polls and models: which numbers to believe?

11
This article was downloaded by: [Universitat Politècnica de València] On: 25 October 2014, At: 07:27 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK The Washington Quarterly Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/rwaq20 Polls and models: Which numbers to believe? Charles E. Cook Jr. a a Political analyst for CNN Published online: 07 Jan 2010. To cite this article: Charles E. Cook Jr. (2000) Polls and models: Which numbers to believe?, The Washington Quarterly, 23:4, 193-202 To link to this article: http://dx.doi.org/10.1162/016366000561303 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/ page/terms-and-conditions

Upload: charles-e

Post on 28-Feb-2017

212 views

Category:

Documents


0 download

TRANSCRIPT

This article was downloaded by: [Universitat Politècnica de València]On: 25 October 2014, At: 07:27Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

The Washington QuarterlyPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/rwaq20

Polls and models: Which numbers tobelieve?Charles E. Cook Jr. aa Political analyst for CNNPublished online: 07 Jan 2010.

To cite this article: Charles E. Cook Jr. (2000) Polls and models: Which numbers to believe?,The Washington Quarterly, 23:4, 193-202

To link to this article: http://dx.doi.org/10.1162/016366000561303

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information(the “Content”) contained in the publications on our platform. However, Taylor& Francis, our agents, and our licensors make no representations or warrantieswhatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions andviews of the authors, and are not the views of or endorsed by Taylor & Francis. Theaccuracy of the Content should not be relied upon and should be independentlyverified with primary sources of information. Taylor and Francis shall not be liablefor any losses, actions, claims, proceedings, demands, costs, expenses, damages,and other liabilities whatsoever or howsoever caused arising directly or indirectly inconnection with, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden.Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

/ -/

// - .

Charles Cook on Washington

Polls say Bush will win,Models say Gore;Our man about WashingtonCan give you more... ,

. |»'.-,

;'%M

'*•-—-pm^

Dow

nloa

ded

by [

Uni

vers

itat P

olitè

cnic

a de

Val

ènci

a] a

t 07:

27 2

5 O

ctob

er 2

014

Dow

nloa

ded

by [

Uni

vers

itat P

olitè

cnic

a de

Val

ènci

a] a

t 07:

27 2

5 O

ctob

er 2

014

Charles E. Cook Jr.

Polls and Models:Which Numbers to Believe?

TJL he general election campaign is formally beginning for what is

arguably the most important U.S. election, in terms of how much is at stake,since the end of World War II. It is not that this election reflects some greatpartisan, ideological, or philosophical fork in the road. Indeed, America'stwo major political parties, almost always relatively centrist, certainly by Eu-ropean standards, are ideologically closer together than any time in memory.Democrats just slightly left of center, Republicans ever so slightly right ofcenter. It is how much is at stake this year that is so unusual.

The House is the closest it's been since 1958—split 50.5 percent Repub-lican to 49.5 percent Democratic. In this extraordinary setting, even the ti-niest movement could tip the House one way or another. At the same time,it's very unlikely that either party will win true, working control of theHouse. For one thing, there are far too few competitive races to make it easyfor either party to pick up a great deal of seats. There is an unusually smallnumber of retirements this year, limiting the volatility because between 88and 98 percent of all incumbents get reelected cycle after cycle. Few openseats usually suggest a narrower trading range for the House. Many of theweakest Democratic incumbents were defeated back in their 1994 tidalwave loss. Similarly, many highly vulnerable Republicans lost in 1996 or1998, two down election years for the GOFJ which culled the herd of manyweaker members. It is also true that neither party is behaving in a suicidalfashion (see Democrats circa 1993 and Republicans circa 1995, 1996, andlate 1998), the kind of conduct that tends to trigger big swings against one

Charles E. Cook Jr. writes weekly columns for the National Journal magazine andCongressDaily AM, published by the National Journal Group. He is a political analystfor CNN and the editor and publisher of the Cook Political Report, a Washington-based,nonpartisan newsletter analyzing U.S. politics and elections.

