Disclaimer

This report is released to inform interested parties of research and to encourage discussion. The views expressed on statistical, methodological, or technical issues are those of the authors and not necessarily those of the U.S. Census Bureau.

Acknowledgments

The authors wish to thank Gregory Weyland in the Demographic Surveys Division, Kimball Jonas in the Demographic Statistical Methods Division, and Jorge del Pinal and Campbell Gibson in the Population Division for their helpful comments on this paper.

I. Hispanic origin.

Table 1. Hispanic Origin in Census 2000 by Hispanic Origin in Current Population Survey. (Edited data; percent by census response categories)

Table 2. Hispanic Origin in Current Population Survey by Hispanic Origin in Census 2000. (Edited data; percent by CPS response categories)

Table 3. Hispanic Origin in Census 2000 by Hispanic Origin in Current Population Survey. (Data before imputation; percent by census response categories)

Table 4. Hispanic Origin in Census 2000 by Hispanic Origin in Current Population Survey. (Data before imputation; percent by CPS response categories)

Table 5. Hispanic Origin in Census 2000 by Hispanic Origin in Current Population Survey. (Data before imputation; percent by total sample response categories)

Table 6. Hispanic Origin in Census 2000 by Hispanic Origin in Current Population Survey. (Date before imputation; percent by Census detailed Hispanic group response categories)

• Mexican, Puerto Rican and Cuban.
• Central and South American.
• Dominican Republic.
• Other Places of Birth.
• Portugal and the Azores.
• Brazil

Table 7. Selected Place of Birth by Hispanic Origin in Census 2000 and Hispanic Origin in Current Population Survey

- Part I(1 page 11, R1) (Edited data by CPS response categories)

- Part II. Data before imputation by CPS response categories)

II. Race

Table 8. Race in Census 2000 by Race in Current Population Survey. (Edited data; percent by census response categories)

Table 9. Race in Census 2000 by Race in Current Population Survey. (Edited data; percent by CPS response categories)

Race Reporting by Non-Hispanics

Table 10. Race Reported by Non-Hispanics^a in Census 2000 by Race in Current Population Survey. (Edited data; percent by census race response)

Table 11. Race Reported by Non-Hispanics in Census 2000 by Race in Current Population Survey. (Edited data; percent by CPS race response)

Race Reporting by Hispanics

Table 12. Race Reported by Hispanics in Census 2000 by Race in Current Population Survey. (Data before imputation; percent by census race response)

Table 13. Race Reported by Hispanics in Census 2000 by Race in Current Population Survey. (Data before imputation; percent by CPS race response)

Conclusion

Appendix A. Population Division Working Paper Series

Current Population Survey (CPS) data provide an important source of information about the socioeconomic status for groups of policy interest in the United States.  For this reason, it is essential to understand how the collection of CPS data by race and Hispanic origin differs from that collected in Census 2000.  An important reason for differences between the two data collections is that the questions used in Census 2000 differ dramatically from those used in CPS as discussed later.  Another is that CPS information is collected by trained interviewers while Census 2000 information was collected using mail questionnaires with some followup by enumerators.  The purpose of this study of matched CPS-Census 2000 data is to quantify how answers to both sets of questions differed for the same respondent in the two data collections.

(Percent distribution. Numbers in parentheses are in thousands. Data are for civilian noninstitutionalized population of the United States. Members of the Armed Forces living off post are included if there is at least one civilian adult in the household.)

Source: Tabulation of edited data from matched Census 2000 and February to May 2000 Current Population Records.(1 page 11, R1)

Table 1 shows the distribution of Hispanic origin in the matched Census 2000/CPS records.  In  these records, 13.1 percent of responses were Hispanic based on the Census 2000 question.  However, of the 34.7 million respondents identified as Hispanic in CPS, only 90.8 percent were also Hispanic in Census 2000 data, and 9.2 percent were not.¹  In contrast, among the 241 million non-Hispanics in CPS, 98.0 percent also identified as non-Hispanic in Census 2000 and 2.0 percent did not.

Table 2 shows the distribution of Hispanic origin in the matched file by CPS responses.  According to the CPS responses, only 12.6 percent of the matched records were Hispanic compared with 13.1 percent in Census 2000 (see Table 1) and 87.4 percent which were not.  Of those identified as Hispanic in Census 2000, 86.8 percent were also identified as Hispanic in CPS data (compare with 90.8 percent in Table 1) and 13.2 percent were not Hispanic.  Among non-Hispanics based on Census 2000, 98.7 percent were also identified as non-Hispanic in CPS and about one percent were not.

In analyzing differences in the new and old Hispanic origin questions using a special May 2002 CPS supplement (not the matched file discussed here), researchers at the Bureau of Labor Statistics (BLS) concluded that “the new question identifies additional and different people as Hispanic” (Bowler, Ilg, Miller, Robinson, and Polivka,  2003:7). (2 page 11, R1) In the matched file (Census 2000 and CPS), we find that Census 2000 identifies 36.3 million people as Hispanic, while the CPS identifies 34.7 million people as Hispanic.²  Thus, Census 2000 recognized 1.6 million more Hispanics in 2000 than did the CPS, confirming that the Census 2000 question identifies more and different respondents than does the CPS question.

(Percent distribution. Numbers in parentheses are in thousands. Data are for civilian noninstitutionalized population of the United States. Members of the Armed Forces living off post are included if there is at least one civilian adult in the household.)

Source: Tabulation of edited data from matched Census 2000 and February to May 2000 Current Population Records.  (1 page 11, R1)

(Percent distribution. Numbers in parentheses are in thousands. Data are for civilian noninstitutionalized population of the United States. Members of the Armed Forces living off post are included if there is at least one civilian adult in the household.)

Source: Tabulation of reported data before imputation from matched Census 2000 and February to May 2000 Current Population Records.  

