Voting error in the 2004 elections

by Eszter Hargittai on December 11, 2004

A friend of mine, Philip Howard, has been taking a very innovative approach to teaching his class on Communication Technology and Politics at the University of Washington this Fall. He and his students have been collecting data about the use of communication technologies in the elections and writing reports about their findings.

The team has released reports on topics from the legalities of voteswapping to the political uses of podcasting. The latest article looks at voting error due to technological errors, residual votes and incident reports. They have collected data on these for all states for the presidential, the gubernatorial and the senate races. They weight the incident-report data by total voting population, eligible voter population and registered voter population. They find that in some cases – see state specifics in the report by type of error – the margin of error was greater than the margin of victory.

What a great way to get students involved, to teach them important skills and to contribute helpful information to the public. They make their data available for those interested in the details. You can download spreadsheets with information off their site. They also offer an extensive list of resources including a pointers to academic literature from the past twelve years on technologies and campaigns.

UPDATE: I should have mentioned that they are posting reports now as white papers and are eager to receive feedback. It looks like they will continue to analyze the data and welcome suggestions.



pedro 12.11.04 at 1:42 am

wow, one degree of separation, Eszter. Philip and I met a few years ago in Chicago. Really nice guy.


Pete 12.13.04 at 4:44 am

isn’t it sort of important to mention that they are calculating error rates using the 2000 MIT/CalTech numbers associated with the various voting techniques (“If we assume that the same machines had the same error rates in 2004 as they did in 2000…”)? It seems to me that 4 years is a long time technologically speaking…
I agree it seems like a great way to keep students involved, but I guess I will provide feedback to them that although they have carefully described the assumptions made in their methodology, these numbers may be mistreated as more fact than guesstimate as the data gets picked up.

Comments on this entry are closed.