Studying Neighborhoods Using Uncertain Data from the American Community Survey: A Contextual Approach

Figures & data

Table 1. 2006 to 2010 American Community Survey estimates of African American median household income in a selected group of proximal tracts in Denver County, Colorado

Tract number	African American median household income ($)	Margin of error ($)
Census Tract 41.01	28,864	8,650
Census Tract 41.02	21,021	4,458
Census Tract 41.03	43,021	14,612
Census Tract 41.04	36,092	3,685
Census Tract 41.06	60,592	68,846

Table 2. Geodemographic conceptual model

Concepts	Domains	Measures
Population
	Age	Age distributions
	Race	Racial categories
	Education	High school graduate, college degree, professional degrees
	Family structure	Same-sex couples, married couples, single females with children
	Language	Only speak English, Spanish at home (with or without good English)
Environment
	Stability	Moved within a year (same town or from another town)
	Housing	Vacancy, structure type, building age, home values, rent
	Density	Population density
Economy
	Commuting	Duration (e.g., 25–29 minutes), commuting by public transport
	Occupation	Types of employment (e.g., agriculture, retail, etc.)
	Wealth	Income, public assistance, vehicle ownership

Figure 1. Dendrogram showing the 250 classes created via k-means and how they are combined into ten groups.

Figure 2. Average silhouette for partitions of the 250-class k-means solution. Vertical lines indicate the ten-class group level of the classification and the fifty-five-class type level of the classification.

Figure 3. Summary of the ten-class group level of the classification by domain.

Figure 4. Group level (ten-class) map of census tracts in New York City and Chicago (maps not to scale).

Figure 5. Group-level (ten-class) map of census tracts in the United States.

Figure 6. Variables from the population domain by group. Numbers represent the percentage of the national average for each variable in each group. For example, the top cell in Panel A indicates that in Group J the number of people with less than a high school education is 72 percent of the national average.

Figure 7. Variables from the economy domain by group. Numbers represent the percentage of the national average for each variable in each group. For example, in Panel B the 493 in the leftmost column indicates that the number of people without a car in Group E is 493 percent of the national average.

Figure 8. Housing variables form the environment domain by group.

Figure 9. The type level of the classification for Group D in Arizona.

Figure 10. Population pyramids showing age distribution by gender for each group of the classification. Bars show five-year age intervals. The width of the bar denotes the amount of variation within each age category in each group. The vertical tick on each bar denotes the median for each age interval. Note the very large numbers of young people in Group J and the inverted pyramid in Group G.

Table 3. Exogenous evaluation: Gini coefficient and index scores for Federal Election Commission profession and contribution amount

	Gini	Group A	Group B	Group C	Group D	Group E	Group F	Group G	Group H	Group I	Group J
Creative	0.68	28	160	22	7	1,031	45	221	68	17	126
Legal	0.59	32	165	30	11	784	74	217	51	36	112
Donations of $5,000 or more	0.59	34	166	31	44	717	71	299	29	27	115
Student	0.58	35	163	30	8	720	73	200	66	24	238
Professor	0.57	21	166	38	7	480	91	159	61	30	421
Donations of $1,000 or more	0.54	36	177	35	30	550	72	280	34	27	102
Finance	0.52	33	180	38	26	577	68	206	44	25	90
Farmer	0.48	112	64	152	202	32	53	78	1	61	58
Home maker	0.48	30	185	51	28	276	76	273	17	35	104
Engineer	0.44	40	210	45	24	195	85	142	27	20	80
Higher professionals	0.43	40	171	50	30	434	81	203	32	36	95
Donations of $250–$1,000	0.43	37	172	52	33	386	80	225	36	31	114
Retired	0.42	34	160	62	49	236	79	359	23	26	129
Medical	0.38	51	164	61	31	244	99	195	28	42	126

* p < 0.001, one-way analysis of variance on group-level means.

Table 4. Crime in Chicago

	Gv	Group A (%)	Group B (%)	Group C (%)	Group D (%)	Group E (%)	Group F (%)	Group G (%)	Group H (%)	Group I (%)	Group J (%)
Manslaughter	0.73	62	0	0	—	0	0	0	278	0	0
Gambling	0.69	8	6	79	—	2	10	9	276	218	0
Prostitution	0.63	90	7	37	—	31	19	13	200	150	79
Drug abuse	0.61	47	16	82	—	32	4	40	221	189	43
Homicides	0.58	61	30	114	—	15	18	16	193	183	0
Weapons violation	0.57	51	18	139	—	24	6	37	193	227	28
Aggravated battery	0.56	61	28	96	—	44	16	45	187	212	83
Disorderly conduct	0.55	82	34	83	—	50	10	44	169	174	48
Public violence	0.55	51	30	146	—	43	5	48	182	193	65
Arson	0.55	104	42	65	—	21	0	29	150	225	60
Aggravated assault	0.54	65	33	114	—	52	18	60	171	213	43
Family offenses	0.54	83	42	94	—	49	21	59	162	157	72
Violent	0.53	60	35	117	—	70	12	65	172	189	125
Battery	0.53	62	38	99	—	69	21	68	173	202	105
Liquor license	0.52	137	88	20	—	145	49	0	87	127	115
Robbery	0.51	57	40	131	—	92	10	79	164	168	162
Sexual assault	0.50	67	39	108	—	91	0	71	160	171	358
Assault	0.50	62	42	115	—	89	21	77	159	203	133
Burglary	0.50	75	61	110	—	99	19	48	146	147	66
Misc. offense	0.49	64	44	99	—	105	24	63	157	206	187
Vandalism	0.49	79	55	111	—	87	30	69	141	166	95
Vehicle theft	0.47	95	68	97	—	102	27	68	122	144	108
Stolen property	0.46	104	61	112	—	90	0	64	104	231	0
Sexual abuse	0.46	96	62	77	—	118	33	78	121	157	291
Property	0.42	69	65	100	—	197	29	90	122	146	326
Forgery	0.40	68	51	114	—	217	26	113	115	125	776
Embezzlement	0.39	69	53	105	—	206	257	0	116	99	899
Fraud	0.38	46	57	96	—	282	15	108	127	154	562
Larceny	0.38	60	66	98	—	253	33	110	115	146	466
Percentage of population		32	14	10	0	7	1	1	28	4	0.1
Percentage of crime		20	7	11	0	9	0.3	1	43	6	0.8

Note: This table shows G_v for each type of crime. Cells contain the percentage of the city-wide crime rate. Values greater than 100% indicate more crime than average and values less than 100% indicate less crime than average. The bottom rows show the percentage of the city's population that lives in each group and the percentage of all crime that occurs in each group. Group D does not occur in the city.