The estimates discussed so far are based on fully edited and imputed data from both sources.  How do differences in response rates affect the estimates?  Table 3 shows the reported Hispanic

origin data distribution before imputation for missing values.  Overall, 3.9 percent of  matched respondents did not answer the Hispanic origin question in Census 2000.  Among those who did not answer the CPS, 91.9 percent were not Hispanic and 3.0 percent were Hispanic based on their answer in Census; 5.1 percent of those who did not answer the CPS question also did not answer the Census 2000 question.  Of those who were Hispanic in CPS, 86.1 percent were also Hispanic in Census 2000, 8.2 percent were non-Hispanic, and 5.8 percent did not answer in Census 2000.³  Among non-Hispanics in CPS, 94.7 percent were also non-Hispanic in Census 2000, 1.7 percent were Hispanic, and 3.6 percent did not answer in Census 2000.                          

(Percent distribution. Numbers in parentheses are in thousands. Data are for civilian noninstitutionalized population of the United States. Members of the Armed Forces living off post are included if there is at least one civilian adult in the household.)

Source: Tabulation of reported data before imputation from matched Census 2000 and February to May 2000 Current Population Records.  (1 page 30, R1)

Table 4 shows the percent distributions of unedited CPS responses by Census 2000 unedited categories. Overall, 1.8 percent of matched respondents did not answer the CPS Hispanic origin question. The Census 2000 nonresponse category results indicate that 78.8 percent of the CPS cases were non-Hispanic,18.8 percent were Hispanic, and 2.4 percent did not respond in either Census 2000 or CPS. Of those who responded as Hispanic in Census 2000, 87.5 percent were also Hispanic in CPS, 12.1 percent were non-Hispanic, and 0.4 percent did not answer.  Among non-Hispanics in Census 2000, 96.8 percent were also non-Hispanic in CPS, 1.2 percent were Hispanic, and 2.0 percent did not answer.⁴

As shown in Table 5, 81.1 percent of all the unedited matched cases were non-Hispanic in both Census and CPS, and 10.8 percent were Hispanic in both.  Only 0.09 percent did not respond in either Census or CPS; 3.8 percent (3.04 + 0.72 = 3.76) did not respond to the Census 2000 question but did answer in CPS; and 1.8 percent (1.69 + 0.06= 1.75) did not respond to the CPS question but did answer in Census 2000.⁵  Finally, about 2.5 percent (1.49 + 1.02 = 2.51) of the matched cases switched between Hispanic and non-Hispanic, or vice versa . 

Percent distribution. Numbers in parentheses are in thousands. Data are for civilian noninstitutionalized population of the United States. Members of the Armed Forces living off post are included if there is at least one civilian adult in the household.)

Source: Tabulation of reported data before imputation from matched Census 2000 and February to May 2000 Current Population Records.  (1 page 30, R1)

Thus, nonresponse in one or collection or the other (3.76 + 1.75 = 5.5 percent of matched cases) may account for more of the inconsistency between Census and CPS responses than does the switching between Hispanic and non-Hispanic categories (2.5 percent).

Table 6 shows selected Hispanic groups as reported in Census 2000 compared with their matched responses in CPS. Only 2.4 percent of those who did not answer the Census 2000  

Hispanic origin question did not report an origin in CPS either.  However, 78.8 percent of the census non-respondents were classified as not Hispanic in CPS and 18.8 percent as Hispanic.

Among those who were reported as not Hispanic in Census 2000, 2.0 percent did not report an origin in CPS; 96.8 percent were reported as not Hispanic in both; and 1.2 percent reported as Hispanic in CPS.⁶ 

Percent distribution. Numbers in parentheses are in thousands. Data are for civilian noninstitutionalized population of the United States. Members of the Armed Forces living off post are included if there is at least one civilian adult in the household.)

Source: Tabulation of reported data before imputation from matched Census 2000 and February to May 2000 Current Population Records.  (3 page 1, R1)

Those identified as Mexican or Cuban Hispanics in Census 2000 exhibited comparatively low nonresponse rates (0.2 percent and 1.1 percent respectively), and were among those exhibiting a higher consistency of reporting (91.3 percent and 88.0 percent, respectively)⁷.  Those reporting Other Hispanic had the lowest consistency of reporting (77.9 percent). Conversely, Other Hispanics had the highest proportion reporting as non-Hispanic in CPS (21.2 percent) while Mexicans and Cubans were among the lowest (8.5 percent and 10.9 percent, respectively).⁸  Clearly, in some cases the CPS origin question generated different answers about Hispanic origin compared with the Census 2000 question.  

Table 7, Part 1, shows selected place of birth by the joint Hispanic origin response in CPS and Census 2000 for a the subset of the Hispanic population born in the listed places (Hispanic US births are excluded with the exception of Puerto Rico). The discussion of Table 7 will proceed in four sections:  1) Hispanic origin categories that appear in both CPS and Census (Mexican, Puerto Rican, and Cuban, not to be confused with Mexico, Puerto Rico and Cuba as birth places); 2) specific Central and South American countries; 3) the Dominican Republic; 4) Spain; 5)  specific non-Hispanic origins who sometimes identify as “Hispanic.”⁹

Mexican, Puerto Rican and Cuban.  There was very high consistency of a “Hispanic” response in both CPS and Census 2000 (about 90 percent) for people born in Mexico, Puerto Rico, and Cuba which may not be surprising because the three Hispanic origin categories based on these places appear in both Census 2000 and CPS questions.  A similar  percent of the respondents born in each of these places reported “not Hispanic” in Census 2000 and “Hispanic” in CPS.¹⁰   There were also some respondents born in Mexico and Puerto Rico who were identified as “not Hispanic” in both the CPS and Census 2000 (Table 7, part 1).¹¹   

Central and South American.  Table 7, Part I shows that about 4 percent of those born in countries specified as Central America or South America (4.2 percent and 3.7 percent, respectively, and not statistically different) were reported as “not Hispanic” in both collections. On the other hand, 83 percent of Central Americans and 78 percent of South Americans reported as “Hispanic” in both collections.¹²  In addition, 4.4 percent of Central Americans and 8.3 percent of South Americans switched between “not Hispanic” in Census 2000 and       “Hispanic” in CPS.  Four point six percent of those born in Central America and 5.3 percent of those born in South America were identified as “Hispanic” in CPS and “not Hispanic” in Census 2000.¹³  Clearly, both the CPS and Census 2000 questions omit some Central or South Americans who identify as Hispanic in some circumstances.  These omissions may indicate some confusion among them about the questions.