Copyright © 2000 by The Center for Strategic and International Studies and theMassachusetts Institute of TechnologyThe Washington Quarterly • 23:4 pp. 195-202.

THE WASHINGTON QUARTERLY • AUTUMN 2000 195

Dow

nloa

ded

by [

Uni

vers

itat P

olitè

cnic

a de

Val

ènci

a] a

t 07:

27 2

5 O

ctob

er 2

014

\ Charles E. Cook Jr.

party or the other. Finally, with the economy strong and voters showingsigns of relative complacency, it seems very unlikely for a large number of in-cumbents to be washed out to sea or for either party to be strongly rewardedor punished. In short, all the signs point to very little volatility. It is thus un-likely that either party can build up the kind of majority that gives a partytrue control of a body.

Increasingly, political observers of the Senate are looking beyond just theNew York race, with a dozen good races out there and the chances of Demo-

crats scoring gains not only real but increas-ing, though winning a majority is still likely

More is at stake inthis election than anyother since the endofWorldWarll.

to be beyond their reach this time. Republi-cans have to be really worried about theiropen seat in Florida, where Connie Mack isretiring, and about incumbents Bill Roth inDelaware, Spencer Abraham in Michigan,and Rod Grams in Minnesota; all three arein very serious races. Facing challenges al-most as formidable are John Ashcroft in Mis-

souri, Conrad Burns in Montana, Rick Santorum in Pennsylvania, and SladeGorton in Washington.

Democrats, though, have their work cut out for them in Nevada, whereDick Bryan is retiring, and Republicans are favored to score a pick up, aswell as other open seats in Nebraska, New Jersey, and New York, where BobKerrey, Frank Lautenberg, and Daniel Patrick Moynihan, respectively, arestepping down. Most estimates today suggest that Republicans will lose asfew as one or two Senate seats to as many as three or four. If Republicanswere to break even or limit their losses to just a single seat, it would be de-moralizing for Democrats, whose minority status in the Senate would be ex-tended to its longest period since 1933. On the other hand, a three—orfour—seat gain, just short of the five- (with a presidential victory) or six-seat gain required for a majority, would still be a very big win for Democrats,putting them within reach of a majority in 2002, when 20 Republican seatsare at risk, compared to only 13 for Democrats.

Then there's the presidential race, with two candidates whose campaignskills are remarkably balanced. Texas governor George W. Bush—a year agohailed as something just short of a second-coming in the Republican Party—is certainly improved as a candidate but is not nearly as accomplished inspeaking before large groups, on television, or in debates as he is in smallgroups. Rather than being a 9 or 10 on a 10-scale, he's turned out to beabout a 6. Conversely, Vice President Al Gore, who had a well-earned repu-tation not only as stiff and uninspiring but also as condescending and lack-

THE WASHINGTON QUARTERLY • AUTUMN 2000196

Dow

nloa

ded

by [

Uni

vers

itat P

olitè

cnic

a de

Val

ènci

a] a

t 07:

27 2

5 O

ctob

er 2

014

Polls and Models

ing a common touch, is much improved but still not great, coming in asabout a six as well.

Swing voters see Gore as smarter, better informed on the issues, and moreexperienced, at least relevant to the presidency, but also as too political, toopartisan, and transparently opportunistic. They tend to like and trust Bushbut still wonder whether he is smart enough, knows enough, and has theright kind of experience. They would rather not vote for Gore, but theyknow he is qualified; they would rather vote for Bush, but they are not surehe is. In the end, the debates may well make the difference.

Schizophrenic Polls?

In late May, conventional wisdom seemed to say that Bush had built up alead sufficient to wrap up the 2000 presidential election. Just a couple ofweeks later, one poll put the race even; another put the race within a point,prompting some to suggest that Gore had made a comeback and that now itwas Bush on the ropes. Both points of view suggest a fundamental misunder-standing of, and/or misuse of, political surveys. In truth, this race has hardlymoved in months. There was one brief period, just after Bush's New Hamp-shire primary loss to Arizona senator John McCain and Gore's quick victoryover former New Jersey senator Bill Bradley, when the race did close up toabout even. Aside from that brief period, Bush has maintained a very steady,though not necessarily wide, lead.

