An Appearance of Corruption
The bogus research undergirding campaign finance reform.
May 26, 2003, Vol. 8, No. 36 • By DAVID TELL
IN EARLY 1999, Brennan Center staffers finalized plans for a research project led by an Arizona State University political scientist named Kenneth Goldstein. With funds to be provided by the Pew Charitable Trusts, Brennan would purchase an archive of "storyboards"--full scripts and selected still-frame images--from thousands of political ads broadcast during calendar year 1998. Goldstein would then assign a team of student volunteers to review those storyboards and classify each ad as either an honest-to-goodness public policy pronouncement or an electioneering message in disguise. Once this "coding" task was finished, Goldstein would assemble a computer database recording his students' subjective judgments about the ads along with documentary information about when and where the ads had aired and whether they'd mentioned a candidate for federal office. Whereupon Brennan Center political scientist Jonathan Krasno would analyze the contents of Goldstein's database against the broadcasting prohibitions contemplated by the Snowe-Jeffords Amendment. Among interest-group-sponsored political ads that targeted a specific federal candidate's voters during the final 60 days of the 1998 general election, how many were "genuine," and how many were not? The smaller the number of "genuine" ads regulated by Snowe-Jeffords, the better would be the law's chances of surviving judicial review.
The successful grant proposal that Krasno wrote for Pew--on instructions from Brennan's president, Joshua Rosenkranz, to emphasize "the pop and sizzle stuff, not the research"--made explicit that the project was principally a hunt for political ammunition. "Issue Advocacy: Amassing the Case for Reform," dated February 19, 1999, explained that "[t]he purpose of our acquiring the data set is not simply to advance knowledge for its own sake, but to fuel a continuous multi-faceted campaign to propel campaign reform forward." Dispassionate academic inquiry was so alien to the spirit of the thing that Brennan promised to suspend its work midstream, pre-publication, if the numbers turned out wrong. "Whether we proceed to phase two will depend on the judgment of whether the data provide a sufficiently powerful boost to the reform movement."
Before that year was out, Goldstein and his students had completed the storyboard-coding effort, and Krasno was busy convincing himself that the "sufficiently powerful boost" Brennan had hoped for was present in the results. Krasno's report to that effect, "Buying Time 1998," was still at the printer when, on April 26, 2000, Rosenkranz testified about its findings at a Senate Rules Committee hearing on the Snowe-Jeffords Amendment. Given "solid empirical data" recently obtained by the Brennan Center, Rosenkranz advised, Congress can be "confident that the major campaign finance proposals currently before it do not inhibit true issue advocacy." Only "7 percent of ads categorized as genuine issue advocacy would have been affected" had Snowe-Jeffords-style restrictions been in place for the last 60 days of the 1998 campaign.
E-mail and suchlike records produced under subpoena during the McConnell trial, however, indicate that Rosenkranz himself must no longer have been "confident" in Jonathan Krasno's "solid empirical data," even as he was describing it that way to the senators. The very next day, Rosenkranz wrote a forceful memo to his donors at Pew, apparently reiterating an earlier SOS about the "Buying Time" report: "I continue to believe that we cannot be put in a position where we Brennan Center staff are expected to rely upon or collaborate with Jon" for further advertising research. "It's a recipe for disaster," Rosenkranz went on, announcing that the Brennan Center would refuse involvement with any such study "in which Jon's role (even if artificially downplayed and insulated) is significant." That would be "an untenable situation." Krasno was then abruptly fired. It was a "personality conflict," he has since insisted.
But it seems to have been more than that, too. It seems that other Brennan Center staffers were having considerable difficulty verifying Krasno's projection that genuine issue ads would make up just "7 percent" of the political broadcasts prohibited by Snowe-Jeffords. A research assistant named Luke McLoughlin ran appropriate calculations through the Goldstein database and repeatedly produced a much higher and thus politically inconvenient figure: 38.5 percent. A professor independently commissioned by Brennan to work with the database arrived at an eerily similar conclusion: 40 percent. As late as January 2001, eight months after "Buying Time 1998" was published and widely circulated, no one in the Brennan Center's ambit had yet found a way to budge this number.