Dominican Republic.  Of those born in the Dominican Republic, 77 percent identified as “Hispanic” in both CPS and Census 2000.  However, 17 percent of Dominicans switched from “Hispanic” in Census 2000 to “Not Hispanic” in CPS, and 2.1 percent went the other way, and 2.8 percent reported “no answer” in Census 2000 and “Hispanic” in CPS

(Table 7, Part II).¹⁴ Clearly, the CPS does not identify a sizeable proportion of those born in the Dominican Republic as Dominican, but at least some of them must have been confused by the Census 2000 question as well.

Spain.  Of these born in Spain, 17.2 percent consistently report as “not Hispanic” while 26.6 percent report consistently as “Hispanic” in both CPS and Census 2000.  About 51 percent of the foreign born from Spain said they were “not Hispanic” in CPS and “Hispanic” in Census 2000 compared with none who went the other way.¹⁵  

Other Places of Birth.  Respondents from other places of birth, such as the Philippines, Portugal, or Brazil, also appear to have some confusion about whether they should identify as “Hispanic” or not Hispanic.¹⁶ For example, 91.9 percent of those born in the Philippines were identified as “not Hispanic” in both CPS and Census 2000, and less than one percent were identified as “Hispanic” in both.  However, 2.4 percent of those born in the Philippines identified as  “Hispanic” in CPS and “not Hispanic” in Census 2000, compared with one percent who said they were “Hispanic” in Census 2000 and “not Hispanic” in the CPS. In addition, 4.2 percent were identified as “Not Hispanic” in CPS, but provided no response in Census 2000.¹⁷

For those born in Portugal and the Azores, 89.3 percent identified as “not Hispanic”and 2.3 percent as “Hispanic” in both CPS and Census 2000.  An additional 3.1 percent identified as “Hispanic” in CPS  and “Not Hispanic” in Census 2000, and for 1.7 percent the reverse was true.¹⁸  The other 3.6 percent were identified as “Not Hispanic” in CPS and did not answer in Census 2000.

Among those born in Brazil, 50.5 percent were identified as “not Hispanic” and 4.5 percent were “Hispanic” in both the CPS and Census 2000.  However, 43.8 percent of Brazilians identified as “Hispanic” in CPS but were identified as “not Hispanic” in Census 2000.  The main reason for this may be that respondents who identified as “Other Spanish/Hispanic/Latino” and wrote-in “Brazilian” were recoded into “not Hispanic” in Census 2000.¹⁹ 

Percent distribution. Numbers in parentheses are in thousands. Data are for civilian noninstitutionalized population of the United States. Members of the Armed Forces living off post are included if there is at least one civilian adult in the household.)

Source: Tabulation of edited data from matched Census 2000 and February to May 2000 Current Population Records.(1 page 11, R1)

Percent distribution. Numbers in parentheses are in thousands. Data are for civilian noninstitutionalized population of the United States. Members of the Armed Forces living off post are included if there is at least one civilian adult in the household.)

Source: Tabulation of reported data before imputation from matched Census 2000 and February to May 2000 Current Population Records.  

(-) Zero or rounds to zero.

Table 8 shows the distribution of edited Census 2000 by edited CPS race categories.  Of respondents who were reported as White in the CPS, 90.4 percent also reported as White in Census 2000.  Similarly about 90.8 percent of CPS Blacks were also Black in Census 2000.²⁰  About 80 percent of CPS Asian and Pacific Islanders (API) were matched with Asian or Native Hawaiian and Other Pacific Islander (NHPI) in Census 2000. There was relatively less concordance with the responses of American Indian and Alaska Natives (AIAN) – at about 52 percent.

Percent distribution. Numbers in parentheses are in thousands. Data are for civilian noninstitutionalized population of the United States. Members of the Armed Forces living off post are included if there is at least one civilian adult in the household.)

Source: Tabulation of edited data from matched Census 2000 and February to May 2000 Current Population Records.

0.0 (-) Zero or rounds to zero.

These differences should not be totally surprising as there were more reporting options available in the Census 2000 questionnaire than there were in the CPS interview.  For example, about six percent of CPS Whites chose “Some other race” (SOR) in Census 2000 – perhaps because SOR was not a CPS option. About seven percent of the CPS AIAN category also reported SOR in Census 2000, and another 10 percent reported “Two or more races” (TOMR).²¹ Similarly, about ten percent of the CPS API reported as TOMR in Census 2000.  Perhaps more importantly, about 26 percent of the CPS AIAN group reported as White in the Census, a result that may be a CPS interviewer effect.

Percent distribution. Numbers in parentheses are in thousands. Data are for civilian noninstitutionalized population of the United States. Members of the Armed Forces living off post are included if there is at least one civilian adult in the household.)

Source: Tabulation of edited data from matched Census 2000 and February to May 2000 Current Population Records.

-) Zero or rounds to zero.

a Non-Hispanics were identified from the reported (not imputed) Hispanic origin response from Census 2000.

Table 9 shows the distribution of edited CPS race responses by edited Census 2000 categories.  In general there is a good concordance between Census 2000 and CPS historical OMB categories. For example, although the Census AIAN matched proportion is relatively smaller at 62.7 percent, notable proportions of the Census White (98.5 percent), Black (94.5 percent), and Asian (87.7 percent) categories retained a corresponding race identity in the CPS.

