About that house of yours -- does it have a toilet and a bathtub or shower?
Does it get natural gas from an underground pipe?
And if it's a mobile home, does it have an installment loan?
These are not trivial questions to the federal government. They're among
the arcane and archaic topics mandated by law to be on the long form of the
U.S. Census that's headed for one in every six American households in the
But when it comes to the electronic plumbing that's reshaping every
facet of American life from education to employment, nobody thought to ask
how many Americans have access to the Internet.
Not one of the 53 questions on the upcoming long form will even remotely
concern digital technology.
Home acreage, yes. Home computers? No.
Mobile homes, yes. Modems? No.
''This is a colossal mistake that must be rectified immediately,'' said
Vanderbilt University Professor Donna Hoffman, one of the country's top
Internet demography experts. ''The bottom line is that if not even a single
question about Internet access or use makes it into the 2000 census, then
someone is asleep at the wheel.''
The decennial census, which will cost an estimated $4 billion to $5
billion, is the government's most significant, detailed and comprehensive
statistical road map. Census 2000 is being developed at a time when
virtually every piece of the government, including the Census Bureau, is
shifting information and services to cyberspace.
The questions on the long form were the result of six years of
discussions and review by federal agencies, the Clinton administration and
Congress. The Census Bureau, working with the Office of Management and
Budget, began soliciting comment from all federal agencies in December
1992. The House and Senate also held several years of hearings. On April 1,
the final draft of the long form was submitted to the House Subcommittee on
the Census and the Senate Committee on Governmental Affairs. While neither
committee is planning any further action on the long form, the last-minute
addition of an Internet question is still a possibility -- albeit a faint
Staffers for both committees say that nowhere in the process can they
recall anyone lobbying for including an Internet question. When asked
earlier this week, many top technology policy-makers within the Clinton
administration -- from the White House to the Commerce Department -- said
privately that they would support the idea of an Internet-access question.
The data from such a question could represent a statistical platform to
shore up the telecom debates the government will tackle in the next few
years. Still, no one within the administration has come forward to advocate
including a Net-access question.
According to the Census Bureau, every question on the form -- from trolley
commuters to telephones -- is mandated in one of three ways. Some
legislation explicitly demands the use of the decennial census to gather
information on a subject. Other legislation requires some statistics that
have historically been harvested from the census or can come only from the
census. Finally, a question can be required by a federal agency that needs
the data to plan, implement or evaluate programs.
Both Congress and the Clinton administration have been planning and
implementing various Net initiatives for several years. Those projects
range from Internet-access subsidies for schools and libraries to junk
e-mail controls and Internet domain-name policies. Those initiatives have
been proceeding despite the fact that the government has only a hazy idea
who the Internet constituency really is.
Until now, the best estimates of digital haves and have-nots have come
from surveys done by private marketing and research firms. While there is
no standard definition of what constitutes Internet use, current estimates
put the number of U.S. users between 41.5 million and 57 million.
Only one major study
The federal government's only major study on digital haves and
have-nots, ''Falling Through the Net,'' was executed in 1994, before the
World Wide Web emerged as a mass medium. That study, done for the National
Telecommunications and Information Administration (NTIA) by the Census
Bureau, polled just 54,000 households and concerned home computer ownership
rather than Internet access. The results showed direct correlations between
high income and computer ownership, and high education and computer
In July, the NTIA will revisit that subject with a study that looks at
trends in technology adoption based on data gathered in 1994 and 1997. The
update to the original ''Falling Through the Net,'' based on a sample of
48,000 households (and containing more specific Internet-access questions),
is not scheduled to be released until mid-1999.
Aside from the NTIA surveys, the only federal statistical tool for
gathering significant technology data is the Current Population Survey, a
monthly study of 52,000 households done by the Census Bureau for the Bureau
of Labor Statistics. That survey, which exists primarily to report
employment data, has carried questions on personal computer ownership in
the past. A Census Bureau staff member said the CPS study would be the most
likely place for an Internet-access question to be broached.
The executive ultimately charged with direction of the census,
Undersecretary of Commerce Robert Shapiro, says that while an
Internet-access question on Census 2000 may be useful for crafting policy,
it's not the only way to get an understanding of haves and have-nots.
''There's a lot of very serious policy-making done on the basis of samples,''
he said. ''Those may not be as comprehensive as the decennial, but there
are many programs that depend on special surveys. (Census 2000) is not the
absolute last opportunity the government will have to ask this question.''
The decennial census has always been the premier vehicle used by
government to build the statistical foundations for its most significant
public policy debates. Historically, it has also served to gauge how fast
new technology is entering the mainstream.
''It does mirror the technological changes of the day. But most often it
follows technological change,'' said Dave Pemberton, the Census Bureau's
historian for the decennial census. ''It's after people begin to adopt
technologies that the government needs substantial data to answer questions
about that technology.''
According to Pemberton, the first question about radio appeared in 1930,
the first about refrigeration in 1940 and the first about television in
Beyond the fact that the Clinton administration and Congress failed to
direct the Census Bureau to add an Internet-access question, there may be a
secondary reason for its omission: The bureau has been under pressure to
compress both the short form that reaches each household and the long form
that reaches one in six. The long form, which the bureau estimates will
take the average household 38 minutes to complete, has been increasingly
viewed as an unduly onerous burden on its recipients. The bureau's chief
concern is that a longer list of questions will lower the rate of accurate
returns on the form.
For Census 2000, the short form has shrunk from 12 subject areas to
seven and will contain just seven individual questions. It is the shortest
questionnaire since 1820. The long form will contain an additional 46
questions, five fewer than in 1990. The only new subject to be added to the
upcoming long form concerns grandparents as care-givers.
Hoffman maintains that Census 2000 represents a unique opportunity and
that waiting until 2010 or beyond to get a comprehensive look at the
Internet constituency would be a grave mistake. ''Failing to address the
issue of Internet use represents what hindsight will almost certainly show
to be an enormous blunder,'' she said. ''It would be mind-boggling,
considering what's happened in the past three or four years since the Net
went commercial, that the census wouldn't address this issue. It's hard to
Not everyone in the Internet demographics field agrees. Nick Donatiello,
president and CEO of the San Francisco research firm Odyssey, said the
federal government was wise to avoid asking a question that much of the
populace may not be able to answer accurately.
''Getting a correct answer is much more complicated than it appears on
the surface,'' he said. ''I think the chances of getting bad data are
highly likely with such a question. In order to get it right, you'd have to
ask an entire battery of questions.''
When the White House needed up-to-date Internet statistics for a recent
presidential address, it turned not to the Census Bureau but to
Donatiello's firm, which publishes Internet demographics studies twice a
Up to Congress
While Congress traditionally gives the Census Bureau a free hand to
craft the form as the bureau sees fit, the final decision ultimately rests
on Capitol Hill. Right now, with no further review planned in either the
House or Senate, sources in the administration and the Census Bureau say it
would take a last-minute directive to get an Internet access question on
''The Internet question is worth asking,'' said Shapiro, the
administration's top executive overseeing the census. '' It's getting very
late to add questions, but the door is not completely closed.''