Krasno's successor at Brennan, Craig Holman, sent Rosenkranz a series of downbeat e-mails about the rather "awkward" situation they confronted. "Luke and I have run over it many, many times," Holman confirmed. The "7 percent" business "can be a little misleading," as he delicately put it. "[I]n reality, according to the 1998 database, about 40 percent of genuine issue ads would be deemed electioneering within a 60 day regulatory period." Rosenkranz wasn't happy. "Here's the problem," he wrote back to Holman. "Our findings are not just misleading; as I read them, they are flat-out false." Luke McLoughlin subsequently managed to "cut some new numbers" and jigger the 40 percent statistic back down into the teens. But Rosenkranz's overbreadth-analysis "comfort" zone was never recovered.
Neither, on the other hand, did the Brennan Center ever publicly express so much as a hint of doubt about the validity or significance of Krasno's original 7 percent analysis. Quite the contrary: In mid-March 2001, barely two months after Rosenkranz had privately dismissed that analysis as "flat-out false"--and, not incidentally, immediately in advance of a crucial Senate debate on the Snowe-Jeffords Amendment--Brennan issued a splashy open letter citing "Buying Time 1998," and nothing else, as proof that the measure was constitutional. "This empirical evidence demonstrates that the Snowe-Jeffords criteria are not 'substantially overbroad,'" the letter proclaimed--on the authority of 88 signatory "constitutional scholars" led by luminaries like Ronald Dworkin and Norman Ornstein.
Two days later, the Senate debate still underway, Rosenkranz and his colleagues pulled off a second successive publicity coup: a press release touting preliminary results from what would eventually become an updated volume of Brennan quasi-scholarship, "Buying Time 2000." Kenneth Goldstein had relocated to the University of Wisconsin. His students there had completed a fresh round of storyboard-coding work. And that work, Brennan's press release suggested, "directly rebuts the dire 'sky is falling' predictions" about Snowe-Jeffords that non-profit groups and labor unions were then beseeching their senators to heed. Specifically: If Snowe-Jeffords had been in force during the waning weeks of the 2000 election campaign, "only three genuine issue ads," representing "fewer than one percent" of total interest-group ad buys, would have been "improperly" suppressed.
These "facts," too, as the Brennan Center knew quite well, were dubious. But they helped do the trick and, just as with "Buying Time 1998," Brennan has never stopped publicly boasting about them. The Center's "real contribution to the debate came with our unprecedented empirical work," Rosenkranz wrote in spring 2002, memorializing McCain-Feingold's enactment into law. "The Congressional Record was rife with references to our data and our analyses at every step of the way. So was the popular press. From the New York Times to the Los Angeles Times, journalists and editorial boards consistently turned to us both for legal comment and empirical support."
That much, at least, was true.
AMONG THE FIRST THINGS the McConnell plaintiffs' attorneys did to prepare their trial case against McCain-Feingold was to recruit James Gibson, a well-known government professor at Washington University in St. Louis, to critique the Brennan "Buying Time" studies as an expert witness. Gibson was perplexed by the computer disks and documents obtained through discovery for his review, especially those related to Jonathan Krasno's work. Brennan's 1998 advertising database, according to the report Gibson prepared for the court, was "riddled with internal inconsistencies and errors." Though multiple iterations of that database were preserved on the disks he'd been given, Gibson was able to replicate "virtually none" of the results "Buying Time 1998" had claimed to derive from it. Gibson did manage to trace the origin of one such result, however--the 7 percent figure that had so irked and mystified Krasno's officemates. The numerator was the expected one: genuine issue-ad broadcasts that the Snowe-Jeffords 60-day blackout would have unfairly swallowed up. But the denominator was an apples-and-oranges anomaly: all issue-ad broadcasts for the entire calendar year.
Krasno's argument for the mildness of Snowe-Jeffords's mischief, in effect, was like a playground bully insisting that the lunch money he'd stolen in May represented only a very small fraction of the lunch money his classmates had collectively carried since September. Retabulated in the proper fashion, so far as Gibson could determine, "Buying Time 1998"'s results actually meant that a shockingly large number of the issue ads silenced by Snowe-Jeffords--44.4 percent of them--wouldn't be "shams" at all.
And there was worse. Gibson was delivered a small sample set of the primary-source material used by Kenneth Goldstein's Arizona State University students: 25 of the 1998 storyboards, matched up with the "coding sheets" on which those students had registered opinions about the presence or absence of an "electioneering" purpose in the broadcasts. Chancing to check the coding sheets against related entries in different versions of the 1998 database, Gibson established, to his amazement, that eight of the original student scorings, almost a third, had been altered after the fact. At some point during the preparation of "Buying Time 1998," probably in late January 2000, someone had accessed the Goldstein/Brennan computer files and switched at least eight ads, representing a significant chunk of the study's ostensible broadcast-airtime baseline, out of the "genuine" category and into the "sham." The "sufficiently powerful boost to the reform movement" referred to in Krasno's Pew grant proposal, and provided by means of his creative 7 percent construction, had itself been boosted rather powerfully--and was phonier, therefore, than even Joshua Rosenkranz had feared.
Correcting for Krasno's numerator/denominator subterfuge and for the demonstrable manipulation of Brennan's underlying data, "Buying Time 1998" should more honestly have reported, Gibson advised the McConnell judges, that the vast majority of political advertisements likely to be kept off the air by the Snowe-Jeffords Amendment, 64 percent of them at minimum, would be perfectly unobjectionable, altogether genuine issue broadcasts. Snowe-Jeffords, it appeared, was unconstitutionally "overbroad," just as those "sky is falling" critics had alleged--to an almost unimaginable, spectacular extent.
BY THE SOUND of the transcripts, last fall's McConnell deposition sessions must have been a sweaty-palms affair for the current and former Brennan Center staffers called to testify. Each proved loyal to the cause, however, and suffered personal embarrassment rather than help corroborate the now obvious-beyond-question First Amendment infirmities of McCain-Feingold's issue-ad provisions. Each, for example, eagerly repudiated his own well-documented misgivings over the accuracy of Jonathan Krasno's 7 percent figure. In retrospect, Craig Holman said, that number was "not untrue." The accounting procedures embodied in Krasno's whole-year denominator were not necessarily unjustified. And, no, the legal and political inferences Congress had drawn from "Buying Time 1998" were not the wrong ones: It "depends on how you measured" things. Asked when it was that he'd revised his thinking on the matter--deciding that Krasno had been right all along--Holman paused a moment and said: "within the last couple of weeks." Asked the same question, Luke McLoughlin, who'd offered much the same sort of self-abasing testimony, replied: "yesterday."
No one representing the Brennan Center acknowledged responsibility for the ex post facto score-switchings Professor Gibson had uncovered in the 1998 advertising database--or claimed to know who'd done the deed. But Gibson had also identified six instances of equivalent corruption in the "Buying Time 2000" computer disks. And Kenneth Goldstein, many reluctant hours into a day-long deposition last October 24, finally fessed up to those. Three of the changes were made in mid-March 2001--while Goldstein was on vacation, standing in a Florida airport. His cell phone rang. The "Brennan Center called me up" and "asked me to give another assessment of three ads," he recounted. "And I gave them that assessment," overruling the judgments of student researchers who had examined the storyboards. Why the urgency, an attorney for the McConnell plaintiffs wondered? "I believe that was the time McCain-Feingold was being debated in the Senate," Goldstein remembered. "Brennan wanted to be able to write a report or have a press release . . . and talk about the 2000 data." As we have seen, Brennan did issue such a press release--claiming that "fewer than one percent" of the issue ads affected by Snowe-Jeffords would be "genuine." Had six ads in the 2000 database not been altered from "genuine" to "sham," Goldstein conceded, that figure would have ballooned to 17 percent, a constitutionally unsustainable level.
Ultimately, even the McConnell defendants' own independent expert proved unwilling to stake his reputation on the integrity of Brennan's "Buying Time" exercises. Arthur Lupia, a comically self-important University of Michigan political scientist--one of the "2000 Outstanding Scientists of the 20th Century," his curriculum vitae points out--buckled under questioning from legendary First Amendment lawyer Floyd Abrams. Are the academic standards on display in Brennan's grant-proposal research prospectus "appropriate," Abrams wanted to know?
LUPIA: In what sense do you mean "appropriate"?
ABRAMS: You don't understand the word?
LUPIA: Well, I do, sir, but there are several senses in which one could ask me the question.
ABRAMS: Do you think it is consistent with scholarly behavior of honor and seriousness as a member of a learned profession?
LUPIA: No, that I do not.
But partisan motivations alone, though generally frowned upon in the learned professions, do not by themselves impeach the results of a scholarly project, Lupia protested. People don't think cancer research is useless simply because the physician conducting that research hopes to eradicate the disease.
ABRAMS: Are there any pro-cancer doctors that you know of?
LUPIA: No, sir.
Karen LeCraft Henderson, roundabout page 90 of the 345-page McConnell v. FEC opinion she signed May 2, offered a thumbnail sketch of the "Buying Time" evidence on which Congress had relied when voting to incorporate the Snowe-Jeffords Amendment into the Bipartisan Campaign Reform Act of 2002. The judge said very little in her own voice about the Brennan Center research. Even a rigorously deadpan description of that research, Henderson understood, would more than adequately explain why she felt the federal courts could not afford to rely on "Buying Time" at all. Columnist George Will noticed Henderson's reference to Brennan's role in the litigation--a subtext of the McCain-Feingold legal battle that has otherwise eluded media attention--and wrote a piece making passing mention of the "Buying Time" project's intellectual "debasement." To which characterization Frederick A.O. Schwarz Jr., the Brennan Center's interim president, vehemently objected. Notwithstanding Will's "resuscitation" of "discredited" attacks against his institution, Schwarz rejoined, the McConnell court--"two of three judges on the panel," at least--had "found the research in the Center's "Buying Time" reports to be credible and useful," and had "followed Congress in embracing [their] central findings."
Mr. Schwarz is misinformed.
Judge Henderson, of course, rejected both "Buying Time" studies, outright, as "flawed" beyond repair. Judge Leon, for his part, rejected "Buying Time 1998"'s Snowe-Jeffords data, as published, because it measured ads "that never would have been regulated by BCRA." Judge Leon did accept "Buying Time 1998'"s Snowe-Jeffords data--as recalculated the way he thought it ought to have been. But he only did so, ironically, because the recalculated data confirmed his suspicion that the measure's 60-day electioneering definition suffered "real and substantial" overbreadth, and was consequently unconstitutional. Judge Leon accepted "Buying Time 2000"'s Snowe-Jeffords data on the same, subversive terms: Properly interpreted, he wrote, the numbers signify a "realistic danger that the statute will significantly compromise recognized First Amendment protections." Even Judge Kollar-Kotelly, who voted to uphold the 60-day test, on the basis of "empirical" evidence in "the record" that she did not name or describe, felt obliged to refuse the specific empirical evidence she'd been offered by the McConnell defendants. She was "troubled" that student coding decisions had been altered in both the "Buying Time" studies. And, she added, "I do not accept" the first study's central conclusion about 1998. And "I cannot accept" the second study's central conclusion about 2000.
The Snowe-Jeffords section of the campaign reform act contains a "backup" electioneering rule that Congress intended would take effect if and when the "primary" broadcast-restriction criteria were invalidated by the courts, which is what's just happened. As edited and endorsed by Judges Leon and Kollar-Kotelly, the backup provision would prohibit unions and corporations from making direct, unregulated expenditures on any broadcast communication that "promotes or supports," or "attacks or opposes," any "candidate for federal office"--anywhere, for any reason, in any month of any year. Judges Leon and Kollar-Kotelly have neglected to cite constitutional precedent for this novel piece of censorship. And there isn't a knowledgeable lawyer in the country who thinks the Supreme Court will stand for it, even in its slightly less draconian, pre-McConnell form. Proponents of the McCain-Feingold "sham ad ban" must hope, instead, that the Supreme Court will look afresh at the trial-record facts, perform a brand-new overbreadth analysis on the original Snowe-Jeffords electioneering definition, persuade itself that the provision is constitutional, and reverse the trial judges' contrary ruling.
Trouble is, the Supreme Court may look afresh at the trial-record facts all it likes; those facts won't change. The credibility of "Buying Time," what the Bush administration's most recent McConnell defense brief called "the only empirical evidence . . . provided to the court" for the Snowe-Jeffords Amendment's constitutionality, is in ruins.
The credibility of New York University's Brennan Center for Justice isn't in such great shape, either.
David Tell is opinion editor of The Weekly Standard.