Conversely, some switching took place for non-traditional categories.  For example, 93.2 percent of the Census 2000 SOR responses were White in CPS, 4.0 percent were Black, and about one percent each were AIAN or API.²²  Amounts for the Census 2000 TOMR category were 64.2 percent White,14.8 percent Black, 4.8 percent AIAN and 16.2 percent API.²³

Race Reporting by Non-Hispanics

Table 10 shows that about one percent of non-Hispanics in the CPS did not answer the Census 2000 race question, 78.8 percent reported as White, 13.3 percent as Black, 4.2 percent reported as Asian and two percent or less each were reported as AIAN, NHPI,  SOR, or TOMR.  Among non-Hispanics who did not answer the CPS race question, 4.9 percent did not answer the Census 2000 question. Forty-seven percent of the CPS non-respondents were White according to Census 2000,

Percent distribution. Numbers in parentheses are in thousands. Data are for civilian noninstitutionalized population of the United States. Members of the Armed Forces living off post are included if there is at least one civilian adult in the household.)

Source: Tabulation of edited data from matched Census 2000 and February to May 2000 Current Population Records.

(-) Zero or rounds to zero.

a. Non-Hispanics were identified from the reported (not imputed) Hispanic origin response from Census 2000.

14.7 percent Black, 0.9 percent AIAN, 13.4 percent Asian, 0.6 percent NHPI, 3.2 percent SOR and 15.4 percent TOMR.²⁴  

Table 10 also reveals that about 97 percent of the non-Hispanics responding to the CPS White  category also reported White in Census 2000; 94 percent of the CPS Blacks responded to the Census 2000 category; 83 percent of CPS API responded to corresponding Census categories (Asian and NHPI); and 59.2 percent of the CPS AIAN were AIAN for Census 2000. 

Comparison of the percent values in Table 10 with analogous statistics in Table 8 shows many of the pairings for race groups produced statistical differences between the edited and unedited estimates; however, when they occurred the differences were usually small, producing a similar pattern for the two sets of results. We expect some consistency between the edited and unedited files because most of the cases were not edited.²⁵ Both Table 8 and Table 10  show that about 24 percent of the CPS non-Hispanic AIAN reported as White in Census 2000 and about 11 percent reported TOMR indicating instability in this particular category.²⁶

Table 11 includes unedited non-Hispanic data and shows Census totals with CPS distributions; one percent of the Census responders did not answer the CPS race question and another one percent gave AIAN, compared with 79.8 percent who reported White, 13.5 percent Black, and 4.6 percent API.  Among the Census non-responders approximately 7 percent also did not answer the CPS question, 69.9 percent reported White, 10.8 percent said they were Black, 0.9 percent reported AIAN, and 11.3 percent API.²⁷  The unedited data in Table 11 also reveal that 98.3 percent of Census Whites were CPS Whites, 95.2 percent of the Census Blacks were CPS Blacks, 73.4 percent of the Census AIAN were CPS AIAN  and 87.9 percent of the Census Asians reported API in CPS. Comparable edited results from Table 9 were 98.5 percent, 94.5 percent, 62.7 percent, and 87.7 percent, respectively, indicating that in at least some instances the final results reflect the relatively small effects of imputation.²⁸

  Although the two sets of figures are mostly statistically different (only the API amounts are not), the differences when they occur are small, and thus the reporting pattern was not much altered by editing, an indication that response choices play a much larger role in determining race responses than Census editing procedures.

Percent distribution. Numbers in parentheses are in thousands. Data are for civilian noninstitutionalized population of the United States. Members of the Armed Forces living off post are included if there is at least one civilian adult in the household.)

Source: Tabulation of edited data from matched Census 2000 and February to May 2000 Current Population Records.(1 page 11, R1)

(-) Zero or rounds to zero.

^a Non-Hispanics were identified from the reported (not imputed) Hispanic origin response from Census 2000.

Table 11 further shows that 23.1 percent of the SOR and 8.5 percent of the TOMR did not answer the CPS race question, perhaps lending further support to the notion that if not given a choice, people will not respond.  Another 4.5 percent of Census NHPI and 3.2 percent of Census Asians also did not answer the CPS race question.  Although their rates differed from each other, some Census Blacks and Whites also failed to respond to the CPS race question.

While the specific group failure rates reveal little evidence of a correspondence between question example and response choice for CPS Non-Hispanics, these data show some race switching occurred. For example, 21.6 percent of the Census 2000 AIAN, 20.9 percent of NHPI, and 6.9 percent of Asians reported White in the CPS.²⁹  Among those who were SOR in Census 2000, there was no statistical difference between those who did not answer the CPS race question (23.1 percent) and those who reported API (14.8 percent). About 45 percent of the Census SOR reported White, and three percent AIAN. Among TOMR respondents in Census 2000, 8.5 percent did not report a race in CPS, 49.3 percent reported as White, 15.9 percent Black, 6.2 percent AIAN, and 20.1 percent API.³⁰

Race Reporting by Hispanics  

Table 12 shows that 13.8 percent of the Hispanics identified by the Census did not answer the Census race question. Another 44.7 percent identified as White, 1.9 percent Black, 1.2 percent AIAN, 32.9 percent SOR, while 5.3 percent TOMR, 0.2 percent as Asian and 0.1 percent as NHPI. Comparable statistics for the CPS race responses of Census Hispanics (Table 13) were: 15.3 percent did not answer the CPS question, 80.0 percent reported as White, 2.9 percent Black, and smaller amounts reported AIAN or API.  Among Census 2000 Hispanics responding as White in the CPS (Table 12), 13.5 percent did not answer the Census race question, 48.8 percent identified as White, 0.9 percent as AIAN, 0.6 percent as Black, 31.3 percent as SOR, and 4.8 percent TOMR.  The comparable proportions for CPS Blacks were: 14.6 percent did not answer the Census 2000 question; 14.6 percent reported White; 35.8 percent reported Black in Census 2000, 20.9 percent reported SOR, and 12.7 percent TOMR.³¹  The distributions for AIAN and API are shown but not discussed because the numbers of observations on which they are based are relatively small.³²

