Drawing Blanks

Premature Optimization is a Prerequisite for Success

Russian elections: Electronic ballot boxes

leave a comment »

I got my hands on a list of precincts where electronic ballot boxes were used.

An electronic ballot box (in Russian they are called KOIB which stands for “complex for processing of electoral ballots”) can scan ballots, store the counts in memory, save them to a flash card, print out the tally. They are typically installed in pairs which share a common database, for redundancy purpose. They can also be connected to some sort of centralized database and transmit the tallies there directly.

Supposedly, the KOIBs should make it more difficult to forge the tallies and to stuff in extra ballots. It’s still possible to stuff, but each ballot would have to be scanned separately, so it becomes really time consuming.

The list of precincts which utilized KOIBs was compiled by Russian fraud buster enthusiasts from some government purchase contract documents. The list contains about 4,000 precincts which is less than 5% of the total number of precincts. Anyway, lets see what the data looks like, it looks really interesting!.

Distribution by precinct size, automated precincts on the left, general population on the right:

Ksizesizedist

As I previously showed, the statistics of voting at the smaller and larger precincts differs drastically. So in order to compare apples to apples, I’m going to break down the automated precincts by the same size criteria as I previously applied to the general population. I’m going to name the category of automated precincts with < 800 registered voters “K2” and the category of automated precincts with 800 to 3000 registered voters “K3” (K for KOIB) and compare them to the corresponding C2 and C3 categories.

K2

Turnout on the left, votes for United Russia on the right:

K2todistK2votedist

And here is the data for the C2 category (all precincts smaller than 800, outside of the ethnic regions):

C2TOdistC2votedist

Average turnout is the same (70%), and support of the ruling party is 57% at the KOIB’d locations vs. 56% at all locations.

Frankly, I did not expect this result at all, and I had to double check what I did. I conclude that at least one of the following is true:

  • The data on the KOIB adoption is unreliable, or maybe the KOIBs were installed but weren’t used at the smaller precincts
  • The KOIBs do not help mitigate fraud at all
  • There was no significant fraud at the C2 locations (I think this one is the least likely explanation of the data. Until this moment I was sure that C2 must have had more voting manipulations than C3).

But let’s move on to K3 and C3…

K3

Turnout on the left, votes for United Russia on the right:

K3todistK3votedist

And here is the data for the C3 category (all precincts with 800-3000 voters, outside of the ethnic regions):

C3TOdistC3votedist

Alright, here a difference is seen.

Average turnout: 55.4% vs. 56.4% Average vote: 35.4% vs. 41.7%

Is this difference significant?

Some would claim, it is ENORMOUS! The probability that a random sample of the same size as K3 picked out of C3 will have the vote % lower than say 37, can be safely described as “never”. So K3 is not a random sample, for sure. And who in the right mind would expect that the KOIBs are scattered randomly. Their placement must correlate with some social factors.

Only about 8% of precincts in C3 had the KOIBs. And 80% of precincts in C3 have the vote percent < 55. And if we randomly sample from those 80%, we will always be getting the average vote around 35%, regardless of how many KOIBs are in the sample.

On the other hand, the 35% figure looks suspiciously close to the number obtained from the observers’ copies of the tallies at http://ruelect.com/en/

Once again, I cannot make any strong conclusions. But if we assume that the KOIBs did help reduce fraud, then we must once again admit that the fraud did not inflate turnout and was not responsible for the vote-turnout correlation.

With respect to the social conformity theory, the automated precincts fit the entropy curve very nicely with no surprises.

Advertisements

Written by bbzippo

01/19/2012 at 7:07 am

Posted in Uncategorized

Tagged with

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: