The Phantom Poll Booth
Who won the 1988 Iowa Democratic caucuses? We'll never know.
By William Saletan
This article was originally published in the June 1988 issue of American Politics.
At 9:13 P.M. on February 8, CBS News announced that Richard Gephardt had won the Iowa Democratic caucuses. Before the night was over, Gephardt had visited all three network sets, brimming with confidence over his comeback victory in the formative Democratic contest of 1988. Runner-up Paul Simon struggled to spin the outcome his way, but was told point-blank in two morning interviews that he had fallen short of expectations. Gephardt promptly jumped up 10 percent in New Hampshire polls, passing Simon for good. A week later, Simon finished behind Gephardt again and was effectively written out of the race.
But wait a minute. Did Paul Simon really lose in Iowa? The networks said he did, based on a mix of entrance polls, samples of key precincts and a county-by-county tally of votes compiled by a cooperative effort called the News Election Service. But a funny thing happened on the way to the White House. And maybe Paul Simon wasn't really the loser he was reported to be.
While ABC and CBS projected Gephardt the winner by three percentage points, NBC aborted—and didn't report until 24 hours later—its own projection showing Gephardt leading by only half a point. Why didn't NBC go public on caucus night with its projection? Because, in the words of NBC director of elections and polling Mary Klette, "you'd be absolutely crazy" to predict a winner with only a half-point margin in the key precinct sample. ABC director of political operations Stan Opotowsky agrees. "The point of key precincts is to tell you who won. A half a point doesn't mean a thing." In other words, if NBC's squelched figures were right, the race was simply too close to call.
To the network political pros, worrying about key precincts in retrospect is a pointless exercise. After all, the News Election Service's count of actual votes showed Gephardt winning by three points. "The official vote is that county vote as it comes in from N.E.S.," explains Klette. "Once that adds up to 100 percent, that's what decides who wins and who loses." Opotowsky concurs.
But questioning the precinct tallies isn't pointless, for three reasons. One, the N.E.S. count was never completed; it stopped with only 70 percent of precincts reporting. Two, it wasn't official. And three, a comedy of errors—attributable in part to the N.E.S., in part to its reporters and in part to the Iowa Democratic Party—rendered the N.E.S. count absolutely worthless as a final authority on the caucus outcome. Who won the Iowa Democratic caucuses? The truth is, nobody knows.
The Iowa Democratic Party thinks it knows who won. On caucus night, the party tabulated and reported the share of county delegates won by each candidate using a system of numerical thresholds and weights that makes a space-based laser missile defense system look like a junior high science fair project. According to the party's count, Gephardt won 31.24 percent of the weighted delegates to Simon's 26.68 percent—a margin of 4.56 points.
That's the sausage; here's how it's made. Say 100 people show up at precinct caucus X. The caucus chair designates eight stations around the room, one for each candidate and one for undecided. Twenty-eight people go to Simon, 27 to Gephardt, 18 to Dukakis, 9 to Jackson, 8 to Babbitt, 1 to Hart and 9 to Uncommitted. Now, caucus rules say you need 15 percent of the total to elect a precinct delegate. So the caucus chair announces that Jackson, Babbitt, Hart, Gore and Uncommitted aren't "viable." Members of those groups can then band together or defect to their second-choice groups. Say six Jackson people move to Gephardt, two to Simon, one to Uncommitted. The Babbitt group moves over to Uncommitted to elect one of its own as an "uncommitted" delegate who can vote for Babbitt at the county level; the Hart oddball defects to Gephardt; two uncommitted people defect to Gephardt, another to Dukakis. Now the chair counts again: Gephardt has 36; Simon, 30; Dukakis, 19; Uncommitted, 15. Of the seven delegates precinct X will send to the county convention, each group wins one delegate just for meeting the 15 percent threshold. Gephardt is awarded two of the remaining delegates; Simon gets one. So the chair phones in the results to county headquarters: Gephardt, three delegates; Simon, two; Dukakis, one; Uncommitted, one.
After all 20 precincts in the county have reported, the county chair phones in the total county delegates to state Democratic headquarters: Gephardt, 51; Simon, 41; Dukakis, 33; Jackson, 7; Uncommitted, 5; Babbitt, 3. Now, due to high Democratic turnout in this county in the last two election years, the Iowa Democratic Party "weights" these delegates to give the county more clout in the caucus process than another county, where Simon has won 30 percent of the delegates to Gephardt's 29 percent. In our hypothetical scenario, when the "weighted" delegates across the state are added up, Gephardt ends up with just over 31 percent, Simon just under 27 percent. That's the official result of the Iowa Democratic caucuses.