Table 13 shows that among Hispanics who did not answer the Census race question, 17.8 percent did not answer the CPS race question, 78.2 percent were reported White in CPS and three percent Black. Of those identified as Hispanic and White in Census 2000, 10.8 percent did not answer race in CPS, 87.3 percent reported White, and around one percent each reported Black or AIAN.  Of those who reported as Black in the Census, 19.3 percent did not answer race in CPS, 25.7 percent reported as White, and 54.6 percent as Black.³³  Among Census AIAN, 12.7 percent did not answer race in CPS, 63.5 percent were reported as White, about two percent Black, and 21.8 percent AIAN.³⁴  Asian and NHPI distributions are shown in the table but are not discussed because of the relatively small amounts. Of the Hispanics who reported SOR in Census 2000, 20.5 percent did not answer the CPS race question, 76.0 percent said they were White and 1.8 percent reported Black. Similarly for those reporting TOMR in Census 2000, 13.5 percent did not respond to the CPS question, 72.3 percent said they were White, and 6.9 percent said they were Black.³⁵

Percent distribution. Numbers in parentheses are in thousands. Data are for civilian noninstitutionalized population of the United States. Members of the Armed Forces living off post are included if there is at least one civilian adult in the household.)

Source: Tabulation of edited data from matched Census 2000 and February to May 2000 Current Population Records.

(-) Zero or rounds to zero.

 Hispanics were identified from the reported (not imputed) Hispanic origin response from Census 2000.

Percent distribution. Numbers in parentheses are in thousands. Data are for civilian noninstitutionalized population of the United States. Members of the Armed Forces living off post are included if there is at least one civilian adult in the household.)

Source: Tabulation of edited data from matched Census 2000 and February to May 2000 Current Population Records.(1 page 11, R1)

(-) Zero or rounds to zero.

Hispanics were identified from the reported (not imputed) Hispanic origin response from Census 2000.

Because there were more race categories in the Census question than in the CPS question, we would expect a lack of correspondence between the two sets of responses. However, among non-Hispanics, we found a fairly good correspondence between the race reported in CPS and Census 2000 (Table 10 and Table 11) compared with a relatively poor correspondence among Hispanics (Table 12 and Table 13).  Among CPS Non-Hispanics, we saw very high correspondence for White (98.3 percent), Black (95.2 percent) and Asian/API (87.9 percent), with a bit less for AIAN  (73.4 percent) (Table 11). Comparable statistics for CPS Hispanics were 87.3 percent, 54.6 percent, 31.7 percent, and 21.8 percent, respectively (Table 13). The possibility of providing TOMR as a Census response with no corresponding possibility in the CPS may have affected some of the observed results. For example, 23.6 percent of the CPS Hispanics reporting API chose SOR in the Census, while another 38.2 percent chose the Census TOMR (Table 12).³⁶ On the other hand, about 16.9 percent of the Hispanic CPS API category selected White as a Census category, perhaps indicating the presence of a CPS  interviewer effects.³⁷  These shifts from one kind of response to another indicate many respondents view race within a cultural context our data do not measure.

¹ Each number and percent in every table is the weighted result of CPS responses to a sample survey.  In some instances we use the word responses in this paper instead of people because the same respondents were asked both Census 2000 and CPS Hispanic origin questions and some respondents who identified as Hispanic in one instance did not do so in the other.  If the respondent identified as Hispanic in the CPS he received the CPS Hispanic sample weight, even if he gave an answer that indicated he was not Hispanic in the Census 2000 question.  The opposite was true for the cases that said they were not Hispanic in the CPS but indicated they were Hispanic in Census 2000.  This adjustment between the two groups is necessary because the census cases were matched to the CPS sample, and the CPS sample includes additional Hispanic respondent cases as part of a minority over-sample.  The Census Bureau’s experience with the CPS Hispanic sample is explained in more detail in “Thirty-five Years of Tracking Hispanic Ethnicity: Evaluation of  Current Population Survey Data Quality for the Question on Hispanic Origin, 1969 to 2004" by Dianne Schmidley and Arthur Cresce (2005), US Census Bureau Population Division Working Paper No.80 (forthcoming). The Census Bureau’s experience with Hispanic data in the decennial censuses is addressed in “Identification of Hispanic Ethnicity in Census 2000: Analysis of Data Quality for the Question on Hispanic Origin" by Arthur R. Cresce Jr., A. Dianne Schmidley, and Roberto Ramirez, US Census Bureau Population Division Working Paper No. 75.     

² 31.5 million people were identified as Hispanic by both Census 2000 and the CPS.

³ The two values, 5.1 percent and 5.8 percent, are not statistically different.  The two amounts, 5.1 percent and 3.9 percent, are not statistically different.

⁴ The two values, 2.4 percent and 2.0 percent, are not statistically different. 

⁵ The two amounts, 0.06 percent and 0.09 percent, are not statistically different.  The amount not responding in either Census 2000 or the CPS is not included in either of these calculations (0.09 percent). 

⁶ The amounts, 2.4 percent and 2.0 percent, are not statistically different from one another.  

⁷ Where differences occur they are relatively small. The amounts for Cuban (88.0 percent) and Mexican (91.3 percent) Hispanics are not statistically different from each other, however the Mexican percent is statistically different from the amount for Central and South American Hispanics (85.7 percent), whereas the Cuban amount is not different from the latter. Similarly, there is no difference between 0.2 percent and 1.1 percent.

⁸ The proportion for Cuban Hispanics (10.9 percent) is not statistically different from the proportion for Mexican

(8.5 percent) or Central or South American Hispanics (13.8 percent).

⁹ Specific Central American places of birth were Costa Rica, El Salvador, Guatemala, Honduras, Nicaragua, Panama, and “Other Central America.” Specific South American places of birth were Argentina, Bolivia, Chile, Colombia, Ecuador, Paraguay, Peru, Uruguay, Venezuela and “Other South America.”

¹⁰ There is no difference between these pairs of values: 3.3 percent (Mexico) and 3.0 percent (Cuba); 3.3 percent and 2.7 percent (Puerto Rico); 3.0 percent and 2.7 percent.

