Show The Graduate Center Menu
 
 

Census 2010 Participation Rate Analysis (Week 2)

Be Counted, America! How are we doing? (Part 2)

Two weeks & counting: An analysis of the 1st & 2nd weeks of participation in the 2010 Census

 

April 8, 2010

The Center for Urban Research (CUR) at the CUNY Graduate Center has analyzed the latest participation rates from the 2010 decennial census, in an ongoing effort to understand which areas are not responding as well and why.

On March 31 we prepared an analysis of the first week of participation, when the nationwide rate was 50%. One week later, on April 6, the nationwide participation rate had risen to 62%. Participation rates also increased locally across the country, but the increases were uneven from one area to the next. Also, while there was a generally consistent correlation between participation rates and demographic characteristics, these relationships varied widely from one city to the next.

In addition to this analysis, CUR continues to provide easy access to the latest participation rates mapped at the census tract, county, and state levels nationwide with CUNY’s Census 2010 Hard to Count mapping site (www.CensusHardToCountMaps.org). Our application provides features that the Census Bureau’s Take 10 map does not have, allowing users to:

  • type in a county and highlight the tracts below a certain participation rate (you can enter whatever threshold you want);
  • sort the resulting list so you can see at a glance the highest and lowest performing tracts (results are highlighted on the map to show geographic concentrations); and
  • compare the 2010 rate map with the 2000 rate map (click the “More…” tab and check the box for “Participation Rate in 2000”).

Key findings from Week 2 (as of April 6)

As with our report for Week 1, we compared trends between urban/non-urban areas and analyzed correlations nationwide at the county-level between participation and race/ethnic population characteristics. New for this week: we looked more closely at race/ethnicity characteristics and participation rate correlations at the tract level on a city-by-city basis (for the 67 largest cities with populations greater than 250,000). We also examined the relationship between county-level unemployment in 2008 and participation rates.

Based on the 2010 participation rates published by the Census Bureau on April 6, we found that:

1. On average nationwide, there continues to be a positive correlation at the county level between participation rates and the percent of the population that is White, while the correlations between participation rate and the percentage of the population that is Hispanic or Black remains negative.

(The county-level race and ethnicity data is from the Census Bureau’s population estimates program (http://www.census.gov/popest/counties/asrh/), as of 2008.)

Correlation of County-level Population Characteristics with Participation Rates
(as of April 6, 2010)

 

Pearson Correlation

Percent White

.453

Percent Hispanic

-.314

Percent Black

-.28

Percent Asian

-.006*

Percent All Others

-.258

(Correlations based on N = 3,112 counties. * indicates finding is not significant at the .01 level.)

 

2. Within counties, however, race and ethnicity are even more strongly negatively correlated with participation rates nationwide, especially regarding the percent of a tract’s population that is Black (the greater percentage of Black population, the lower the participation rate).

(Note that the tract-level analysis uses race/ethnicity data from the 2000 Census (SF3, table P7), the latest data available for that level of geography.)

Correlation of Tract-level Population Characteristics with Participation Rates
(as of April 6, 2010)

 

Pearson Correlation

Percent White

.588

Percent Hispanic

-.344

Percent Black

-.428

Percent Asian

-.062

Percent All Others

-.258

(Correlations based on N = 63,653 tracts. All r values significant at the 2-tail, .01 level.)

While the relationship between participation rate and the percent of a county’s population that is Asian is not statistically significant, the tract level correlation is significant.

3. In Week 1 (as of March 30), the correlations between participation and race/ethnicity population characteristics were of similar strength and direction (we did not test the statistical significance of the differences in the correlations from Week 1 to Week 2).