Gallup polling done for CNN and USA Today shows that, between mid-February and mid-June, only one of the nine polls put Bush outside of a verynarrow range of 48 to 52 percent of the vote in a two-way matchup. In eachof the nine, Gore fell within a 41 to 45 percent range. Aside from the onesurvey that showed Bush ahead of Gore by a single point, taken from March30 to April 1, each of the polls put Bush ahead by between four and ninepercentage points.

There has long been a natural tendency for people to seize on the resultsof the public opinion poll that most closely reflects what they would like tosee happen, even if the preponderance of the polling data suggests some-thing else. This poll is "right," all others are wrong. Consider it a form of se-lective perception, very commonly found among partisans, whether they areoperatives or journalists.

There is another tendency to pick the latest poll, sometimes regardless ofthe source, track record, and credibility of that survey, even if it is not con-sistent with the overall pattern of recent polls. Those guilty of this practice,including many reporters, producers, columnists, and editors, read far morevolatility into political campaigns than is often the case. The public often

THE WASHINGTON QUARTERLY • AUTUMN 2000 197

Dow

nloa

ded

by [

Uni

vers

itat P

olitè

cnic

a de

Val

ènci

a] a

t 07:

27 2

5 O

ctob

er 2

014

j Charles E. Cook Jr.

looks neurotic as a result of this kind of reporting, producing seemingly con-tradictory results reported almost every day. It is not the public that is neu-rotic; rather, it is flawed reporting.

A June 8-13 Los Angeles Times survey gave Bush a 10-point lead overGore, 50 to 40 percent, triggering much of the "Gore is dead" talk. Whilethe Los Angeles Times polls had a solid reputation, at that point, the onlyother poll showing anything like it was a Fox News Channel/Opinion Dy-namics poll putting the lead at 8 points. The Fox poll consistently showsbetter numbers for Bush and/or Republicans. Yet many read too much intothese numbers, ignoring contradictory data from other polls. One example isthe fact that in the Los Angeles Times poll, Bush led Gore among women.The fact is, if Bush is ahead by 10 points overall, he will almost surely leadeven among women. Conversely, if Gore were ahead of Bush by 10 points,he would probably be ahead in this group. Any poll showing the race in the3-7 point range—which is where the race seems to have been for sometime—would show Bush ahead among men while Gore would lead amongwomen. Those that seized the Zogby poll done in late May and a corre-sponding Newsweek poll done days later in early June that both showedBush up just a single point are guilty of exclusivley focusing on their favor-ite, or the latest, poll.

Prudent observers should either compute or keep in their heads a mentalmoving average of the most recent 5 or 10 national polls, paying attentionto that average or the pattern rather than any single poll. A variation is toemulate the Olympic scoring method: throw out the high and low and aver-age the rest. Care should also be given to identify which polls rather consis-tently reflect results at variance with the larger pattern of results.

It's the Economy, Stupid...

If you listen to the political scientists, the 2000 presidential election is aforegone conclusion. A recent Washington Post survey of a half dozen politi-cal scientists who have run political/economic models forecasting the resultsof the 2000 election all found that Gore should beat Bush. The model's es-timates of Gore's election percentage range from as low as 53 percent to ashigh as 60 percent. This is interesting, given that Gore has not lead in apublicly released national poll in over two months and that the conven-tional wisdom gives Bush a steady, though hardly insurmountable, advan-tage. The political scientists point to the fact that early polls have oftenbeen wrong.

So what gives? Political scientists and their economist collaborators arguethat fundamentals matter, that certain inarguable factors are far more impor-

THE WASHINGTON QUARTERLY • AUTUMN 2000198

Dow

nloa

ded

by [

Uni

vers

itat P

olitè

cnic

a de

Val

ènci

a] a

t 07:

27 2

5 O

ctob

er 2

014

Polls and Models

It is very unlikelythat either party willwin true, workingcontrol of theHouse.

tant than the whimsical nature of political campaigns. After having studiedthousands of possible factors that could theoretically predict the outcome of apresidential election, they have identified a few that have worked and usethese criteria to evaluate future elections. These findings indicate that themost important factors have been the state of the economy, often measured bythe gross domestic product (GDP) growth rate, and how popular the incum-bent president is, measured by national job approval ratings in the polls. Typi-cally, if the economy is in marginal or poorshape, people believe it is time for a change;similarly, if the president is politically in mar-ginal or poor shape, his party will fare poorly inthe next election, whether his name is on theballot or not.

