A. Introduction

This section provides an in-depth examination of Internet (or Net) access and usage. In contrast to Part I, which looks at household access, Part II focuses primarily on trends among individuals. This is a new analysis in the Falling Through the Net series, which we have included for at least two related reasons. First, given the Internet's robust growth, the Internet has assumed an importance in Americans' everyday lives that compels us to probe more deeply into this new medium. Second, a sufficient number of people are now online, enabling meaningful surveying and statistically significant analyses.

Many of the findings in this new section will be useful to the stakeholders in the new Information Age. They may be particularly useful for policymakers concerned with ensuring affordable access to the Internet. Key findings include:

The Internet is a nascent, rapidly diffusing technology that promises to become the economic underpinning for all successful countries in the new global economy. Understanding who is connected to the Net, and how it is being used, is critical to the development of sound policies in this area. In the sections that follow, we examine both Internet access and its usage through a variety of measurements.

B. Where People Access the Internet

1. General Access to the Internet

Many people have the option of accessing the Internet from more than one place. A person can connect from home; select another site, such as at work, a school, library, or community center; or use a combination of the two. Among all Americans, 22.2% currently use the Internet at home, and 17.0% use it at some site outside the home. Almost one-third (32.7%) use the Internet somewhere, while approximately two-thirds (67.3%) do not use it at all. (Chart II-1)

Demographic and Geographic Variables. Levels of Internet access differ dramatically among different groups and geographic areas. A cross-sectional analysis based on the seven variables set forth below illustrates this theme. Where a given variable is cross-tabulated with the degree of urbanization (such as rural or central city), significant differentials also typically occur.

This discussion has attempted to present a broad assessment of which Americans access the Internet. In the section that follows, we narrow the focus to patterns of access for those who go online at sites other than home.

2. Points of Access Outside the Home

Of those people who go online outside the home, there are significant differences as to where people access the Internet. Certain demographic groups are particularly likely to have access at work. Those same groups are far less likely to access the Internet at schools, public libraries, or through someone else's computer. The converse is also true. Those groups with lower access rates at work or at home are far more likely to use the Internet at a public place, such as a school, library, or a community center. These findings suggest that Americans without ready access to the Internet (at home or at work) are making use of public resources.

Access At Work

By far the most popular place to access the Internet outside the home is at work. Of those who access the Internet outside of home, more than half (56.3%) of Americans access it at work,

particularly in urban (58.8%) and central city (58.7%) areas. (Chart II-15) Certain groups have particularly high rates of access at work. Those with college or advanced degrees are the most likely to have access at work -- about ten times more likely than those with only some high school education (87.2% versus 8.7%). (Chart II-20) Similarly, those earning at least $75,000 are nearly six times more likely to have work access than those earning less than $5,000 (72.9% versus 12.3%). (Chart II-16) Families without children and non-family households who access the Internet externally also rank high in accessing the Internet from work -- 69.4% and 68.4%, respectively (compared to 32.8% for male-headed households, 29.0% for female-headed households, and 50.4% for dual-parent households). (Chart II-21)

There are also notable disparities based on race. Whites and Asians/Pacific Islanders who use the Net outside the home are more likely to be connected at work (58.8% and 56.6%, respectively), compared to American Indians/Eskimos/Aleuts, Hispanics, and Blacks (34.8%, 39.1%, and 49.3%, respectively). (Chart II-17) Men are also more likely than women to access the Internet at work (58.7% compared to 53.8%). (Chart II-22)

Not surprisingly, those demographic groups with higher access from work tend to be the same groups that have higher rates of access at home. (See discussion above.) They are also the same groups that exhibit lower usage rates from public access points, such as schools, libraries, or community centers.

