New to the issue? Need a refresher course? Need to read every book about race and politics? Or do you just want to see how all those conversations with Skerry over the years were summarized? If any of these apply, this is the book for you. Skerry's new book is the perfect beach reading for census policy wonks for the summer of 2000. You can ponder the following as your head slowly shifts, eyes attracted to some "eye-magnets" shaking their way down the sand.
Really, the book is an easy read, by wonk standards that is. More importantly, it deals with a serious issue and digests years of research, dozens of reports from dozens of sources, and hundreds of hours of hearings into an understandable summary.
Of course, Skerry doesn't hit all the angles. For the proponents, he glosses over their charge that this is the Civil Rights issue of the decade and that the only reason Republicans are against adjustment is thinly disguised racism. On the opponents' side, he forgets to mention that it is the Clinton Administration's own previous actions (e.g., Filegate and the expedited processing at the INS) that form one of the main reasons why Republicans in Congress feel they can't trust these folks to implement a complex and untested system as adjustment which, by its very nature, is infused with opportunities for judgement calls. "[P]olitics permeates even the most technical aspects of sampling and adjustment."
These few oversights aside, Skerry's review of the issue is a useful study of the causes of the undercount, the measurement of it, and the reasons for or against the remedy to 'fix' it.
Skerry points out that the "scientific sampling" to be used in the post-census survey that forms the basis for the adjustment (the ICM), is in reality a different kind of sampling with which we are all familiar. "Typically, sampling is used to draw inferences about average properties of the whole from its parts. But this is not the case with sampling for the census adjustment." Quoting one of dozens of statisticians, he points out the difference. "'Ordinarily, samples are used to extrapolate upwards, from the part to the whole. Census adjustment extrapolates sideways, from 60,000 sample blocks to each and every one of 5 million inhabited blocks in the U.S.'" Likewise, he points out that the model used by the Bureau is not, in the minds of several statisticians, "robust" enough to complete the mission at hand.
He also addresses a perspective central to the entire debate in our modern polity. We frequently read that the census should not be politicized and should be "a politically neutral government task". This involves politics within and without the government. Skerry's discussion reviews the bureaucratic problems associated with viewing the Bureau as an apolitical entity and also points to the compromises that are made by the Bureau to accommodate political pressures from Capitol Hill and elsewhere. Yet, Skerry points out the very nature of the census is political, just another issue in an untidy world of politics in the U.S., just as the Framers intended it to be. He also looks at the "two pillars" of adjustment advocates, "science and rights. Americans have long relied on both to get around the untidiness of politics. But when it comes to the census, the invocation of rights claims and appeals to the authority of science are particularly wrongheaded."
I did find a few inconsistencies. For example, "Georgia would almost certainly had lost funding on account of adjustment" even though a table subsequently lists the undercount rate for Georgia as being higher than the nation's for both versions of the 1990 undercount rates. Likewise, my own review of the issue leads me to disagreement with Skerry's take on a few perspectives. For example, in his specific analysis about California, his minimalization of the potential impact of adjustment on the actual drawing of lines for Congressional districts, and his stray language about the failure of the uncounted to vote. Or, that "no one has ever received any affirmative action benefit from self-identifying as a member of a specific minority group..." This depends upon ones definition of affirmative action and the application of the Voting Rights Act.
Skerry, who now teaches Political Science at Claremont-McKenna, doesn't neglect his discipline. Aside from several quotations from political scientist and current Bureau director Kenneth Prewitt, he also includes a straight political essay on the role of interest groups in the politics of today. He argues that not all minority groups have been fully behind the adjustment effort and points outs an eventual drawback of any adjustment. "[I]f statistical adjustment were implemented, then the political action would shift away from local jurisdictions and minority groups pressuring the Census Bureau to reach out to their constituents (and pressuring their constituents to stand up and be counted), and towards experts and bureaucrats haggling over complex technical issues... Even local leaders and officials would be enveloped by a much tidier but more restricted insider politics."
The book also brings to the front numerous tidbits of information that are often forgotten, or were never heard, by those deeply involved in the process. Perhaps the most interesting is the Bureau's own assessment of the adjustment process for 1990 (which was not implemented). The so-called CAPE report brought the news that computer errors had revised the original assessment of the national net undercount from 2.1% down to 1.6%. Probably most of us had heard then Secretary of Commerce Mosbacher, who headed the "Bush Gardens" of political appointees, decided against the use of adjustment for the 1990 results even though his Census Bureau Director, Barbara Bryant, advocated it. However, Skerry states that after the CAPE report was released in 1992, Director Bryant "intimated that on the basis of the 1.58 percent undercount figure she, too, might have changed her position." Quoting Bryant, "The lower the undercount, the harder it is to be sure you're doing the right thing on adjustment."
Another useful fact deals with the commonly heard statement that the 1990 census was the only one in history to be worse than the one before it. This was based on the undercount rate from 1980 as being 1.2% as measured by Demographic Analysis. As Skerry's table indicates, and the accompanying chart illustrates, 1980 was the anomaly, the statistical outlier. This fact involves the techniques used in the nascent days of adjustment in 1980. "Although it registered a very low undercount, the 1980 census involved considerable double-counting such that most experts regard it as a near fiasco. At the bureau the old hands say: 'In 1970 we made sure everyone was counted only once, but in 1980 we made sure everyone was counted at least once.'"
Skerry has done an admirable job of wresting some structure from the sometimes acrimonious debate which still continues in Congress. Just last month (May 2000) the House Subcommittee on the Census held another oversight hearing with Bureau Director Dr. Kenneth Prewitt as the chief witness. The hearing dragged up much of the pro and con on the issue from the past few years and rehashed it for yet another time. It reminded be of some talk show that aired briefly sometime ago with George Hamilton and his former wife, Alana Stewart. Perhaps a new talk show can be aired, the Dan Miller and Carolyn Maloney show.
Until then, read Skerry's book and arm yourself with the facts.