¹¹ The percent for Cuba (1.1 percent) was not significantly different from zero.

¹² The two values, 78.0 percent and 83.0 percent, are not statistically different.

¹³ The two values, 4.6 percent and 5.3 percent, are not statistically different. Implied comparisons between 4.2 percent, 4.6 percent, and 4.4 percent are also not statistically different.  Comparisons between these pairs: 3.7 percent and 5.3 percent; and 5.3 percent and 8.3 percent; yield no statistical differences.

¹⁴ The values, 2.1 percent and 2.8 percent are not statistically different from each other.  

¹⁵ Because the sample base is so small for Spain, fairly large apparent differences in the estimates may nevertheless fail to be statistically different. For example, there is no significant difference between the numbers 26.6 and 51.4 nor between 17.2 percent and 26.6 percent.

¹⁶ Some of these cases may have identified as Hispanic in Census 2000, but the data edit rules specified that write-ins such as “Philippines” or “Filipino” in the “other Spanish/Hispanic/ Latino,” category were not considered “Hispanic” but rather “not Hispanic” responses. These rules also affected other groups in this section.

¹⁷ The following pairs of numbers are not statistically different from each other: (0.2 percent and 1.0 percent), (2.4 percent and 1.0 percent), and (2.4 percent and 4.1 percent). In fact, 0.2 percent and 1.0 percent are not statistically different from zero, i.e. they don’t necessarily represent positive numbers. 

¹⁸ All the comparisons in this paragraph are not statistically different from each other with the exception of comparisons with 89.3 percent.  For example, the two values, 3.1 percent and 1.7 percent, are not statistically different while 89.3 and 2.3 are statistically different.

¹⁹ Because the sample base is so small for Brazil, fairly large apparent differences in the estimates may nevertheless fail to be statistically different. For example, there is no statistical difference between the values 50.5 and 43.8.

²⁰ 90.4 percent and 90.8 percent are not statistically different.

²¹ The two values 6.1 percent and 7.0 percent are not statistically different.

²²  The two values, 1.5 percent and 1.3 percent, are not statistically different from each other or one percent. 

²³ The following are not statistically different:14.8 percent and 16.2 percent.

²⁴ A number of non-differences occurred in the “no answer” in CPS column of Table 10: 14.7 percent, 13.4 percent and 15.4 percent were not different from each other; similarly, 13.3 percent compared with 14.7 percent, 4.9 percent compared with 3.2 percent; and 0.6 percent compared with 0.9 percent were all not statistically different.

²⁵ About 84 percent of the total population is not Hispanic.

²⁶  Twenty-four percent and 11 percent were not statistically different from their analogous values of 26 percent and 11 percent in Table 8

²⁷ There is no statistical difference between 10.8 percent and 11.3 percent.

²⁸ There is no statistical difference between 87.7 percent and 87.9 percent.

²⁹ There is no statistical difference between 20.9 and 21.6.

³⁰  There is no statistical difference between the Census SOR reporting API (14.8 percent) or TOMR reporting API (20.1 percent). There is no statistical difference between the Census SOR reporting White in CPS (45.1 percent) or TOMR reporting White in CPS (49.3 percent). Other implied comparisons based on the CPS ‘No Answer” column of Table 11  are not significantly different, including: 4.5 percent and 3.2 percent; 4.5 percent and 0.6 percent; 4.5 percent and 1.1 percent; and 4.5 and 8.5 percent.

³¹  The proportion of CPS Blacks who did not answer the Census 2000 question (14.6 percent) is not statistically different from the proportion of CPS Whites who did not answer the Census question (13.5 percent), and neither is statistically different from the Total CPS respondents who did not answer the Census race question (13.8 percent). Furthermore, neither the proportion of CPS Blacks who said they were White in Census (14.6 percent), nor the proportion of Blacks who did not answer the Census question (14.6 percent) are statistically different from 20.9 percent or 12.7 percent. The two values 20.9 percent and 12.7 percent are statistically different from each other, however. 

³² These values are not statistically different from each other: 13.8 percent (CPS Total and Census No Answer) and 13.5 percent (CPS White and Census No Answer); 13.8 percent (CPS Total and Census No Answer) and 14.6 percent (CPS Black and Census No Answer); 13.5 percent (CPS White and Census No Answer) and 14.6 percent (CPS Black and Census No Answer). 

³³ The two values, 19.3 percent and 25.7 percent are not statistically different from each other.

³⁴ The two values, 12.7 percent and 21.8 percent are not statistically different from each other.

³⁵  Indirect comparisons from the “No answer in CPS” column include these not statistically different percent amounts: 17.8 and 19.3; 17.8 and 12.7; 17.8 and 20.5; 10.8 and 12.7; 10.8 and 13.5; 19.3 and 20.5; 19.3 and 13.5;  19.3 and 12.7; and 12.7 and 13.5. Other indirect and not statistically different amounts from the “White in CPS” column include these percent comparisons: 78.2 and 76.0, 63.5 and 72.3 76.0 and 72.3 ; and, from the “Black in CPS” column, these percent comparisons: 3.0 and 2.0, 0.9 and  2.0, and,  2.0 and 1.8.

³⁶ The two values 23.6 percent and 38.2 percent are not statistically different.

³⁷ The two values 23.6 percent and 16.9 percent are not statistically different.

October 3, 2005

No. 1

The Census Bureau Approach for Allocating International Migration to States, Counties, and Places:  1981-1991.  David L. Word.  Issued October 1992.

No.  2

Geographic Coding of Administrative Records--Past Experience and Current Research. Douglas K.Sater.  Issued April 1993.

No.  3

Postcensal Population Estimates:  States, Counties, and Places.  John F. Long. Issued August 1993.

No.  4

Evaluating the Passel-Word Spanish Surname List:  1990 Decennial Census Post Enumeration Survey Results.  R. Colby Perkins.  Issued August 1993.

No.  5

Evaluation of Postcensal County Estimates for the 1980s.  Sam T.Davis.  Issued March 1994.

