Democrats used good data to win midterms
Party scrapped worn-out computers and old voter files for newer technology
- By Dibya Sarkar
- Feb 09, 2007
The Democratic Party's victories in the midterm elections that enabled it to take control of the House and Senate were partly because of a new Democratic National Committee information technology system that improved its get-out-the-vote initiative, officials say.
Designed by Boston-based Intelligent Integration Systems (IISi), the system was used last spring to sort through the 200 million voter records and more than 900 fields associated with each record to target prospective Democratic voters.
The core component of the system is a data warehouse ' developed by Netezza, a company based in Framingham, Mass. ' that helped users sift through data faster and more accurately, DNC and IISi representatives said. The new system includes data quality software from FirstLogic, which Business Objects acquired last spring. In addition, it features extract, transform and load applications from Sunopsis and SPSS' modeling software, representatives said. Officials estimate the system's cost to be less than $1 million.
Jim Baum, Netezza's president and chief operating officer, described the company's flagship Performance Server data warehouse product as an appliance because it bundles the database software with server and storage capabilities in a prepackaged device that is quick to install and inexpensive to maintain.
Although Netezza's bundled capability was a factor, Richard Zimmerman, IISi's president and chief technology officer and the system's chief architect, said the appliance's performance was the primary consideration. He said it ranked higher than other vendor products they reviewed.
'The Netezza system runs literally two orders of magnitude faster than a comparable traditional architecture system, like an Oracle system, and it's cheaper than that,' he said.
The DNC's former system could hardly be described as a system, Zimmerman said. It was a bunch of computers housing numerous voter lists ' provided by state parties, secretaries of state and other sources ' that workers tried to reconcile. That was no easy task, he said.
'Every time the state parties would give them another list, it would have different sets of fields,' Zimmerman said. 'And one time, they'd get them and they would have a bunch of phone numbers. The next time, half the phone numbers would be gone, and the third time, the first name and last name were in one field [but] the next time, they were split out in two different fields.'
The result was low quality and inconsistent data, he added.
Gus Bickford, a database consultant and former executive director at the Massachusetts Democratic Party for five years, said a crucial factor in the success of the party in November was cleansing the data to match potential voters' names with the right contact and demographic information.
Such data is created at the county or precinct level and usually is typed in by clerks or others, who send it to the DNC. A great deal of work then goes into standardizing the data so that names, addresses and other pieces of information are consistent. Data cleanup usually occurs twice before the fall elections, Bickford said, and people check the data against several sources, including consumer information and the national change of address file. Then they send the data back to the state parties. He said the new system allowed the DNC to get quality voter data out the door two to three times faster.
'And then you [clean] it a third time in either late September or early October because you want to get all the newly registered voters,' Bickford said. 'And so that's why the turnaround time is so critical' for processing the data. n