Transcript Working with Multi-Year ACS Products: Arun Peter Lobo, NYC Dept

American Community Survey Five-Year
Estimates
New York State Fall Affiliate Meeting
October 28, 2010
1
Arun Peter Lobo
Population Division
Overview
 Three major concerns with 2005-2009
ACS census tract data:
 The use of 2000 Geography
 Validity of data
 Reliability of data
New York City
2010 Census Tracts
2000 – 2010 CT Changes
 Validity of Age Data at the PUMA level
And what it says about tract-level data
Bronx Public Use
Microdata Areas
(PUMAs)
3701
3702
3706
3704
3705
Bronx
Other, NH
5.8%
3703
3707
3709
White, NH
14.5%
Hispanic
48.4%
Black, NH
31.2%
3710
3708
5
Method: The Cohort Component (CC)
Model
• The basic building blocks in a CC model are
age and sex “cohorts” in a given population.
• The CC model applies birth, death, and
migration rates to these age and sex cohorts
to project them into the future.
2000
20-24
2005
2010
2015
2020
2025
2030
20-24
25-29
30-34
35-39
40-44
45-49
50-54
Arrow shows 20-24 year old cohort aging through time
50-54
Age/Sex Distributions for Bronx PUMA 3701:
2005-07 ACS, and 2006 DCP Baseline and DCP Estimates
TOTAL
0-4
5-19
20-44
45-64
65+
2005-2007 ACS
Male
Female
51,082
59,128
3,567
4,083
11,030
9,881
17,966
19,649
11,790
14,358
6,729
11,157
2006 DCP Baseline
Male
Female
49,623
60,387
3,525
3,384
11,156
10,574
17,844
20,962
11,127
14,056
5,972
11,411
2006 DCP Estimates
Male
Female
51,081
59,125
3,487
3,350
11,495
10,348
18,389
20,541
11,497
13,740
6,213
11,145
7
Bronx PUMA 3701: DCP Baseline and DCP Modified AgeSpecific Migration Rates for Males, 2000-2006
300
+ 5 per 1,000
Baseline
-21 per 1,000
100
0
65+
60-64
55-59
50-54
45-49
40-44
35-39
30-34
25-29
20-24
15-19
10-14
5-9
-100
0-4
Migration Rate
200
DCP Modified
Age Group
8
Age/Sex Distributions for Bronx PUMA 3701:
2005-07 ACS and 2006 DCP Absolute Estimates
TOTAL
0-4
5-19
20-44
45-64
65+
2005-2007 ACS
Male
Female
51,082
59,128
3,567
4,083
11,030
9,881
17,966
19,649
11,790
14,358
6,729
11,157
2006 DCP Estimates
Male
Female
51,081
59,125
3,487
3,350
11,495
10,348
18,389
20,541
11,497
13,740
6,213
11,145
100 age/sex estimates: 27 significant differences
9
Population 0 to 4 by Sex for Bronx PUMAs:
Comparing 2005-2007 ACS Estimates to DCP Estimates
Males Females
2005-2007 ACS
2006 DCP Estimate
2006 Baseline
9,000
Population 0 to 4
8,000
7,000
6,000
5,000
4,000
3,000
2,000
1,000
0
M
F* M* F* M
3701
3702
F
3703
M
F
3704
M* F
3705
M
3706
F
M
3707
F
M
3708
F
M
3709
F
M
3710
PUMA
*Difference between the ACS and DCP estimates is significant at the .10 level.