No.  6

Metropolitan Growth and Expansion in the 1980s.  Richard L. Forstall and James D. Fitzsimmons.  Issued April 1993.

No.  7

Geographic Coding of Administrative Records -- Current Research in ZIP/Sector-to-County Coding Process.  Douglas K. Sater.  Issued June 1994.

No.  8

Illustrative Ranges of the Distribution of Undocumented Immigrants by State.  Edward W. Fernandez and J. Gregory Robinson.  Issued October 1994.

No.  9

Estimates of Emigration of the Foreign-Born Population:  1980-l990.  Bashir Ahmed and J. Gregory Robinson.  Issued December 1994.

No. 10

Estimation of the Annual Emigration of U.S. Born Persons by Using Foreign Censuses and Selected Administrative Data:  Circa 1980.  Edward W. Fernandez.  Issued January 1995.

No. 11

Using Analytic Techniques to Evaluate the 1990 Census Coverage of Young Hispanics.  Edward Fernandez.  Issued May 1995.

No. 12

Metropolitan and Nonmetropolitan Areas:  New Approaches to Geographical Definition.  Donald C. Dahmann and James D. Fitzsimmons.  Issued October 1995.

No. 13

Building a Spanish  Surname List for the 1990's--New Approach to An Old Problem. David L. Word and R. Colby Perkins, Jr.  Issued February 1996.

No. 14

Fertility of American Men.  Amara Bachu.  Issued March 1996.

No. 15

Comparisons of Selected Social and Economic Characteristics Between Asians, Hawaiians, Pacific Islanders, and American Indians (Including Alaskan Natives).  Edward W. Fernandez.  Issued June 1996.

No. 16

Findings on Questions on Race and Hispanic Origin Tested in the 1996 National Content Survey.  Prepared in the Population Division by the Racial Statistics Branch and the Ethnic and Hispanic Statistics Branch.  Issued December 1996.

No. 17

Race and Ethnicity Classification Consistency Between the Census Bureau and the National Center for Health Statistics.  Larry Sink.  Issued February 1997.

No. 18

Results of the 1996 Race and Ethnic Targeted Test.  Issued May 1997.

No. 19

Who Responds/Who Doesn’t?  Analyzing Variation in Mail Response Rates During the 1990 Census. David L. Word.  Issued July 1997. 

No. 20

Trends in Marital Status of U.S. Women at First Birth:  1930 to 1994.  Amara Bachu.  Issued March 1998.

No. 21

State Estimates of Organized Child Care Facilities.  Lynne Casper and Martin O’Connell.  Issued March 1998.

No. 22

How Well Does the Current Population Survey Measure the Foreign-Born Population in the United States  Dianne Schmidley and J. Gregory Robinson.  Issued April 1998

No. 23

Poverty, Family Structure, and Child Well-Being:  Indicators from the SIPP.  Jason Fields and Kristin Smith.  Issued April 1998

No. 24

Child Well-Being Indicators from the SIPP.  Kristin Smith, Loretta Bass, and Jason Fields.  Issued April 1998

No. 25

Timing of First Births:  1930-34, 1990-94.  Amara Bachu.  Issued May 1998

No. 26

Co-Resident Grandparents and Grandchildren:  Grandparent Maintained Families.  Lynne Casper and Ken Bryson.  Issued March 1998.

No. 27

Population of the 100 Largest Cities and Other Urban Places in the United States: 1790 to 1990.  Campbell Gibson.  Issued June 1998

No. 28

Are There Differences in Voting Behavior Between Naturalized and Native-born Americans?  Loretta E. Bass and Lynne M. Casper.  Issued March 1999.

No. 29

Historical Census Statistics on the Foreign-born Population of the United States: 1850-1990.  Campbell J. Gibson.  Issued February 1999

No. 30

Direct Measures of Poverty as Indicators of Economic Need:  Evidence from the Survey of Income and Program Participation.  Kurt J. Bauman.  November 1998

No. 31

American Community Survey and Intercensal Population Estimates:  Where Are the Cross-roads?  Amy Symens.  Issued December 1998

No. 32

Women’s Labor Force Attachment Patterns and Maternity Leave:  A Review of the Literature.  Kristen Smith and Amara Bachu.  Issued January 1999    

No. 33

Evaluation of Relationship, Marital Status, and Grandparents Items on the Census 2000 Dress Rehearsal Charles Clark and Jason Fields.  Issued April 1999.

No. 34

Unbinding the Ties:  Edit Effects of Marital Status on Same Gender Couples.  Charles Clark and Jason Fields.  Issued April 1999.

No. 35

Racial-Ethnic and Gender Differences in Returns to Cohabitation and Marriage: Evidence from the Current Population Survey.  Philip N. Cohen.  Issued May 1999.

No. 36

How Does POSSLQ Measure Up?  Historical Estimates of Cohabitation.  Lynne Casper, Philip N. Cohen, and Tavia Simmons.  Issued May 1999.

No. 37

Childlessness Among American Women On the Rise?  Amara Bachu.  Issued May 1999.

No. 38

Methodology and Assumptions for the Population Projections of the United States: 1999 to 2100.  Frederick Hollman, Tammany Mulder, and Jeffrey Kallan.   Issued January 1999.

No. 39

What Do We Know About the Undercount of Children?  Kirsten K. West and J. Gregory Robinson.  Issued August 1999.

No. 40

Canceled

No. 41

Canceled

No. 42

Measures of Help Available to Households in Need:  Their  Relationship to Well-Being, Welfare, and Work.  Kurt Bauman and Barbara Downs.  Issued May 2000

No. 43

Have We Reached the Top?  Educational Attainment Projections of the U.S. Population.  Jennifer Cheeseman Day and Kurt Bauman.  Issued May 2000.

No. 44

The Emerging American Voter:  An Examination of the Increase in the Black Vote in November 1998.  Avalaura L. Gaither and Eric C. Newburger.  Issued June 2000

No. 45

An Analysis of State and County Population Changes by Characteristics:  1990-1999.  Amy Symens Smith, Bashir Ahmed, and Larry Sink.  Issued November 2000.

