration Patterns and Mover Characteristics from the 2005 ACS Gulf Coast Area Special Products Kin Koerber Housing and Household Economic Statistics Division.
The American Community Survey (ACS) Lisa Neidert NPC Workshop: Analyzing Poverty and Socioeconomic Trends Using the American Community Survey July 12 –
Using American FactFinder John DeWitt Project Manager Social Science Data Analysis Network Lisa Neidert Data Services Population Studies Center.
American Community Survey 2006 Data Release WEBINAR A ugust 15, 2007.
Technical Issues Associated with the American Community Survey Lisa Neidert NPC Poverty/American Community Survey Workshop June 22-26, 2009.
Measures of Income, Poverty and Health Insurance Wesley Basel, U.S. Census Bureau Presented at the Walter Cronkite School of Journalism June 17, :00.
1 U.S. Census Bureau Data Availability for Geographic Areas March 25, 2008.
1 The American Community Survey HSUG-West Conference October 1, 2004 Berkeley, CA.
The American Community Survey (ACS) Lisa Neidert McCormick Specialized Training Institute October , 2009.
The American Community Survey (ACS) Lisa Neidert NPC Workshop: Analyzing Poverty and Socioeconomic Trends Using the American Community Survey June 22 –
Technical Issues Associated with the American Community Survey Lisa Neidert NPC Poverty/American Community Survey Workshop July , 2010.
Technical Issues Associated with the American Community Survey Lisa Neidert NPC Poverty/American Community Survey Workshop June 23-27, 2008.
The American Community Survey (ACS) Lisa Neidert NPC Workshop: Analyzing Poverty and Socioeconomic Trends Using the American Community Survey June 23 –
11 American Community Survey Data Products. 2 What do I need to know before using ACS data and data products?
1 The American Community Survey (ACS) 2005 Data Release.
APDU Webinar User Needs for Calculating Standard Errors in the ACS OR What is a Statistical Calculator? Presented by Doug Hillmer, Independent Consultant.
Economics and Statistics Administration U.S. CENSUS BUREAU U.S. Department of Commerce Comparing IRS Exemptions to 2010 Census Population Counts Esther.
2014 SDC and CIC Annual Training Conference: Accessing ACS PUMS Data Tim Gilbert U.S. Census Bureau April 2, 2014.
11 The American Community Survey Steve Murdock, Ph.D. Director, Hobby Center for the Study of Texas Rice University.
Household Surveys ACS – CPS - AHS INFO 7470 / ECON 8500 Warren A. Brown University of Georgia February 22,
U.S. Census Bureau census.gov Census Data Immersion From A Novice to A Skilled Data Miner Infopeople Webinar August 7,
The American Community Survey The American Community Survey Accessing Information for Hawaii from the 2006 American Community Survey (ACS) Jerry Wong Information.
Employment and Earnings Outcomes for Young Adult Bachelor’s Degree Holders: Findings From the American Community Survey 25th Annual STATS-DC 2012 Data.
The ACS: Fulfilling its Promise to Data Users Alfredo Navarro US Census Bureau APDU 2010 Annual Conference Washington, DC September 21, 2010.
American Community Survey Presented at the Meeting of the National Neighborhood Indicators Partnership Susan Schechter May
Issues Related to Data Dissemination in Official Statistics Presented at the European Conference On Quality in Official Statistics Helsinki, Finland May.
111 American Community Survey Fundamentals 2009 Population Association of America ACS Workshop April 29, 2009.
Saadia GreenbergElena Fazio Office of Performance and Evaluation Administration on Aging US Department.
Microdata Simulation for Confidentiality of Tax Returns Using Quantile Regression and Hot Deck Jennifer Huckett Iowa State University June 20, 2007.
Screening Data for Disclosure Risk and the Research behind One Possible Tool Kristine Witkowski Research support from the National Institute of Child Health.
Adaptive Kernel Density in Demographic Analysis Richard Lycan Institute on Aging Portland State University.
1 Journey-to-Work Data in the American Community Survey (ACS) May 17, 2009 TRB Transportation Planning Applications Conference Federal Data for Modelers.
1/26/09 1 Community Health Assessment in Small Populations: Tools for Working With “Small Numbers” Region 2 Quarterly Meeting January 26, 2009.
C2ER 52 nd Annual Conference & LMI Training Institute Annual Forum Regional Socioeconomic Statistics Update on U.S. Census Bureau Programs June 8, 2012.
The American Community Survey: An Overview
Using IPUMS.org Katie Genadek Minnesota Population Center University of Minnesota The IPUMS projects are funded by the National Science.
Introduction to the Public Use Microdata Sample (PUMS) File from the American Community Survey Updated February 2013.
Integrating ACS with the World’s Census Data: ACS Microdata and the IPUMS Presented at the Pre-ALAP ACS/IPUMS Workshop November 16, 2010 Trent Alexander.
Using the American Community Survey (ACS) Maryland Sate Data Center Affiliate Meeting April 4, 2007.
Using the ACS: Issues with studying small areas and change over time Presented to Association of Public Data Users January 20, 2011.
American Community Survey Overview September 4, 2013 Tim Gilbert American Community Survey Office.
1 Things That May Affect Estimates from the American Community Survey.
American Community Survey Maryland State Data Center Affiliate Meeting June 17, 2008.
American Community Survey Getting the Most Out of ACS Jane Traynham Maryland State Data Center.
American Community Survey. Outline American Community Survey basics Accessing ACS data products Resources for learning more 2.
American Community Survey Maryland State Data Center Affiliate Meeting September 16, 2010.
1 The American Community Survey An Update Pamela Klein American Community Survey Office Washington Metropolitan Council on Governments Cooperative Forecasting.
American Community Survey (ACS) 1 Oregon State Data Center Meeting Portland State University April 14,
Comparisons of Synthetic Populations Generated From Census 2000 and American Community Survey (ACS) Public Use Microdata Sample (PUMS) 13 th TRB Application.
Using ACS and Census 2010 in Communities and Neighborhoods: Guidelines and Tools POPULATION REFERENCE BUREAU | PRESENTATION BY MARK MATHER.
U.S. Census Bureau’s Population Estimates Program Victoria Velkoff Population Division U.S. Census Bureau APDU 2010 Annual Conference Public Data 2010:
Developing Survey Handbooks as Educational Tools for Data Users Presented at the European Conference on Quality in Official Statistics May 2010 Deborah.
Some ACS Data Issues and Statistical Significance (MOEs) Table Release Rules Statistical Filtering & Collapsing Disclosure Review Board Statistical Significance.
Data on the Foreign Born in 2010: Accessing Information on Immigrants and Immigration from the U.S. Census Bureau’s American Community Survey Thomas A.
The U.S. Census Bureau Population Estimates Program Victoria A. Velkoff U.S. Census Bureau APDU Annual Conference September 25, 2008.
 Public Use Microdata Sample – sample file of unaggregated raw data with no identifying information about an individual person or household (no addresses,
American Community Survey “It Don’t Come Easy”, Ringo Starr Jane Traynham Maryland State Data Center March 15, 2011.
Things that May Affect the Estimates from the American Community Survey Updated February 2013.
Disclosure Avoidance at Statistics Canada INFO747 Session on Confidentiality Protection April 19, 2007 Jean-Louis Tambay, Statistics Canada
American Community Survey (ACS) Product Types: Tables and Maps Samples Revised
1 ACS Statistical Issues and Challenges: One-, Three-, and Five-year Period Estimates Alfredo Navarro U.S. Census Bureau Association of Professional Data.
Household Surveys: American Community Survey & American Housing Survey Warren A. Brown February 8, 2007.
The U.S. Census Bureau’s Postcensal and Intercensal Population Estimates Alexa Jones-Puthoff Population Division National Conference on Health Statistics.
Can We Trust Data Users to Consider Data Quality? Presented at the 2008 European Conference on Quality in Official Statistics.
1 Population Controls for the American Community Survey Alexa Kennedy-Puthoff Population Division Prepared for the 2009 SDC Annual Training Conference,
ASDC Annual Meeting November 10, 2011 Kathleen Gabler Socioeconomic Research Associate Center for Business and Economic Research Culverhouse College of.
Census Data-Strictly Business?:
Presentation transcript:

