9 20,513
1,419
9,634
61
5,954
45
74
755
4 20,725
12,430
774
586
2,408
30
23
1,106
2,571Alameda
Amador 2 6,759 1,130 1,628 13 432 18 23 173 3,342Calaveras Colusa 1 3,849 2,779 20 177 215 2 8 197 451Contra Costa Del Norte 24 129,136 16,821 14,531 1,056 16,379 140 115 3,121 76,973El Dorado Fresno 4 20,914 15,641 295 1,002 507 48 124 1,436 1,861Glenn 7 33,900 3,869 586 1,041 413 48 21 284 27,638Humboldt Imperial 1 2,524 1,493 — 122 23 3 5 43 835Inyo Kern 9 50,889 29,706 2,387 804 318 58 80 1,383 16,153Kings 3 20,830 15,438 887 466 253 42 26 710 3,008Lake Lassen 1 5,679 2,293 1,430 66 31 4 95 31 1,729Los Angeles Madera 173 810,743 42,471 122,256 2,550 76,192 1,403 1,165 10,611 554,095Mariposa Mendocino 3 16,496 10,522 105 499 158 19 22 455 4,716Merced 1 4,008 1,552 264 29 200 2 37 62 1,862Modoc Monterey 1 3,538 2,881 8 109 37 4 15 72 412Napa Nevada 1 2,547 399 79 12 93 6 — 51 1,907Orange 1 3,723 3,412 25 6 116 3 — 20 141Placer Plumas 19 68,114 34,778 2,983 614 1,596 125 40 1,440 26,538Riverside Sacramento 14 67,166 15,668 10,904 598 11,103 759 141 3,931 24,062San Bernardino San Diego no 9 23,809 5,870 2,345 448 752 65 38 598 13,693San Francisco 20 92,161 7,101 5,632 251 11,103 289 97 2,017 65,671San Joaquin San Luis Obispo o 8 29,050 4,470 4,758 132 16,306 535 66 DECEMBER 872 20081,911Santa Barbara Santa Clara 11 55,072 7,320 5,599 344 9,402 88 54 1,675 30,590Santa Cruz spo 3,476 2,567 67 28 177 3 16 94 524Shasta Sierra 6 24,746 19,179 414 630 1,185 56 56 1,158 2,068Siskiyou Solano 3 8,491 6,629 73 689 117 5 — 337 641Sonoma Stanislaus 1 3,275 595 1,218 23 483 16 — 167 773Sutter ALI MODARRES, Ph. D. Tehama 1 2,539 1,366 156 47Edmund21 7 8 of Public 74 Affairs 860 G. “Pat” Brown Institute Trinity 1 5,072 3,024 171 59 83California26 2 122 1,585 State University, Los Angeles Tulare Tuolumne 2 3,836 3,164 2 247 23 2 11 196 191Ventura 341 1,543,580 275,987 189,231 12,709 156,080 3,851 2,362 33,191 870,169Yolo Yuba
otal100.00100.0017.88
3,368Butte
In Search of Digital Equity: Assessing the Geography of Digital Divide in California
12.26
0.82
10.11
0.25
0.15
2.15
56.37
Acknowledgment
This research and publication were made possible by a grant from the Community Partnership Committee through its Applied Research Initiative on access to telecommunications services in California’s underserved communities, with support from ZeroDivide. The Community Partnership Committee was formed by eight coalitions of 134 community-based organizations and SBC (now AT&T) to serve underserved communities throughout California after the SBC/Pacific Telesis merger in 1997.
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
2
Introduction
1
topic of digital divide, as well as recent initiatives by
allowed us to (1) gain a first glimpse into the coincidence of socioeconomic and technology indicators at the census-tract level, (2) develop a methodology that can be used at some point when the actual subscription and access data from the telecommunication providers become available, and (3) begin to explore how this approach may be utilized to develop area-based policies to diminish the emerging digital divide.
a number of public and private entities to implement
Relying on the findings of our research on Los
policies that attempt to diminish this widening gap. To
Angeles County, we engaged in this study to examine
further contextualize this issue, a number of scholars,
the phenomenon of digital divide in the state of
policymakers, and community advocates have begun
California (at the census-tract level). This analysis
to ask how new technologies might play a role in
relied on a 2007 dataset, acquired from a commercial
furthering the goals of community and economic
data provider, Claritas, Inc. Here, we examine the
development and provision of particular services.
geography of estimated access to technology, pattern,
Recent elections have also shown that politicians,
and type of usage and contextualize this information
policymakers, and policy advocates have begun to
within a sociodemographic context. The dataset
view ICTs as possible tools for advancing the cause
acquired for this research contains a larger number of
of democracy, political dialogue, and the construction
variables that include the following:
The growing prevalence of Information and Communication Technologies (ICTs) in everyday life makes it crucial that we continue to assess issues of access to these technologies, particularly as they affect lower-socioeconomic groups in urban and rural areas. Concerns about emerging ICT inequities and what their consequences may entail have resulted in the burgeoning of research and publications on the
of a more engaged civil society. This growing reliance on ICTs has infused the concern over an emerging digital divide with other issues, such as economic development, public safety, health, transportation, and other quality of life indicators. This is especially problematic because digital divide manifests not only across generational and gender differences but also
• Access to landlines and cell phones • Type of access to the Internet (e.g., dial-up, DSL, and cable modem)
• Reasons for accessing Internet (e.g., e-mail, banking, shopping, and gaming)
These
variables,
along
with
estimated
2007
sociodemographic variables, were used to create spatial,
within social and economic divides.
has illustrated that given the geography of race, ethnicity, and socioeconomic status, digital divide has a clear spatial dimension, requiring urgent attention from policymakers. Though our previous study relied on a commercially available dataset (as opposed to actual subscription data from the telecommunication companies), it Our research on Los Angeles County
• Computer ownership (desktop and laptop)
1
statistical, and visual assessment of how access to technology varies across the state and within individual counties. During the first phase of the analysis, we mapped the individual variables to create a visual assessment of access to technology and how this pattern may be related to various sociodemographic indicators. To provide a better visual tool, we developed a set of 35 maps for each county (see Appendix). These county
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
3
Introduction
maps provide a more detailed view of spatial variations
methodological groundwork for formulating the first
at the local level.
steps toward the development of equitable policies on
This initial visual assessment was followed by a
access to technology. In a few years, when the actual
statistical analysis during the second phase, which
data become available at the census-tract level (from
included the creation of two different indexes. These
the telecommunication industry), our study can be
methodologies, which will be explained, were based
repeated to create a more accurate approximation of
on our previous research on Los Angeles County. Our
digital divide in the state. This will also allow us to
techniques allowed us to analyze the spatial variation
create a monitoring process that enables the state to
of access to technology as related socioeconomic
continually fine tune its digital equity policies and
characteristics, providing some context for observed
implementation processes.
geographic differences within urban and rural areas. With a final goal of identifying priority areas in the
ORGANIZATION OF THIS REPORT
state, we concluded the study with an overall grouping
While this report provides an overall presentation
of census tracts by their socioeconomic status and
of the findings, it also includes a comprehensive atlas
access technology. These final products identify
of selected variables and other composite indicators
particular areas that should become the target of
for the state and each of the 58 counties in California.
policy intervention, which includes infrastructural
This, we hope, will provide a roadmap to an area-based
investment as well as an expansion of educational
approach to ameliorate the inequitable digital divide
activities that aim to increase the level of access to
conditions in the state and, more important, in particular
multiple ICTs. Although actual subscription data at
geographies. Since we have created over 2,000 such
the census-tract level could help us formulate policies
maps, they can only be made available electronically.
on firmer empirical grounds, we believe that the data
Those interested in this particular product should visit
from Claritas, Inc., can adequately provide us with the
our Web site at http://www.patbrowninstitute.org/.
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
4
2
Methodology and Findings
To make the content of this report accessible to a
County. The Native American population was mostly
wide range of readers, we have chosen to discuss our
in Northern and Eastern California counties, as well
methodologies within the context of our findings and in
as a few Southern California counties, including San
a manner that is, hopefully, accessible to nonacademics.
Diego. Non-Hispanic Asian Americans were mostly
Even though we briefly mention the types of statistical
residing in the Bay Area, extending to Sacramento, as
analyses employed in this research, we avoid discussing
well as Los Angeles, Orange, and San Diego counties.
such techniques in great detail. We trust that those
Geographic distribution of the Latino population is
who are familiar with the techniques will rely on their
strikingly different from others. As Figure 7 illustrates,
knowledge and those who are not can simply read the
Latinos are highly concentrated in the Central Valley
results and benefit equally from what they reveal. What
communities and in Ventura, Los Angeles, Orange, San
follows is an attempt at describing our findings in a
Bernardino, Riverside, and Imperial counties. These
comprehendible manner, answering many questions
areas represent both rural and urban settings with
that we had at the beginning of this research or that
particular socioeconomic characteristics, which will be
were developed as a result of it. We have also tried
described. However, it is noteworthy that a majority
to anticipate what other questions might exist on this
of census tracts, where a large number of Latinos are
important topic. We hope our findings shine some
found, have a population that is significantly younger
light on where we are and how we can build toward
(see Figure 8). This may point to difference in the
a more equitable future for accessing information and
family structure (as well as stage in life and family
communication technologies.
size) of Latinos. This is particularly important since these census tracts also portray lower-socioeconomic
SOCIODEMOGRAPHIC LANDSCAPE
status, as measured by the median household income
OF THE GOLDEN STATE
(see Figure 9).
Race and Ethnicity In 2007, there were slightly more than 37 million individuals living in California (see Table 1). Among
Table 1. Race and Ethnicity in California Source: Claritas Inc., Computations by A. Modarres
them, more than 42.3% were Non-Hispanic White and
RACE AND ETHNIC CATEGORIES
35.8% were Latinos. Non-Hispanic Asian and NonHispanic African Americans made up 12% and 6% of the population, respectively. However, as Figures 1 through 7 illustrate, these racial and ethnic populations were distributed unevenly across the state. For example, while the Non-Hispanic White population was mainly concentrated in coastal areas in the west and the Sierra
2007 Population 2007 Non-Hispanic White Population 2007 Non-Hispanic African American Population 2007 Non-Hispanic Native American Population 2007 Non-Hispanic Asian Population 2007 Non-Hispanic Pacific Island Population 2007 Non-Hispanic Other 2007 Non-Hispanic 2 races or more 2007 Latino Population
TOTAL
PERCENT
37,075,982 15,678,282 2,239,278 185,990 4,433,354 120,668 73,918 1,061,524 13,282,968
100.0 42.3 6.0 0.5 12.0 0.3 0.2 2.9 35.8
communities to the east, the Non-Hispanic African American population was residing mainly in South Los Angeles and East Bay communities in Alameda
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
5
Methodology and Findings
Figure 1. 2007 Non-Hispanic White Population
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
6
Methodology and Findings
Figure 2. 2007 Non-Hispanic African American Population
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
7
Methodology and Findings
Figure 3. 2007 Non-Hispanic Native American Population
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
8
Methodology and Findings
Figure 4. 2007 Non-Hispanic Asian American Population
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
9
Methodology and Findings
Figure 5. 2007 Non-Hispanic Pacific Islander Population
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
10
Methodology and Findings
Figure 6. 2007 Non-Hispanic 2 or More Races
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
11
Methodology and Findings
Figure 7. 2007 Latino Population
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
12
Methodology and Findings
Figure 8. 2007 Percent Population 18 or Older
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
13
Methodology and Findings
Figure 9. 2007 Median Household Income
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
14
Methodology and Findings
As Figure 9a suggests, the degree of demographic
also, more than half of census tracts are home to a
integration and segregation varies greatly across the
highly homogenous population. Third, given that this
state. This visualization relies on a technique known as
index simply reveals the degree of population mixing
an entropy index. This index is calculated to measure
among the seven racial/ethnic groups, less integrated
the diversity of population in any one tract, giving each
areas can appear across various socioeconomic status,
a value that ranges from zero to the logarithmic value
including smaller and more well-to-do regions. This
of the number of groups included in the analysis. In
is made apparent by Figure 9a, where it is possible to
this analysis, a total of seven groups were included,
see that in Southern California and in the Bay Area,
which means that the value could range from zero
African American and Latino neighborhoods, as well as
(made up of mostly one racial/ethnic group) to 0.845
well-to-do communities, are marked by “segregation”
(housing an equal number of people from each of the
or lack of racial and ethnic diversity. As Table 3
seven groups). To make the interpretation of these
suggests, the entropy value is negatively correlated
values more readily understood, the values across
with the presence of Latinos. This is similarly true
all census tracts in the state were grouped into four
for the Non-Hispanic White population, but with a
categories: (1) integrated (those with values larger
lesser impact. These results suggest that demographic
than one standard deviation from the mean of entropy
integration levels increase as the number of African
index), (2) moderately integrated (those with values
American, Asian, Pacific Islander, and “Other” racial
that fall within one standard deviation above the mean,
and ethnic groups grows in a census tract. However,
inclusive of the mean value), (3) moderately segregated
as Latinos and Non-Hispanic Whites grow in numbers,
(those within one standard deviation below the mean),
the likelihood of integration or diversity of racial and
and (4) segregated (those smaller than one standard
ethnic groups diminishes. In other words, Latino and
deviation below the mean). Note that, here, the words
Non-Hispanic White neighborhoods are more likely
integration and segregation simply suggest the degree of
to lack a significant presence of other racial and ethnic
demographic diversity in any one census tract. Figure
groups. Given that Latino and Non-Hispanic White
9a provides a snapshot of how these values vary across
percentages are negatively correlated at -0.79 (Pearson
the state. Combining this figure with Table 2 allows us
correlation value, significant at 0.00), it is clear that
to see what this indexing reveals. First, a number of
these two groups are also less likely to co-reside in a
rural counties with a significant concentration of one
census tract.
