Printer Friendly

Blind data; why the life's gone out of the government's vital statistics.

Among the millions of documents spewed forth each year by the federal government is a glossy, 692-page tome from the Department of Health and Human Services (HHS) called Healthy People 2000: National Health Promotion and Disease Prevention Objectives. The document lists hundreds of laudable health care goals for the coming millennium, such as reducing deaths from work-related injuries to 4 per 100,000 workers, or limiting HIV infections to no more than 800 per 100,000 people. Trouble is, the department doesn't say how it intends to get there from here. And even if the department's officials did have a plan, they couldn't possibly put it into effect--since by their own admission, they can't even guess at the size of one quarter of the problems they have nevertheless promised to address, and for a full two thirds lack the data to figure out if they are helping matters or making them worse.

For instance, the department wants to reduce drug-abuse-related emergency room visits by 20 percent over the next time years. Great idea. But how many such visits are there now? The department doesn't know. Nor does it know the number of adolescent suicide attempts (which the deparment wants to reduce by 15 percent) or the percentage of pediatricians who screen children for developmental problems (the goal: 80 percent) or the number of people with "inappropriately stored" weapons (Healthy People 2000 will cut that number, whatever it is, by 20 percent).

If it all sounds to you a bit like another episode of "Yes, Minister," bear in mind that the numbers those bureaucrats are merrily tossing about represent real people: teenagers shutting themselves in a New Jersey garage to drink in carbon monoxide, or a four-year-old Bronx boy shooting himself dead with his daddy's pistol. You'd think these are the sorts of problems Uncle Sam would be keeping his eye on. But then again, if he were, we'd all be able to tell when he wasn't doing anything to address them. Of course, statistics can't tell you everything about tragedies like these. But without numbers to give them some sense of direction and accountability, as the empty promises of Healthy People 2000 amply demonstrate, the government programs that might prevent such tragedies simply won't get anywhere.

In fact, it's a telling demonstration of the power of statistics that when they do exist, they can be more dangerous than when they don't. It's like the difference between having no bridge and having a bridge that drivers don't realize is about to collapse. If you pick apart any federal number, you'll probably find that it's derived from faulty assumptions or incomplete data, which means we are basing all sorts of decisions on lies. For instance, according to a recent story in The New York Times Magazine, the Food and Drug Administration (FDA) reports that it rejects 27 percent of all inspected seafood as spoiled or contaminated--a measure that suggests it's doing a thorough job of protecting consumers. So, feeling safe and sound, we cheerfully indulge in kippers for breakfast, tuna for lunch, sushi for supper ... until we read the fine print and learn that the FDA inspects no more than 2 to 4 percent of the total catch, presumably allowing large quantities of rotten fish to slip through.

Counted out

But maybe you never expected much from HHS in the first place, and maybe you don't like seafood. Missing or misleading statistics are hurting you anyway. The figures generated by the federal government's $1.9 billion statistics-collecting enterprise make their way into every nook and cranny of the government, from entitlement programs to law enforcement to trade policy--not to mention private businesses, local and regional economic planning, private health care administration, and magazine pieces like this one. Name an organized activity at random, and chances are good that it depends in some fashion or another on federal numbers. So wouldn't it make sense to keep the system in sharp working order?

Sure it would. Unfortunately, no one's been at the switch for at least a decade--and in some cases, far longer. If someone had been, perhaps we wouldn't have witnessed missteps like these:

* Consider a single mother of two in Washington who scrubs floors and empties wastebaskets in gleaming office towers for $5.10 an hour. Guess what? She's not officially considered poor, despite the fact that she has to raise two kids on barely $10,600 a year in one of the most expensive cities in the nation. Now, that alone doesn't keep her from receiving all federal assistance, since the poverty line is so badly calculated that Congress has instructed most federal welfare programs to set their eligibility levels a third above it. But there are glaring exceptions to this policy, such as Medicaid, which is typically available only to those whose incomes fall far below the poverty line.

Why does such a widely cited figure bear so little relation to reality? Because no one is paying attention to how it's calculated. Back in the sixties, Molly Orshansky--an obscure Social Security Administration statistician who has probably had a greater impact on federal welfare policy than most U.S. congressmen--looked up a decade-old study that said poor families generally spent one third of their income on food, multiplied the cost of minimum annual food supply by three, and--presto!--created the poverty line. Since 1969, the Census Bureau hasn't even calculated Orshansky's figure anew, instead simply adjusting the previous year's figure for inflation. Of course, such crude assumptions aren't even close anymore; these days a poor family typically spends ony one sixth of its income on food, as other expenses like rent and utilities have taken a larger and larger bite. Patricia Ruggles, an Urban Institute economist, estimates that a more realistic measure might raise the current poverty line by 50 percent--a change that, with the stroke of a pen, would boost the national poverty rate from 13 to 25 percent. Of course, those poor people exist no matter what the numbers say. But the numbers help determine whether and how the federal government tries to help them, and, at a more profound level, how Americans feel about the state of their society: Wouldn't it shock you to learn that one in four Americans is poor?