The network brass who deal with this process point out that it's incomprehensible, slow and seriously impure as an exercise in democracy. They spawned the N.E.S. in 1983 to eliminate these problems and to tell them who "really" won the caucuses. The N.E.S. was supposed to count how many Democrats went to the caucuses to pick each candidate as their first choice for the nomination—a simple straw vote. Above all, the N.E.S. was supposed to report those results quickly, so the networks could announce the winners during prime time on caucus night.
With nearly 2,500 precincts to cover, N.E.S. president Bob Flaherty sought the cooperation of the Iowa Democrats, the one organization that would have a representative—namely, the caucus chair—at each caucus anyway and, of course, the very organization whose electoral policies the N.E.S. would undermine. At a January 1987 meeting, Flaherty asked top party officials to make a straw vote part of the official process. The Democrats refused to sanction the N.E.S. count or to cooperate with the N.E.S. in reporting the initial head count. From the party's point of view, the N.E.S. was sloppy and had no business revising the caucus system for the sake of expediency. From the N.E.S.'s standpoint, the party was obstructing all attempts to render the caucuses fair and intelligible.
"Some of our people were excused from the caucuses because they were under 18 or because they weren't Democrats, which is not normally a test that one gives to reporters," says Flaherty. It would also be a breach of the state's open-meeting law. But Flaherty refuses to cite any incidents in which N.E.S. reporters were kicked out, and he declines to discuss specific complaints until more "research" is done.
In caucus training sessions across the state in January and in a letter just before the caucuses, the state party instructed precinct leaders to bar N.E.S. reporters from leaving to make phone calls to the N.E.S. (for fear their absence would affect viability counts) and from checking the registry of caucus-goers' initial preferences. But on caucus night, some counties allowed the N.E.S. reporters to caucus and collect the pre-viability count; many others didn't. Marty Ryan is the Democratic co-chair in Crawford County; his wife was hired by the N.E.S. as a reporter. When the N.E.S. sent her instructions to report the initial head count, Marty Ryan recalls, "I crossed out that part and sent it out to all the N.E.S. reporters in the county and said, Do not report these results." Only four precincts in the county defied him. Elsewhere, N.E.S. reporters showed up late, missed the initial counts and asked to peek at the registries; caucus chairs blew them off.
Aside from the party's attitude, four systematic problems rendered the N.E.S. figures useless.
2. Early second choices. "Right away, when they saw they didn't have, enough for viability, people started moving," says one N.E.S. reporter. "It was really hard to count them before they switched." Interviews indicate that this happened at about one of every three caucuses—generally the smaller ones. Here, the first preferences of many voters disappeared before N.E.S. reporters could record them—and an equivalent number of second choices contaminated what was supposed to be a pre-viability count.
3. Chaos. "There were so many people in the room," explains one exasperated Des Moines caucus-goer, "you couldn't tell where one group stopped and the other group began." One western Iowa county leader says that at most caucuses he's seen, "Unless you're standing five feet above everybody, you can't see the actual movement." This was the case in a third of the caucuses—generally the larger ones.
4. Incomplete counting. "If there's 100 people in a caucus, and there's 50 people over there, you know that's a viable group," says Crawford County's Ryan. "They're not even gonna count that until final alignment." This happened in about a fourth of the caucuses.
5. Procedural quirks. One caucus erupted into a fistfight; another caucus consisted of 132 participants who decided to count themselves as only 129; another voted to start all over after some Hart people complained they had been betrayed by one of their own. One N.E.S. reporter was told she couldn't obtain any figures from her caucus because the results could appear on television sets at living-room caucuses and prejudice their results.
N.E.S. corner-cutting only made matters worse. Many of the people hired as reporters were minors recruited by high school government teachers and youth group directors. Most of the kids were told they could get the raw vote from the caucus registry—precisely what the Democrats had already ruled out. Some kids tried to read the registries upside down. Some freaked out and left.
Union County youth-group director Les Sallee says the N.E.S. asked him to find kids to cover the county's 17 precincts as part of a "citizenship development" program. Were the kids up to the task? "More or less," he ventures. Apparently, less. "Some of these kids came to me and wanted to know what they were supposed to do," says Lee Campbell, a local teacher. The N.E.S. reporters he saw ranged from 15 to 17 years old. (Interestingly, 15 to 17 was also the ratio of Union County caucuses reported to caucuses held.) "I'm not sure that the young people that they had out there were getting the information that they were supposed to," says Campbell. Another 4-H director asked each of her kids to cover two caucuses at once, gambling that the initial counts wouldn't be taken simultaneously; they missed two precincts. (A 4-H officer in a third county missed six.) A sorority president in Osceola tried the same stunt. She missed two of three initial counts and ended up reporting only 3 of 10 precincts to the N.E.S.