population group score very highly on segregation. These
include
Amador,
Calaveras,
El
Dorado,
Education
Mariposa, Nevada, Plumas, Shasta, Sierra, Siskiyou,
Slightly more than half of all Californians have
Trinity, and Tuolumne counties. More than 50% of
an educational attainment level that ranges from a
census tracts in these counties were categorized as
high school diploma to an associate degree (see Table
demographically segregated (or consisting mostly
4). Those with a bachelor’s degree or better make up
of one racial/ethnic group). Second, rural areas are
about a quarter of all Californians. This educational
not the only ones with such status. In Marin County,
attainment level, if it were equally distributed among
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
15
Methodology and Findings
Table 2. Integration and Segregation Patterns by California Counties (Values Represent the number of Census Tracts in Each Category) Source: Claritas Inc., Computations by A. Modarres 2007 INTEGRATION STATUS
2007 INTEGRATION STATUS Integrated
Moderately Integrated
Moderately Segregated
Total No. of Segregated Census Tracts
Alameda County Alpine County Amador County Butte County Calaveras County Colusa County Contra Costa County
150 0 1 2 0 0 55
115 0 0 11 0 2 61
51 1 0 14 3 3 39
5 0 6 15 4 0 13
321 1 7 42 7 5 168
Del Norte County El Dorado County Fresno County Glenn County Humboldt County Imperial County Inyo County Kern County Kings County Lake County Lassen County Los Angeles County Madera County Marin County Mariposa County Mendocino County Merced County Modoc County Mono County Monterey County Napa County Nevada County
0 0 31 0 0 0 0 8 3 0 1 282 1 2 0 0 1 0 0 9 3 0
3 4 63 1 3 6 1 51 13 1 1 696 3 5 0 5 30 0 0 22 3 0
3 8 50 5 12 11 3 47 9 10 2 594 8 15 1 10 14 3 2 25 17 1
0 24 14 0 12 12 3 34 0 1 3 476 7 29 3 4 2 1 0 27 4 17
6 36 158 6 27 29 7 140 25 12 7 2048 19 51 4 19 47 4 2 83 27 18
Note: 12 Census tracts had no values.
Integrated
Moderately Integrated
Moderately Segregated
61 0 0 51 131 0 46 100 34 56 1 29 1 63 0 0 0 0 53 1 4 1 0 0 0 0 6 4 2
237 4 0 148 70 2 98 217 95 37 5 73 31 188 7 2 0 0 17 20 49 13 1 0 20 1 39 27 7
183 25 1 94 69 5 80 191 38 27 25 40 36 82 26 14 0 6 10 46 33 4 6 1 44 0 83 6 1
96 22 5 49 9 1 19 97 9 1 13 12 18 8 19 17 1 8 0 19 3 0 4 3 12 9 27 0 2
577 51 6 342 279 8 243 605 176 121 44 154 86 341 52 33 1 14 80 86 89 18 11 4 76 10 155 37 12
Total (State of California) 1193
2508
2137
1199
7037
Orange County Placer County Plumas County Riverside County Sacramento County San Benito County San Bernardino County San Diego County San Francisco County San Joaquin County San Luis Obispo County San Mateo County Santa Barbara County Santa Clara County Santa Cruz County Shasta County Sierra County Siskiyou County Solano County Sonoma County Stanislaus County Sutter County Tehama County Trinity County Tulare County Tuolumne County Ventura County Yolo County Yuba County
Total No. of Segregated Census Tracts
Table 3. Correlation between entropy value and percentage of various racial and ethnic groups Source: Claritas Inc., Computations by A. Modarres
Entropy
Pearson Correlation Sig. (2-tailed) N
Entropy
Percent Latino
1.00 — 7037
-0.225 0.00 7037
Percent Non-Hispanic White
-0.144 0.00 7037
Percent Non-Hispanic African American
0.293 0.00 7037
Percent Non-Hispanic Native Americans
0.017 0.16 7037
Percent Non-Hispanic Asian
0.404 0.00 7037
Percent Non-Hispanic Pacific Islander
0.341 0.00 7037
Percent other racial & ethnic groups
0.465 0.00 7037
Bold values indicate that correlation is significant at the 0.01 level (2-tailed).
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
16
Methodology and Findings
Figure 9a. 2007 Diversity Status
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
17
Methodology and Findings
all social and demographic groups (and geographies),
Table 4. 2007 Educational Attainment, Age 25+ Source: Claritas Inc., Computations by A. Modarres
Less than 9th Grade Some High School, No Diploma High School Graduate (Includes Equivalency) Some College, No Degree Associate Degree Bachelor’s Degree Master’s Degree Professional School Degree Doctorate Degree Total
POPULATION
PERCENT
2,736,903 2,780,088 4,757,863 5,410,587 1,682,009 3,970,877 1,394,201 532,023
11.63 11.81 20.22 22.99 7.15 16.87 5.92 2.26
266,820
1.13
23,531,371
100.00
could suggest a significant achievement for the state. However, as Figures 10 through 17 and Table 5 suggest, this educational attainment level is far from equitable. As Table 5 illustrates, the percentage of Latinos in a census tract is highly and positively correlated with lower educational attainment (and, negatively, with higher educational attainment). This phenomenon is reversed in the case of Non-Hispanic White population. While African Americans display similar conditions as Latinos, the correlation values are smaller (possibly
Table 5. Correlation between geographic distribution of racial and ethnic groups and educational attainment (2007 Population Age 25+)Source: Claritas Inc., Computations by A. Modarres Less than 9th Grade
Some High School No Diploma
High School Graduate (Incl. Equivalency)
Some College No Degree
Associate Degree
Bachelor’s Degree
Master’s Degree
Professional School Degree
Doctorate Degree
Percent Latino Pearson Correlation Sig. (2-tailed) N
0.769 0.000 7037
0.558 0.000 7037
0.071 0.000 7037
-0.255 0.000 7037
-0.334 0.000 7037
-0.552 0.000 7037
-0.552 0.000 7037
-0.459 0.000 7037
-0.408 0.000 7037
Percent Non-Hispanic White Pearson Correlation Sig. (2-tailed) N
-0.670 0.000 7037
-0.503 0.000 7037
-0.050 0.000 7037
0.248 0.000 7037
0.259 0.000 7037
0.436 0.000 7037
0.450 0.000 7037
0.441 0.000 7037
0.340 0.000 7037
Percent Non-Hispanic African American Pearson Correlation Sig. (2-tailed) N
0.027 0.023 7037
0.154 0.000 7037
0.051 0.000 7037
0.004 0.708 7037
-0.040 0.001 7037
-0.157 0.000 7037
-0.156 0.000 7037
-0.158 0.000 7037
-0.134 0.000 7037
Percent Non-Hispanic Native Americans Pearson Correlation -0.0520 Sig. (2-tailed) 0.0000 N 7037
0.0172 0.1482 7037
0.0394 0.0010 7037
-0.0033 0.7801 7037
-0.0344 0.0039 7037
-0.1135 0.0000 7037
-0.1101 0.0000 7037
-0.0952 0.0000 7037
-0.0815 0.0000 7037
Percent Non-Hispanic Asian Pearson Correlation Sig. (2-tailed) N
-0.083 0.000 7037
-0.135 0.000 7037
-0.082 0.000 7037
-0.036 0.003 7037
0.116 0.000 7037
0.269 0.000 7037
0.248 0.000 7037
0.104 0.000 7037
0.184 0.000 7037
Percent Non-Hispanic Pacific Islander Pearson Correlation Sig. (2-tailed) N
0.0007 0.9554 7037
0.0562 0.0000 7037
0.0592 0.0000 7037
0.0231 0.0527 7037
0.0120 0.3130 7037
-0.0777 0.0000 7037
-0.1150 0.0000 7037
-0.1241 0.0000 7037
-0.0998 0.0000 7037
Percent other racial and ethnic groups Pearson Correlation Sig. (2-tailed) N
-0.352 0.000 7037
-0.226 0.000 7037
0.030 0.013 7037
0.150 0.000 7037
0.195 0.000 7037
0.205 0.000 7037
0.169 0.000 7037
0.101 0.000 7037
0.123 0.000 7037
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
18
Methodology and Findings
Figure 10. 2007 Percent Population 25 Years and Older with Less than 9th Grade Education
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
19
Methodology and Findings
Figure 11. 2007 Percent Population 25 Years and Older with some High School Education
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
20
Methodology and Findings
Figure 12. 2007 Percent Population 25 Years and Older with High School Education
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
21
Methodology and Findings
Figure 13. 2007 Percent Population 25 Years and Older with some College Education
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
22
Methodology and Findings
Figure 14. 2007 Percent Population 25 Years and Older with Associate Degree
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
23
Methodology and Findings
Figure 15. 2007 Percent Population 25 Years and Older with Bachelor’s Degree
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
24
Methodology and Findings
Figure 16. 2007 Percent Population 25 Years and Older with Master’s Degree
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
25
Methodology and Findings
Figure 17. 2007 Percent Population 25 Years and Older with Doctoral Degree
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
26
Methodology and Findings
caused by having a smaller population). For Asian
Occupation
population, a higher level of concentration is positively
Given the educational achievement and median
correlated with the presence of individuals with a
household income patterns, it should not come
bachelor’s, master’s, or doctoral degree. Due to their
as a surprise that the geography of occupation in
smaller population, Native American and Pacific Islander
California reveals a significant level of divide, as
population do not show such a correlation pattern.
well. As Figure 18 illustrates, counties in Central
The correlation pattern displayed in Table 5 has a
California, from Kern to Glenn, a few coastal
particular geographic consequence (Figures 10 through
counties, and the Imperial Valley house the largest
17). Central California, where Latinos are highly
concentration of those employed in farming, fishing,
concentrated, has some of the largest concentrations of
and forestry occupations. These are also areas where
people with less than a ninth-grade education (compare
Latinos are highly concentrated, median household
Figures 7 and 10). Conversely, areas with the highest
incomes are relatively low, and very few census
presence of people with college degrees are where
tracts portray a high level of educational attainment.
Non-Hispanic Whites and people with high median
While construction, but more obviously production,
household incomes can be found (compare Figures 9,
jobs seem to be concentrated in low-income urban
15, 16, and 17).
neighborhoods (see Figures 19 and 20), management, business, and financial occupations coincide closely with high-income neighborhoods, where minorities
Table 6. 2007 Employed Civilian Population, Age 16+ by Occupation Source: Claritas Inc., Computations by A. Modarres POPULATION
PERCENT
Management, Business and Financial Operations Occupations
2,383,214
14.60
Professional and Related Occupations
3,470,745
21.26
Service Occupations
2,406,519
14.74
Sales and Office Occupations
4,370,751
26.77
224,977
1.38
Farming, Fishing, and Forestry Occupations Construction, Extraction and Maintenance Occupations
1,384,034
8.48
Production Transportation and Material Moving Occupations 2,083,955
12.77
Total
16,324,195
are minimally represented (see Figure 22). This pattern is repeated in the case of professional jobs (see Figure 23) but reversed for service-oriented occupations (see Figure 24).
100.00
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
27
Methodology and Findings
Figure 18. 2007 Percent Population Working in Farming, Fishing, and Forestry Occupations
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
28
Methodology and Findings
Figure 19. 2007 Percent Population Working in Construction, Extraction, and Maintenance Occupations
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
29
Methodology and Findings
Figure 20. 2007 Percent Population Working in Production Transport and Material Moving Occupations
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
30
Methodology and Findings
Figure 21. 2007 Percent Population Working in Sales and Office Occupations
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
31
Methodology and Findings
Figure 22. 2007 Percent Population Working in Management, Business, and Financial Occupations
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
32
Methodology and Findings
Figure 23. 2007 Percent Population Working in Professional and Related Occupations
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
33
Methodology and Findings
Figure 24. 2007 Percent Population Working in Service Occupations
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
34
Methodology and Findings
is a common methodology for constructing and
Socioeconomic Index As this and other analyses of race and ethnicity
mapping such indexes in geographic analyses. For
in California have suggested, neighborhood-level
California, I have conducted a similar analysis, using
socioeconomic status appears to be highly correlated
educational attainment, occupation, and median
with demographic conditions. This spatial covariation
household income. The result of this analysis is
does suggest that creating composite socioeconomic
displayed in Table 7. This analysis was able to create 3
indexes not only is possible but also, for the purpose of
composite variables from the original 17 socioeconomic
studies such as this, will allow us to examine the social
variables. The three indexes, collectively, explain
and economic structure of a phenomenon (e.g., digital
79.6% of the variation in the original data. As the
divide) in a more substantial and comprehensive
loading patterns suggest, the first index is capable
manner. The use of factor analysis or factorial ecology
of identifying those areas with a population that has
Table 7. Factor Analysis Source: Claritas Inc., Computations by A. Modarres SOCIOECONOMIC INDEXES
VARIABLES
LOWER EDUCATIONAL ATTAINMENT WITH SALES, SERVICE, AND BLUE-COLLAR OCCUPATIONS
Age 25+: < 9th Grade Age 25+: Some High School, No Diploma Age 25+: High School Graduate (Includes Equivalency) Age 25+: Some College, No Degree Age 25+: Associate Degree Age 25+: Bachelor’s Degree Age 25+: Master’s Degree Age 25+: Professional School Degree Age 25+: Doctorate Degree Age 16+: Management, Business and Financial Operations Occupations Age 16+: Professional and Related Occupations Age 16+: Service Occupations Age 16+: Sales and Office Occupations Age 16+: Farming, Fishing, and Forestry Occupations Age 16+: Construction, Extraction and Maintenance Occupations Age 16+: Production Transportation and Material Moving Occupations Median Household Income Initial Eigenvalue Percent of Variance Rotation Sums of Squared Loadings Percent of Variance
— 0.577 0.895 0.891 0.801 0.305 — — — 0.397 0.374 0.807 0.834 — 0.853 0.658 — 6.9 40.7 5.5 32.6
EDUCATED WITH PROFESSIONAL OCCUPATIONS
LOWER EDUCATIONAL ACHIEVEMENT AND LOW INCOME POPULATION
— — — 0.313 0.429 0.879 0.934 0.848 0.809 0.834 0.863 — 0.438 — — — 0.599 5.3 31.1 5.5 32.3
0.888 0.616 — — — — — — — — — 0.330 — 0.734 — 0.534 -0.331 1.3 7.8 2.5 14.6
Extraction Method: Principal Component Analysis. Rotation Method: Equamax with Kaiser Normalization. Rotation converged in 5 iterations.
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
35
Methodology and Findings
Figure 25. 2007 SES I Lower Education with Sales, Service and Blue-Collar Occupations
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
36
Methodology and Findings
Figure 26. 2007 SES II Educated with Professional Occupations
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
37
Methodology and Findings
Figure 27. 2007 SES III Low Education, Low Income Population
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
38
Methodology and Findings
achieved a mid-range level of educational attainment,
ACCESS TO INFORMATION AND
and it works in sales, service, construction, extraction,
COMMUNICATION TECHNOLOGIES
maintenance, production transportation, or material
To assess the geography of access to ICTs, I relied
moving occupations. The second index is associated
on the information provided by Claritas, Inc, for 2007.
with areas that house a highly-educated population
This database provided the number of households per
with occupations such as management, business and
census tracts having access to a particular technology
financial operations, and related professions. The
or using a selected application. I selected 28 variables to
third index identifies typically rural areas with lowest
conduct my analyses. These were as follows:
educational attainment among a population who is mainly working in farming, fishing, and forestry
1. Owning a desktop
occupations and achieves an overall low median
2. Owning a laptop
household income. Figures 25 through 27 illustrate how these three
These two variables provide an indication for
indexes map across the state. The first index (SES I)
the basic level of access to computer technology.
attains higher values in urban and rural communities, where sales and service occupations are major sources
3. Owning one cell phone
of employment. Geographically, these are found in
4. Owning two cell phones
a number of eastern counties, from Riverside to the
5. Owning more than two cell phones
Sierras, and in some of the central and coastal counties.
6. Owning more than three cell phones
Most significant is the fact that in central California
7. Owning more than four cell phones
counties, where farming exceeds other occupations,
8. Owning more than five cell phones
this index scores low. The second index (SES II) clearly identifies high-
Based on our previous analysis of Los Angeles County,
income areas of Southern California, the Bay Area,
we were keenly aware of the importance of cell phones
and other coastal communities. It is striking that the
as an important piece of communication technology.
third index (SES III) clearly highlights the farming
These devices can easily replace traditional (landline)
communities in central California and elsewhere
phones and provide access to the Internet and other
(despite
the
Web-based information and services. In our focus
suitability of this index for evaluating how equitably
group meetings, the various participants helped
particular services and public goods are distributed.
increase awareness about the functional aspects of
For the purpose of this study, I rely on this index
cell phones across socioeconomic sectors and, more
to assess the degree of our regional equity in the
pronouncedly, among the younger population. For
distribution of ICTs.
that reason, I made sure that these variables were
its
lower
eigenvalue),
suggesting
included in the database. Unlike other variables, having information on the number of cell phones available to a household allows us both to assess
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
39
Methodology and Findings
the level of access to this technology to all household
23. Use Internet to visit/publish to an online community
members and to determine the socioeconomic status
24. Use Internet to download/purchase games
associated with having multiple phones. Though
25. Use Internet to play games alone
family plans, offered by many providers, have reduced
26. Use Internet to play multiplayer games
the cost associated with having multiple cell phones,
27. Use Internet to listen to streaming audio
it is nonetheless a costly affair that may distinguish
28. Use Internet to watch streaming video
between lower-socioeconomic groups and others.
29. Use Internet to watch TV
9. Having dial-up Internet
These 12 variables provide a reasonable indication
10. Having cable Internet
of how Internet is used based on content and interest
11. Having DSL
of the user. With the exception of a few of them, the significant majority of these uses require high-speed
This information allows us to see how households
connection and access to well-equipped computers.
access the Internet and whether the geographic patterns of these variables will reveal the infrastructural backbone of access to the broadband.
These 29 variables were mapped to provide an overall assessment of access to technology. This was followed with detailed analysis of how these variables,
12. Having zero wire line
individually and cumulatively, relate to socioeconomic and demographic indicators. The following provides an
This singular variable provides us with an indication
overview of these findings. Readers may wish to review
of how badly some areas may remain disconnected.
(in addition to the maps included here) the Appendix to this report, which contains 35 selected maps for each
13. Use Internet to send and receive e-mail
county in the state.
14. Use Internet for text-based chat or instant messaging 15. Use Internet for voice-based chat or instant messaging 16. Use Internet to send pictures 17. Use Internet to send video e-mails
Spatial Patterns of Access to Technology Figures 28 and 29 reveal patterns of desktop and laptop ownership by census tract. As expected, while desktops are more abundantly available throughout the
These five variables allow us to examine how Internet
state, achieving rates above 90% in many neighborhoods
is used for communication purposes.
and sliding to rates below 50% in low-income areas, nearly two thirds of all census tracts have household-
18. Use Internet for online banking
laptop ownership rates below 25% and, more important,
19. Use Internet for shopping
400 census tracts have laptop ownership rates below
20. Use Internet to search online Yellow Pages
10% (for households). These tracts are mostly located
21. Use Internet to download/purchase music
in low-income urban and rural areas. The significant
22. Use Internet to download video
difference in desktop and laptop ownership rates by
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
40
Methodology and Findings
Figure 28. 2007 Percent of Households Own PC
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
41
Methodology and Findings
Figure 29. 2007 Percent Households with Laptops
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
42
Methodology and Findings
Figure 30. 2007 Percent Households with 1 Cell Phone
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
43
Methodology and Findings
Figure 31. 2007 Percent Households with 2 Cell Phone
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
44
Methodology and Findings
Figure 32. 2007 Percent Households with 2+ Cell Phone
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
45
Methodology and Findings
Figure 33. 2007 Percent Households with 3+ Cell Phone
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
46
Methodology and Findings
Figure 34. 2007 Percent Households with 4+ Cell Phone
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
47
Methodology and Findings
Figure 35. 2007 Percent Households with 5+ Cell Phone
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
48
Methodology and Findings
census tract suggests that access to mobile computing is
perspective, this may require a reassessment of pricing
more clearly affected by socioeconomic status.
plans or subscription fees.
The observed socioeconomic pattern in desktop and
Beyond PCs and cell phones, access to the Internet,
laptop ownership is equally manifested in patterns of
its pattern of usage, and the level of access to various
access to cell phones. Figures 30 through 35 illustrate
Internet contents provides a more complete picture of
how the number of cell phones per household varies
the nature of digital divide and how it may be related
across the state. While having one cell phone seems
to the underlying geography of race, ethnicity, and
to be prevalent in low- income areas, having multiple
socioeconomic status. Furthermore, an evaluation
cell phones per household occurs in areas with higher-
of these patterns allows us to develop a better
socioeconomic status. There are at least three lessons to
understanding of how market forces may be shaping
be learned immediately. First, in low- income areas, the
the geography of access to ICTs.
use of cell phones is becoming more common, perhaps
Figures 36 through 38 illustrate how Californians
replacing the traditional landline phones. Second, while
are connected to the Internet. While dial-up rates
one cell phone per household may meet the minimum
rarely exceed 25% in any one census tract, this type of
need of a household for communication purposes,
connection remains a viable option in more rural areas
having access to multiple phones, which improves the
with minimal availability of cable and DSL, or where the
communication ability of multiple household members,
price for these faster modes of connection is prohibitive.
is highly related to the socioeconomic status. Third,
Northeastern and western sections of Santa Barbara and
the greatest shift in access to multiple phones occurs
southwestern areas of Colusa and Butte counties are
between households having 2+ or 3+ cell phones (see
among the highest users of dial-up services. However,
Figures 32 and 33). While at 2+ level, over 1,400 census
as Figure 36 indicates, in less populated areas of King,
tracts had 60% or better rates, at 3+ level, none achieved
Fresno, and other counties in Central California and the
this rate. In fact, only slightly over 200 tracts achieved
Sierras, many households rely on this service to connect
rates of about 30%. At 4+ or 5+ number of cell phones
to the Internet also.
per household, census tract level rates declined to 15% and 5%, respectively, for the highest categories.
Contrary to the observed pattern of dial-up usage, cable appears to be an important choice for less
This pattern of access to cell phones is of particular
economically strapped urban neighborhoods. This is
importance to those concerned with digital divide.
similarly true for DSL services. This suggests that location
Clearly, as mobile devices supplement or replace
is not only a good predictor of one’s socioeconomic
computers for accessing the Internet and the information
status but also access to the infrastructural backbone
it provides, as well as for engaging in multiple modes
and service nodes within our society. To illustrate this
of communication, such as sending e-mails and text
point, Figure 38a was constructed to map the level of
messages, it becomes crucial that policies regarding
Internet disconnectivity by geographic location. For
the expansion of broadband and access to ICTs include
each census tract, I have calculated the percentage of
full consideration of how we may increase access to
households that do not have access to the Internet,
cell phones and smart phones. From a private sector
whether by dial-up, cable, or DSL. Assuming that, given
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
49
Methodology and Findings
Figure 36. 2007 SES II Educated with Professional Occupations
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
50
Methodology and Findings
Figure 37. 2007 Percent Households with Cable Internet
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
51
Methodology and Findings
Figure 38. 2007 Percent Households with DSL Internet
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
52
Methodology and Findings
Figure 38a. 2007 Percent Households without Dial-up, DSL, or Cable Internet
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
53
Methodology and Findings
their socioeconomic status, it is improbable that these
that was slightly more than half of the average for all
households have their own sophisticated technologies,
census tracts in the state. By comparison, areas that
such as a personal dish or Internet-ready cell phones
did not meet this criterion housed a population that
that can act as a modem for a computer. We can also
was 42.5% Non-Hispanic White, 5.9% Non-Hispanic
assume that the calculated figures are a reasonable
African American, 12% Non-Hispanic Asian American,
approximation of not having access to the Internet.
and 35.7% Latino, earning median household incomes
Figure 38a clearly identifies portions of the state that
that were on the average larger than those in all tracts
are being left behind in the digital age. These locations
in the state.