4. At the county level, unemployment rates were negatively correlated with participation rates. This finding was statistically significant, though the association was not substantial (r = -.078). (Unemployment data was downloaded from ftp://ftp.bls.gov/pub/special.requests/la/laucnty08.txt)

Local Patterns in Major Cities

We also analyzed the relationship between participation rate as of April 6 and race/ethnicity characteristics of tracts for the country’s 67 largest cities (those with populations greater than 250,000, based on 2007 population estimates). The results were quite varied. These local exceptions to the nationwide patterns underscore that our overall findings noted above need to be understood within a local context – and census outreach efforts need to be responsive to these local trends.

Because the tract-level analysis uses race/ethnicity data from the 2000 Census (SF3, table P7), the latest data available at that level of geography, the relationships described below may not be as reliable in local communities whose racial and ethnic (Hispanic) composition has changed substantially since 2000. Otherwise, they provide useful indicators regarding associations at the neighborhood level between race/ethnicity and participation in the 2010 Census.

Here are some interesting variations identified by our analysis of tract population characteristics and participation rates within these cities. The results listed below are only for cities in which the correlations were found to be statistically significant. The data used for this analysis is available in this spreadsheet [Excel].

1. In 19 major cities (especially St. Louis, Boston, and Oakland), the negative correlation between participation rates and the percent of the population that is Black is stronger than the national pattern (nationally, the tract-level correlation statistic = -.428). In these cities, participation rates are more likely to be lower in census tracts compared to a given percent of a tract’s population that is Black than what we found nationally. (The list below is limited to cities with a Black population greater than 20% of total population.)

 

City Pearson Correlation
St. Louis, MO (strongest)
Boston, MA
Toledo, OH
Oakland, CA
New Orleans, LA
Buffalo, NY
Pittsburgh, PA
Cincinnati, OH
Philadelphia, PA
Milwaukee, WI
Columbus, OH
Kansas City, MO
Raleigh, NC
Indianapolis city (balance), IN
Jacksonville, FL
Charlotte, NC
Nashville-Davidson (balance), TN
Washington, DC
Tampa, FL
-.709
-.708
-.692
-.669
-.634
-.595
-.587
-.581
-.572
-.557
-.503
-.486
-.477
-.473
-.457
-.454
-.452
-.441
-.439

 

2. Only 9 cities (Houston, Dallas, and 7 others) had negative correlations at the tract level between participation rates and a tract’s Black population that were weaker than the national pattern (r = -.428). (The list below is limited to cities with a Black population of 20% or more.)

 

City Pearson Correlation
Houston, TX (weakest)
Dallas, TX
Atlanta, GA
Baltimore, MD
Cleveland, OH
Chicago, IL
New York, NY
Memphis, TN
Newark, NJ
-.206
-.250
-.282
-.305
-.318
-.349
-.381
-.406
-.423

 

3. In 34 cities, the positive correlation between census participation and the percent of a tract’s population that is White is stronger than the overall national pattern (r = .588).

 

City Pearson Correlation
St. Paul, MN (strongest)
Minneapolis, MN
Arlington, TX
Omaha, NE
Oakland, CA
Colorado Springs, CO
Milwaukee, WI
Toledo, OH
Wichita, KS
St. Louis, MO
Las Vegas, NV
Stockton, CA
Boston, MA
Anchorage, AK
Anaheim, CA
Phoenix, AZ
Fresno, CA
Long Beach, CA
Philadelphia, PA
Riverside, CA
Mesa, AZ
Buffalo, NY
Chicago, IL
New Orleans, LA
Seattle, WA
Pittsburgh, PA
Charlotte, NC
Cincinnati, OH
Kansas City, MO
Tulsa, OK
Santa Ana, CA
Raleigh, NC
Albuquerque, NM
Bakersfield, CA
.837
.834
.832
.807
.799
.799
.788
.788
.744
.740
.739
.737
.705
.702
.698
.694
.679
.670
.667
.656
.642
.642
.636
.635
.621
.618
.613
.612
.602
.598
.598
.598
.594
.593

 

4. The positive correlation between census participation rates and the percent of a tract’s population that is White is weaker than the national pattern (r = .588) in 27 other cities.

