News organizations retain NORC to survey election ballotsA group of the nations largest news organizations has retained the National Opinion Research Center to conduct an in-depth inventory of uncounted ballots from the presidential race in Florida.
The group, which includes The New York Times, The Wall Street Journal, The Washington Post, Tribune Publishing, CNN, the Associated Press, the St. Petersburg Times and the Palm Beach Post, plans to produce a database that will describe in detail the 180,000 Florida ballots that did not register a vote on machine countsincluding both undervotes (no vote for president recorded) and overvotes (two or more votes for president).
NORC teams will view the ballots and abstract information about the marks, or lack of marks, on each. NORC will not attempt to assess whether any particular ballot contains a vote but simply describe the marks.
The intention of the project from NORCs viewpoint is not to identify who got more votes, but rather to examine closely the variabilities in the voting systems themselves, said Kirk Wolter, NORCs Senior Vice President for Statistics and Methodology and Professor in Statistics. This information will be helpful to state and local governments in selecting balloting systems that count ballots with a high degree of reliability, he said.
The project will produce the definitive historical archive of Florida ballots that did not register a vote, using the highest standards of scientific accuracy.
NORC will screen and train teams of three independent coders to classify each ballot into categories based on the varying interpretations canvassing boards have confronted in manual recounts of machine-readable ballots. For instance, on punch-card ballots, NORCs three coders will record independently whether a ballots key chads were missing, hanging by one or more corners, or dimpled; whether light is visible around the dimple; and whether the remainder of the ballot is similarly marked.
The use of three coders insures the highest level of accuracy and reliability in abstracting information from ballots. It makes it possible to guard against partisan biases and to judge the difficulty that canvassing boards encounter in trying to assess voter intent on machine ballots. One coder may view over-voted ballots, if an initial test using three coders shows they are easier to read than the under-voted ballots.
The pool has requested elec-tion officials in all 67 counties to segregate all ballots on which machines could find no vote, or more than one vote for president. Most of those ballots were not included in the certified vote totals.
Each news organization will use the raw data to produce its own analyses and stories about what the uncounted ballots reveal. The data also will be released publicly within a few days after completion to allow academics, or any other interested parties, a chance to assess it as well.
The pool has invited other news organizations willing to share costs to participate. Organizations sharing costs will get the first crack at viewing and using the database. Plans call for the database to be completed within eight to 10 weeks.