Available online at www.sciencedirect.

com

International Journal of Forecasting 24 (2008) 301 309

www.elsevier.com/locate/ijforecast

The keys to the white house: An index forecast for 2008

Allan J. Lichtman
Professor of History, American University, Washington, DC 20016

Abstract
The Keys to the White House are an index-based prediction system that retrospectively accounts for the popular-vote
winners of every American presidential election from 1860 to 1980, and prospectively forecast the winners of every presidential
election from 1984 through 2004 well ahead of time. The Keys give specificity to the theory that presidential election results
turn primarily on the performance of the party controlling the White House. The Keys include no polling data and consider a
much wider range of performance indicators than economic concerns. Already, the Keys are lining up for 2008, showing how
changes in the structure of politics will produce a Democratic victory, in a dramatic reversal from 2004. The Keys also suggest
that candidates need not follow the empty scripted campaigns of the recent past, but should instead be liberated to offer
forthright discussions of the issues and ideas that will shape America's future.
2008 International Institute of Forecasters. Published by Elsevier B.V. All rights reserved.

1. The winds of political change

The Keys to the White House are an index
forecasting model that I first developed in 1981, in
collaboration with Vladimir Keilis-Borok, founder of
the International Institute of Earthquake Prediction
Theory and Mathematical Geophysics. Retrospectively,
the model accounts for the popular vote winners of every
American presidential election since 1860. Prospectively, it has correctly forecast the popular vote winners
of all six presidential elections from 1984 to 2004,
usually months or even years prior to Election Day. In
2004, for example, the Keys forecast George W. Bush's

re-election in April 2003, nearly a year before any other

academic model (Jones, 2007). In a paper presented at
the annual International Symposium on Forecasting in
June 2005, I noted that Already the Keys are lining up
for 2008, demonstrating surprisingly bright prospects
for Democrats to recapture the White House. That
forecast was published in early 2006 (Lichtman, 2006,
p. 5). Subsequent events have reinforced the original
forecast. The Keys to the White House, as of this writing
in November 2007, indicate that the Democrats will
recapture the White House in 2008. The thirteen Keys,
along with a brief description of each and a decision rule
for their application, are listed in Table 1.
2. The keys model

Tel.: +1 202 885 2411.

E-mail address: lichtman@american.edu.

Unlike nearly all other models for predicting election

outcomes, the Keys to the White House relies on an

0169-2070/$ - see front matter 2008 International Institute of Forecasters. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.ijforecast.2008.02.004

302

A.J. Lichtman / International Journal of Forecasting 24 (2008) 301309

Table 1
Keys To The White House
KEY

DEFINITION

1: Party Mandate

After the midterm elections, the incumbent party holds more seats in the U.S. House of Representatives than it did
after the previous midterm elections.
2: Party Contest
The candidate is nominated on the first ballot and wins at least two-thirds of the delegate votes.
3: Incumbency
The sitting president is the party candidate.
4: Third Party
A third-party candidate wins at least 5 percent of the popular vote.
5: Short-term Economy The National Bureau of Economic Research has either not declared a recession, or has declared it over prior to the
election.
6: Long-term Economy Real per-capita economic growth during the term equals or exceeds the mean growth during the previous two terms.
7: Policy Change
The administration achieves a major policy change during the term comparable to the New Deal or the first-term
Reagan Revolution.
8: Social Unrest
There is no social unrest during the term that is comparable to the upheavals of the post-civil war Reconstruction or of
the 1960s, and is sustained or raises deep concerns about the unraveling of society.
9: Scandal
There is no broad recognition of a scandal that directly touches upon the president.
10: Foreign or Military
There is no major failure during the term comparable to Pearl Harbor or the Iran hostage crisis that appears to
Failure
significantly undermine America's national interests or threaten its standing in the world.
11: Foreign or Military There is a major success during the term comparable to the winning of World War II or the Camp David Accords that
Success
significantly advances America's national interests or its standing in the world.
12: Incumbent Charisma/ The incumbent party candidate is a national hero comparable to Ulysses Grant or Dwight Eisenhower or is an
Hero
inspirational candidate comparable to Franklin Roosevelt or Ronald Reagan.
13: Challenger
The challenger party candidate is not a national hero comparable to Ulysses Grant or Dwight Eisenhower and is not an
Charisma/Hero
inspirational candidate comparable to Franklin Roosevelt or Ronald Reagan.