Access At K-12 Schools

The second most frequently used access point is the Kindergarten through 12th grade (K-12) school, particularly in rural areas (30.0%). (Chart II-15) These figures may be higher than other public access points because they include school-aged children, many of whom use the Internet at school. The inclusion of children who access the Internet at school could account, in part, for the particularly high levels of usage among those with lower education levels, lower incomes, and those "not in the labor force." (Charts II-20, II-16, and II-23)

Nevertheless, certain groups who access the Internet outside the home are particularly likely to go online at K-12 schools. American Indians/Eskimos/Aleuts and Hispanics are particularly high users (36.5% and 35.1%, respectively), compared to Asians/Pacific Islanders (19.4%), Whites (20.0%), and Blacks (26.6%). (Chart II-17) Hispanics and American Indians/Eskimos/Aleuts are especially likely to use schools for access if they live in rural areas (e.g., 46.6% for Hispanics). (Chart II-18) Single-parent households are also far more likely to use K-12 schools (43.6% for female-headed households, 38.5% for male-headed households), than are dual parent households (33.7%), families without children (5.8%), or non-family households (4.3%). (Chart II-21)

Access At Public Libraries and Community Centers

Many Americans who obtain Internet access outside the home rely on such places as public libraries (8.2%) and community centers (0.6%). (Chart II-15)(2) Public libraries, in particular, are used by certain groups with some regularity. Unemployed persons who access the Internet outside their homes are nearly three times more likely to use public libraries as the national average (21.9% versus 8.2%). (Chart II-15, II-23) Those Americans who are "not in the labor force," such as retirees or homemakers, are twice as likely to use the public libraries for access (16.1%). Both groups are even more likely to use public libraries in urban, as opposed to central city or rural, areas (22.8% and 17.9%, respectively).

Other groups that also use public libraries more frequently include those earning less than $25,000 (Chart II-16), those with less than a high-school education (Chart II-20), those in female-headed households (Chart II-21), and American Indians/Eskimos/Aleuts, Blacks, and Hispanics. (Chart II-17) Of these groups, American Indians/Eskimos/Aleuts are especially likely to use libraries in urban areas (17.3%), while Blacks are more likely to use libraries in rural areas (16.3%). Those in female-headed households are also more likely to gain Internet access in libraries in central cities (16.4%).

Using a logistic regression analysis, we also compared the likelihood of a group's using public libraries or community centers for online access. Our analysis pertained only to those people who reported usage of the Internet from outside the home.(3) This regression analysis revealed the additional interesting comparisons:

These findings support our general conclusion that those who are less likely to have Internet access at home or work (e.g., those earning less than $20,000, certain minorities, and those without a college degree) are relying on the resources of public facilities.

C. How Households Access the Internet

1. Type of Internet Access Device

The 1998 Census survey also asked how people access the Internet from home. Personal computers with modem capability have been historically, and are currently, the mode of choice for Internet access. Among those households that have a computer or a WebTV®, 61.0% connect to the Internet via PCs, 1% obtain access via WebTV®, and 38.0% do not use the Internet at all. (Chart II-24a)(4) Already there are signs that alternative modes -- for example, Internet phones -- will soon become available for browsing the web or e-mailing.

Whether or not a household with a PC is also an Internet user depends on various demographic characteristics. For instance, usage varies by race/origin, ranging from high levels of use by Asians/Pacific Islanders (65.0%) and Whites (63.5%) to lower usage levels by American Indians/Eskimos/Aleuts (53.2%) to Hispanics (48.7%) and Blacks (47.4%). Breakdowns by type of household also reveal differing usage rates, ranging from married couples with children (63.4%) to single-parent households with children, either male-headed (55.2%) or female-headed (46.4%) families.

2. Type of Internet Service Provider

An important part of the linkage in being able to go online is to connect to an Internet Service Provider (ISP). Currently there is a tremendous variation in market share among the types of ISPs to whom households choose to subscribe. (Chart II-24b) National service providers have captured the bulk (69.0%) of the market. Local phone companies rank second (14.0%), followed by long distance companies (4.0%), cable TV systems (2.0%), and wireless firms (1.0%). An "other" category, comprised of types of Internet Service Providers (ISPs) that are too small to be broken out, accounts for the rest of the total (10.0%).