11 ACS Public Use Microdata Samples of 2005 and 2006 – How to Use the Replicate Weights B. Dale Garrett and Michael Starsinic U.S. Census Bureau AAPOR Conference, New Orleans May 16, 2008

2 Public Data The American Community Survey (ACS) produces an annual Public Use Microdata Sample (PUMS) file. You can download these files for free. Write your own program to tally and analyze data.

3 Key Points PUMS data users want to know the reliability of an estimate. This paper explains how to use PUMS replicate weights to estimate standard errors.

44 Outline the American Community Survey (ACS) the Public Use Microdata Sample (PUMS) –sample design –confidentiality –weights –standard errors –issues with standard errors

55 The American Community Survey The 2005 ACS –Sample of 250,000 housing units per month. –Every county represented in the fifty states, District of Columbia and Puerto Rico. –Collects population and housing characteristics The 2006 ACS was similar but added –A sample of both institutional and noninstitutional Group Quarters population. –GQ sample size was 16,000 persons per month

66 PUMS Sample Design PUMS is a subsample of ACS –Sort the ACS interviews on geography, mode of interview, types of housing units, demographics –Sample size: one percent of the total HUs and HH persons in 2005 and one percent of total GQ persons in 2006 –Systematic sampling at the state and PUMA level.

7 PUMA Definition PUMA - Public Use Microdata Area –Designed for public release of information by local state officials. –Large enough to achieve disclosure avoidance. An area of 100,000 population or more as of the 2000 Census.