are mostly in rural northern California, eastern portions
This unequal pattern of access to the Internet is
of Imperial, Riverside, and San Bernardino counties, as
further exacerbated by the inadequacy of available
well as isolated tracts from Inyo and King to Tuolumne,
wirelines in low-income neighborhoods, located either
Mariposa, and Amador counties. Overall, there are 252
in urban areas or rural communities in central and
census tracts (or 3.6% of all tracts) where 60% or more of
northern California (see Figure 39). Cumulatively,
the resident households do not have access to dial-up,
this translates to a significant digital divide in the
cable, or DSL services (see Table 8). In 2007, these tracts
world of ICTs, creating obstacles to economic and
housed over 990,000 individuals who were racially and
community development efforts. Among the 252 tracts,
ethnically identified as follows: 36.4% Non-Hispanic
where more than 60% of the households did not have
White, 9.8% Non-Hispanic African American, 1.3%
access to dial-up, cable, or DSL, 114 were estimated to
Non-Hispanic Native American, 8.9% Non-Hispanic
have zero wirelines for at least 20% of their resident
Asian, and 40.9% Latino. On average, households
households. These tracts are located across multiple
in these tracts achieved a median household income
counties, including Alameda, Butte, Contra Costa,
Table 8. Demographic Indicators for the Prevalence of Access to the Internet Source: Claritas Inc., Computations by A. Modarres PERCENT HOUSEHOLDS WITHOUT DIAL-UP, CABLE, AND DELL DOES NOT EXCEED 60% NO. OF TRACTS
ALL CENSUS TRACTS
EXCEEDS 60%
SUM
MEAN
NO. OF TRACTS
TOTAL
SUM
MEAN
NO. OF TRACTS
SUM
MEAN
2007 Population
6774
36,045,808
5,321
252
990,647
3,931
7026
37,036,455
5,271
2007 Non-Hispanic White Population
6774
15,304,796
2,259
252
360,463
1,430
7026
15,665,259
2,230
2007 Non-Hispanic African American Population
6774
2,132,614
315
252
97,126
385
7026
2,229,740
317
2007 Non-Hispanic Native American Population
6774
172,883
26
252
12,742
51
7026
185,625
26
2007 Non-Hispanic Asian Population
6774
4,342,967
641
252
88,722
352
7026
4,431,689
631
2007 Non-Hispanic Pacific Island Population
6774
118,183
17
252
2,357
9
7026
120,540
17
2007 Non-Hispanic Other
6774
72,094
11
252
1,473
6
7026
73,567
10
2007 Non-Hispanic 2 races or more
6774
1,038,753
153
252
22,467
89
7026
1,061,220
151
2007 Median Household Income
6774
418,760,443 61,819
252
8,172,241 32,430
7026
426,932,684
60,765
2007 Latino Population
6774
7026
13,268,815
1,889
12,863,518
1,899
252
405,297
1,608
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
54
Methodology and Findings
Figure 39. 2007 Percent Households without Wireline
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
55
Methodology and Findings
Figure 40. 2007 Percent Households use Internet to Send and Receive Email
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
56
Methodology and Findings
Figure 41. 2007 Percent Households use Internet for Chat/IM
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
57
Methodology and Findings
Figure 42. 2007 Percent Households use Internet for Chat/IM with Voice
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
58
Methodology and Findings
Figure 43. 2007 Percent Households use Internet to Send Pictures
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
59
Methodology and Findings
Fresno, Imperial, Los Angeles, Mendocino, Monterey,
e-mailing, text-based instant messaging, sending
Sacramento, San Bernardino, San Diego, San Francisco,
pictures,
San Joaquin, San Luis Obispo, Shasta, Solano, and
Yellow Pages, playing games (alone), and listening
Stanislaus. However, 73 of them are to be found in
to streaming audio are among the most popular
Los Angeles County. To confirm the socioeconomic
uses of the Internet, attracting as many as 40% or
status of these tracts, it is sufficient to know that only
more of the households in some census tracts. In
31 of these tracts reported a median household income
fact, e-mailing and shopping are among the top two
exceeding $25,000 and none exceeded $45,000.
activities in this regard. Spatial patterns of usage
online
banking,
shopping,
searching
For those with access to the Internet, content
across the high-ranking activities remain similar,
and type of usage are important considerations. As
with urban and suburban middle-class communities
mentioned earlier, 17 variables were included in this
and high-income areas dominating the landscape of
analysis to assess how households use the Internet.
usage. However, two particular applications, playing
Figures 40 through 56 provide maps of these indicators
games (alone) and listening to streaming audio,
for the state. Overall, it is clear that functions such as
display
a
geographically
concentrated
pattern.
Table 9. Technology Indicators (number and percent of Households using a particuar service or utility) Source: Claritas Inc., Computations by A. Modarres
Household Base for Technology Variables Has Internet–Cable Has Internet–DSL Has Internet–Dial Up Wireline Zero Chat/Instant Messaging with Text Chat/Instant Messaging with Voice Download Video Content Download/Purchase Games Download/Purchase Music Listen to Streaming Audio Online Banking Play Games Alone Play Multi-Player Games Send Pictures Send Videos Send Email Online Shopping Visit/Publish to Online Community Watch Internet TV Watch Streaming Video Online Yellow Pages
Max. No. of Households in any one Census Tract
Total No. of Households
Percent of Households
Average No. Households
Standard Deviation
14,025 5,074 4,073 2,189 1,518 4,807 1,801 997 1,558 3,414 4,302 7,610 5,717 2,614 7,211 2,024 9,570 8,396 2,730 1,285 4,296 6,778
12,461,651 3,250,859 2,953,657 1,777,351 1,446,597 3,638,645 1,345,462 741,418 1,094,786 2,082,354 2,884,833 5,180,702 4,134,046 1,801,332 4,749,423 1,366,233 7,014,779 5,655,815 2,148,152 961,679 2,967,346 4,533,519
100.0 26.1 23.7 14.3 11.6 29.2 10.8 5.9 8.8 16.7 23.1 41.6 33.2 14.5 38.1 11.0 56.3 45.4 17.2 7.7 23.8 36.4
1768 461 419 252 205 516 191 105 155 295 409 735 586 256 674 194 995 802 305 136 421 643
933 309 261 145 144 324 121 72 102 212 280 482 364 165 445 128 605 521 208 91 286 421
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
60
Methodology and Findings
Figure 44. 2007 Percent Households use Internet to Send Video Email
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
61
Methodology and Findings
Figure 45. 2007 Percent Households use Internet for Online Banking
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
62
Methodology and Findings
Figure 46. 2007 Percent Households use Internet to Shop
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
63
Methodology and Findings
Figure 47. 2007 Percent Households use Internet to Search Online Yellow Pages
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
64
Methodology and Findings
Figure 48. 2007 Percent Households use Internet to Download/Purchase Music
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
65
Methodology and Findings
Figure 49. 2007 Percent Households use Internet to Download Video
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
66
Methodology and Findings
Figure 50. 2007 Percent Households use Internet to Visit/Publish to Online Community
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
67
Methodology and Findings
As Figures 52 and 54 illustrate, only a small set of
tracts) counties. These tracts house 2.7 million people,
census tracts in Southern California and the Bay area
consisting mostly of 1.3 million Non-Hispanic Whites,
has a large proportion of its households engaged
0.5 million Non-Hispanic Asian Americans, and 0.6
in these activities (above 50% for games and 40%
million Latinos. The median household incomes in
for streaming audio). At rates below 40% and 30%,
these 427 census tracts are mostly high, with the highest
respectively, the geography of usage more closely
one exceeding $240,000.
resembles other high-ranking categories. Among
the
emerging
new
Internet
The enabling role of high-speed connection in utilities,
promoting the use of particular Internet application and
instant messaging with voice, sending videos by e-
content is also visible for activities such as downloading
mail, downloading/purchasing music, downloading
large video files or watching Internet TV. As Table 9
video, visiting and publishing to online communities,
illustrates, only 7.7% of all households in California
downloading and purchasing games, watching streaming
use their Internet connection to watch TV. At 5.9%,
video, engaging in multiplayer games, and watching
downloading video contents from the Internet occurs
Internet TV are included in this study, not only to
at even a lower rate. As speed and bandwidth improve,
assess their usage patterns but also to further illustrate
these functions will attract a higher number of users.
the degree to which socioeconomic status and the
However, if this improvement occurs simply according
speed of connectivity affect how advanced applications
to the existing patterns of access and usage, low-income
are utilized. For example, although voice-based instant
urban and rural communities will not be able to benefit
messaging and sending videos by e-mail act as more
from these services.
sophisticated modes of communication, they are highly reliant on uninterrupted high-speed connection.
Statistical Analysis of the Geography
However, downloading and purchasing music requires
of Access to Technology
a particular socioeconomic status and consumption
In the previous section, the overall patterns of access
habit, in addition to high-speed connectivity. This
to ICTs were discussed, suggesting how these patterns
difference is illustrated by a comparison of Figures
may reveal particular rural-urban, socioeconomic, and
44 and 48. While a large number of tracts fall in the
racial/ethnic digital divides. In this section, I examine
10% to 15% category (the middle range) for sending
these patterns statistically and attempt to develop a
video e-mails, downloading music seems to narrow
better understanding of how various variables may
to smaller set of tracts, starting with the middle-range
explain the observed patterns of access to various
category. In fact, tracts where 25% or more of the
information and communication technologies. This
households download/purchase music are mostly
analysis is done in two phases. First, I use the results
found in well-to-do urban and suburban sections of the
of a correlation analysis to establish the basic level of
Bay Area and Southern California. There are only 427
relationship between various variables. The second
such tracts (about 6% of all tracts), located mostly in
section builds on these findings to arrive at more
Los Angeles (79 tracts), San Francisco (64 tracts), Santa
refined conclusions, relying on various multivariate
Clara (56 tracts), Orange (37 tracts), and Riverside (34
statistical techniques.
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
68
Methodology and Findings
Figure 51. 2007 Percent Households use Internet to Download/Purchase Games
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
69
Methodology and Findings
Figure 52. 2007 Percent Households use Internet to Play Games Alone
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
70
Methodology and Findings
Figure 53. 2007 Percent Households use Internet to Play Multi-player Games
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
71
Methodology and Findings
Figure 54. 2007 Percent Households use Internet to Listen to Streaming Audio
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
72
Methodology and Findings
As Table 102 indicates, SES I has relatively
low correlation values across all variables. As implied previously, this variable has lower ability to provide a clear indication of socioeconomic divide in the state. This is partially caused by the fact that it includes service occupations, which could include a range of income potentials and spatial distribution. As such, while the index may be useful for understanding how socioeconomic variables are structurally related, it is less useful for understanding the emergent fractures within our social geography. Contrary to this variable, SES II and SES III are able to more clearly provide the observed social and economic differences. As indicated before, SES II is capable of identifying areas with a significant concentration of highincome population engaged in professional occupations, and SES III can identify rural areas with a large number of people employed in farming, fishing, and forestry occupations. Looking across Table 10, it is clear that SES II is positively correlated with desktop and laptop ownership (slightly higher for the latter), having more than two cell phones, access to the Internet via cable and DSL, using various modes of instant messaging, e-mailing, and all types of Internet usage. It is interesting that ownership of one cell phone is correlated negatively with this variable. This suggests that in areas with highsocioeconomic status, households are not likely to house a population that relies on a single cell phone and each member will have his or her own cell phone. Surprisingly, this variable (having access to only one cell phone) appears to be capable of distinguishing areas with a large number of lowersocioeconomic households from others.
