Statistics Of, By, and For the People
Census director compares use, quality of statistical frameworks
Robert Groves, director of the U.S. Census Bureau,
discusses the 2010 Census at his IPR
Distinguished Public Policy Lecture.
In recounting his first days at the helm of the U.S. Census Bureau, Robert Groves recalled the eye-opening swath of operations surrounding this “unique, public event,” in which “everyone is the target audience.”
Groves delivered IPR’s 2011 Distinguished Public Policy Lecture on May 2 at Northwestern University.
“We mailed out hundreds of millions of paper question-naires,” Groves said, referring to the most recent census. “We had 600,000 people knocking on 47 million household doors. I figured that we knocked 100 million times in all.”
Groves, who was appointed by President Obama as the nation’s 23rd director, moved from a 35-year career as one the world’s top survey methodologists to become the nation’s head statistician in July 2009.
For the more than 100 attendees, the nation’s census director also traced how his new job has changed his thinking about the fundamental uses and quality of statistical information in a democracy such as ours.
“In a democracy, the statistics produced by a government agency belong to the people,” he said. “It’s the only way you can keep the government honest.”
The 2010 Census
Mandated by Article I, Section 2 of the U.S. Constitution, the Census has taken place every decade since 1790. Its data are used for many purposes, the most newsworthy of which is the apportionment of seats in the House of Representatives.
More than $13 billion were spent over a 13-year cycle to get the 2010 Census off the ground. It counted 308,745,538 residents in some 134 million homes around the country. The data will eventually produce “billions of statistics,” Groves said.
More than 100 people attended the lecture.
Census forms were available in six languages, with language assistance in 59 more, bringing it within the linguistic reach of 99.7 percent of the U.S. population. The national, award-winning outreach effort involved Super Bowl advertising, hip-hop and Karl Rove videos on YouTube, and K-12 education programs, among others.
For the first time, the Census also paired with external partners, such as FedEx, Google, Best Buy, and Telemundo, who provided free advertising and informational placements like casting a census worker on a telenovela, a
Spanish-language TV soap opera.
“It was a massive campaign, quite contrary to my experience in academia on large-scale surveys,” Groves said.
The outreach effort, which also includes a regular blog by the director, has led him to think a lot about surveys and statistical information in general.
Statistics in Government and Academia
Groves flashed a slide of a widely used academic paradigm of statistical thinking on “total survey error.” He pointed to the four commonly accepted sources of statistical uncertainty: the mode and administration of questionnaires, the interviewer, and the respondent. Bookshelves overflow with books on each of these topics, he noted.
As a social scientist, Groves described how it was his job to inform his peers and colleagues about the numbers—good and bad—and then include lots of additional qualifying information about them.
Yet as a government statistician, his job is to produce unbiased data for the entire population, directly delivering simple statistics such as means and percentages. Since these direct estimates are what are most often used for policy decisions and become official statistics, they alone become the focus—not the uncertainty attached to them as in academia, he said. For the public, he explained, “It’s an odd thing to have someone fill in all sorts of qualifications about what it is that we know.”
“If you examine the last five issues of the leading journals in social statistics, there is much more emphasis on analysis of existing data and construction of data,” he said. “In my new life, I worry about measurement a lot more.”
This has also led him to become more acquainted with a “revolutionary” quality framework called “fitness of use,” in which the same number is used for completely different purposes. Thus, the quality of a statistic is driven by its end user, not its producer.
Keep It Simple
Rather than creating complex statistical experiments, Groves said that he now spends more time worrying about how to keep things simple so that they can be easily communicated to the media and American public.
From left, Provost Daniel Linzer of Northwestern University,
Census Director Robert Groves, and IPR Director Fay Lomax Cook
Simplicity is crucial as it allows Americans to understand how such figures might relate to their daily lives, Groves added. Studies have shown that when statistics are presented to people in ways that appear more relevant, the numbers themselves are seen as more believable. Plus, it’s important that the process of giving information remains credible and transparent so that people feel comfortable providing their personal data—and they trust that the data are nonpartisan, he continued.
Credibility and transparency also rest upon fundamental assumptions that people have about the role of government in their lives, Groves noted. In the United States, low levels of statistical knowledge among citizens—even on hot topics like unemployment—have been compounded by a current wave of distrust in government.
The Census needs to remain relevant, Groves said, yet in such a rapidly changing, heterogeneous country as the United States, this can be a tall task. Is it more important to make apples-to-apples comparisons, evaluating the same things consistently over time? Or, he asked, should we try to measure new things to capture the different components of a changing economy?
“That’s what makes relevance hard for me at the Census Bureau,” he said.
He later concluded, “My hunch is that relevance is the gateway to credibility. We have to make sure that what we do is relevant to peoples’ lives, and we have to be transparent.”
In commending Groves, IPR Director Fay Lomax Cook said, “There are few people who are as uniquely qualified to run the Census Bureau as Bob Groves—and even fewer who can straddle the worlds of policymaking and social science research with such ease. His talk perfectly illustrated the usefulness—and importance—of social science research expertise in national discourse and decision making.”
Photos by Jason Reblando.