The Web site. (Andrew Harrer/Bloomberg)

The director of the Census Bureau said he is confident that new changes in the wording of survey questions on health insurance will not diminish its ability to measure the impact of the Affordable Care Act.

John H. Thompson, responding to critics, said the survey questions were rephrased specifically to establish a benchmark for how many people were uninsured in 2013, the year before the health insurance mandate took effect, so comparisons can be made in future years.

Concerns, first reported in the New York Times, have arisen that changes in the phrasing of one survey’s questions may make future comparisons more difficult. Critics of the health-care law have accused the Census Bureau of working to fudge the numbers so the White House can claim more success than is merited.

Thompson said that Census Bureau demographers had been preparing for many years to modify the questions and that the changes should make it easier to measure the law’s impact, not harder.

“I can assure you, I have had no discussions of this with the White House or with anyone else in the administration,” Thompson said in a phone interview. “This has been a scientific process, and that’s the way we operate. I don’t understand where all these concerns come from suddenly. We pride ourselves on being a statistical agency that produces objective, nonpartisan and high-quality information. That’s our mission.”

The concerns have emerged over changes made to one of three separate surveys the government conducts with questions about who has health insurance.

The most current and complete measure of how many people are uninsured comes from the Health Interview Survey, which is conducted by the National Center for Health Statistics. Researchers focusing on geographical comparisons find the American Community Survey helpful for exploring differences down to the neighborhood level.

But the annual Social and Economic Supplement of the Current Population Survey is considered the most useful for looking at differences among people with different income levels. It is conducted over three months, every spring, and the statistics are released every fall as part of an annual report on income, poverty and health insurance. No other survey provides the same level of detail on income.

Until this year, the Current Population Survey asked people whether they had been uninsured at any point in the previous year. Someone uninsured for the month of January, for example, was counted as uninsured for the entire year.

Census Bureau tests concluded that the question effectively inflated estimates of the uninsured. Tests run in 2010 and again last year showed that the share of people saying they were uninsured dropped almost two percentage points, from 12.5 percent to 10.6 percent, when they were asked to elaborate whether they were uninsured for just part of the year.

So this year, the Current Population Survey asked anyone who said they had been uninsured in 2013 to be more precise, noting what months they went without insurance. That conforms with the phrasing of questions asked in the American Community Survey and the Health Interview Survey — questions that have not been changed and have been asked for years — so comparisons can more readily be made.

Thompson said the three surveys can supplement each other in measuring the law’s impact.

“We’ll have a pretty good baseline,” he said. “And the American Community Survey provides more information to understand the dynamics of changes at the state and substate levels. I’m confident with all the information available we’ll be able to really track the changes.”