Internet usage variables with the highest correlation values with SES II are online shopping, e-mailing, online banking, and sending pictures. This suggests a high reliance on the Internet for communication and economic transactions among the well-to-do households in California. As Table 10 indicates, these households live in areas with a particular racial and ethnic structure. The high positive correlation between SES II and NonHispanic White, a smaller, but positive, correlation with the Non-Hispanic Asian population, and a negative correlation with all other racial and ethnic groups confirms that the emergent digital divide in California may have a particular racial/ethnic dimension, as well. This is further illustrated by SES III, which appears to be the opposite of SES II in its patterns of correlation with technology variables and racial/ethnic indicators. In fact, most technology variables are negatively correlated with this index, while concentration of Latinos is positively correlated. Since SES III increases in value where rural populations are found, the results can clearly suggest that a ruralurban divide is exacerbated by social class and racial/ethnic differences. SES III correlates positively with having one cell phone, zero wirleines, and a Latino population, providing a particular picture of digital divide in California. As the Latino population in a rural census tract increases, access to cable and DSL, as well as any chance of using the broadband for any Internet activity, diminishes. Households in these areas are more likely to rely on a single cell phone, having no access or no need for wirelines (note the positive correlation between having one cell phone and zero wirelines). Once again,
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
73
Methodology and Findings
Figure 55. 2007 Percent Households use Internet to Watch Streaming Video
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
74
Methodology and Findings
Figure 56. 2007 Percent Households use Internet to Watch Internet TV
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
75
Methodology and Findings
0.13
0.17
Banking
-0.02
Streaming Audio
% IM Voice
0.13
Music
% IM Text
0.07
Games
% Zero Wirline
0.05
Video
% Dial-up Internet
% with 2+ cell phones
0.06
% DSL Internet
-0.06
% with 2 cell phones
% with 1 cell phone
-0.07
% Internet Use % Cable Internet
SES I
% Own Laptop
% Own PC
Table 10. Correlations Source: Claritas Inc., Computations by A. Modarres
0.09
0.15
0.12
0.14
0.11
SES II
0.53
0.59
-0.42
0.48
0.50
0.55
0.45
-0.04
-0.39
0.39
0.30
0.20
0.25
0.36
0.37
SES III
-0.38
-0.36
0.20
-0.26
-0.27
-0.35
-0.29
0.05
0.39
-0.16
-0.11
-0.04
-0.05
-0.13
-0.19 -0.33
% Own PC
1.00
0.90
-0.80
0.87
0.92
0.92
0.85
0.19
-0.87
0.71
0.60
0.30
0.46
0.59
0.66
0.90
1.00
-0.77
0.74
0.88
0.87
0.88
-0.70
0.69
0.57
0.49
0.51
0.66
0.69
0.85
1.00
-0.82
-0.93
-0.83
-0.72
-0.12
0.62
-0.74
-0.66
-0.37
-0.59
-0.72
-0.70 -0.79
1.00
0.91
0.85
0.69
0.35
-0.72
0.69
0.60
0.18
0.45
0.55
0.61
0.81
1.00
0.90
0.81
0.22
-0.73
0.76
0.65
0.35
0.56
0.68
0.70
0.86
1.00
0.87
-0.69
0.84
0.76
0.47
0.67
0.77
0.81
0.95
0.09
-0.60
0.87
0.80
0.72
0.74
0.83
0.89
0.94
1.00
-0.21
0.13
0.13
-0.12
0.05
0.15
1.00
-0.37
-0.26
0.07
-0.08
-0.23
-0.31 -0.65
1.00
0.96
0.75
0.91
0.94
0.96
0.92
1.00
0.76
0.92
0.92
0.94
0.85
1.00
0.86
0.85
0.84
0.60
1.00
0.96
0.93
0.75
1.00
0.97
0.84
1.00
0.90
% Own Laptop % with 1 cell phone % with 2 cell phones % with 2+ cell phones % Cable Internet % DSL Internet % Dial-up Internet % Zero Wirline % IM Text % IM Voice % Internet Video % Internet Games % Internet Music
1.00
% Internet Streaming Audio % Internet Banking
0.49
1.00
% Internet Play Game Alone % Internet Play Game Multi-player % Internet Send Picture % Internet Send Video % Internet Email % Internet Shop % Internet Publish to Comm. % Internet TV % Internet Streaming Video % Internet Yellow Pages % Latino % NH White % NH African American % NH Native Americans % NH Asian % NH Pacific Islander
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
76
Methodology and Findings
0.12
0.11
0.05
-0.04
% Non-Hispanic Pacific Islander
0.10
% Non-Hispanic Asian
0.18
% Non-Hispanic Native American
Yellow Pages
0.07
% Non-Hispanic White
Streaming Video
0.08
% Latino
Internet TV
0.17
Publish to Community
0.12
Shop
0.22
Email
Send Video
0.18
Send Picture
SES I
Play Game Multi-player
Play Game Alone
% Internet Use
% Non-Hispanic African American
Table 10. Correlations—CONTINUED Source: Claritas Inc., Computations by A. Modarres
0.08
SES II
0.36
0.18
0.49
0.36
0.51
0.52
0.23
0.30
0.40
0.48
-0.50
0.43
-0.20
-0.12
0.23 -0.13
SES III
-0.21
-0.03
-0.30
-0.24
-0.33
-0.35
-0.08
-0.11
-0.23
-0.30
0.62
-0.51
-0.03
-0.04
-0.09 -0.06
% Own PC
0.80
0.50
0.90
0.75
0.94
0.94
0.42
0.50
0.71
0.88
-0.53
0.47
-0.23
-0.14
0.22
% Own Laptop
0.66
0.44
0.82
0.73
0.89
0.91
0.45
0.62
0.75
0.84
-0.52
0.38
-0.17
-0.18
0.33
% with 1 cell phone
-0.80
-0.70
-0.83
-0.73
-0.80
-0.81
-0.50
-0.57
-0.69
-0.80
0.27
-0.25
0.17
0.14
-0.15
% with 2 cell phones
0.83
0.58
0.87
0.69
0.84
0.84
0.38
0.41
0.62
0.80
-0.42
0.45
-0.24
-0.12
0.06 -0.04
% with 2+ cell phones
0.83
0.64
0.89
0.75
0.90
0.90
0.47
0.56
0.72
0.85
-0.39
0.36
-0.22
-0.16
0.18
% Cable Internet
0.88
0.67
0.94
0.83
0.93
0.94
0.61
0.67
0.83
0.94
-0.47
0.39
-0.19
-0.18
0.24
% DSL Internet
0.82
0.69
0.90
0.91
0.94
0.94
0.72
0.82
0.92
0.94
-0.35
0.21
-0.14
-0.20
0.33
% Dial-up Internet
0.28
0.22
0.22
0.18
0.20
0.18
0.05
-0.08
0.19
-0.17
0.09
-0.21 -0.04
% Zero Wirline
-0.57
-0.19
-0.67
-0.47
-0.70
-0.72
-0.05
-0.14
-0.38
-0.61
0.53
-0.52
0.21
0.04
-0.11
% IM Text
0.93
0.89
0.91
0.92
0.88
0.87
0.91
0.89
0.96
0.94
-0.22
0.13
-0.13
-0.17
0.23
0.05
% IM Voice
0.89
0.92
0.84
0.90
0.80
0.78
0.93
0.88
0.92
0.87
-0.13
0.06
-0.10
-0.16
0.18
0.06
% Internet Video
0.51
0.69
0.53
0.73
0.55
0.54
0.86
0.94
0.83
0.64
-0.18
0.31
0.11
% Internet Games
0.78
0.91
0.72
0.83
0.68
0.67
0.94
0.92
0.90
0.79
-0.04
-0.06
-0.05
-0.17
0.22
0.06
% Internet Music
0.82
0.90
0.81
0.88
0.78
0.78
0.92
0.93
0.94
0.86
-0.15
0.03
-0.08
-0.18
0.27
0.05
% Internet Streaming Audio
0.86
0.88
0.87
0.92
0.85
0.84
0.92
0.93
0.98
0.92
-0.21
0.09
-0.10
-0.18
0.28
0.08
% Internet Banking
0.92
0.75
0.98
0.92
0.98
0.98
0.75
0.77
0.92
0.99
-0.43
0.34
-0.17
-0.17
0.26
0.04
% Internet Play Game Alone
1.00
0.87
0.95
0.90
0.90
0.89
0.76
0.71
0.86
0.93
-0.27
0.25
-0.17
-0.14
0.12
1.00
0.76
0.82
0.69
0.67
0.86
0.79
0.83
0.78
0.04
-0.06
-0.09
-0.14
0.11
1.00
0.91
0.97
0.97
0.72
0.72
0.89
0.98
-0.43
0.37
-0.19
-0.14
0.20
1.00
0.89
0.89
0.83
0.86
0.93
0.93
-0.27
0.19
-0.14
-0.16
0.23
1.00
0.99
0.68
0.72
0.88
0.97
-0.45
0.36
-0.20
-0.17
0.25
1.00
0.66
0.71
0.88
0.97
-0.48
0.38
-0.20
-0.16
0.26
1.00
0.90
0.90
0.78
-0.07
-0.04
-0.03
-0.15
0.22
0.08
1.00
0.93
0.79
-0.10
-0.04
-0.05
-0.19
0.30
0.09
1.00
0.94
-0.28
0.14
-0.11
-0.18
0.30
0.07
1.00
-0.39
0.29
-0.16
-0.17
0.26
0.05
1.00
-0.79
0.04
-0.05
-0.27
1.00
-0.38
0.07
-0.25 -0.18
1.00
-0.04
-0.06
1.00
-0.12
% Internet Play Game Multi-player % Internet Send Picture % Internet Send Video % Internet Email % Internet Shop % Internet Publish to Comm. % Internet TV % Internet Streaming Video % Internet Yellow Pages % Latino % NH White % NH African American % NH Native Americans
0.14
-0.19
% NH Asian % NH Pacific Islander
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
1.00
0.00
0.08
0.05 0.07
0.20 0.13 1.00
77
Methodology and Findings
this confirms that having a single or multiple cell phones is an important predictor of socioeconomic status. In fact, as Table 10 illustrates, while having one cell phone is negatively correlated with all other technology access indicators, it is positively correlated with percentage of Latino, NonHispanic African American, and Non-Hispanic Native American populations. Since owning a desktop or a laptop is the first step toward using the Internet, it is important to note that there are some differences between the two. As the number of households with laptops increases in an area, the level of advanced Internet usage increases. This is particularly visible in the higher correlation values for video downloading and watching Internet TV. While this may be driven by the underlying socioeconomic factors (laptop owners versus desktop owners), the difference nonetheless translates to issues of mobility and higher utility, factors that can be improved as the price for laptops approaches those of desktops. In terms of connectivity, Table 10 makes it clear that in areas where a large number of households employ dial-up services to access the Internet, usage of various Internet services remains minimal. At such low speeds, it would be nearly impossible to engage in sophisticated usage of important online functions, such as shopping and banking. Internally, variables defining various types of Internet usage are highly correlated. This suggests that Internet users who use cable or DSL are more likely to consume all that the Internet can provide. The only exceptions are those who rely on a large bandwidth, such as downloading videos or watching Internet TV. As such, selected technology
variables can be used to create a typology (or grouping) of Internet usage (by census tract). To illustrate this, I ran statistical analyses to examine the structural relationship between these variables. Readers need to be reminded that since the data used for this report represent census geography, the grouping simply suggests how areas can be grouped, based on how their residents use various Internet services. Table 11 represents the results of a factor analysis on 15 variables. Because this is indeed a variable reduction technique, we can use the results to understand how particular variables group together (through their loading levels on each new complex Table 11. Interent Usage Typology at Census Tract Level Source: Claritas Inc., Computations by A. Modarres COMMON INTERNET USAGE
SPECIALIZED HIGH SPEED INTERNET USAGE
— 0.424 0.546 0.633 0.893 0.847 0.520 0.921 0.744 0.921 0.927 0.418 0.427 0.676 0.869 12.8 85.3 7.4 49.2
0.932 0.879 0.812 0.763 0.434 0.441 0.731 0.376 0.608 0.348 0.335 0.867 0.872 0.716 0.480 1.4 9.3 6.8 45.3
Percent Internet Video Percent Internet Games Percent Internet Music Percent Internet Streaming Audio Percent Internet Banking Percent Internet Play Game Alone Percent Internet Play Game Multi-player Percent Internet Send Picture Percent Internet Send Video Percent Internet Email Percent Internet Shop Percent Internet Publish to Community Percent Internet TV Perecnt Internet Streaming Video Percent Internet Yellow Pages Initial Eigenvalue Percent of Variance Rotation Sums of Squared Loadings Percent of Variance Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. Rotation converged in 3 iterations
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
78
Methodology and Findings
variable or index). The result of this analysis suggests that the 15 variables could help construct two indexes, collectively explaining close to 94.5% of the variance in the original data. These variable groupings reflect two types of Internet usage: common applications (such as e-mailing, banking, shopping, sending pictures and videos, and playing games alone) and specialized usage that requires high-speed connection (Internet videos, games, music, streaming audio, multiplayer games, visiting and publishing to community, Internet TV, and streaming video). Since the unit of analysis is a census tract, this grouping pattern suggests that not only is there a distinct geography of Internet usage but also this pattern is driven by socioeconomic status of an area, which affects its prevalent mode of connection to the Internet, the degree of need for particular application, and cost associated with more advanced applications (and technologies). Based on initial findings, I ran a second factor analysis, including all the pertinent technology variables, attempting to create a composite technology index that can provide an overview of the geography of access to technology the state. Table 12 provides the results of this analysis, highlighting the degree to which a typology of access to technology can be created. The three extracted composite indexes collectively explain 96.6% of the variance in the 40 variables in this study.3
The three indexes provide an interesting grouping of census tracts by technology indicators. The first index includes the largest number of technology variables and, as such, should be treated as the technology index for mapping and further analysis. It portrays a high loading pattern by all variables other than 11. These 11 variables have higher loading values on Indexes II and III, suggesting a clear division between neighborhoods and their
pattern of access to technology. For example, the second index includes having access to one cell phone, one desktop, cable connection to the Internet, and zero or one wireline. This confirms the earlier findings regarding the number of available cell phones per household. Given that having only one cell phone per household produces the highest loading value on the second index and is missing from the first index (having achieved a loading value below 0.3), we can state with certainty that this variable is an important identifier for the level of access to technology. In other words, in neighborhoods where a large number of households have only one cell phone, access to other technologies seems to score low. Note that all the Internet usage variables scored lower on the second index than on the other two. The third index provides further evidence to what was discussed regarding the typology of use (see Table 11). Once again, this statistical analysis illustrates how high-end uses tend to group together, creating an index that not only distinguishes itself from the second index but also from the first. In other words, there is a distinct grouping of census tracts that differentiates neighborhoods with users of instant messaging with voice, downloading video content, downloading/purchasing games, downloading/ purchasing music, listening to streaming audio, visiting/publishing to online community, and watching Internet TV live from others. Figures 57 through 57b illustrate the geographic structure of these three indexes. The first index, which appears in Figure 57, helps identify areas where a significant number of residents have access to a host of technologies and Internet usage. Areas with higher
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
79
Methodology and Findings
Figure 57. 2007 Use of Various Information and Communication Technologies
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
80
Methodology and Findings
Figure 57a. 2007 Technology Index II Reliance on Single Cell Phone per Households & with Minimal Usage of other Technologies
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
81
Methodology and Findings
Figure 57b. 2007 Technology Index III High End Users of the Internet
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
82
Methodology and Findings
Table 12. Technology Indexes Source: Claritas Inc., Computations by A. Modarres Tech. Index I
Tech. Index II
Tech. Index III
HAS CELL PHONE - NUM: 1 HAS CELL PHONE - NUM: 2 HAS CELL PHONE - NUM: 2+ HAS CELL PHONE - NUM: 3 HAS CELL PHONE - NUM: 3+ HAS CELL PHONE - NUM: 4 HAS CELL PHONE - NUM: 4+
— 0.76 0.82 0.81 0.85 0.88 0.88
0.94 0.54 0.44 0.31 — — —
— 0.31 0.36 0.43 0.42 0.39 0.38
HAS CELL PHONE - NUM: 5+ HAS DESKTOP - HOME USE HAS DESKTOP - HOME USE - NUM: 1 HAS DESKTOP - HOME USE - NUM: 2+ HAS DESKTOP - HOME USE - NUM: 4+ HAS INTERNET AT HOME: DIALUP HAS INTERNET AT HOME: DSL HAS INTERNET AT HOME: CABLE OWN LAPTOP/NOTEBOOK PC HAS PC - LAPTOP - NUM: 1 HAS PC - LAPTOP - NUM: 2 HAS PC - LAPTOP - NUM: 3+ HAS WIRELINE - NUMBER: 0 HAS WIRELINE - NUMBER: 1 HAS WIRELINE - NUMBER: 2 HAS WIRELINE - NUMBER: 3+ INTERNET USE: CHAT/INSTANT MESSAGING WITH TEXT INTERNET USE: CHAT/INSTANT MESSAGING WITH VOICE INTERNET USE: DOWNLOAD VIDEO CONTENT INTERNET USE: DOWNLOAD/PURCHASE GAMES INTERNET USE: DOWNLOAD/PURCHASE MUSIC INTERNET USE: LISTEN TO STREAMING AUDIO INTERNET USE: ONLINE BANKING INTERNET USE: PLAY GAMES ALONE INTERNET USE: PLAY MULTI-PLAYER GAMES INTERNET USE: SEND PICTURES INTERNET USE: SEND VIDEO EMAIL INTERNET USE: SEND/RECEIVE EMAIL INTERNET USE: SHOP INTERNET USE: VISIT/PUBLISH TO ONLINE COMMUNITY INTERNET USE: WATCH INTERNET TV INTERNET USE: WATCH STREAMING VIDEO INTERNET USE: YELLOW PAGES Initial Eigenvalue Percent of Variance Rotation Sums of Squared Loadings Percent of Variance
0.82 0.58 0.39 0.81 0.86 0.78 0.72 0.53 0.80 0.81 0.87 0.93 — 0.52 0.82 0.88 0.65 0.60 0.48 0.58 0.64 0.64 0.73 0.67 0.61 0.74 0.67 0.71 0.76 0.52 0.58 0.66 0.72 34.9 82.3 20.0 50.1
— 0.75 0.87 0.51 0.31 0.43 0.46 0.77 0.36 0.42 — — 0.76 0.80 0.45 0.33 0.51 0.52 0.39 0.47 0.36 0.42 0.47 0.56 0.51 0.48 0.47 0.55 0.47 0.47 0.41 0.42 0.46 2.7 6.7 9.9 24.7
0.35 0.30 0.30 — 0.31 0.43 0.50 — 0.41 0.37 0.35 — 0.52 — — — 0.56 0.60 0.77 0.67 0.67 0.64 0.48 0.46 0.58 0.46 0.56 0.43 0.44 0.71 0.70 0.62 0.50 1.0 2.6 8.7 21.8
Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. Rotation converged in 7 iterations.
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
83
Methodology and Findings
scores on this index typically have higher socioeconomic
which depicts areas with households having access
status, but they are not exclusively urban. In fact, a
to one cell phone and a minimal number of wirelines,
number of census tracts in rural and less populated areas
correlated
of California scored relatively high on this score. The
significant at the 0.01 level, 2-tailed) with the first
second and third technology indexes more decisively
socioeconomic index, which represented areas where
identify the geographic divide between high-end users
households with lower education, sales, service, and
and those with only entry-level access to basic ICTs.
blue-collar occupations can be found.
highly
(i.e.,
Pearson
correlation=0.61,
While Figure 57a identifies mostly rural areas with
Though factor analysis produces indexes that are
high scores on the second technology index (associated
created through regression-like models, the results
with reliance on a single cell phone per household and
are not easily understood by people unfamiliar with
minimal usage of other technologies), the third index
multivariate statistics and interpretation their results.
(Figure 57b) identifies mostly areas that house high-end
To create a more understandable index, I followed the
users of the Internet (suggesting better connectivity and
methodology that was developed during our last study
higher socioeconomic status). In Southern California,
of Los Angeles. For this, I created a ranking of 1 to 3 for
portions of Ontario (an unincorporated area of San
each tract on each of the 26 technology variables.4,5 This
Bernardino) south of Ontario and portions of Chino
meant that if a tract scored uniformly low across all variables, it could only achieve a value of 26 on the new index. If it achieved the highest score on all variables, a census tract would receive an index value of 78. The result for all census tracts in California revealed that the lowest achieved score is 29 and the highest is 75. These values represent the Cumulative Technology Index for the State. Figure 58 illustrates the geographic distribution of this index, illustrating that areas with the highest scores are more likely to be located closer to the coastal regions in the Bay Area, extending from Contra Costa to Santa Clara and Marin to San Mateo counties, and in Southern California, extending from Ventura to San Diego, including the southwestern region of San Bernardino and the western section of Riverside counties. With a mean score of 52 and a standard deviation of 11 for the state, it is expected that about two thirds of all census tracts achieved a score that ranges from 41 to 63.6 However, 979 census tracts had values
score very high on this score. This is also true for two small urban corridors: one extends from Hollywood through West Hollywood, Beverly Hills, and along Wilshire Blvd. to UCLA and Santa Monica and the other, along the 101 freeway corridor from Burbank to Sherman Oaks. Other pockets of high scores also appear in a number of other cities in Los Angeles, Orange, San Diego, Ventura, and Riverside counties. In the Bay Area, few census tracts in the cities of Fremont, San Francisco, and San Jose score high on this index. As Figure 57b illustrates, very few tracts achieved a high score on this index, clearly distinguishing various concentrations of high-end users. From a statistical perspective, while the first technology index, which includes access to a significant majority of ICTs, is highly correlated (i.e., Pearson correlation=0.71, significant at the 0.01 level, 2-tailed) with the second socioeconomic status index (i.e., areas with a large number of educated households with professional occupations), the second technology index,
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
84
Methodology and Findings
Figure 58. 2007 Cumulative Index Access to Information and Communication Technologies
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
85
Methodology and Findings
exceeding 63 (or one standard deviation above the mean), 404 of which were located in the three counties of Los Angeles, Orange, and Santa Clara (see Table 13). Collectively, the 979 tracts had a population of close to 6 million, which accounts for 16.2% of the total population in the state. Nearly half of the residents of these tracts were NonHispanic White, while 5.3% were Non-Hispanic
African American, 18.4% were Non-Hispanic Asian American, and 22.6% were Latino. Comparing these values with the overall racial and ethnic structure of the state (see Table 1) suggests that the population residing in tracts with the highest levels of access to technology is disproportionately NonHispanic White and Asian. While only 26 counties appear on the list of
Table 13. Areas with High Scores on Access to Technology Index Source: Claritas Inc., Computations by A. Modarres
COUNTY
NO. OF CENSUS TRACTS
2007 POPULATION
2007 NON-HISPANIC WHITE POPULATION
Alameda Contra Costa El Dorado Fresno Imperial Kern Los Angeles Merced Monterey Orange Placer Riverside Sacramento San Benito San Bernardino San Diego San Francisco San Joaquin San Mateo Santa Barbara Santa Clara Solano Sonoma Stanislaus Ventura Yolo Total Percent of Total
70 32 2 10 1 12 155 2 10 101 10 66 35 2 46 90 69 12 21 6 148 21 3 4 48 3 979 100.00
400,670 274,740 10,856 62,492 3,443 75,856 791,767 17,122 48,706 562,006 78,047 472,100 248,505 20,864 532,400 552,151 269,115 129,769 88,223 31,122 803,165 131,746 23,981 32,929 319,112 13,328 5,994,215 100.00
176,795 140,262 8,733 40,284 1,139 47,020 441,650 6,701 13,920 357,754 61,202 218,025 124,800 9,926 182,603 303,579 171,492 55,925 45,566 17,946 277,532 50,305 17,851 17,218 164,822 5,933 2,958,983 49.36
2007 2007 NON-HISPANIC NON-HISPANIC AFRICAN AMER. NATIVE AMER. POPULATION POPULATION
13,841 24,018 168 1,840 19 3,233 39,900 935 1,162 8,559 1,266 47,892 23,355 290 53,277 21,122 10,247 11,842 1,874 408 21,754 20,364 340 1,196 5,471 197 314,570 5.25
1,316 919 23 258 27 593 2,151 81 186 1,300 384 2,009 1,142 100 1,660 1,451 626 765 214 98 1,941 516 114 146 1,177 48 19,245 0.32
2007 NON-HISPANIC ASIAN POPULATION
131,216 43,522 642 7,116 218 4,639 116,378 433 5,753 101,670 6,230 38,756 45,161 918 52,230 102,769 43,881 18,055 25,082 2,764 296,543 27,718 1,759 3,571 23,728 2,061 1,102,813 18.40
2007 NON-HISPANIC 2007 PAC. ISLNDR NON-HISPANIC POPULATION OTHER
1,536 1,054 25 145 -— 122 1,525 80 146 1,043 150 1,793 1,055 54 1,250 1,947 505 827 555 53 3,126 895 87 172 704 107 18,956 0.32
1,136 697 23 192 7 92 2,198 67 67 1,073 151 888 711 38 930 1,082 1,030 432 273 57 1,794 336 36 155 501 25 13,991 0.23
2007 NON-HISPANIC 2 RACES OR MORE
2007 LATINO POPULATION
18,325 12,725 329 1,965 36 1,919 20,893 495 1,000 20,033 2,709 15,608 13,899 405 14,957 20,014 7,724 6,913 3,250 1,083 27,641 7,519 846 1,517 6,675 507 208,987 3.49
56,505 51,543 913 10,692 1,997 18,238 167,072 8,330 26,472 70,574 5,955 147,129 38,382 9,133 225,493 100,187 33,610 35,010 11,409 8,713 172,834 24,093 2,948 8,954 116,034 4,450 1,356,670 22.63
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
86
Methodology and Findings
Table 14. Areas with Low Scores on Access to Technology Index Source: Claritas Inc., Computations by A. Modarres
COUNTY
Alameda Amador Butte Calaveras Colusa Contra Costa Del Norte El Dorado Fresno Glenn Humboldt Imperial Inyo Kern Kings Lake Lassen Los Angeles Madera Mariposa Mendocino Merced Modoc Monterey Napa Nevada Orange Placer Plumas Riverside Sacramento San Bernardino San Diego San Francisco San Joaquin San Luis Obispo Santa Barbara Santa Clara Santa Cruz Shasta
NO. OF CENSUS TRACTS
2007 POPULATION
2007 NON-HISPANIC WHITE POPULATION
2007 2007 NON-HISPANIC NON-HISPANIC AFRICAN AMER. NATIVE AMER. POPULATION POPULATION
2007 NON-HISPANIC ASIAN POPULATION
2007 NON-HISPANIC 2007 PAC. ISLNDR. NON-HISPANIC POPULATION OTHER
35 5 29 4 2 9
111,562 23,926 141,761 24,837 5,715 50,169
9,143 21,394 107,602 21,125 3,484 15,119
47,619 78 2,254 261 100 10,722
410 210 2,384 367 109 191
15,239 251 6,693 293 52 3,384
659 10 252 21 13 273
5 4 46 3 19 15 5 32 4 9 2 466 6 3 13 10 4 2 1 7 26 7 5 89 41 45 70 12 28 7 2 1 2 23
22,215 16,849 270,370 12,128 95,869 79,147 15,205 157,242 25,351 48,615 8,571 2,196,272 39,807 16,172 63,472 42,169 9,682 11,579 5,179 38,862 143,833 39,651 19,617 445,961 192,175 216,185 335,710 39,482 142,559 38,925 14,802 328 14,649 116,648
16,646 13,918 55,691 8,009 74,486 9,708 10,235 64,212 7,311 36,584 4,631 144,799 15,434 13,665 45,356 10,156 7,852 3,145 2,575 34,074 41,187 30,212 17,144 223,524 57,432 116,917 80,798 6,504 25,887 31,195 2,316 186 2,201 95,779
120 38 22,966 141 1,024 4,524 51 10,040 771 1,218 1,495 303,562 535 140 426 2,133 73 2,228 21 176 1,181 297 179 18,179 30,193 13,393 20,794 9,782 14,479 574 503 7 85 1,085
1,364 233 2,331 276 5,294 1,223 1,492 1,841 419 1,145 171 6,356 676 407 2,375 233 315 79 31 279 416 265 382 3,723 1,501 2,083 1,962 148 926 214 112 — 120 2,891
777 207 25,335 559 1,822 753 175 1,580 412 520 40 172,928 238 163 563 4,058 67 407 87 423 8,957 689 137 10,296 30,497 8,804 25,028 18,091 21,268 1,486 322 72 326 2,373
12 19 362 10 180 54 8 194 48 66 6 5,162 113 13 110 41 7 81 2 39 157 72 15 813 2,032 585 1,159 582 299 40 30 — 8 148
2007 NON-HISPANIC 2 RACES OR MORE
2007 LATINO POPULATION
253 22 257 21 11 126
4,172 452 5,664 657 96 1,493
34,067 1,509 16,655 2,092 1,850 18,861
38 16 373 14 329 50 21 197 18 44 97 3,145 134 18 120 75 27 262 3 83 85 39 32 483 507 373 414 99 230 50 6 — 25 144
869 573 6,038 268 4,705 457 434 3,432 447 1,338 131 30,374 706 439 1,768 903 174 222 68 1,180 1,289 1,034 470 8,447 11,510 6,824 8,951 1,234 4,641 854 258 8 163 4,566
2,389 1,845 157,274 2,851 8,029 62,378 2,789 75,746 15,925 7,700 2,000 1,529,946 21,971 1,327 12,754 24,570 1,167 5,155 2,392 2,608 90,561 7,043 1,258 180,496 58,503 67,206 196,604 3,042 74,829 4,512 11,255 55 11,721 9,662
CONTINUED
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
87
Methodology and Findings
Table 14. Areas with Low Scores on Access to Technology Index—CONTINUED Source: Claritas Inc., Computations by A. Modarres
COUNTY
NO. OF CENSUS TRACTS
2007 POPULATION
2007 NON-HISPANIC WHITE POPULATION
Sierra Siskiyou Solano Sonoma Stanislaus Sutter Tehama Trinity Tulare Tuolumne Ventura Yolo Yuba Total Percent of Total
1 14 3 2 13 5 10 4 19 8 5 7 2 1,191 100.00
3,427 46,108 5,235 8,113 60,663 18,550 57,427 13,958 94,547 46,459 13,316 35,003 11,695 5,707,752 100.00
3,017 37,737 1,181 6,912 22,513 10,268 42,164 11,938 21,599 37,003 8,494 16,223 7,739 1,714,424 30.04
2007 2007 NON-HISPANIC NON-HISPANIC AFRICAN AMER. NATIVE AMER. POPULATION POPULATION
7 603 1,426 54 2,114 402 414 60 1,641 1,899 156 1,401 156 533,780 9.35
46 1,444 60 49 558 213 941 541 1,032 575 87 371 264 51,135 0.90
2007 NON-HISPANIC ASIAN POPULATION
3 558 904 180 2,315 1,031 479 116 3,416 454 384 4,343 973 380,528 6.67
2007 NON-HISPANIC 2007 PAC. ISLNDR. NON-HISPANIC POPULATION OTHER
3 39 24 13 209 40 38 16 154 73 7 147 8 14,466 0.25
2 43 8 15 104 20 115 14 113 38 13 78 14 8,818 0.15
2007 NON-HISPANIC 2 RACES OR MORE
2007 LATINO POPULATION
43 1,413 264 247 2,032 496 1,401 585 1,610 1,142 264 1,579 559 128,944 2.26
306 4,271 1,368 643 30,818 6,080 11,875 688 64,982 5,275 3,911 10,861 1,982 2,875,657 50.38
census tracts with the high scoring values on the technology index (see Table 13), the list for census tracts with low score values includes 53 counties (see Table 14), missing only Alpine, Marin, Mono, San Benito, and San Mateo counties. Of these, only San Benito and San Mateo show up on Table 13. This suggests that census tracts in these counties are entirely in the high-scoring category. The other three counties have census tracts that fall entirely in the middle range for the Cumulative Technology Index.