3. Why Households with Computers Have Never Had Internet Access

Multiple reasons exist as to why households with computers at home have never used the Internet there.(5) (Chart II-25) In the 1998 CPS supplement survey, the most common response given was that the household's occupants "don't want" such access (25.7%). The second major reason among respondents concerns "cost" (16.8%), which is further disaggregated into the monthly service charge (9.7%), the need to make a toll call in order to reach one's ISP (4.8%), and other costs (2.3%). Following cost are such categories as "can use elsewhere" (9.6%), "no time" (8.7%), computer not capable"(8.3%), "future access planned" (but none at home currently) (7.5%), "concern with children" (6.0%), and "not useful" (5.6%). Some people gave "not user friendly" (2.7%) and "problem with service provider" (1.3 %) as reasons for not having Internet access at home. Myriad other responses whose percentages are quite small appear under the headings "other cost" and "other". Id.

These profiles were gleaned from an analysis of Internet "non-users" by demographic variables.

In sum, the most important reasons why certain households have never used the Internet is that they "don't want it" or it is too expensive. Although the former is the more important reason overall, the cost factor dominates among low-income groups, Hispanics, single-parent families, the youngest householders, and the unemployed. Policymakers should therefore consider the role of cost as a deterrent to expanding online access.

4. Why Households with Computers Have Discontinued Internet Use

Internet churn -- the incidence of households discontinuing Internet use -- represents another area that policymakers have begun to examine. In the 1998 CPS supplement survey, respondents identified "cost, too expensive" (15.0%) as the most important reason for dropping off the Internet. (Chart II-31) The second most compelling reason is " no longer owns computer" (14.0%), followed by "not enough time to use it" (10.0%), followed by "can use elsewhere" (9.0%), "don't want it" (7.0%), and "moved" (7.0%). "Other" accounts for the rest (17%). Id.

D. How People Use the Internet

The 1998 data also reveal that demographic characteristics not only determine whether and where one uses the Internet, but how a person uses the Internet. Income, education, race, and gender, among other characteristics, strongly influence what a person does online. They can affect not only the types of Internet activities and searches, but even the nature of a person's e-mail. This is true regarding both Internet use at home and Internet use outside the home.

Most significantly, people are using the Internet to improve and advance their current status. For example, those who are unemployed are using the Internet to find jobs, and those with lower incomes and many minorities are using the Internet to take courses or do school research. The data therefore show the Internet is becoming not only a source of information, communication, and entertainment, but also a tool that can help users help themselves.

1. Internet Use at Home

E-mail is one Internet use, however, that transcends all demographic and geographic boundaries. E-mail is clearly the "killer application" of the Internet for the 1990s. Of Americans who use the Internet, nearly 80 percent (77.9%) use it to send e-mail, and over half (53.6%) of people with Internet access outside the home use the Internet for e-mailing. (Charts II-32, II-37) The numbers are consistently high, regardless of income, race, gender, age, or any other characteristic.

Apart from e-mail, however, there are distinct differences in the ways people use the Internet at home. Using the Internet for "job-related tasks" is far more common, for example, for those at incomes higher than $25,000 and for those at higher education levels. (Charts II-33, II-35) Job-related uses are also higher for men (38.7%) than for women (26.4%). (Chart II-49)

Taking courses or finding jobs are important activities, on the other hand, among minorities, the young, Americans with lower incomes, and the unemployed. At home, minorities, for example, are taking courses or conducting school research online at rates higher than the national average (36.1%) or than Whites (at 35.3%). Blacks and Hispanics rank highest at 43.5%, followed by American Indians/Eskimos/Aleuts (42.9%), and Asians/Pacific Islanders (41.1%). (Chart II-33, II-34) Minorities are much more likely than Whites to use the Internet to search for jobs (19.1% for Blacks, 18.1% for Asian/Pacific Islanders, 17.4% for Hispanics, compared to 13.2% for Whites). Id.

Similarly, well over half of unemployed persons using the Internet at home are searching for jobs online (53.9%). (Chart II-36) They are also using the Internet to take courses at far higher rates than individuals who are who are employed (40.1% versus 26.8%). Id. Taking courses and searching for jobs also rank high among the young, those with lower education levels, and those with lower incomes. These groups probably have relatively higher rates, however, because they include a large number of students.

