Comparison of Estimates of SameSex Couple Households from the ACS and CPS
by
Daphne Lofquist and Renee Ellis US Bureau of the Census
Annual Meeting of the Population Association of America, Washington, D.C. March 31-April 2, 2011
*This report is released to inform interested parties of ongoing research and to encourage discussion of work in progress. The views expressed on statistical and methodological issues are those of the authors and not necessarily those of the US Census Bureau.
Background
There has been a growing demand for estimates of same-sex couples. Much of the academic and federal research has focused on the American Community Survey (ACS). Research has yet to focus on the variety of Census Bureau surveys, which could be useful if researchers are interested in questions not included on the ACS. We use the 2010 Current Population Survey (CPS) as a test comparison to the 2009 ACS.
Prior to 2010, all reported married same-sex couples in the CPS were changed to be their opposite-sex counterpart. In 2010, for the first time, they were all edited to be same-sex unmarried partners. This change in edits allows us to be able to estimate same-sex couple households in the CPS. We believe that comparing the demographics of same-sex couple households using these two surveys will allow us to evaluate the quality of the CPS estimates, along with verifying the overall reliability of the estimates in both surveys.
Surveys
American Community Survey Current Population Survey
A mandatory survey that is conducted annually over a 12-month calendar period. Sample of approximately 3 million households, with a monthly sample size of about 250,000 households. The questionnaire is administered through a mailout/mailback paper form, and followed up with computer assisted telephone and in-person interviews (CATI/CAPI). The ACS provides state and national data on households. Same-sex couples are identified in the ACS using both the household roster and sex item. A monthly survey conducted, for the Bureau of Labor Statistics, to assess the employment level in the country. Most annual demographic data comes from the Annual Social and Economic supplement of the CPS. The ASEC combines respondents from February, March, and April into a sample size of 99,000 households.* The questionnaire is administered through telephone and in-person interviews (CATI/CAPI). Same-sex couples are identified using both the household roster and sex item and a subsequent probe on cohabitation.
*Statistics from surveys are subject to sampling and nonsampling error. For more information on the source of the data and accuracy of the estimates, including standard errors and confidence intervals, see Appendix G at [Link]
Socio-demographic variables
Age of Householder Educational Attainment 6 category breakdown Householder with at least a Bachelors degree Race of Householder Both householder and partner White with at least a Bachelors Black or African American degree Asian Employment Status Some other race Householder employed Hispanic/Latino origin of Both householder and spouse Householder employed Hispanic or Latino of any Household Income race 5 category breakdown White alone, not Hispanic or Latino
Survey Estimates of Same-Sex Couple Households
Our analyses compares the socio-demographic characteristics of same-sex couple households between the 2009 ACS and 2010 CPS.
All Same-sex households Male-male households Female-female households
ACS 581,300 a 253,576 300,890
CPS 516,396* b 262,820 274923*
*Significantly different from the ACS. a Male-male households are statistically different from female-female households in the ACS b Male-male households are not statistically different from female-female households in the CPS
Age of Householder
(Percent distribution)
35
30
25
20
15
10
0 15 to 24 years 25 to 34 years 35 to 44 years 45 to 54 years 55 to 64 years 65 years and over
ACS Same-sex couples
Source: ACS 2009 and CPS 2010
CPS Same-sex couples
Significantly different from ACS
Race and Hispanic or Latino Origin of Householder
(Percent distribution)
100 90 80 70 60 50 40 30 20 10 0 White Black or African American Asian
*
Some Other Race Hispanic or Latino origin White, not Hispanic
ACS Same-sex couples
Source: ACS 2009 and CPS 2010
CPS Same-sex couples
Significantly different from ACS
Household Income
(Percent distribution)
45 40 35 30 25 20 15 10 5 0 Less than $35,000 $35,000 to $49,999 $50,000 to $74,999 $75,000 to $99,999 $100,000 or more
ACS Same-sex couples
Source: ACS 2009 and CPS 2010
CPS Same-sex couples
Significantly different from ACS
Educational Attainment
(Percent with stated educational level)
60
50
40
30
20
10
0 Householder has at least a Bachelor's degree Both partners with at least a Bachelor's degree
ACS Same-sex couples
Source: ACS 2009 and CPS 2010
CPS Same-sex couples
Significantly different from ACS
Age of Householder by Sex
(Percent distribution)
40 35 30 25 20 15 10 5 0
+ * +
15 to 24 years
25 to 34 years
35 to 44 years
45 to 54 years
55 to 64 years
65 years and over
ACS Unmarried male-male ACS Unmarried female-female
Source: ACS 2009 and CPS 2010
CPS Unmarried male-male CPS Unmarried female-female
* Significantly different from male-male CPS households. + Significantly different from female-female CPS households
Race and Hispanic or Latino Origin of Householder by Sex
(Percent distribution)
100 90 80 70 60 50 40 30 20 10 0 White Black or African American
*
Asian
*
Some Other Race Hispanic or Latino origin White alone, not Hispanic
ACS Unmarried male-male ACS Unmarried female-female
Source: ACS 2009 and CPS 2010
CPS Unmarried male-male CPS Unmarried female-female
* Significantly different from male-male CPS households. + Significantly different from female-female CPS households
Educational Attainment by Sex
(Percent with stated educational level)
60
50
40
30
20
10
0 Householder has at least a Bachelor's degree Both partners with at least a Bachelor's degree
ACS Unmarried male-male ACS Unmarried female-female
Source: ACS 2009 and CPS 2010
CPS Unmarried male-male CPS Unmarried female-female
* Significantly different from male-male CPS households. + Significantly different from female-female CPS households
Summary of ACS 2009 and CPS 2010 Comparison
There are significantly more householders aged 65 years and older in the ACS. This is true for both male-male and female-female relationships. The ACS reports 5.1% of householders of some other race compared to 3.3% of householders in the CPS. There are significantly more householders in male-male relationships in the CPS who report being black or African American, Asian, or of some other race than in the ACS. No statistical differences between the two surveys for female-female relationships are noted by race.
