[EL] Marathon bombing suspects not registered to vote

[Michael P McDonald]

We really don't know exactly what the commercial list vendors do since their techniques are proprietary. My impression from working with commercial vendors' data and talking with employees is that they clean the voter registration data by supplementing it with data from other sources. When they clean the data, they might add information like a phone number, and they might correct what they consider to be errors in the name field identified by matching -- through a fuzzy matching algorithm -- data in another database. They also improve the data through contacts made by the campaigns that are feed back into the system. This may be the weakest link since the data source is often volunteers, who might not be trustworthy. I wouldn't call it fraud, since the vendors believe that they have enhanced the product that they are selling. I'd call it limitations with the data and algorithms that can in some cases fix errors in the voter registration files and in others introduce new errors.

============
Dr. Michael P. McDonald
Associate Professor
George Mason University
4400 University Drive - 3F4
Fairfax, VA 22030-4444

phone:   703-993-4191 (office)
e-mail:  mmcdon at gmu.edu               
web:     http://elections.gmu.edu
twitter: @ElectProject

-----Original Message-----
From: Larry Levine [mailto:larrylevine at earthlink.net] 
Sent: Tuesday, April 23, 2013 11:35 AM
To: Michael P McDonald; 'Rick Hasen'; law-election at uci.edu
Subject: RE: [EL] Marathon bombing suspects not registered to vote

As long as the vendor uses the official voter registration file as a base and simply adds information about voters on that file there should be no problem and no one who isn't registered should be added to the list. If a vendor is selling voter data from a list that adds non-voters that would be fraud, as list rentals usually are based on a cost per thousand names.
Larry

-----Original Message-----
From: law-election-bounces at department-lists.uci.edu
[mailto:law-election-bounces at department-lists.uci.edu] On Behalf Of Michael P McDonald
Sent: Tuesday, April 23, 2013 8:24 AM
To: Rick Hasen; law-election at uci.edu
Subject: Re: [EL] Marathon bombing suspects not registered to vote

Does anyone know which commercial databases indicated incorrectly that the Tsarnaev brothers were registered to vote? 

Many on the list are using commercial databases in their scholarly work. I was unaware that reporters have been using these commercial databases, too.
I've been critical of using these black box data since the vendors use opaque proprietary algorithms to manipulate and clean the data. While I've been more worried about vendors purging records, this case may indicate an instance where the commercial matching algorithms created a false positive match by identifying persons on the registration list with data in another database that contained records for the Tsarnaev brothers. Anyone using these commercial databases uncritically in their work -- scholarly, reporting, or otherwise -- would be wise to learn more about the limitations of these data. And I'm not just talking about the commercial data: the canonical voter registration files -- or more generally, election management databases -- also have errors in them, as do any large-scale databases. When there are millions of reco  rds, errors creep into the system. This is something that anyone who has worked with these data for a long time already know, but many who are just starting to work with so-called "Big Data" are just beginning to learn.

============
Dr. Michael P. McDonald
Associate Professor
George Mason University
4400 University Drive - 3F4
Fairfax, VA 22030-4444

phone:   703-993-4191 (office)
e-mail:  mmcdon at gmu.edu               
web:     http://elections.gmu.edu
twitter: @ElectProject
"Marathon bombing suspects not registered to vote" 
Posted on April 22, 2013 8:48 am by Rick Hasen Josh Gerstein: "Though there have been reports that Dzhokhar and Tamerlan Tsarnaev were registered to vote and there are entries in commercial databases indicating they were, they were not according to the documents available at the Cambridge Elections Commission."

Posted in Uncategorized | Comments Off
--
Rick Hasen
Chancellor's Professor of Law and Political Science UC Irvine School of Law
401 E. Peltason Dr., Suite 1000
Irvine, CA 92697-8000
949.824.3072 - office
949.824.0495 - fax
rhasen at law.uci.edu
http://law.uci.edu/faculty/page1_r_hasen.html
http://electionlawblog.org

_______________________________________________
Law-election mailing list
Law-election at department-lists.uci.edu
http://department-lists.uci.edu/mailman/listinfo/law-election