tems subject to state and federal regulation. We made FOIA requests to all 50 states and the federal government. We spent hundreds of hours compiling, standardizing and scrutinizing data. Our analysis was so thorough and unassailable that the U.S. Environmental Protection Agency inquired about hiring our database editor, Barbara Hansen.
Matters of "significance"
Some of our findings:
• About 40,000 water systems violate safe drinking water laws each year; 9,500 of them have "significant" problems that pose "serious threats to public health," according to the EPA. Yet just one in 10 "significant" violators face binding enforcement action such as an administrative order, fine or lawsuit. One-quarter of all "significant" violators have been in that category for at least three years.
• Eleven states have not implemented all the Safe Drinking Water Act's contamination limits, which should have been in place by 1995. At least 13 states don't meet federal guidelines by inspecting water systems every three to five years. A half-dozen have not given their drinking water regulators the power to levy fines.
• The EPA is more than a year behind in completing required assessments of drinking water programs in at least 11 states. What's more, the agency's Safe Drinking Water Information System (SDWIS), the main national database used to monitor water systems' performance, is terribly inaccurate.
The stories got a strong response. We met requests for more than 30,000 reprints. In Congress, the project prompted an inquiry to the EPA by Reps. Henry Waxman, D-Calif., and Sherrod Brown, D-Ohio. Waxman's office says he intends to raise the issue with EPA officials at hearings this year.
A path full of hurdles
The project began simply, but it didn't stay that way. First, we went about learning how drinking water rules work. Then we set measures for evaluating their success: Are states implementing required regulations? Are states conducting regular sanitary surveys (inspections) of water systems? Are state and federal officials taking action against systems that violate health and safety rules?
We sent FOIA requests to the EPA and the states for the records we needed. The states' data were wildly inconsistent — each state has different rules on what constitutes a violation of drinking water laws and what warrants an enforcement action.
More troublesome: the states' data didn't jibe with much of what the EPA had in SDWIS. The federal database was riddled with errors. Some we were able to resolve, but we had to adjust our expectations. We used SDWIS to draw conclusions at the state and national levels, but information on specific water systems had to be verified independently.
We finished with four databases: SDWIS, which we imported into Paradox using Monarch parsing software, and three we built ourselves. The first, based on sanitary survey records, allowed us to calculate average inspection rates for water systems in each state. The second, based on census data and information from EPA's infrastructure needs survey, allowed us to calculate per-capita costs for bringing water systems into compliance with federal requirements. The third, based on state budget data, allowed us to determine each state's per-capita spending on drinking water oversight.
Lessons learned
• Know the rules: You can't interpret data without a keen understanding of how it was compiled and categorized. Plus, you'll win over sources with your technical expertise.
• Check data: To confirm findings' validity, develop alternate queries and different approaches that should yield the same result.
• Be honest: Explain your data's limits. We wrote a separate story on the problems we found — especially with SDWIS — to help readers understand what conclusions could and could not be reached.
• Allow time: When you're getting data from varied sources, budget time for cleanup and verification. Also, many states aren't used to handling requests for computerized records, so there's a lot of haggling.
Peter Eisler can be reached by e-mail at peisler@usatoday.com