 

City Pearson Correlation
San Antonio, TX (weakest)
New York, NY
Baltimore, MD
Atlanta, GA
Newark, NJ
Houston, TX
Tucson, AZ
Cleveland, OH
Tampa, FL
San Jose, CA
Austin, TX
Washington, DC
Memphis, TN
Sacramento, CA
Fort Worth, TX
Los Angeles, CA
San Diego, CA
Jacksonville, FL
Denver, CO
Portland, OR
Lexington-Fayette, KY
Virginia Beach, VA
Dallas, TX
Columbus, OH
Indianapolis city (balance), IN
Nashville-Davidson (balance), TN
Oklahoma City, OK
.163
.274
.333
.337
.337
.382
.386
.399
.404
.460
.469
.475
.477
.477
.480
.497
.501
.510
.510
.513
.527
.529
.550
.553
.560
.578
.587

 

In contrast to the national tract-level pattern of positive correlation for Whites, a greater percent of a tract’s population that is White in Honolulu, HI tends to lower participation rates (r = -.298).

5. The only major city in which the patterns of White and Black population and census participation are reversed is Detroit. Participation rates in Detroit as of April 6 tended to be lower in tracts with a greater percentage of Whites (r = -.154), and tended to be higher in tracts with a greater percentage of Blacks (r = .235), though the statistical strength of these associations is only moderate.

6. Several cities had stronger correlations for tract-level White and Black populations and census participation. (The list below omits cities with Black population less than 20%. It also omits New Orleans, given the problems of attempting to analyze post-Katrina population patterns using pre-Katrina data from the 2000 Census.)

 

City, State

Pearson Correlation

(for % White, non-Hispanic in 2000)

City, State

Pearson Correlation

(for % Black, non-Hispanic in 2000)

Boston, MA

.705

Boston, MA

-.708

Toledo, OH

.788

Toledo, OH

-.692

St. Louis, MO

.740

St. Louis, MO

-.709

Oakland, CA

.799

Oakland, CA

-.669

 

7. In Miami, Newark, NJ and New York, tract-level Hispanic population concentrations had the opposite effect on census participation than the national pattern. In tracts in these 3 cities, greater Hispanic populations tended to increase participation rates. The strength of this relationship is greatest in Miami (r = .631), and only moderate in Newark (r = .340) and New York (r = .138). Nationally, tracts with greater Hispanic populations tended to lower participation rates (r = -.344).

8. The correlation between participation rates and the percent of a tract’s population that is Asian was slightly negative on a national basis (r = -.062). But the tract-level percent of Asians had a much stronger association with census participation, either positive or negative, in several cities. The list below is limited to cities with an Asian population greater than 5% of total population.

 

Positive

Negative

Santa Ana, CA (strongest)

.626

Honolulu, HI

.487

San Francisco, CA

.254

 

St. Paul, MN (strongest)

-.654

Minneapolis, MN

-.630

Seattle, WA

-.466

Anchorage, AK

-.385

Riverside, CA

-.374

Fresno, CA

-.357

Arlington, TX

-.297

 

 

9. The negative correlation between participation rates and the percent of the population that is Hispanic is stronger than the national correlation (r = -.344) in 17 major cities (limited to cities with Hispanic population greater than 20%).

 

City Pearson Correlation
Santa Ana, CA (strongest)
Long Beach, CA
Phoenix, AZ
San Jose, CA
Anaheim, CA
Fresno, CA
Las Vegas, NV
Bakersfield, CA
San Diego, CA
Stockton, CA
Albuquerque, NM
Oakland, CA
Sacramento, CA
Los Angeles, CA
Austin, TX
Riverside, CA
Denver, CO
-.783
-.720
-.664
-.617
-.592
-.586
-.561
-.525
-.518
-.512
-.471
-.464
-.422
-.399
-.381
-.378
-.376

 

The New York exception

We also examined the relationship between census participation and Hard-to-Count (HTC) scores in these 67 major cities. Generally there was strong negative correlation at the tract-level. Census tracts with higher scores (i.e., characterized by the Census Bureau as harder to count) tended to have lower participation rates.