So why do the computer models point onedirection, while the polls and conventionalwisdom point the other? President BillClinton's job approval ratings are typically inthe mid-to-high 50s, remarkably good for apresident in his last year. Similarly, theeconomy continues to run like a top, with no end of this recovery in sight.Therefore, voters should be in the mood to "stay the course," as RonaldReagan's ad makers liked to say, or "keep on keeping on," in a more youthfulvernacular. By any objective measure, people should be inclined to keep theClinton/Gore team in place, in whatever form they are constitutionally ableto do.

The fundamental forces that these political scientists and economistspoint to are certainly very important. Indeed in most presidential elections,they may be the only thing that is important. Looking back at the 1996 cam-paign, one can argue that whether Bob Dole was or was not a good candi-date, whether Jack Kemp was or was not a good running mate, and whetherthe Dole campaign was or was not well planned and executed may not havebeen terribly important. The economy was strong, the country was doingwell, and people simply were not in the mood for change. Similarly, a reces-sion (though it technically ended before the election) and public gloominessover the economy may have doomed President George Bush to his 1992loss, more than his lethargic campaign.

The contrarian view is that Clinton and the Democratic National Com-mittee, in their dubiously financed television advertising campaign thatcommenced in key markets soon after Labor Day 1995, effectively framedthe campaign in such a way that, by the time Dole won his nomination, hewas done. Devotees of this view might add that Clinton's flawlessly executed

THE WASHINGTON QUARTERLY • AUTUMN 2000 199

Dow

nloa

ded

by [

Uni

vers

itat P

olitè

cnic

a de

Val

ènci

a] a

t 07:

27 2

5 O

ctob

er 2

014

\CharksE.Cook]r.

campaigns in both 1992 and 1996 campaigns made a key difference. It canalso be argued that, had President Bush quickly refocused his attention ontothe economy and domestic policy concerns after his 1991 Persian Gulf vie'tory, the outcome may have been different.

The academics, however, point to their numbers and their impressivetrack records forecasting past elections and say hogwash. Although these ex-perts may ultimately prove themselves right again, there does seem to bemore to this election than just the economy and what kind of job Clintonhas done.

Despite the fact that we are in the midst of the longest period of eco-nomic growth in over a half-century and consumer confidence about theeconomy is extremely high, Americans are in a less than optimistic mood. Inthe latest Voter.com/Battleground national survey, taken in early May byDemocratic pollster Celinda Lake and Republican polltaker Ed Goeas, only39 percent of the American people felt that the country was going in "theright direction," while 49 percent felt that things had "gotten off on thewrong track." Indeed when intensity is taken into account, Americans areeven more pessimistic, with those feeling strongly that the country washeaded off on the wrong track outnumbering those strongly optimistic bybetter than a two-to-one margin—32 percent to 15 percent.

This same Lake/Goeas Voter.com survey, like most others, peggedClinton's job approval ratings high: 58 percent approving the job he hasdone, 38 percent disapproving; but when asked to rate him as a person, only29 percent approved of him, while 62 percent disapproved. Those stronglydisapproving of Clinton outnumbered those strongly approving by almostfour-to-one—45 percent to 12 percent. Such numbers are reminiscent of afocus group of 12 Democrats held in Towson, Maryland, last October, con-ducted by Peter Hart. Hart asked them whether they approved or disap-proved of the job Clinton had done in office; nine out of twelve said theyapproved. He then asked that if Clinton could run for a third term as presi-dent, whether they would vote for him or not; only three said they would.For some, they like his policies but no longer like him. For others, they likehis policies but have grown weary of him and those around him—which mayor may not include Gore. Welcome to the world of Clinton fatigue.

Making Sense of Conflicting Signals

There is no doubt at all that if Americans judge this election solely on thebasis of the state of the economy and how Bill Clinton is perceived as hav-ing managed the country, the political scientists and their computer modelsare right—Gore wins. Indeed his margin would fall then anywhere from

THE WASHINGTON QUARTERLY • AUTUMN 2000200

Dow

nloa

ded

by [

Uni

vers

itat P

olitè

cnic

a de

Val

ènci

a] a

t 07:

27 2

5 O

ctob

er 2

014

Polls and Models

In the end, thedebates may wellmake the difference.

comfortably, say six to eight points, to by a landslide, 10 points or more.But is that all there is? First, many believe that Americans may have

grown more liberal in how they distribute the credit for our strong economy.Much of it certainly goes to the strong stewardship of Federal Reserve BoardChairman Alan Greenspan and his monetary policies. (Appointed by Bush,reappointed by Clinton, where does the partisan credit go?) Then there isthe remarkable decline in federal budget defi-cits, something many of us thought we wouldnever see. President Bush surely deservessome of the credit for going along with con-gressional Democrats in raising taxes back inthe fall of 1990, a move that may well have

contributed to his reelection loss but started „__«««»»»»»»__««___the federal government on the road to fiscalresponsibility. Congressional Democrats canget some credit, at least for balancing the budget with their deep, perhapsunwisely deep, defense spending cuts, while Republicans can surely claimcredit for their insistence on domestic spending cuts. Clinton and congres-sional Democrats also raised taxes in 1993, further contributing to a reduc-tion of the deficit.

Then there is the extrapolitical side of the equation, the painfuldownsizing and reengineering of our economy by the private sector that hasmade U.S. business leaner and meaner, as well as, more importantly, extraor-dinarily competitive. The jobs we are losing are the jobs of the past, but inthe information and technology sectors, the wave of the future, the UnitedStates is surely on the cutting edge. Thus credit can be awarded on bothsides of the aisle, so Gore does not necessarily reap all of the benefits of astrong economy that the sitting vice president theoretically should.

Then there is Clinton. Just because Americans overwhelmingly opposedhis impeachment and removal from office, this should not be interpreted asan enthusiastic endorsement for the totality of his presidency. The wearinessfound among the Democrats in the Towson focus group is pervasive. Theyseem to be looking beyond economic indicators such as inflation, interestrates, unemployment, GDP growth rates, and Dow Jones industrial averages.Maybe they are taking the strong economy for granted, having gone nownine years without a recession. When voters complain about the countrylosing its moral compass and fearing for the futures of their children andgrandchildren, at a time when jobs are plentiful and economic futures oftenbright, it suggests something else is at work besides the numbers fed intothese models.

This is a complicated election. In terms of substantive policies and han-

THE WASHINGTON QUARTERLY • AUTUMN 2000 201

Dow

nloa

ded

by [

Uni

vers

itat P

olitè

cnic

a de

Val

ènci

a] a

t 07:

27 2

5 O

ctob

er 2

014

\ Charles E. Cook Jr.

dling of the economy, Americans are certainly not seeking change. That iswhy Bush has to be careful not to represent himself as advocating too muchchange at a time when much is going well. But at the same time, Gore cannot allow himself to be seen as Bill Clinton's identical twin, given voters'lack of trust of the president and their weariness of embarrassment andscandal. If the word in recent days of major changes in the strategies andtactics of the Gore campaign is true, then they too are banking on thesemodels and predictions.

Do not dismiss these models; the fundamentals that they represent arepowerful forces that have in fact been determinative in the past. But do notbe a prisoner of them either. Things happen, circumstances change. If thecampaigns were that certain, all of this money wouldn't be spent.

THE WASHINGTON QUARTERLY • AUTUMN 2000202

Dow

nloa

ded

by [

Uni

vers

itat P

olitè

cnic

a de

Val

ènci

a] a

t 07:

27 2

5 O

ctob

er 2

014