2. Internet Use Outside the Home

Americans who use the Internet tend to use it for somewhat different purposes when outside the home than at home. Nationwide, people are far less likely to use the Internet outside the home to e-mail (53.6% versus 77.9% at home), check the news (23.3% compared to 45.9%), search for jobs (8.5% outside home, 13.8% at home), or pay bills or shop (7.5% , compared to 24.6% at home). (Charts II-34, II-37) On the other hand -- not surprisingly, given the numbers who access the Internet at work -- conducting job-related tasks online is much more likely outside the home than at home (44.6%, compared to 29.0%). (Chart II-37) Again, this use is far more common among people of higher incomes and education levels. (Charts II-38, II-40)

The one activity that is equally popular for both at-home or outside-the-home Internet users is pursuing online courses and school research (38.8% outside home, compared to 36.1% at home). Again, minorities are more likely users and pursue online courses and school research at even higher rates outside the home (50.3% for Hispanics, 47.0% for American Indians/Eskimos/Aleuts, and 46.3% for Blacks). (Charts II-32, II-37, II-39) People in rural areas also taking courses at higher rates (45.4%), compared to those in central cities (36.8%) or urban areas (36.9%). Finally, households with children are also more likely to take online courses than those without children, with female-headed households being the leading users (57.3%). (Chart II-41)

While Americans as a whole are unlikely to use the Internet outside the home to search for jobs, there is a particularly notable exception -- the unemployed. As a whole, this group is more than three times more likely to use the Internet for job searching than the national average (29.1% v. 8.5%), and is more than four times more likely (34.9%) to do so in central cities. (Charts II-32, II-36) This finding is especially significant, given that this group does not have the option to access the Internet at work and must rely on other access points, such as public libraries.

3. E-mail Use

As noted above, most Americans with Internet access are using the Internet to send e-mail. The nature of those e-mails, however, can again vary widely by demographic characteristics: certain people are more likely to use it for educational purposes; others are more likely to use it to buy goods; still others are more likely to use it for job-related purposes.

Nevertheless, there is one constant: almost all Americans who use e-mail at home are using it to communicate with family and friends (93.6%), and a significant percentage (59.7%) are using it outside the home for the same purpose. (Charts II-43, II-50) At home, the same high usage rate generally holds true across all income levels, education levels, races, ages, genders, and locations. (Charts II-43-49) For Americans who access the Internet outside the home, the rate of sending e-mail to family and friends declines as income rises: more than 65% of those earning less than $25,000 use e-mail for this purpose, while that rate declines at higher income levels. (Chart II-51)

E-mailing for job-related purposes is also popular. It is obviously common for outside-the-home users (70.6%), but is also frequent for at-home users as well (32.8%). (Charts II-50, II-43) As with job-related Internet use, this type of e-mailing occurs more often at higher income levels (Charts II-44, II-51), and higher education levels. (Charts II-46, II-53) It also occurs at higher rates for men than women, both at home (38.7% versus 26.4%) and outside the home (74.6% versus 66.2%). (Charts II-49, II-56)

Other distinctions have also emerged between the way men and women use e-mail. Men are more likely to e-mail regarding hobbies or special interests -- whether at home (34.5% versus 28.5%) or outside the home (13.0% versus 9.9%). (Charts II-49, II-56) They are also using e-mail for commercial uses more than women: 13.7% compared to 10.0% for at-home uses, and 11.1% compared to 7.6% for outside-the-home uses. Id.

4. Online Confidentiality Concerns

The frequency with which Americans purchase goods online, send e-mail, conduct research, or undertake any other Internet activity may be affected by their concerns about confidentiality online. As a whole, Americans are very wary about confidentiality on the Net. When asked about these concerns, by far the highest percentage of those polled (40.0%) stated that they were "very concerned" about confidentiality on the Internet. (Chart II-57) Equal numbers (24.0%) responded that they were "somewhat concerned" or "not concerned." Id.

The level of concern varies by race/ethnic origin, as well as other demographic characteristics. Indeed, Whites (40.5%) and Blacks (39.1%) are "very concerned," while Asians/Pacific Islanders are the least likely to be "very concerned" (31.5%), followed by Hispanics (34.5%). (Chart II-58) Of all groups, Hispanics are most likely (31.5%) to say they are "not concerned" at all about confidentiality. Id.

The level of education and income are also factors in levels of concern about online confidentiality. Those with a high school diploma or some college education are the most likely to be "very concerned" (41.8% and 43.5%, respectively). (Chart II-59) Similarly, those earning between $25,000 and $50,000 expressed high levels of concern (above 44.0%).

To sum up, the groups that express more serious concern are the same groups that are using the Internet more frequently. However, those using the Internet the most -- those with college degrees, those earning $75,000 or more, and Asians/Pacific Islanders -- have expressed slightly lower degrees of concern. Whatever the reasons for these patterns, it appears that concern arises among those with mid-level usage rates, while there is a higher comfort level among those using the Internet most often.

E. Conclusion

For the first time in our Falling Through the Net series, the Commerce Department has collected and analyzed wide-ranging data with respect to Internet usage by Americans. These statistics will advance our knowledge base with respect to whether, where, and how people in this country are making use of the Internet.

These data provide concrete evidence that the Internet is being used by an increasing number of Americans. More than one-third of Americans go online from any point, either at home or outside the home. Approximately one-quarter access the Internet at home. For those households with a computer, approximately two-thirds have Internet access. Households that do not have Internet access most frequently explain that they either do not want it or that it is too expensive; for those households that have dropped off the Net, cost is the most important reason.

While Americans are becoming increasingly connected, there are still significant discrepancies in access: Blacks and Hispanics, for example, are less connected anywhere than Whites are at home. Those groups with lower access rates at work or home are much more likely to use the Internet at a public place such as a school, library, or community center. They are also more likely to use the Internet to take courses or to conduct job searches than other groups. These and other findings -- present and future -- will provide an important factual foundation for the sound policymaking needed to ensure socioeconomic success in the Information Age.

<-- Part I: Household Access
Table of Contents

Part III: Challenges Ahead -->

1. For the first time in these studies, NTIA sought information regarding means of accessing the Internet other than personal computers. The Census CPS survey asked respondents whether they owned a WebTV®, which is the most widely used system for accessing the Internet through television sets. A WebTV® unit connects to a television set, much like a VCR, and to a telephone line to send and receive data. This data is then displayed on the television, rather than a computer monitor. WebTV® Networks, Inc. is a subsidiary of the Microsoft Corporation. We note that WebTV® is not the only vendor of non-PC based access to the Internet.

2. Community centers are generally a new and growing point of access to the Internet. We are unable to conduct an independent, meaningful analysis of those Americans using community centers, however, because the numbers involved generally fall below statistical levels of significance. Nevertheless, the Census Bureau data suggest that -- as with public libraries -- community centers are used more often by low-income persons, Blacks, American Indians/Eskimos/Aleuts, and the unemployed, than by other groups.

3. The logistic regression analysis was conducted according to the model set forth in the methodology section of the report. Logistic regression analysis requires a base reference group for purposes of comparison. The base group in this analysis is higher income, White-non Hispanic, suburban, computer owner, and college educated. The dependent variable indicates whether someone has used the Internet from a public library or community center. The number of respondents who used the Internet at a community center was relatively small by itself. By combining them with public libraries, community centers could be included in the analysis.

The sample was broken into two groups by income, those making above $20,000 a year and those making below. The $20,000 threshold isolates households in poverty from those with higher incomes. The race variables were disaggregated to separate Hispanics from all racial groups. Hispanics were then added into the regression. In the model "Minorities" are considered to be all non-Hispanics not included in Black or White racial groups.

The sample was also broken down into two segments: by suburban and non-suburban, with non-suburban combining non-metropolitan (rural) and central cities. Education was broken down into two categories, one for those who graduated from college and another for those who did not. The "computer at home" variable indicates whether or not households have a computer at home.

The fit of the model is significant with the Hosmer and Lemeshow Goodness-of-fit test statistic of 7.6733 with 7 degrees of freedom (p = 0.3623). The model also shows relatively low collinearity due to the breakdown of the binary variables.

4. Because of the small percentage of households with WebTV®, the term "PCs" includes WebTVs® for purposes of this section.

5. The CPS supplement questionnaire asked respondents to provide the main reason for non-use

<-- Part I: Household Access
Table of Contents

Part III: Challenges Ahead -->