The one major city with a very weak correlation between participation rate and HTC score is New York City. New York had a correlation statistic of -.111 between participation and HTC score. At the same time, the relationships between race/ethnicity and participation tended to be weak in New York (the percent of an average tract’s population that is Black or White had a weaker relationship with participation rates; the trend was in the opposite direction for the Hispanic population).

For example, the following two maps show the HTC scores compared with the latest participation rates for tracts in southeast Queens:

 

April 6, 2010 participation rate by tract

(dark blue/green color = low participation)

Tracts with HTC scores > 60

(yellow = HTC score 61-70; orange = 71-75; red = 76+)
Census_Queens_PartRate Census_Queens_HTC

(Map source: www.CensusHardToCountMaps.org)

While the Mayor’s office in New York City has noted that the city has “the highest percent of hard-to-count residents of any city in the nation”, the hard-to-count scores may not have as much bearing on why New Yorkers are not participating as fully in the Census as residents of other cities. (Source: “Mayor Bloomberg Details Neighborhood by Neighborhood 2010 Census Response Rates in Final Push to Increase Participation”, News Release PR- 147-10, April 7, 2010. Downloaded from www.nyc.gov.)

Follow up from Week 1

1. Census tracts in cities have much lower participation rates than non-urbanized areas. The median participation rate as of April 6 in tracts located in major cities was 56%, while the median participation rate in tracts in non-urban areas was ten points higher, at 66%. ( We used ArcGIS geographic information system (GIS) software to determine which tracts were located in urbanized areas, based on Census Bureau geographic classifications (www.census.gov/geo/www/ua/uaucbndy.html). We also separately determined which tracts were located in central cities based on metropolitan statistical areas. Central cities can be inside or outside urbanized areas.)

2. Rates continued to rise, however, across urban and non-urban areas. In the week between March 30 and April 6, the median participation rate in major cities rose by 13 points (from 53%). In non-urban areas it rose somewhat slower, by 12 points (from 54%).

3. Generally, hard to count tracts continued to have lower participation rates – regardless of urban/non-urban location – than tracts with lower HTC scores. See chart and table below.

CensusWeek2barchart

Median Participation Rate by HTC score and Type of Central City/Urban Indicator

HTC Score

Tracts in Non Urban areas

Tracts in Urban Areas & Outside Central City

Tracts in Urban areas & Inside Central City

Nationwide median

1 to 30

69%

69%

67%

68%

31 to 60

59%

59%

57%

58%

61 to 70

54%

54%

52%

53%

71 to 75

53%

52%

51%

51%

76 to 100

50%

50%

48%

49%

101 to hi

44%

43%

46%

46%

Median across
all scores

65%

65%

56%

62%

 

We mapped the latest participation rates nationwide by county, and compared the latest rates to the county-level rates in 2000:

  • The Midwest continues to have the highest rates, consistent with that region’s participation rate in the 2000 Census.
  • But county-level participation in several states in the South, along the east coast, upper Midwest, and Pacific Northwest continues to grow. This map [PDF] highlights these increases.
  • A growing number of counties – 538 of them as of April 6 – have met or exceeded their 2000 participation rates. On March 30, only 44 counties had met or exceeded their 2000 rates. (This analysis excludes the 29 counties for which the Census Bureau was not able to calculate a participation rate in 2000.)

For more information, contact:

Center for Urban Research

at the Graduate Center, City University of New York


www.CensusHardToCountMaps.org

cunymapping@gc.cuny.edu

All work and materials are supported by a grant from the Hagedorn Foundation

and coordinated by the Funders Census Initiative © 2010.