BBV - New Hillsborough question

[harmonyguy]

I've been staring at these Hillsborough numbers for a while and they're starting to really bug me.

Now, keep in mind that I'm not in Florida, and don't have a clue as to the political sentiment down there.
I guessing that in the Aug 31 primary, the top of ticket races would have been the Democratic and Republican Senatorial races. Is that a correct guess?

Were there other races contentious(sp) enough to generate interest from as many as ten thousand OTHER voters, who would NOT declare a party preference, and as a result would NOT vote for either a Republican OR a Democratic senatorial candidate?

IS there anything particularly notable about Circuit 13? Both the Judge's race and the State Attorney's race have anomalies in them, but I can't develop any explanation for them. Is anyone familiar with the issues at that end of the country?
Thanks
HG

[NVMojo]

1. sounds interesting

we should all know more about Florida's primaries ...

[shraby]

2. Where are you getting your figures from?

[harmonyguy]

10. Drawing numbers from a combination of sources.

Hillsborough's website www.votehillsborough.org
Florida Secretary of State Website enight.dos.state.fl.us

The key thing is having taken a number of snapshots of the sites throughout the night, as results came in. It gives me more datapoints to work with.

HG

[claudiajean]

3. I must be looking in the wrong place, HG...

...I have the spreadsheet open, and I'm not seeing the same types of anomalies that you indicate.

I'm seeing the usual drop-off rate as the races progress down-ticket.

10% drop-off (undervotes) would be normal in the top ticket race.

Now, I may be looking at something wrong, and am ready to stand corrected if I missed the issue.

On a related note: The whole issue surrounding "undervotes" is vastly overblown. The CalTech-MIT report lumped overs and unders all together as if they were both "problems". Overvotes are true problems. Undervotes are completely normal. Less than 15% of voters routinely fill out their entire ballot. People tend to vote only in the races in which they feel suitably informed to do so. Once you get down to dogcatcher and cemetery commissioner, the voter participation rates are pretty low, and that's perfectly normal.

Unless caused by ballot mis-design, undervotes are to be expected.

Again, my apologies if I'm missing the correct anomaly.

:)

[meg]

4. 3 irregularities from Hillsborough this primary

1. Elections site posts errant tally
2. The 245 votes, cast in early voting for two weeks prior to the primary, don't count and can't be awarded to the candidates who deserved them, since they were found after state certification. They become a footnote to the primary election of 2004: votes cast but not counted.
3. Johnson had another vote tabulation problem after the polls closed on Aug. 31, when his computer servers slowed to a crawl, resulting in a final tally that was delayed until 5:10 a.m.


link

[RedEagle]

6. What causes the servers to slow?

"3. Johnson had another vote tabulation problem after the polls closed on Aug. 31, when his computer servers slowed to a crawl, resulting in a final tally that was delayed until 5:10 a.m."

Any ideas, anyone?

[harmonyguy]

7. Apparently the 'indicies' failed

from:
http://tampatrib.com/politics/MGBM5JF8PYD.html

"Sequoia has a procedure that can test whether indexes are operational, but that was not used during pre-election testing or the tabulation delay. "

"The elections office did not use the procedure because Sequoia had not run into the problem before and did not recommended the test as part of the system's protocol, Johnson said. "

That's like saying '...we didn't run the test 'cus they didn't tell us to, duh!'
So just who exactly IS running the elections, the County or the voting machine company?

"I made the choice to choose accuracy over speed."
Too bad they blew it on both counts. Later, another 245 votes were 'found', and then a reporter pointed out that more votes were counted in the State Attorney's race than there were people who cast ballots.

No hysteria, no boogiemen, just plain facts. Some of the so-called 'election professionals' who are paid to get it right, still don't get it at all.
Would you keep an accountant on staff, who couldn't add?
Would you keep a plumber on staff who couldn't plumb without checking with the manufacturer first?
It's time to delete these high-paid numbskulls from the payroll, and get in someone who knows enough about systems and processes to be able to implement some.

Grrrrrrrrrr.......
HG

[RedEagle]

5. Kick

What voting systems was in use? I believe in the original post you mentioned it wasn't Diebold.

[harmonyguy]

9. Sequoia along with WinEMS and SOEweb

I wonder if either WinEMS or SOEWeb should be certified by someone?

Mind you, the only folks they'd get to certify them would be the same ones who 'certify' the existing stuff (in the dark - all secret-like)

Maybe *I* should be certified for even thinking that........

HG

[harmonyguy]

8. More Background

http://www.sptimes.com/2004/09/23/Hillsborough/Elections_site_posts_.shtml

While this article does talk about the oddity in the State Attorney's race, it doesn't mention the same problems that occurred with other races.

When Hillsborough's website was reporting a total of 118699 ballots cast, the totals in each of the following races exceeded or equalled that number.
State Attorney Circuit 13 - over by 7192
Property Appraiser - over by 4681
Judge Circuit Court 13 - exactly equal
Judge Group 11 - over by 583
School Board District 7 - over by 671

Now in the above article, the Supervisor of Elections is quick to point out that the website totals were nothing more than human error.
"It's a human error. We input this. We'll fix it today."

Now human error is a plausible explanation for an occasional error (heaven knows, none of us are completely infallible) but the facts of the situation suggest that this was more than just an input error.

First off, the ballots cast total from the Hillsborough website exactly reflected the total of votes cast in the Judge Circuit Court 13 race. While one could accept this as an input error once, could one accept it more than once? Readings taken from the Hillsborough website earlier in the evening show the same pattern - the ballots cast total from the website exactly relected the total of votes cast in the Judge Circuit Court 13 race. If this was human error, its repetitive nature suggests that this was more like a programming error than an input error.

Apparently Hillsborough runs SOEWeb and WinEMS software from a company called SOE Sofware. According to SOE's website (http://soesoftware.com) these products simplify the gathering and disemination of election data, including results. Without knowing anything about the SOE sofware, I would hazard a guess that something in the setup of the interface between SOE and the Sequoia WinEDS tabulator didn't get tested.

Something that was really odd with the Hillsborough tallies, was the fact that throughout the night, they were reporting totals of individual races to the Secretary of State's office, but were NOT reporting ANY totals of votes cast, not even the incorrect ones that they posted on their own site. So, not only were they reporting erroneous totals on their own site but there appears to have been no attempt to ensure that the numbers sent to the SoS were even remotely close.

Am I suggesting that something nefarious was going on? Of course not, I don't have enough information to do that. I AM suggesting that there was insufficient testing done on the systems to ensure that when garbage goes in, at least processed garbage comes out, instead of different garbage.

Now to the question of undervotes.
After some quick adjustments by Hillsborough staff, they have apparently determined that there were a total of 138389 ballots cast.

Assuming that the top-of-ticket races were the Republican and Democratic Senate seats, there were only 127178 ballots cast in these two races combined. Assuming that people wouldn't be able to vote for a candidate from BOTH parties, it means that there were 11211 people that didn't vote on the top contests. I don't know if that's high or low or normal.

I'll be the first to admit that I know absolutely nothing of the local issues, but from my distance it would seem that the votes cast for State Attorney, Property Appraiser, School Board and Circuit Court Judge should be significantly less than those cast for the top races. It's kinda like rushing to vote for the local dog-catcher, and skipping over the vote for president.

HG