Songwriter Dan Hunter relates his conversation with a high school student reporting for the N.E.S. in Des Moines.
Hunter: What are you doing?
N.E.S. rep: I don't know.
Hunter: Why do you have a Simon sticker on?
N.E.S. rep: I'm with his group.
Hunter: The hell you are. You can't caucus with them if you're an N.E.S. counter.
N.E.S. rep: What am I supposed to do?
Hunter: Well, you dumb shit, didn't they tell you what to do?
N.E.S. rep: Well, I was just supposed to count things, right?
Hunter: What are you getting paid for this?
N.E.S. rep: I don't know.
Hunter: Why are you here?
N.E.S. rep: I'm raising money for my wrestling team.
Out of sheer pity, Hunter gave him the count—"The final count, the delegate count. If I had had an axe to grind, I could have told him anything."
Not that the adult N.E.S. reporters couldn't have used some training, too. "It was a very sloppy business," one says. "I did not know how to do it. I thought that the secretary was keeping track of those numbers at the very beginning, and she did have some figures down. But when I went to pick them up, some of those people had moved. I know it wasn't exact. So I had to just kind of guess. There were some real large groups, and I couldn't be absolutely sure of the numbers." This reporter ended up using the caucus registry, which she and several others hadn't signed, to report the raw vote.
"We were just sort of dumped out there," says another. "The News Election Service had not asked for any training, as far as I know, and I didn't see that any would be needed." So how did it go? She laughs: "Actually, I ended up not calling in the initial results, because we forgot about it." This reporter was a caucus chair. She and her friends, none of whom were trained, had been selected by the N.E.S. as a particularly able group deserving first choice among the precincts in her county. "I thought we had problems," she recalls. "Then I heard about some of the other precincts and I thought, Well, we're not too bad off."
A third N.E.S. reporter, confused by what one of her reporters told her after the caucuses, recalls: "He said the figures were all messed up that were reported to the News Election Service."
All N.E.S. reporters were given badges and were instructed to identify themselves to the caucus chair when they arrived, so they should have been noticed wherever they were. But county and caucus chairs say the N.E.S. simply missed some precincts—in Davenport, in Council Bluffs, in South Iowa and, according to state Democratic officials, in other places across the state. Several N.E.S. county officers admit that some of their people just never called in the results, and perhaps didn't even show up.
Aside from the chaotic fieldwork, N.E.S. headquarters managed some bumbling of its own. One western county chair says he and his wife, an N.E.S. reporter, tabulated identical counts of the initial preferences in his precincts and relayed them to the N.E.S. But although the figures the N.E.S. passed on to the networks were based on reports from only 9 of 20 precincts, they showed more people present than voted in all 20 precincts. The county chair says he investigated until he located the N.E.S.'s mistake, but that the N.E.S. has refused to cooperate. He and other county chairs suspect that the local figures published by the N.E.S. were "totally fictitious."
Ironically, network political chiefs don't even seriously consider the possibility that the N.E.S. screwed up the count. "Where do you think N.E.S. got it?" asks ABC's Opotowsky. "From the party. The N.E.S. reporter's job is to phone them in. He doesn't count anything. The party officially tells him what the count is." And what if the reporter gets confused? Opotowsky sighs. "One thing you have to understand," he explains gently, "the N.E.S. reporter's not somebody who just wandered in out of the cold. This is probably his tenth Iowa caucus, in each case."
The contrast with reality could not be any more stark. If the networks thought they were getting some kind of official count of the caucuses, they were ripped off. And the viewers who watched Brokaw, Rather and Jennings rattle off N.E.S. numbers all night on February 8 were ripped off. It's possible that Paul Simon was ripped off too. Skeptics, including some Simon aides, say Gephardt's four-point lead in the final delegate count shows he probably had more raw votes. But the hypothetical scenario sketched out above shows how a Simon win in the raw vote could turn into a four-point Simon loss in the delegate count. Crawford County chair Ryan thinks that's implausible. Yet, in one of his precincts, Simon had three-fourths of the caucus-goers but was awarded only one of two delegates. And in the translation of final head counts into Crawford County delegates—a single step in the complex chain of equations—Simon dropped from 26 percent to 23 percent, while Gephardt rose from 46 percent to 47 percent. Little by little, these things add up.
Maybe Paul Simon did lose the Iowa straw vote after all. But at the very least, that devastating verdict, if it was to be reported at all, should have been based on more than a probability curve.
In the last days before the caucuses, buoyed by the Des Moines Register's endorsement, Simon aides echoed the Babbitt campaign's mocking line on Gary Hart: "Let the media decide." On February 8, that's what happened.