index rather than a regression-based method of forecasting. This approach to election forecasting differs from
the regression models in several ways. First, it forecasts
not percentage votes, but wins and losses by the party
holding the White House, based upon an index comprised of true or false responses to set of questions, each
of which is a key to the White House. Second, it uses
pattern recognition, not regression, to select relevant
keys and develop a decision rule for distinguishing
incumbent from challenging party victories. Third, it
includes a much wider array of predictor variables than
the regression models. Fourth, the model provides for
very long-term forecasts of an upcoming election. For
analyses of methods for forecasting presidential election
results, see Campbell and Garand (2000) and Jones
(2002).
To develop the predictive model, Keilis-Borok and I
applied the simple pattern recognition algorithm
known as Hemming's Distance to two binary vectors.
First, we coded elections between 1860 and 1980 that
fell into Class I the incumbent party prevailed as 0,
and elections that fell into Class C the challenging
party prevailed as 1. In two elections, 1876 and 1888,
the tally of electoral votes reversed the preference set

by the popular vote. Given that our model is based on

national indicators, we put 1876 in Class C, reflecting
the popular vote plurality for the challenging party candidate, Democrat Samuel J. Tilden, rather than the
electoral vote victory for incumbent party candidate,
Republican Rutherford B. Hayes. We put 1888 in Class I,
reflecting the popular vote plurality for incumbent
Democratic President Grover Cleveland, rather than
the electoral vote victory for the challenger, Republican
Benjamin Harrison. In classifying the subsequent elections of 1880 and 1892, the party that actually gained the
presidency in the previous election was considered the
incumbent party: the Republicans in both years. The
incumbent party won in 1880 and lost in 1892 (see
Tables 2 and 3).
The second vector consisted of answers of true or
false to a set of questions that can be answered prior
to an upcoming election. The questions or keys were
based upon a version of the theory of retrospective
voting that a pragmatic electorate chooses a president
according to the performance of the party holding the
White House, as gauged by the consequential events
and episodes of a term. Our questionnaire initially included twelve questions, but was modified to include

A.J. Lichtman / International Journal of Forecasting 24 (2008) 301309

303

Table 2
Keys To The White House: Historical Results, 1860-1980 Incumbent Party Victories, Popular Vote: Class I
YEAR

K1

K2

K3

K4

K5

K6

K7

K8

K9

K 10

K 11

K 12

K13

1864
1868
1872
1880
1888
1900
1904
1908
1916
1924
1928
1936
1940
1944
1948
1956
1964
1972
P(i/I)

0
0
1
0
1
1
0
0
1
1
0
0
1
1
1
0
1
1
0.56

0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0.06

0
1
0
1
0
0
0
1
0
0
1
0
0
0
0
0
0
0
0.22

0
0
0
0
0
0
0
0
0
1
0
0
0
0
1
0
0
0
0.11

0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0

1
0
0
0
0
0
0
1
1
0
0
0
0
0
1
0
0
1
0.28

0
0
1
0
1
0
0
0
0
0
1
0
0
0
0
1
0
1
0.28

1
1
1
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0.22

0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0.06

0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
0
1
0
0.17

0
0
0
1
1
0
0
0
0
0
0
1
1
0
0
0
0
0
0.22

1
0
0
1
1
1
0
1
1
1
1
0
0
0
1
0
1
1
0.61

0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0.06

3
2
3
4
5
3
0
3
3
4
3
1
2
2
5
1
3
4

The popular vote and the Electoral College vote diverged.

false answers as 1. When a key is coded as 1, it is

in effect turned against the party holding the White
House, whereas a coding of 0 indicates that it is turned
in favor of the incumbent party.
The results from the elections of 1860 to 1980 are
reported in Tables 2 and 3, partitioned into elections of
Class I and Class C. Each election in the tables is

thirteen questions prior to its use for advance prediction. As is indicated in Table 1, each question is phrased
so that an answer of true favors reelection of the party
in power, and an answer of false favors its defeat. For
example, Key 13 is phrased as The challenging-party
candidate is not charismatic or a national hero. True
answers for any given election are coded as 0, and