White, over 50% are Latino and 9% are African
Table 14 illustrates an interesting geography and
Beyond these racial and ethnic dynamics, it is
demographic structure for low-scoring census tracts.
important to note that many rural census tracts are
For example, about 39% of all tracts in this category
also affected by the emerging geography of digital
(466 of 1,191 census tracts) fall in Los Angeles County,
divide in the state. While northern, central, and eastern
housing also about 39% of the 5.7 million people who
counties are clearly less populated, they nonetheless
live in such tracts in the state. Overall, while slightly
score very low on their level of access to technology.
over 30% of residents in these tracts are Non-Hispanic
Given the limited resources in the state, it is crucial
American, rates that are disproportionate to the racial and ethnic structure of the population in the state (see Table 1). This pattern is more severe at the county level. For example, in Los Angeles, only about 144,000 of the residents in the low- scoring tracts were Non-Hispanic White. This is slightly over 1% of the total population and about 5% of all Non-Hispanic White residents of the county. In contrast, these low-scoring tracts house over 1.5 million Latinos, making up 15% of the county population and about 32% of its total Latino residents.
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
88
Methodology and Findings
that we prioritize our intervention policies based on
area/population prioritization needs to take a phased
a hierarchy that includes geographic location and
approach that helps some neighborhoods reach the
socioeconomic status. After all, the computed index
middle range quickly (i.e., those near a score of 41) and
of access to technology is significantly (and positively)
invests in low-scoring areas by building the needed
correlated with the second socioeconomic index (i.e.,
physical infrastructure and human capital to achieve
educated population with professional occupations)
higher levels of connectivity in the future.
and negatively with the third socioeconomic index
To provide one such example of an area-based
(i.e., lower education and income). Given that these
prioritization, I identified two groups of census tracts:
geographies cover both urban and rural California, the
those that are between one and one-and-a-half standard
Table 15. Category I Areas for Possible Policy Intervention Source: Claritas Inc., Computations by A. Modarres
COUNTY
Alameda Butte Fresno Humboldt Imperial Kern Kings Los Angeles Madera Merced Orange Riverside Sacramento San Bernardino San Diego San Francisco San Joaquin San Luis Obispo Santa Barbara Siskiyou Solano Stanislaus Sutter Tulare Ventura Yolo Yuba Total Percent of Total
NO. OF CENSUS TRACTS
2007 POPULATION
2007 NON-HISPANIC WHITE POPULATION
2007 2007 NON-HISPANIC NON-HISPANIC AFRICAN AMER. NATIVE AMER. POPULATION POPULATION
2007 NON-HISPANIC ASIAN POPULATION
2007 NON-HISPANIC 2007 PAC. ISLNDR. NON-HISPANIC POPULATION OTHER
4 7 7 3 1 14 2 113 2 4 1 11 5 6
9,934 30,128 42,394 12,779 2,942 73,758 15,075 529,427 13,142 18,630 2,843 65,728 21,490 24,067
1,249 20,953 7,018 7,476 495 11,865 4,196 35,052 1,715 1,005 714 12,169 6,410 3,878
3,928 872 2,029 310 12 7,171 592 62,261 149 1,004 38 3,466 2,615 4,363
68 372 334 2,675 13 573 111 1,432 115 60 1 455 166 132
940 1,956 3,721 287 6 936 312 40,538 42 2,972 1,723 1,254 4,456 475
68 108 80 23 — 60 32 1,591 32 21 4 139 129 74
15 1 6 1 2 1 1 6 1 9 1 3 1 228 100.00
86,018 5,130 34,896 5,802 14,802 1,401 193 28,967 4,116 46,686 2,426 15,919 6,518 1,115,211 100.00
16,071 1,771 3,400 4,490 2,316 644 55 7,982 1,853 7,697 1,539 7,268 3,227 172,508 15.47
8,019 1,919 3,852 39 503 2 45 683 86 1,091 37 716 131 105,933 9.50
478 10 136 17 112 8 — 232 83 376 28 151 161 8,299 0.74
4,935 831 6,326 607 322 5 51 970 57 1,836 117 2,990 923 79,588 7.14
341 9 95 8 30 — — 75 3 62 2 45 5 3,036 0.27
2007 NON-HISPANIC 2 RACES OR MORE
2007 LATINO POPULATION
15 83 54 52 — 60 15 660 13 22 14 53 76 28
430 1,285 913 572 21 1,172 262 8,100 64 346 143 1,195 1,332 511
3,236 4,499 28,245 1,384 2,395 51,921 9,555 379,793 11,012 13,200 206 46,997 6,306 14,606
85 17 111 6 6 3 4 34 4 38 5 19 2 1,479 0.13
3,126 230 1,281 152 258 15 20 875 82 669 107 753 326 24,240 2.17
52,963 343 19,695 483 11,255 724 18 18,116 1,948 34,917 591 3,977 1,743 720,128 64.57
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
89
Methodology and Findings
Figure 59. Category I Priority Areas & 2007 Cumulative Technology Index
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
90
Methodology and Findings
deviation below the mean (i.e., areas where scores are
low on access to either cable or DSL for connecting to
close to the middle range and smaller investments
the Internet. They also rank low on having PCs and
could bring about the needed transition more swiftly)
more than one cell phone per household. In terms of
and areas where the scores are significantly low (i.e.,
Internet usage, the majority of the 228 census tracts
areas with a score below 35). It is interesting that the
rank low on e-mailing, shopping, banking, and other
first category identified 467 tracts in the state (about 48%
modes of information gathering and communication.
of the 979 census tracts that were below one standard
Given lack of access to high-speed Internet, the low
deviation). A significant majority of these tracts (197 or
ranking for these usage categories is understandable. A
42%) is located in Los Angeles County. Among them,
reasonable public policy approach to these tracts could
228 report median household incomes below $30,000,
include the following:
suggesting that they may need a more immediate policy intervention. As Table 15 illustrates, 27 or half of all counties in California show up on this list, including a mixture of rural and urban areas (also see Figure 59,
Table 16. Patterns of Access to Technology in Category I Areas (228 tracts); Values represent the number of census tracts in each ranking category Source: Claritas Inc., Computations by A. Modarres
which identifies these tracts visually). They dot counties
RANKING ON VARIOUS TECHNOLOGY SCORES
in Southern California and Imperial Valley and a chain of them appears from central to northern California, highlighting some of the more rural areas of the state, which may require more immediate attention. Collectively, these 228 tracts house 1.1 million individuals, who are largely Latino (64.6%) and NonHispanic African American (9.5%). However, in counties such as Butte, Humboldt, San Luis Obispo, and Ventura, more than half of the resident population of these tracts is Non-Hispanic White. Counties where Latinos make up more than half of the population in the identified tracts include Fresno, Imperial, Kern, King, Los Angeles, Madera, Merced, Riverside, San Bernardino, San Diego, San Joaquin, Santa Barbara, Siskiyou, Stanislaus, and Tulare. Among these, Los Angeles, San Diego, Kern, Riverside, Tulare, and Fresno house the largest number of Latinos (i.e., close to 83% of the Latinos in the selected 228 census tracts, or 600,000, live in these six counties). The aggregate ranking pattern on the 26 technology variables for these 228 tracts are shown in Table 16. It is clear from this table that these areas collectively rank
Own PC Own Laptop 1 cell phone 2 cell phones 2+ cell phones Cable Internet DSL Internet Dial Up Internet Zero Wirleine IM Text IM Voice Interent Video Internet Games Internet Music Interent Steaming Audio Interen Banking Interent Play Game Alone Internet Plaly Game Multi-player Internet Send Picture Internet Send Video Internet Email Interent Shopping Internet Visit/Publish to Online Community Interent TV Internet Streaming Audio Internet Yellow Pages
LOW
MEDIUM
HIGH
218 191 — 161 117 217 218 32 — 95 34 14 4 13 67 222 188 18 216 216 228 228 18 25 140 227
10 37 142 67 111 11 10 186 23 133 194 214 224 215 161 6 40 210 12 12 — — 210 203 88 1
— — 86 — — — — 10 205 — — — — — — — — — — — — — — — — —
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
91
Methodology and Findings
a)
making sure that they do have the adequate
infrastructural
backbone
to
provide
high-speed
population is significantly smaller and geographically concentrated (in fewer tracts) than that of Latinos, its
connectivity to the Internet,
experience with digital divide may not be as readily
b)
obvious (especially in statewide and large regional
assuring that access to this service is not hindered
by cost, and
analyses). As Table 17 illustrates, 47% of the population
c)
of Alameda County, who live in census tracts that fall
providing educational resources regarding the use
and benefits of the Internet.
in the second category (i.e., the lowest level of access to almost all forms of ICTs), are Non-Hispanic African
Additionally, it is crucial that steps are taken to
American. Similarly, 24% of the selected census tracts
expand subscription and use of cell phones in these
in Contra Costa County, 11% of Fresno, 24% of Lassen,
tracts. As this study has shown, the number of cell
15% of Los Angeles, 16% of Sacramento, 16% of San
phones per household is an important factor in the
Francisco, 10% of San Joaquin, and 37% of Solano are
emerging patterns of digital divide. Increasing the level
Non-Hispanic African American. Hence, focusing
of access to cell phones and smart phones (i.e., more
on these priority areas would not only improve our
than one per household) could help us expand the
digital divide patterns but also take major steps toward
level of access to the Internet in a more immediate (and
improving the status of access in African American and
perhaps) less costly manner. Through a public-private
Latino neighborhoods in the state.
