Census data scandal 'not as bad as we initially thought'

A congressional report found no evidence to substantiate a November 2013 New York Post story that linked falsification of data by the Philadelphia Regional Office of the Census Bureau with a drop in the unemployment rate before the 2012 election. But the report was nonetheless critical of the way the bureau ensures the integrity of its data.

The House Oversight and Government Reform Committee and the Joint Economic Committee investigated the allegations and concluded that "documents and testimony obtained by the committees did not show a link between the data falsification that occurred in the Philadelphia Regional Office and the national unemployment rate."

At the same time, the panels' report states that "the documents and testimony did show, however, that the Current Population Survey is vulnerable to data falsification and that the Census Bureau needs to make common sense reforms to protect the integrity of survey data."

Furthermore, a report released in May by the Commerce Department's inspector general "did not substantiate the existence of widespread falsification" of data but "did identify several vulnerabilities with respect to the Census Bureau policies and processes for detecting and preventing data falsification."

"The situation is not as bad we initially thought," said Texas Republican Rep. Blake Farenthold, chairman of the Oversight and Government Reform Committee's Federal Workforce, U.S. Postal Service and the Census Subcommittee. He added that the bureau would benefit from following the committees' and IG's recommendations.

The congressional report found multiple vulnerabilities in the quality assurance process for collecting unemployment data for the Current Population Survey and concluded that the bureau's lack of recordkeeping and deficient data collection system enabled an environment in which data falsification could occur. Because there is no one master record of a case, it is difficult to "determine the full history and corresponding chain of custody of a particular case," the report states.

Commerce IG Todd Zinser said he largely agreed with those conclusions in his testimony before the Census subcommittee on Sept. 18.

He told lawmakers his office had developed a half-dozen recommendations to reduce the chances of data falsification and to breed a culture of zero tolerance for such activities. His recommendations are:

• Implement a mechanism for reporting confirmed data falsification to survey sponsors.
• Implement a formal policy that prohibits employees suspected of falsification from collecting survey data during the investigative process.
• Update procedural manuals and training materials to reflect current regional office field structure and inform field representatives about survey data falsification and the consequences of committing falsification.
• Implement an independent quality assurance process for all survey operations.
• Ensure that all survey supervisors tasked with detecting and preventing survey data falsification are properly using all available tools to safeguard against such misconduct.
• Implement internal controls to effectively monitor and limit field representative workloads in order to reduce the risk of falsification.

Massachusetts Democratic Rep. Stephen Lynch, the subcommittee's ranking member, said the investigation and ensuing hearings provided an opportunity to clear the names of the accused Census employees.

"It was important to lay this all out because there has been a steady drumbeat of criticism for government employees and agencies," Lynch said. "In the last few years, government employees have endured furloughs, a three-year pay freeze, pension contributions have increased. It's very important to make sure that people understand that these reports were baseless."

The Post was unrepentant.

In a column posted online the morning of the hearing, John Crudele, who wrote the original story alleging the falsification, wrote that although "the report did not find evidence of political pressure in regard to the surveys," investigators had failed to "address the fact that 120 Census laptop computers on which data is gathered were unaccounted for in the Philadelphia region just weeks before the 2012 election."