Summary of ACS 2009 and CPS 2010 Comparison
The CPS reports significantly more households with an income between $35,000-$49,999, while the ACS reports significantly more households with an income of $100,000 or more. Female-female relationships in the ACS report significantly more households with an income of $100,000 or more than in the CPS. There are no significant household income differences between the two surveys for male-male relationships. No statistical differences between the two surveys are noted by Hispanic origin or educational levels or employment status.
Exploring Reasons for Differences
The differences between ACS 2009 and CPS 2010 are fairly small. The following analysis explores the possibility that the differences are explained by the mode of data collection.
ACS CATI/CAPI 152,761*+
ACS Mailout/mailback 428,539*
CPS 516,396
* Significantly different from the CPS. + Significantly different from the ACS mailout/mailback.
Age of Householder by Mode
(Percent distribution)
35 30 25 20 15 10 5 0 15 to 24 years 25 to 34 years 35 to 44 years 45 to 54 years 55 to 64 years 65 years and over
+ +
* *+ + + *
ACS CATI/CAPI
Source: ACS 2009 and CPS 2010
ACS Mailout/Mailback
CPS CATI/CAPI
* Significantly different from the CPS. + Significantly different from the ACS mailout/mailback
Race and Hispanic or Latino Origin of Householder by Mode
(Percent distribution)
100 90 80 70 60 50 40 30 20 10 0 White Black or African American Asian Some Other Race Hispanic or Latino origin White alone, not Hispanic
*+ *+
*+
*+ *+
ACS CATI/CAPI
Source: ACS 2009 and CPS 2010
ACS Mailout/Mailback
CPS CATI/CAPI
* Significantly different from the CPS. + Significantly different from the ACS mailout/mailback
Educational Attainment by Mode
(Percent with stated educational level)
60
50
40
*+
30
20
*+
10
0 Householder has at least a Bachelor's degree Both partners with at least a Bachelor's degree
ACS CATI/CAPI
Source: ACS 2009 and CPS 2010
ACS Mailout/Mailback
CPS CATI/CAPI
* Significantly different from the CPS. + Significantly different from the ACS mailout/mailback
Summary of Differences by Mode
There is a greater percentage of householders who are aged 65 years and over in the ACS mailout/mailback than in both the ACS CATI/CAPI and the CPS. The percentage of householders reporting to be white is lower in the ACS CATI/CAPI than in the other survey types. The ACS mailout/mailback has a higher percentage of white alone, not Hispanic or Latino than the ACS CATI/CAPI. The ACS CATI/CAPI has a higher percentage of householders who are black or African American or of some other race, and also respondents of Hispanic or Latino origin than the other surveys. Those in the ACS CATI/CAPI report lower educational attainment for the householder and for both partners than in both the ACS mailout/mailback and CPS. These comparisons show that the ACS CATI/CAPI respondents are distinctly different from both the ACS mailout/mailback and CPS respondents, the latter two samples are much closer in demographic comparison.
Overall Conclusions
1. Although the ACS is considered to be the best survey in the Census Bureau
for estimates of same-sex couples, the findings of this study lend support to the idea that other Census Bureau surveys (like the CPS) can produce reasonable and consistent estimates of same-sex couples.
By producing reasonable and consistent estimates, the CPS will allow for us to extend research on same-sex couple households through the use of additional economic measures not currently available in the ACS. For example: More detailed questions on income More detailed questions on labor force participation The official poverty measurement A follow-up cohabitation question, which allows us to identify not only householders who are cohabiting but also other cohabiting couples in the household.
2. The ACSs two modal types are very different from each other. These differences reflect those respondents who are willing to complete the survey by mail and those who require subsequent follow-up interviews (ACS CATI/CAPI). This may explain why the CPS, a voluntary survey collected via in-person and telephone interviews, is more similar to the ACS mailout/mailback mode than the ACS CATI/CAPI.
Contact Information
Fertility and Family Statistics Branch Housing and Household Economic Statistics Division US Census Bureau 301-763-2416 Daphne Lofquist [Link]@[Link] Renee Ellis [Link]@[Link]