Voting Machines

[jmknapp]

The voting machine allocation by precinct for Cuyahoga County (Cleveland) presents a very different picture than that of Franklin county (Columbus). In the latter case the inner city (Kerry) precincts were shorted compared to suburban (Bush) precincts:



So based on the standard of active voters per machine, the Democratic precincts were relatively overloaded.

A fair allocation algorithm for Franklin based solely on the active voters per machine standard yields the following graph:



I'm in correspondence with someone at the Franklin County BOE who counters that the "optimal" allocation can't be done, because there are other issues such as differing ballot complexity across the county, and things like school levies that affect local turnout, which they proactively accommodate.

To me, that strikes as handwaving, as I'd bet that an analysis of such factors would still show a bias towards the suburbs.

But enter the Cuyahoga picture of machine allocation. Here is the data on machines per precinct:

http://copperas.com/cuyahoga/CuyahogaVotingDevices.zip

I don't have the figures for "active voters" per precinct in Cuyahoga, but just as a first cut I plotted the number of registered voters per machine as a function of Kerry support:



Now that is remarkable. Neglecting the relatively few outliers, the precincts fall within a narrow band between roughly 100-130 registered voters per machine. It really appears as if the Cuyahoga BOE used the optimal allocation algorithm that the Franklin BOE eschews.

Moreover, it was done on the basis of registered voters. This would tend to favor Democratic precincts, since Democratic turnout is for many reasons (beyond suppression) lower on the average. Indeed, looking at the number of actual voters processed by the machine on election day, it can be seen that the Democratic precincts saw a lighter load:



Generally, there was a load of 50-100 actual voters per machine. In the 13-hour (780 minute) polling day, that represents an average voting time available of 8-15 minutes per voter, which is more than generous. Cuyahoga used punch card voting machines.

In Franklin County in contrast, the higher-tech Danaher electronic machines were dearer, and precincts averaged over 200 active voters per machine, and an average of 184 actual voter per machine indicating an overloaded system operating at maximum capacity. At 184 actual voters per machine, each voter had 780/184 or 4.2 minutes to vote on the average, and peak rush hours saw ballooning lines, deterring many would-be voters.

It's true that the voteprotect.org EIRS system logged many reports of long lines in Cuyahoga too, but these may have been more isolated than the general overloaded picture in Franklin. The Cuyahoga graph above doesn't show a few of the worst outliers, such as Cleveland 4D at Woodbury Elementary School, with 450 registered voters per machine. Here was one EIRS case reported for Woodbury Elementary:

"2 hr waits, need more machines and people.  Inspector reluctant to deal with it -- in over her head. ;Rumors that precinct judge quit this morning. Superintendant said fine to call downtown." (EIRS #34588)

And then another for Woodbury ES later in the day:

"3 machines were added around 12:15pm, but still incredible backlog.  Need more machines.  People in line more than 3 hours.;9 people left at least." (EIRS #51422)

[truehawk]

1. kick

[jamboi]

2. Cool research. Hope you're posting this stuff at Demopedia or someplace

where it can have a permanent, public home.