Sex Ratio
2000 Census
Jul 01-Jul 06 Births
2006 DCP
2005-2007ACS
3701
103.6
104.7
104.1
87.4
3702
104.5
103.6
105.5
129.2
3703
102.1
107.0
104.0
131.8
3704
103.0
104.4
103.2
102.2
3705
102.6
104.1
102.3
112.9
3706
103.4
104.1
104.7
106.5
3707
106.4
103.3
108.8
102.5
3708
103.6
103.2
104.0
92.9
3709
104.8
102.3
105.2
93.3
3710
99.9
104.5
10
100.4
96.0
F
Age/Sex Distributions for Bronx County, New York:
2000 Census, and 2005-07 ACS and 2006 and DCP
Estimates
TOTAL
0-4
5-19
20-44
45-64
65+
2000 Census
Male
Female
620,171
712,479
55,807
53,925
167,541
161,250
236,690
272,441
111,321
139,727
48,812
85,136
2005-2007 ACS
Male
Female
640,981
728,878
55,038
53,043
165,180
159,101
238,650
266,843
129,007
160,234
53,106
89,657
2006 DCP Estimates
Male
Female
640,959
728,887
51,623
49,543
181,003
168,940
238,501
274,463
121,101
152,715
48,732
83,227
11
Bronx County: ACS and DCP Age-Specific Migration
Rates for Females, 2000-2006
300
ACS
200
100
48 per 1,000
0
-100
-82 per 1,000
65+
60-64
55-59
50-54
45-49
40-44
35-39
30-34
25-29
20-24
15-19
10-14
5-9
-200
0-4
Migration Rate
DCP
Age Group
12
Population 65 and Over by Sex for Bronx PUMAs:
Comparing 2005-2007 ACS Estimates to DCP Estimates
Males Females
2005-2007 ACS
2006 DCP Estimate
2006 Baseline
Population 65 and Over
16,000
12,000
8,000
4,000
0
M
3701
F
M
F
3702
M*
3703
F*
M
3704
F*
M
F
M*
3705 3706
PUMA
F*
M
3707
F
M
3708
F
M
3709
F*
M
F
3710
13
*Difference between the ACS and DCP estimates is significant at the .10 level.
Age/Sex Distributions for Bronx PUMA 3701:
2005-07 ACS and 2006 DCP Percent Distributions
TOTAL
0-4
5-19
20-44
45-64
65+
2005-2007 ACS
Male
Female
51,082
59,128
7.0%
6.9%
21.6%
16.7%
35.2%
33.2%
23.1%
24.3%
13.2%
18.9%
2006 DCP Estimates
Male
Female
51,081
59,125
6.8%
5.7%
22.5%
17.5%
36.0%
34.7%
22.5%
23.2%
12.2%
18.9%
100 age/sex estimates: 41 significant differences
24 of the differences were 2 points or greater, including 10 that were 3 points or greater
14
Hispanic Diversity in
Bronx PUMAs in 2000
Hispanics
Puerto Rican 34%
Poverty
43%
3707
Bronx
Other, NH
5.8%
White, NH
14.5%
Hispanic
48.4%
3703
Black, NH
31.2%
Hispanics
Puerto Rican 72%
Poverty
17%
15
Black Diversity in Bronx
PUMAs in 2000
Black nonhispanics
Foreign-born 42%
Poverty
17%
3702
Bronx
Other, NH
5.8%
White, NH
14.5%
Hispanic
48.4%
3708
Black, NH
31.2%
Black nonhispanics
Foreign-born 21%
Poverty
36%
16
 Low Reliability of ACS Estimates at the
Census Tract Level
And a solution for some users
ACS Initial Sample and Final Response as a Percent of
all Bronx Housing Units, 2001-2005
2.8%
2001
1.5%
% of Bronx Housing Units
2.2%
2002
1.1%
Initial Sample
Final Interviews
Housing Unit Average: 496,517
2.2%
2003
1.2%
2.2%
2004
1.1%
2.1%
2005
1.2%
0
5,000
10,000
Housing Units
ACS SAMPLE
5 Year Initial Sample Total: 56,761
(11.4% of annual avg. housing units)
5 Year Final Response Total: 29,896
(6.0% of annual avg. housing units)
2000 CENSUS LONG FORM
Final Response was 11.3% of housing
units
15,000
43 Demographic, Social, Economic, and Housing Variables
Bronx Census Tracts and Neighborhoods, 2001-2005 ACS
Demographic & Social
Economic
Housing
1) Children 0 to 4 years of age
2) Nonrelatives
3) Married couple families
4) Nonfamily households
5) High school graduate only
6) Bachelor’s degree
7) Never married
8) Divorced females
9) Civilian veterans
10) Foreign-born
11) Naturalized U.S. citizen
12) Speaks lang. other than Eng.
13) Speaks Eng. less than "very well"
14) Speaks Spanish at home
1) Persons employed
2) Females employed
3) Use public transp.
4) Mgmt., prof. & related occp.
5) Construction occp.
6) Retail industry
7) Education industry
Class of workers:
8)
Government
9)
Self-employed
Households:
10)
w/ inc. & benefits < $10,000
11)
w/ earnings
12)
w/ Social Security
13)
w/ cash public assist. inc.