No. 46

The Effect of Work and Welfare on Living Conditions in Single Parent Households. Kurt Bauman.  Issued August 2000.

No. 47

Canceled

No. 48

Canceled

No. 49

Variations in State Mortality From 1960 to 1990.  Monique Oosse.  Issued September 2003.

No. 50

Accuracy of the U.S. Census Bureau National Population Projections and Their Respective Components of Change.  Tammany Mulder.  Issued November 2002.

No. 51

U.S. Census Bureau Measurement of Net International Migration to the United States: 1990 to 2000.  Tammany Mulder, Frederick Hollmann, Lisa Lollock, Rachel Cassidy, Joseph Costanzo, and Josephine Baker.  Issued February 2002. 

No. 52

At-Risk Conditions of U.S. School-Age Children.  Robert Kominski, Amie Jamieson, and Gladys Martinez.  Issued June 2001.

No. 53

Home Schooling in the United States:  Trends and Characteristics.  K. J. Bauman. Issued August 2001.

No. 54

Evaluation of the 1990 School District Level Population Estimates Based on the Synthetic Ratio Approach.  E. R. Miller.  Issued September 2001.

No. 55

State Estimates of Child Care Establishments 1977-1997.  Grace O’Neil and Martin O’Connell.  Issued August 2001.

No. 56

Historical Census Statistics on Population Totals by Race, 1790 to 1990, and by Hispanic Origin, 1970 to 1990, for the United States, Regions, Divisions, and States.   Campbell Gibson and Kay Jung.  Issued September 2002.

No. 57

Evaluating Forecast Error in State Population Projections Using Census 2000 Counts.   Paul Campbell.  Issued July 2002.

No. 58

Evaluating Components of International Migration:  Estimates of the Foreign-Born Population by Migrant Status:  2000.  Kevin Deardorff and Lisa Blumerman.  Issued November 2002.

No. 59

Evaluating Components of International Migration:  Legal Migrants.  Marc Perry, Barbara Van der Vate, Lea Auman, and Kathy Morris.  Issued November 2001.

No. 60

Evaluating Components of International Migration:  Temporary (Legal) Migrants.  Rachel Cassidy and Lucinda Pearson.  Issued November 2001

No. 61

Evaluating Components of International Migration:  The Residual Foreign Born.  Joe Costanzo, Cynthia Davis, Caribert Irazi, Daniel Goodkind, Roberto Ramirez.  Issued January 2002.

No. 62

Evaluating Components of International Migration:  Foreign-Born Emigrants. Tammany Mulder, Betsy Guzmán, and Angela Brittingham.  Issued June 2002.

No. 63

Evaluating Components of Internation Migration: Native-Born Emigrants. Jim C. Gibbs, Gregory S. Harper, Marc J. Rubin, and Hyon B. Shin. Issued January 2003.

No. 64

Evaluating Components of International Migration:  Migration Between Puerto Rico and the United States.  Matt Christenson.  Issued January 2002.

No. 65

Evaluating Components of International Migration:  Quality of Foreign-Born and Hispanic Population Data.  Art Cresce, Roberto Ramirez, and Gregory Spencer.  Issued July 2002.

No. 66

Evaluating Components of International Migration:  Consistency of 2000 Nativity Data.  Nolan Malone.  Issued January 2002.

No. 67

Evaluation of Census Bureau’s1995-2025 State Population Projections. Ching Li Wang.  Issued October 2002.

No. 68

Guide To International Migration Statistics: The Sources, Collection, and Processing of Foreign-Born Population Data at the U.S. Census Bureau.  Joseph M. Costanzo, Cynthia J. Davis, and Nolan Malone.  Issued October  2002.

No. 69

Seasonality of Moves and Duration and Tenure of Residence: 1996. Jason Schachter and Jeffrey Kuenzi.  Issued December 2002.

No. 70

Evaluation of 2000 Subcounty Population Estimates.  Greg Harper, Chuck Coleman, and Jason Devine.  Issued May 2003.

No. 71

People Might Move Out but Housing Units Don’t: An Evaluation of the State and County Housing Unit Estimates.  Jason Devine and Charles Coleman.  Issued April 2003.

No. 72

Analysis of General Hispanic Responses in Census 2000. Arthur R. Cresce and Roberto R. Ramirez. Issued September 2003.

No. 73

Measuring the Foreign-Born Population in the United States With the Current Population Survey: 1994-2002. A. Dianne Schmidley and J. Gregory Robinson.  Issued October 2003.

No. 74

Evaluation of April 1, 2000 School District Population Estimates Based on the Synthetic Ratio Method.  Monique Oosse.  Issued June 2004.

No. 75

 Identification of Hispanic Ethnicity in Census 2000: Analysis of Data Quality for the Question on Hispanic Origin. Arthur R. Cresce, Audrey Dianne Schmidley and Robert R. Ramirez Issued July 2004.

No. 76

Historical Census Statistics On Population Totals By Race, 1790 To 1990, And By Hispanic Origin, 1970 To 1990, For Large Cities And Other Urban Places In The United States. Campbell Gibson and Kay Jung. Issued July 2005.

No. 77

Analysis of Multiple Origin Reporting to the Hispanic Origin Question in Census 2000. Roberto Ramirez. Issued November 2005.

No. 78

Changes in the Lives of U.S. Children: 1990-2000. Julia Overturf Johnson. Issued July 2005.

No. 79

Matched Race and Hispanic Origin Responses from Census 2000 and Current Population Survey February to May 2000. Jorge del Pinal and A. Dianne Schmidley. Issued December 2005.

No. 80

Thirty-Five Years of Tracking Hispanic Ethnicity: Evaluation of Current Population Survey Data Quality for the Question of Hispanic Origin, 1969 to 2004. A. Dianne Schmidley and Arthur R. Cresce. Release Pending.