* The Centers for Disease Control (CDC) determine how many people have died of AIDS by examining mortality data and adding up the number of patients for whom HIV or AIDS was the "underlying cause" of death. But as a team of CDC researchers reported in the Journal of the American Medical Association last year, this conservative definition probably understates the true size of the epidemic. The problem is that, although the CDC maintains what is known as a "surveillance definition" of AIDS--a list that includes a number of opportunistic infections almost always associated with HIV infection, such as Pneumocystic carinii pneumonia or Kaposi's sacroma--it does not use that information in calculating its aggregate AIDS death figures. When the researchers conducting the JAMA study counted deaths due to these infections in their sample, HIV-related mortality jumped by 28 percent. If that figure held up within the population as a whole, the number of deaths due to AIDS--23,300 in 1990--would be about 6,500 people higher.

* Until recently, Republicans enjoyed boasting about how their economic policies brought about 92 months of uninterrupted growth--what George Bush used to call "the longest peacetime expansion in history." There's just one problem: It never happened. A little-noticed 1988 revision in the GNP figures for the second quarter of 1986 turned a quarter of healthy growth into one in which the GNP actually fell. A less dramatic, but still serious, revision affected the last quarter of 1984, when an initial estimate of a 4.3 percent growth rate dove to an anemic 0.6 percent. The hidden weakness of the economy did more than bolster Republican political fortunes: Data collected by the National Association of Business Economists (NABE) suggests that before the Bureau of Economic Analysis (BEA) lowered its initial 1989 GNP growth figures by an average of 1.2 percent, or roughly $600 billion, the Federal Reserve was misled into holding down the growth of the money supply, thus choking off an expansion of credit that might have produced higher growth and possibly even helped avert the recent recession.

* Legislative redistricting is already under way, predicated on the preliminary figures produced by the 1990 census--the one that missed about 2.1 percent of the entire country and more than 5 percent of the black and Hispanic population. Still to come is the redistribution of nearly $30 billion in federal "revenues sharing" that goes to the states based upon population. But neither congressional seats nor federal funds will be distributed according to the nation's true population, because last month Commerce Secretary Robert Mosbacher decided not to adjust the flwed tally. While Mosbacher claimed his decision was based on the scientific uncertainty of the adjustment procedure--an expert panel convened by the Commerce Department split evenly on the question--hopes of helping the Republican party seem a more probably explanation, especially if you consider the arguments advanced by Mosbacher's chief statistician, Barbara Bryant. Prior to Mosbacher's decision, Bryant, the director of the Census Bureau, noted that "statistical adjustment, while far from a perfect procedure, will on average increase the accuracy of the 1990 census. . . . [N]ot adjusting would be denying that these 5 million persons exist. That denial would be a greateer inaccuracy than any inaccuracies that adjustment may introduce." But denial, as Mosbacher went on to prove, is what right-wing number-crunching is all about.

Stat of the union

Since it's impossible to measure how much bad data is out there (they just don't keep statistics on this), no one really knows what this wealth of misinformation costs the United States. Studies such as one conducted by the Office of Technology Assessment in 1989 have concluded that "the cost of a poorly run government program may be many times higher than the cost of improvements to statistical agencies. Unlike other government purchases that can be postponed, statistics cannot be turned on and off--once a gap is created it cannot be easily eliminated." And that doesn't even begin to account for the fortune in lost business opportunities, misdirected health care, and misguided economic policies.

The root of all these holes, inconsistencies, and outright contradictions is clear: Statistics aren't considered important enough to warrant sustained attention and support. Just look at the numbers. Since the early eighties, statistical agencies have been starved of money and personnel--from 1980 to 1988, the six major statistical agencies in the federal government lost 13 percent of their constant-dollar funding and more than 10 percent of their staff, leading to sharp drops in the level of data collection and analysis. Even now, after some growth in these budgets in the late eighties, the federal statistical budget is no more than 1 percent higher in real terms than it was in 1980--a rate that amounts to a shrinkage of the total statistical effort against the growth of the nation's population. Since so much of this information collection is essentially invisible, the cutbacks provoked no anguished outcries from lobbyists or letter-writing campaigns to Congress. "You can squeeze the statistical system a lot without apparent damage," says Norman Bradburn, director of the National Center for Opinion Research and a professor at the University of Chicago. "It isn't the kind of think that rises up and hits you in the face."

Except for every now and then. Last year's budget negotiators, for instance, discovered what a statistical fog they were operating in when the five-year deficit projections were suddenly revised upward by $100 billion--all because the guesses BEA had made about unavailable salary and wage data turned out to be grossly wrong. Once again, bad data almost wound up being worse than no information at all, simply because no one had any idea there could be anything wrong with it. Such "imputations"--the estimates an agency will make when it lacks hard data--hide the informational vacuum at the heart of many important economic statistics.