Housing units in:
1)
Struct. w/ 20+ units
2)
Struct. blt.1990 or later
3)
Struct. blt.1980-89
4)
Struct. blt.1970-79
5)
Struct. blt. before 1940
6) Rooms in unit = 1
7) Rooms in unit = 5
8) Vehicles available = 0
9) Vehicles available = 1
Heating fuel:
10)
Utility gas
11)
Electric
12)
Oil, etc.
13) No telephone service
14) 1.5+ occupants per room
Gross rent as % of HH inc.:
15)
less than 15%
16)
35% or more
Margins of Error
• Standard Error (SE): Measures the variability
of an estimate due to sampling.
• Margin of Error (MOE) : Describes the
precision of an estimate at a given level of
confidence. For the ACS, since 90 percent
confidence used, MOE= SE*1.645
• Say we estimate the number of carpenters to
be 10,000, with an MOE of 1,000 persons.
• If we had the patience and the dollars to
create the same estimate of carpenters
several thousand times from several
thousand samples, 90% of the estimates of
carpenters would be between 9,000and
11,000.
Coefficient of Variation
•
Coefficient of Variation (CV) : Standardized
indicator of the reliability of an estimate.
CV=SE/Estimate
•
Step 1 - Derive the standard error
MOE= SE*1.645
SE = MOE/1.645
=1,000/1.645=608
•
Step 2 - Express the SE as a percent of the
estimate CV= SE/Estimate * 100
CV = 608/10,000 * 100, or 6.0 percent
•
Step 3 – Evaluate: Our rule of thumb is that
a CV in excess of 15 percent warrants
caution
21
Median Overall Coefficients of Variation for
Bronx Census Tracts, 2000 Census and 2001-2005 ACS
Ratio of ACS
to Census = 1.8
30
1.6
27.4
26.8
Coefficient of Variation
1.5
21.1
20
16.6
17.9
11.5
10
2001-2005 ACS
2000 Census
0
Demographic & Social
Variables
Economic Variables
Housing Variables
Neighborhood Aggregates of Bronx Census Tracts
Riverdale
Fieldston
Woodlawn
Wakefield
Van
Cortlandt
Park
Williamsbridge
Olinville
Spuyten Duyvil
Kingsbridge
Van
Cortlandt
Village
Eastchester
Edenwald
Baychester
Norwood
Co-Op City
Bedford Park
Fordham North
Kingsbridge
Heights
Bronxdale
Fordham
South
Morris
Heights
Pelham
Bay
Park
Belmont
Allerton
Pelham Gardens
Bronx
Park
Pelham Parkway
Pelham
Bay
Park
Mount Hope
Bathgate
Claremont
Highbridge
Claremont
Park
Crotona
Park
East
West Farms
Bronx River
Westchester
Unionport
Soundview - Bruckner
Morrisania
Melrose
Schuylerville
Throgs Neck
Edgewater Park
Longwood
Soundview
Park
Melrose South
Mott Haven North
Mott Haven
Port Morris
Pelham Bay
Country Club
City Island
Parkchester
Crotona
Park
East Concourse
Concourse Village
West
Concourse
Van Nest
Morris Park
Westchester Sq.
East
Tremont
Hunts Point
Riker's
Island
Soundview
Castle Hill
Clason Point
Pelham Bay
Country Club
City Island
Median Overall Coefficients of Variation for
Bronx Census Tracts and Neighborhoods, 2001-2005 ACS
Ratio of Neighborhoods
to Tracts =
0.36
0.31
0.33
40
Coefficient of Variation
32.9
30
26.8
21.6
20
10.9
10
7.7
8.4
Census Tracts
Neighborhoods
0
Demographic & Social
Variables
Economic Variables
Housing Variables
How Can We Assist Users?
• Important things to keep in mind:
–
–
–
–
–
–
What the ACS is and isn’t
What a 5 year estimate means
Area with abundant change
Explain measures that have a “look back” period
Depict Confidence Intervals pictorially
Resist establishing “hard and fast” rules
• Distinguish the ACS from the decennial count
Create a sense that there is a “cycle” at work
Developing an Interactive Relationship
with the Census Bureau
• Creation of a User Feedback Loop
• Keep track of inquiries and develop practical
strategies to respond
• Identify knowledge gaps that need to be
addressed through special efforts
WEB SITE
WWW.NYC.GOV/PLANNING