88 PUMS Protects Confidentiality PUMS does not reveal: –Names of persons. –Address. –Detailed Type of group quarters. –Geographic data below the PUMA level. The respondent’s identity is protected. –Top-coding of age, income and other variables. –Data swapping –Synthetic data –Perturbation of data

9 Rural PUMAs in KY 9

10 PUMAs in Baltimore Co., MD 10

11 PUMS Weighting The PUMS initial weight was equal to the ACS final weight times the sampling interval. The 2006 PUMS file was ratio-estimated to ACS –persons in households by sex by PUMA –housing units by vacant/occupied by PUMA –persons in group quarters by institutional/ noninstitutional by state

12 How to Program an Estimate – Counts, Aggregates, Ratios, Medians Totals (counts) –Sum the PUMS weights (for the characteristic). Aggregates –Sum the product of the PUMS weight times the value Ratios –Form the total or aggregate for the numerator –Sum the PUMS weights for the characteristic in the denominator –Divide Medians – use weighted distributions

13 ACS Standard Errors The ACS uses the successive difference model of replicate weights to estimate standard errors. The successive difference model of Kirk Wolter was developed for ACS by Robert Fay and George Train.

14 Two Methods for PUMS Standard Errors Design factor method –Design factors are factors to multiply times the standard error of a simple random sample. –Easier to use than the replicate weights Replicate weight method –Generally, you get a more accurate standard error estimate by using the replicate weights. –Somewhat more work than design factors.

15 Three Steps to Standard Errors Using Replicate Weights Write a program to derive an estimate using the PUMS weight. Run the program 80 more times using each of the 80 replicate weights. Use the PUMS estimate and the 80 replicate estimates in the Standard Error formula.

16 ACS PUMS Replicate Weight Formula for a Standard Error where: –X is the estimate formed from the PUMS weight –X r is the estimate formed from the r th replicate weight.

17 Standard Errors of Differences There are two estimates, A and B. You want to use a Z-test to see if the difference (A – B) is significant. The Z-test requires the standard error of the difference.

18 For Independent Estimates SE A-B – the standard error of (A – B) SE A – the standard error of estimate A SE B – the standard error of estimate B Use the standard errors of the two estimates to estimate the standard error of the difference.

19 For Correlated Estimates Directly use the replicate weights to calculate the standard error of the difference. –Let X = (A - B) = the difference –Let X r = (A r – B r ) for the 80 replicate differences X 1 … X 80 Use the replicate weight formula ( seen earlier ).

20 Replicate Weight Issues Estimate is zero, standard error is not zero. –Cannot use replicate weights to estimate the standard error. –See the PUMS Accuracy document for a formula. The replicate standard error is zero, estimate is not zero. –Zero means that if you reselected the sample the answer would be the same. –Acceptable if estimate controlled in the weighting. –Not acceptable if the estimate is a median. Often a direct median gives a zero standard error.

21 Standard Error Options for Medians Direct median with replicate weights may give a zero standard error. This is not good. Categorical median with replicate weights will give a more stable standard error, but still some zero standard errors. Design factor method – Start with either the direct or categorical median, use design factors for the standard error.

22 Conclusion Replicate weights for ACS PUMS are: –Available for 2005 PUMS and later. –Easy to use for most estimates. –Few issues For medians –Replicate weight standard errors may be zeros. –To avoid the zeros use the design factor method.

23 References US Census Bureau: Accuracy of the Data (2006) for ACS is found at: – US Census Bureau: PUMS Accuracy of the Data (2006) is found at: – pdfhttp://acsweb2.acs.census.gov/acs/www/Downloads/2006/AccuracyPUMS. pdf US Census Bureau: Design and Methodology: American Community Survey, Technical Paper 67, May 2006, – Fay & Train, Aspects of Survey and Model-Based Postcensal Estimation of Income and Poverty Characteristics for States and Counties, 1995 –

24 Contact Information For questions about this presentation or for an example program to generate standard errors. Contact me at Views expressed in this paper are those of the authors and not necessarily those of the U.S. Census Bureau.

25 How to Derive an Estimate – Direct Medians The direct median is the weighted sample median or the distributional median. Sum the weights for the characteristic total. Sort the file on the value of interest. Sum the weights until the 50% point. The direct median is the value of the record which crosses the 50% point. Or a point between the values of two records that divide the file into two exact halves.

26 How to Derive an Estimate – Categorical Medians Categorical or interpolated medians. –Used for published ACS statistics in Factfinder. Categorical medians are interpolations: –A weighted distribution of the characteristic. –Each bin or row is assigned a range of values. –Uses linear interpolation for most variables.

27 Direct Median Example Based on 5 Records Record # Percent of Total Income from record Direct median 1 18% 18, % 33, % 41, % 49, % 62,000

28 Direct and Categorical Medians Example Based on 5 Records Income Range Record # Percent of Total Income from record Direct median Categorical median -59,000 to 20, % 18,000 20,000 to 40, % 33,000 40,000 to 60, % 41,000 45, % 49,000 60, % 62,000