partnership, we could bring about less costly services
Not surprisingly, almost all of the 341 census tracts
and offer more education about how these devices can
in this category scored low on each of the 26 technology
play the dual role of providing personal communication
variables. In fact, these areas were selected based on the
and access to digital information.
two criteria of having a score of 35 or lower and having
The second category, which identifies the least
a median household income of less than $30,000. This
connected census tracts, includes both rural and urban
means that the cumulative technology index for these
areas; however, as Figure 60 suggests, a larger number
tracts could range from a low of 29 to a high of 35 (note
of these tracts are located in Northern California. The 341
that to receive a score of 35, a tract must score at least a
census tracts in this category house 1.5 million people,
2 in 9 categories – based on 26 variables). An assessment
56.4% of whom are Latino; 12.3%, Non-Hispanic African
of scores for various variables reveals that this was
American; and another 17.9%, Non-Hispanic White
achieved by scoring higher than the lower values
(see Table 17). It is interesting that this category has the
on a small number of indicators, including having
highest representation of African Americans, compared
access to one cell phone and having dial-up services
with all other groupings previously discussed. This
for connecting to the Internet. Overall, these census
may suggest that while Latino neighborhoods remain
tracts need a significant infrastructural and human/
among the most technologically disconnected in the
social capital development. This can be best achieved,
state, African American neighborhoods are equally and,
perhaps, by a mixture of educational and infrastructural
in some cases, more drastically affected by the same
policies. While the latter would focus on improving
phenomenon. However, since the African American
access to ICTs, especially access to the broadband, the
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
92
Methodology and Findings
Table 17. Category II Areas for Possible Policy Intervention Source: Claritas Inc., Computations by A. Modarres
COUNTY
Alameda Butte Contra Costa Del Norte Fresno Humboldt Imperial Inyo Kern Lake Lassen Los Angeles Mendocino Merced Modoc Monterey Orange Riverside Sacramento San Bernardino San Diego San Francisco San Joaquin San Luis Obispo Shasta Siskiyou Solano Stanislaus Sutter Trinity Total Percent of Total
NO. OF CENSUS TRACTS
2007 POPULATION
2007 NON-HISPANIC WHITE POPULATION
2007 2007 NON-HISPANIC NON-HISPANIC AFRICAN AMER. NATIVE AMER. POPULATION POPULATION
2007 NON-HISPANIC ASIAN POPULATION
2007 NON-HISPANIC 2007 PAC. ISLNDR. NON-HISPANIC POPULATION OTHER
9 4 2 1 24
20,513 20,725 6,759 3,849 129,136
1,419 12,430 1,130 2,779 16,821
9,634 774 1,628 20 14,531
61 586 13 177 1,056
5,954 2,408 432 215 16,379
45 30 18 2 140
4 7 1 9 3 1 173 3 1 1 1 1 19 14 9 20 8 11 1 6 3 1 1 1 2 341 100.00
20,914 33,900 2,524 50,889 20,830 5,679 810,743 16,496 4,008 3,538 2,547 3,723 68,114 67,166 23,809 92,161 29,050 55,072 3,476 24,746 8,491 3,275 2,539 5,072 3,836 1,543,580 100.00
15,641 3,869 1,493 29,706 15,438 2,293 42,471 10,522 1,552 2,881 399 3,412 34,778 15,668 5,870 7,101 4,470 7,320 2,567 19,179 6,629 595 1,366 3,024 3,164 275,987 17.88
295 586 — 2,387 887 1,430 122,256 105 264 8 79 25 2,983 10,904 2,345 5,632 4,758 5,599 67 414 73 1,218 156 171 2 189,231 12.26
1,002 1,041 122 804 466 66 2,550 499 29 109 12 6 614 598 448 251 132 344 28 630 689 23 47 59 247 12,709 0.82
507 413 23 318 253 31 76,192 158 200 37 93 116 1,596 11,103 752 11,103 16,306 9,402 177 1,185 117 483 21 83 23 156,080 10.11
48 48 3 58 42 4 1,403 19 2 4 6 3 125 759 65 289 535 88 3 56 5 16 7 26 2 3,851 0.25
2007 NON-HISPANIC 2 RACES OR MORE
2007 LATINO POPULATION
74 23 23 8 115
755 1,106 173 197 3,121
2,571 3,368 3,342 451 76,973
124 21 5 80 26 95 1,165 22 37 15 — — 40 141 38 97 66 54 16 56 — — 8 2 11 2,362 0.15
1,436 284 43 1,383 710 31 10,611 455 62 72 51 20 1,440 3,931 598 2,017 872 1,675 94 1,158 337 167 74 122 196 33,191 2.15
1,861 27,638 835 16,153 3,008 1,729 554,095 4,716 1,862 412 1,907 141 26,538 24,062 13,693 65,671 1,911 30,590 524 2,068 641 773 860 1,585 191 870,169 56.37
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
93
Methodology and Findings
Figure 60. Category II Priority Areas & 2007 Cumulative Technology Index
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
94
Methodology and Findings
former would help enable the population to utilize
capital of these neighborhoods as their physical and
these services to expand their social and economic
economic structures are enhanced. In the end, to
opportunities. This would mean that in addition to
improve the access to technology in the most severely
the private-public partnership for making resources
disconnected places, it will take more than making a
available, nonprofit and grass-root groups would need
few technologies available. We need to prepare and
to be included for the full diffusion of the technology.
cultivate the conditions that make technology relevant
This would also provide the needed education and
to the life of residents and sustain their access to these
community development efforts to build the social
tools and services.
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
95
Concluding Remarks
This study focused on an in-depth analysis of access to information and communication technologies (ICTs) in California. Using census tract data from 2007, we were able to provide a detailed visual and statistical assessment of the nature of digital divide in the state. This allowed us to create statewide and county-level maps and analyses that identified the patterns of inequity by social, economic, racial, ethnic, and geographic indictors. While the main body of the report displayed maps at the state level, we prepared an Appendix that contains a set of 35 maps for each county in the state. In all, over 2,000 maps have been created to provide readers with a visual tool to assess the spatial structure of access to ICTs in the state.7
Our findings illustrate the degree to which Latinos and Non-Hispanic African Americans remain isolated from advancements in the use of ICTs and the information and services they offer. In the case of Latinos, this takes both a rural and an urban dimension, suggesting that as a group, they face many obstacles in accessing technology, regardless of where they live. It was also illustrated that, within an urban context, African Americans in particular locations, such as Alameda and Los Angeles, remain equally and, in some cases, more pronouncedly isolated from what the information technology can offer. The racial/ethnic dimension of the digital divide is an important concern, especially when we consider the degree to which this factor has correlated with socioeconomic status. As indicated in this report, while less homogenous places are not necessarily always low-income minority neighborhoods, many African American and Latino majority areas experience lower-socioeconomic status. This was clearly illustrated when we discussed the concept
3
of diversity (i.e., entropy index). The lingering question of race/ethnicity and its geography becomes clearly important as we consider the issue of digital divide. For a state that has attracted many immigrants and minorities over the last few decades, allowing it to become one of the most diverse places in the world, the paradox of segregating in the midst of diversity is an ongoing challenge. In the case of digital divide, then, it should not come as a surprise that the emergent spatial patterns are strongly influenced by the geography of race and ethnicity. In fact, what is ironic about the state of digital divide in California is the degree to which diversity status in a census tract is related to the observed level of access to technology. As discussed earlier, an entropy or diversity index allows us to measure the degree to which various racial and ethnic groups cohabit in an area. The larger the value of this index, the higher the level of diversity would be. To be sure, this index was shown to be negatively correlated with Latino and Non-Hispanic White populations and positively with Non-Hispanic Asian and African American populations. This meant that census tracts with a high-diversity index were more likely to house a large number of the latter groups and less of the former. With that information in mind, it was surprising to discover that our cumulative index of access to technology was positively correlated with the diversity index! In other words, the higher the diversity level, the more likely an area was to receive a high score for access to various technologies. Interpreting this in a positive manner, it means that in areas with a higher socioeconomic status, in which a mixture of racial and ethnic groups, particularly Asian
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
96
Concluding Remarks
Americans and African Americans, live together, access to technology is more prevalent. Interpreting it negatively, less diverse places, where lowincome Latinos are more likely to reside, are more likely to experience low levels of access to ICTs. This means that the path to digital equity is not that different from the path to social justice. Space has become the container of our social, cultural, and economic relationships, encapsulating our structural differences and inequities. To ameliorate these sociospatial injustices, we need to accept that “place” matters, places are marred by the nature of our past and present relationships and sociopolitical dynamics, and places reproduce these conditions due to years of disinvestment and neglect. From a policy perspective, it means that to improve the state of digital divide, we need to understand its social, cultural, economic, and demographic underpinnings; and, we need to construct our solutions in a systematic manner that dovetails social justice efforts, economic development plans, educational reforms, and all other progressive social policies. Digital divide does not occur in a vacuum, unaffected by social processes or a social context. In fact, it would be a great mistake to assume that digital divide is merely a technological problem. The geography of digital divide, as presented by this research, suggests that to produce sustainable solutions for the existing patterns of inequitable conditions, we must deal directly with the sociospatial contexts that produce them. Without changing these contexts, a lasting change cannot occur. For this reason, and based on the findings of this research, we offered a particular methodology and
a set of recommendations for to dealing with the problem of digital divide in the state. In summary, these are as follows: I.
Form public-private partnerships to assess
the actual patterns of access to technology, using subscription and technology application data. II. Identify and prioritize areas for short-term and long-term policy interventions. A. For this study, we identified two groups of census tracts: 1) One group houses a population, whose access to ICTs is only slightly below the state average (see Figure 59). For these tracts, we recommend the following strategies: – Address infrastructural inequities to assure high-speed connectivity – Ensure that access is not hindered by cost – Provide educational resources regarding the use and benefits of the Internet – Expand subscription and use of cell (through a public-private partnership that brings about less costly services and wider geographic coverage) 2) The second group of tracts housed a population with some of the lowest levels of access to the ICTs. For these tracts, we recommend strategies that focus on expanding the existing physical infrastructure, access to services, and enhancing the social capital of
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
97
Concluding Remarks
the affected communities. This means that in
the level of access and usage in the most severely
addition to the four suggested strategies for the
disconnected places, the strategy needs to move
first group, assistance of nonprofit and grass-
beyond simply making broadband and various ICTs
root groups has to be sought, in order to create a
available. We need to prepare, improve, and cultivate
wider diffusion of available technologies and to
the conditions that make technological products and
offer the needed education to improve their use.
services relevant to the life of those who have been left behind in every phase of progress and development.
We believe that a place-based approach with an
For that reason, we believe that digital equity needs
eye on social, cultural, economic, racial, and ethnic
to be made a logical and articulated component of
indicators can provide the best and most measurable
community and economic development efforts in the
results in overcoming the current patterns of digital
least connected places. It is through the convergence
divide. For that to occur, areas with minimal
of these policy arenas that we can create the conditions
connection need to receive a boost in their digital
that will lead to an improved quality of life for all
infrastructure, while residents are provided with
residents, enriched with sustainable use of ICTs and
economically feasible services. However, to improve
the benefits they can provide.
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
98
Notes
1 This study was published in 2006 and can be found at:
4 This excludes three cell phone variables: having 3+,
http://www.patbrowninstitute.org/publications/
4+, and 5+ cell phones. Since there is a variable for
documents/CTF_Report.pdf
households with 2+ phones, these data are already
2 Table 10, which appears across three pages,
captured by others. By excluding these variables,
provides correlation values between technology
we are making sure that cell phone ownership is
variables, socioeconomic indexes, and racial and
not having an undue influence on the outcome of
ethnic variables. To improve the readability of
our index calculation.
the results, the variables that appear in columns
5 Each census tract received a value of three for being
have been grouped so that content indicators and
one standard deviation above the mean, a two
race and ethnicity variables appear together on
for being within one standard deviation from the
a single page. Furthermore, since variables are
mean, and one for being more than one standard
duplicated in the table (in a diagonal manner),
deviation below the mean. In cases where being
it is unnecessary to show the entire table. In the
one standard deviation above the mean meant
triangular upper half, where correlation values
that a census tract was worse off (i.e., number of
appear, blank cells contained values that were not
households with dial-up Internet or zero wirelines),
statistically significant and, hence, were removed
the order was reversed.
from the table. 3 Note that for the purpose of this analysis, a larger number of variables was used. These additional variables provide more frequency information for
6 Actually, 4,856 census tracts (or 68.9% of all tracts) were in this range. 7 Due to its size, the Appendix is made available as a supplemental CD/file.
access to particular technologies (e.g., desktops, laptops, and cell phones).
DECEMBER 2008 • IN SEARCH OF DIGITAL EQUITY: ASSESSING THE GEOGRAPHY OF DIGITAL DIVIDE IN CALIFORNIA
99