2000 Presidential Election

 

Background

 

One of the many claims made while determining the result of the 2000 Presidential election was that Palm Beach County's "butterfly ballot" caused people to mistakenly vote for Pat Buchanan rather than Al Gore, and that this changed the result of the election.  In this project you will analyze the data, using linear regression, to determine whether there is any evidence to support this hypothesis.

 

Data

 

Below is a table of the results of the voting in each of Florida's 67 counties.  Listed for each county are the number of votes cast for Gore, Bush, Buchanan and all other candidates, and the total number of votes cast.  Data was obtained from http://madison.hss.cmu.edu/, by way of Prof. Kevin Lanning.

 

County

GORE

BUSH

BUCHANAN

Other

Total

ALACHUA

47365

34124

263

3977

85729

BAKER

2392

5610

73

79

8154

BAY

18850

38637

248

1070

58805

BRADFORD

3075

5414

65

119

8673

BREVARD

97318

115185

570

5322

218395

BROWARD

386561

177323

788

8724

573396

CALHOUN

2155

2873

90

56

5174

CHARLOTTE

29645

35426

182

1643

66896

CITRUS

25525

29765

270

1640

57200

CLAY

14632

41736

186

799

57353

COLLIER

29918

60433

122

1668

92141

COLUMBIA

7047

10964

89

408

18508

DADE

328764

289492

560

6546

625362

DE SOTO

3320

4256

36

193

7805

DIXIE

1826

2697

29

114

4666

DUVAL

107864

152098

652

4022

264636

ESCAMBIA

40943

73017

502

2186

116648

FLAGLER

13897

12613

83

518

27111

FRANKLIN

2046

2454

33

111

4644

GADSDEN

9735

4767

38

187

14727

GILCHRIST

1910

3300

29

156

5395

GLADES

1442

1841

9

73

3365

GULF

2397

3550

71

126

6144

HAMILTON

1722

2146

23

73

3964

HARDEE

2339

3765

30

99

6233

HENDRY

3240

4747

22

129

8138

HERNANDO

32644

30646

242

1687

65219

HIGHLANDS

14167

20206

127

649

35149

HILLSBOROUGH

169557

180760

847

9131

360295

HOLMES

2177

5011

76

131

7395

INDIAN RIVER

19768

28635

105

1114

49622

JACKSON

6868

9138

102

192

16300

JEFFERSON

3041

2478

29

94

5642

LAFAYETTE

789

1670

10

36

2505

LAKE

36571

50010

289

1741

88611

LEE

73560

106141

305

4371

184377

LEON

61425

39053

282

2353

103113

LEVY

5398

6858

67

401

12724

LIBERTY

1017

1317

39

37

2410

MADISON

3014

3038

29

81

6162

MANATEE

49177

57952

271

2821

110221

MARION

44665

55141

563

2587

102956

MARTIN

26620

33970

112

1311

62013

MONROE

16483

16059

47

1289

33878

NASSAU

6879

16280

90

332

23581

OKALOOSA

16948

52093

267

1372

70680

OKEECHOBEE

4588

5057

43

165

9853

ORANGE

140220

134517

446

4942

280125

OSCEOLA

28181

26212

145

1119

55657

PALM BEACH

268945

152846

3407

7088

432286

PASCO

69564

68582

570

4015

142731

PINELLAS

200629

184823

1013

12004

398469

POLK

75193

90180

532

2581

168486

PUTNAM

12102

13447

148

525

26222

ST. JOHNS

12802

36274

311

932

50319

ST. LUCIE

72853

83100

305

4684

160942

SANTA ROSA

59174

75677

194

2589

137634

SARASOTA

19502

39546

229

1469

60746

SEMINOLE

41559

34705

124

1601

77989

SUMTER

9637

12127

114

383

22261

SUWANNEE

4075

8006

108

252

12441

TAYLOR

2649

4056

27

76

6808

UNION

1407

2332

37

50

3826

VOLUSIA

97063

82214

496

3483

183256

WAKULLA

3838

4512

46

191

8587

WALTON

5642

12182

120

374

18318

WASHINGTON

2798

4994

88

141

8021

 

Problems

 

1.      Plot Buchanan's votes against each of the other columns:  Gore, Bush, Other and Total.  Do you see any relationships in the plots?  Are there any outliers?  What do you think is the best explanation for the relationships you see?

2.      Compute the regression line for each of the plots.  How good is the fit?

3.      Plot the residuals for each plot.  Are any observations particularly influential?

4.      Now remove the outliers, and compute the new regression lines.  How good is the fit?

5.      Using the regression lines in Part 3, estimate the number of votes Buchanan might have expected to receive in Palm Beach County.

6.      It has been argued that Palm Beach and Broward counties are similar demographically, and so the distribution of votes should have been similar.  How do the data points for Palm Beach and Broward compare on each plot?

7.      Bush officially won Florida by 537 votes.  Does your work support the hypothesis that the butterfly ballot cost Gore the election?

8.      What problems might there be with this analysis?  See http://madison.hss.cmu.edu/ for some ideas (there are also links to several other analyses, if you are interested in looking into the issue further).  Do these problems seem severe enough to affect your conclusions?