Informatica Perspectives

In Chris Cingrani's recent post the question: "Data quality, does anyone care?" was posed. The answer is yes, of course people care about data quality – in fact, there are a lot of good reasons why a lot of people should care very deeply about data quality. Let’s look at the most recent example of where data quality makes a big difference, and that is in the federal election process.

A recent Washington Post article "Thousands Face Mix-Ups In Voter Registrations", published just prior to the recent presidential election, points to all the problems and confusion that have been arising around managing data about hundreds of millions of voters. A major problem has been that state voter registration systems have been denying eligibility to thousands of voters, resulting in a lot of acrimony, as well as lawsuits. One of those cases, involving data on 200,000 voters in Ohio whose voter records did not match those in other state databases such as drivers’ licenses and Social Security records, went all the way to the US Supreme Court. In other places, eligible voters’ names have erroneously been flagged as those of convicted felons.

All this confusion started with the best intentions. Following the hotly contested and deadlocked election of 2000, Congress passed the Help America Vote Act (HAVA), which funded the establishment of, among other things, centralized databases within states to maintain digitized voter records. The states were required to authenticate these records against driver’s licenses, Social Security numbers, and other records.

As Mary Pat Flaherty wrote in the Washington Post:

“As the databases are implemented, voters' names and other information are verified against state driver's license records or Social Security records to determine their eligibility. Federal law allows each state to decide what constitutes a match — whether it will accept nicknames, for example. But states are not using "the best scientific knowledge known today" when they verify the information, said Herbert Lin, who is studying the issue for the federal Election Assistance Commission, which oversees election reforms.”

The mismatches are occurring because voter records may vary from database to database. “The problem is that databases are prone to errors such as misspellings and transposed numbers and, as a result, thousands of new voters across the country are coming up as non-matches and having to take additional steps to verify their identity,” observes Kim Zetter, writing in the Wired article "Voter Registration Databases and Purges — Part I". “Voters who have hyphenated names, names with non-traditional spellings or names transliterated from other alphabets (such as those from Middle East and Asian languages) are most likely to show up with discrepancies among databases and wind up on mismatch lists.”

Sound familiar? What these articles and posts are touching upon are the same issues that enterprises have been wrestling with for a number of years now – rationalizing and integrating information from silos of data that have arisen across their organizations. Many organizations have set out to integrate information from various parts of their organizations, only to find tangles of data.

As Lin said in the Washington Post article, state-of-the-art technology is not being applied against this problem when it comes to election management. The federal and state governments could probably learn a thing or two from the experiences other organizations have been having in dealing with data quality issues. Perhaps a good place to start for these government entities is data governance. Voter registration systems need common information models, combined with common integration platforms, robust identity resolution, and ultimately, perhaps even enterprise data warehouses and master data management to ensure that every eligible voter can exercise the right to vote.