But lack of resources alone can't explain why so many federal statistics bear so little relation to reality. Beyond the issue of funding, the diffuse nature of the information collection system breeds wasteful duplication, interagency rivalry, and gaps in statistical coverage. In 1984, for instance, deregulators succeeded in abolishing the Civil Aviation Board, which oversaw the airlines, and scaled back the Interstate Commerce Commission, which regulated the railroads and trucking. Both moves were hailed as great victories for the consumer, but somehow no one anticipated that regulatory data formerly collected by the two agencies would no longer be available to the BEA for income and product accounts that, among other things, help determine the GNP measure. Although other agencies have since begun collecting the information themselves, it took years for them to pick up the slack.

Go figure

Such obstacles are hardly insurmountable. An office with a broad perspective on the statistical system could iron them out, right? Well, sure. And such an office exists, at least in name. Buried deep within the Office of Information and Regulatory Affairs of the Office of Management and Budget (OMB) sits the office of statistical policy--the entity charged with coordinating all 70-plus statistical agencies. But if you were expecting a muscular operation to handle all these weighty responsibilities, think again. Although the office had more than 40 employees in the late seventies, it now has 5.

There are signs that the Bush administration is beginning to wake up to the crisis in the statistical system. Earlier this year, Michael Boskin, chairman of the Council of Economic Advisors, announced a plan to increase funding of several major statistical agencies by $30 million in order to improve the collection and analysis of information like inflation rates, productivity statistics, and trade between firms in various industries (a competitive statistic that the Japanese currently measure better than the Americans). That's a good start. But the new initiative does little to correct the glaring deficiencies in health, poverty, and demographic statistics--information on subjects that can affect the nation's economic health every bit as much as the trade deficit or the size of the money supply. In fact, Boskin did propose to increase spending on poverty statistics last year, but there's no mention of it in the current proposal.

The Boskin initiative also completely overlooks the structural weaknesses of the data collection effort. Granted, it's not an easy problem to solve. But it doesn't take much imagination to realize that seriourly beefing up the office of statistical policy could yield some major results. Moving it out of the lower levels of the OMB swamp and into a position where its director would report directly to the OMB director would have several advantages, not the least of which would be to make a single person accountable for the statistical system. Furthermore, by enhancing the stature of the nation's statistical advocate, such a move would provide him the clout to nudge agencies into closer coordination, to convince Congress to align the different legislative mandates of the various agencies, and in general to fight the hard battles for accurate and timely information that simply aren't being fought today.

An even more audacious plan would be to abolish the crazy quilt of federal statistical agencies altogether by merging them into a single office. And Boskin wouldn't have to travel far to find a good model. "Statistics Canada is a perfect example of an agency run on centralized data," says a Census Bureau official who requested anomymity. StatCan, as it's informally known, is a $294-million behemoth that collects information on everything from health to demographics to crime to epidemiology. It not only carries out a population count every five years, in contrast to the decennial U.S. census, it turns around economic data such as trade figures much faster than comparable American agencies. Although it's true that you can't compare the Canadian and American model directly--Canada, after all, has only one tenth the population of the U.S.--the proof of StatCan's efficiency is demonstrated by those who use its information. U.S. companies, for instance, frequently use StatCan data to figure out the dollar volume of goods exported to Canada by eact state. Such information is also available from the BEA--but it's generally at least five and a half years out of date.

Like the physical infrastructure, the statistical infrastructure can suffer neglect for only so long before it begins to take a heavy toll on the country. When American companies don't have access to the kinds of data foreign firms rely on, they are that much less able to compete. And when a flimsy statistical bridge gives way, as did the absurd accounting methods regulators used to monitor the savings and loans, it can suck an awful lot of unsuspecting Americans down with it. Who knows--with real reforms, the wheel-spinning bureaucrats at HHS might have enough information to make the next Healthy People 2000 somewhat less ludicrous than its predecessor. And the rest of us might have enough information to hold them to their promises. And then the federal bureaucracy might begin to get some real work done, by the numbers--the correct ones, that is.

David Hamilton is a reporter for Science. Research assistance was provided by Jennifer Bradley.
COPYRIGHT 1991 Washington Monthly Company
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1991, Gale Group. All rights reserved. Gale Group is a Thomson Corporation Company.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:dealing with Department of Health and Human Services documents
Author:Hamilton, David
Publication:Washington Monthly
Date:Oct 1, 1991
Previous Article:How Congress won the war in the gulf.
Next Article:Socialized medicine now - without the wait.

Related Articles
U.S. health: the good and the bad.
C-Section Rate Reduction.
Quality of Health Canada's statistics questioned by Auditor General. (Health).
Public policy. (Resources).
Pushing doors open: while addressing the problems of people with disabilities is on government agendas now, advocates say they're not moving fast...

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters