Berkeley - Electronic and optical scan voting
systems did the best job of recording and tabulating votes
cast in the 2000 presidential election, while punch card ballots
performed the worst, according to a new University of California,
The finding, released today (Monday, Oct. 1) by the campus's
Survey Research Center and its Institute of Governmental
Studies, found that, among the nation's 100 largest voting
jurisdictions, which served 40 million voters in the 2000
election, electronic and optical scan machines outperformed
all other machines, producing fewer overcounted or undercounted
The report, "Counting all the Votes: the Performance of
Voting Technology in the United States," concludes that
"concerted efforts should be made to move away from Votomatic-style
punch card systems to other, more appropriate ones. In most
cases this will mean either optical scan or electronic systems."
Henry Brady, a UC Berkeley professor of political science
and public policy, directed the study, which was launched
in the wake of calls for a recount of Florida votes cast
during the Bush-Gore presidential contest. Many Florida
counties used punch-card style systems in 2000.
Brady and a team of researchers sorted through presidential
election results from 38 states and 2,219 counties, rejecting
unreliable data and cross-checking results using sophisticated
"Optical scan and electronic systems perform better than
other systems across all U.S. counties and especially in
large counties," said Brady, "and the newer versions of
both systems give voters feedback that 'checks their work'
before they submit their ballot."
Brady also noted that touch-screen electronic voting systems
allow individuals to select a ballot translated into a foreign
language. Some of these systems also allow disabled persons
to vote by listening to recordings.
The UC Berkeley study differs from a Cal Tech/MIT Voting
Technology Project report issued in July 2001 that concluded
that electronic systems, though improving, might be harder
to use than other voting systems. The UC Berkeley study
relies on the most recent data and finds no evidence that
electronic systems are harder to use.
In addition to considering voting system performance across
all U.S. election jurisdictions, the report also addresses
their performance in the nation's 100 largest counties.
Optical scan and electronic systems, which produced an overcount
or undercount of votes less than 1 percent of the time,
received the top rating of "good." Lever machines, which
had such problems 1 to 2 percent of the time, were rated
"adequate." Punch card ballots, with such problems 2 to
3 percent of the time, were rated "worrying."
UC Berkeley researchers noted, however, that all systems
will perform less effectively when they are not properly
implemented by election officials or precinct workers at
the local level. In addition, some voting systems may work
well in certain counties due to that population's long history
using a particular system, due to demographic characteristics
of a particular county, or due to the particular variety
of the system used.
Electronic systems, the newest of voting systems, are now
used in more than 300 counties across the country. Voters
push a button or touch the face of an ATM-type machine and
choose a candidate.
Optical scan systems, in which an infrared or other scanner
records the markings on a paper ballot, are used in more
than 1,300 counties and more than 40 states.
Mechanical voting machines such as lever systems no longer
are manufactured but are heavily used in seven states -
most of them on the East Coast. These machines, developed
in the 1890s, involve large displays of the entire ballot
and small levers next to each choice. The voter flips a
small lever to choose a candidate and, once finished voting,
pulls a large lever, which counts each vote.
Paper ballots, the oldest and simplest method, are used
in about 300 mostly small, rural counties and townships.
A voter simply places a check next to the name of a preferred
candidate on a piece of paper. All votes are counted by
In addition to Brady, the study was co-authored by UC Berkeley
Institute of Governmental Studies researchers Justin Buchler,
Matt Jarvis, and John McNulty. It was funded, in part, by
Sequoia Voting Systems, an Oakland-based manufacturer of
punch card, optical scan and electronic voting systems.