-------------------------- eGroups Sponsor -------------------------~-~> <FONT COLOR="#000099">eGroups eLerts It's Easy. It's Fun. Best of All, it's Free! </FONT><A HREF="http://click.egroups.com/1/9698/2/_/1406/_/974343346/"><B>Click Here!</B></A> ---------------------------------------------------------------------_-> Please send as far and wide as possible. Thanks, Robert Sterling Editor, The Konformist http://www.konformist.com http://www.netrinsics.com/Duval/Duval.html Analysis of Precinct Return Data for Duval County, Florida Introduction The raw precinct return data for Duval County, Florida, for the general election of November 7, is available from the county web page for election results. This data is in a format which is not spreadsheet friendly. I reformatted the data into a standard CSV format file (which you can download here), and loaded it into a standard spreadsheet program. >From this point it is a straightforward matter to analyze the raw data. President and Senator ballot analysis Media reports on the Duval County returns have focused on two major issues: Predominantly Democratic precincts had a disproportionate number of ballots undervoted or overvoted for the presidential race. An unusually high number of ballots reflected a valid vote for senator, but not for president. The following two charts illustrate these two issues: (See URL above) In the charts above, each red diamond or grey triangle represents data for one precinct (out of 268) in Duval County. The red diamonds represent the percentage of ballots counted in each precinct that reflect a valid vote for president. The grey triangles represent the percentage of ballots counted in each precinct that reflect a valid vote for senator for the state of Florida. The first chart plots the data against the percentage of ballots in each precinct that reflect a valid vote for George Bush. One can reasonably infer that this percentage is higher for predominantly Republican precincts. The second chart plots the data against the percentage of ballots in each precinct that reflect a valid vote for Al Gore. One can reasonably infer that this percentage is higher for predominantly Democratic precincts. Interpretation These charts clearly illustrate a strong correlation between Democratic precincts and missing votes for a presidential ticket. Almost all precincts with more than 50% Al Gore votes have more than 15% missing presidential votes. Almost all precincts with more than 50% George Bush votes have less than 15% missing presidential votes. With respect to the senate votes, the charts demonstrate two saliant pieces of information: In the majority of all precincts, there were more valid votes cast for senator than for president. This suggests that the "two page" presidential ballot was problematic for voters across the political spectrum. The decline in valid presidential votes in predominantly democratic precincts was much sharper (from 95% to 75%) than the decline in valid senatorial votes (from 97% to 87%). It is very difficult to explain the high rate of non-valid presidential ballots in Democratic precincts purely in terms of voter confusion. Another interesting feature is the cluster of presidential ballot data that runs from 90%/90% to 65%/65% on the Gore chart. This line represents the "no Bush" limit, i.e., in a precinct where only 75% of the ballots hold a valid presidential vote, Gore can receive no more than 75% of votes cast. Therefore, there can be no precincts that fall to the right of this line. The fact that so many precincts are clustered near this line suggests that there are a sizeable number of precincts in Duvel County that are so strongly Democratic that nearly every valid vote cast will be a vote for the Democratic candidate. These "no Bush" precincts show, by far, the widest range of valid presidential ballots, ranging from 87% down to 69%, an 18% spread overall. This spread is very difficult to explain by way of any of the "usual" voting irregularities commonly discussed in the context of Florida elections, and definitely warrants closer scrutiny. Another View: Vote Distribution The following two charts answer a slightly different question: of all the ballots cast in any given precinct, how many of those ballots were marked for Bush, how many were marked for Gore, how many were marked for "Other", and how many ballots were missing a valid selection for president? These two charts group Duval County precincts according to the percentage of ballots cast for Bush; that is, precincts are grouped according to their Republican/Democratic balance. This chart shows the distribution of total ballots cast: (See URL above) This chart shows the distribution as a percentage of ballots cast: (See URL above) These charts clearly illustrate the strong bias for invalid presidential votes in Democratic precincts. Half of the "missing" votes for president in the county are located in precincts which account for only one quarter of the total Duval County vote: precincts where Bush received less than 50% of ballots cast. Digging Deeper: Regression Analysis One of the apparent features of the percentage distribution chart (above) is that there is no noticeable jump in the percentage of missing presidential votes from the Democratic to the Republican end of the scale. This suggests that the bias in the missing votes is not located in fixed number of particular precincts, but is rather related to the relative number of Democratic and Republican votes overall. This chart illustrates this relationship: (See URL above) In the chart above, all 268 precincts are arranged from left to right according to the number of presidential votes missing in that precinct. Each precinct has a red diamond, which indicates the number of ballots cast for George Bush in that precinct, and a grey triangle, which represents the number of ballots cast for Al Gore in that precinct. This chart shows a striking correlation between missing presidential votes and votes for for Al Gore (correlation coefficient of 0.79), and almost no correlation at all between missing presidential votes and votes for George Bush (correlation coefficient of -0.15). The blue line represents an ideal 1-to-4 ratio (the actual ratio of missing votes to Gore votes in Duval county overall is 1/4.01, or 24.95%). The Gore votes are clustered along this line. What this means, in practical terms, is that for every four ballots cast for Gore in Duval County, there is approximately one missing presidential vote. This ratio holds true regardless of whether the precinct is large, or small, and it holds true whether the precinct is mostly Republican or mostly Democratic. No such relationship exists for ballots cast for Bush. If the tabulating equipment used by Duval County miscounted one out of every five valid votes cast for Gore, but counted Bush votes within the normal range of accuracy for that technology, then this is what the data would look like. Or if, on average, one out of every five paper ballots cast for Gore countywide were defective, then this is also what the data would look like. There are other possible explanations, as well. For example, if all Democrats countywide relied on voting guides to vote, and one out of every five accidentally left theirs at home and didn't vote for president because of that, then this is what the data would look like. Or, if one out of every four Democrats countywide brought along another voter as part of a "get out the vote" effort, and none of these extra voters voted for president, then this is what the data would look like. Regardless, it is clear that the missing presidential ballots are tied to some unknown factor related to Gore votes, and are almost completely independent of factors related to Bush votes. If you are interested in a free subscription to The Konformist Newswire, please visit http://www.eGroups.com/list/konformist/ and sign up. Or, e-mail [EMAIL PROTECTED] with the subject: "I NEED 2 KONFORM!!!" (Okay, you can use something else, but it's a kool catch phrase.) Visit the Klub Konformist at Yahoo!: http://clubs.yahoo.com/clubs/klubkonformist