Table 3
Keys To The White House: Historical Results, 1860-1980 Challenging Party Victories, Popular Vote: Class C
YEAR

K1

K2

K3

K4

K5

K6

K7

K8

K9

K 10

K 11

K 12

K 13

1860
1876
1884
1892
1896
1912
1920
1932
1952
1960
1968
1976
1980
P(i/C)

0
1
1
1
1
1
1
1
0
1
1
1
1
0.85

1
1
1
1
1
1
1
0
1
0
1
1
1
0.85

1
1
1
0
1
0
1
0
1
1
1
0
0
0.62

1
0
0
1
0
1
0
0
0
0
1
0
1
0.38

0
1
1
0
1
0
1
1
0
1
0
0
1
0.54

0
1
1
0
1
0
1
1
1
1
0
1
0
0.62

1
1
1
0
1
1
0
1
1
1
0
1
1
0.77

1
0
0
1
1
0
1
1
0
0
1
0
0
0.46

0
1
0
0
0
0
0
0
1
0
0
1
0
0.23

0
0
0
0
0
0
1
0
1
1
1
1
1
0.46

1
1
1
1
1
1
0
1
0
1
1
1
0
0.77

1
1
0
1
0
1
1
1
1
1
1
1
1
0.85

0
0
0
0
0
0
0
1
1
1
0
0
1
0.31

7
9
7
6
8
6
8
8
8
9
8
8
8

The popular vote and the Electoral College vote diverged.

304

A.J. Lichtman / International Journal of Forecasting 24 (2008) 301309

described by the number of keys turned against the

incumbent party. Technically, this is represented by
the binary vector Yi(X1, X2 Xn), where Xi = 0 or
Xi = 1 is the answer to the ith question. For each
question, statistics are computed that indicate the percentage of negative keys for elections won by incumbent and challenging parties, respectively.

party victories, and must add to the degree of distance between these classifications, as measured by
mD(C) mD(I), where mD(C) is the mean Hemming's Distance for elections classified as challenger
victories and mD(I) is the mean Hemming's Distance
for elections classified as incumbent victories. These
criteria resulted in the exclusion of about 15 initially
proposed questions, including whether the economy is
in a state of war or peace, whether the incumbent or
challenging candidate is more centrist in policies,
whether the incumbent party has held office for more
than one term, whether the incumbent party gained
more than 50 percent of votes cast in the previous
election, and whether the incumbent party is Republican or Democratic. For additional elaboration on the
method, see Gvishiani, Zelevinsky, Keilis-Borok, and
Kosobokov (1980) and Keilis-Borok and Lichtman
(1981).
Unlike many alternative models, the Keys include
no polling data, but are based on the big picture of how
well the party in power and the country are faring prior
to an election. In addition, the Keys do not presume that
voters are driven by economic concerns alone. Voters
are less narrow-minded and more sophisticated than
that; they decide presidential elections based on a wideranging assessment of the performance of incumbent
parties, all of which factors are reflected in one or more
keys.
As is indicated in Table 4, we subsequently used the
Keys model to correctly forecast the popular-vote
outcomes of all six presidential elections between 1984
and 2000. As is reported in Table 5, the keys lined up
definitely for or against the party in power anywhere
between 32 months and 1 month prior to the general
elections. Events subsequent to the publication of these

P(i/I) = n(i/I)/n(I) and P(i/C) = n(i/C)/n(C), where

n(i/I) = the number of elections in which Xi = 1 for
Class I;
n(i/C) = the number of elections in which Xi = 1 for
Class C;
n(I) = the number of elections of Class I; and
n(C) = the number of elections of Class C.
From the results reported in Tables 2 and 3, we
computed Hemming's Distance for each individual
election, defined as D RXia , which is equal to the
number of keys turned against the party in power, with
no differential weighting. The value of D varies from
0 to 13, with higher numbers indicating a greater
likelihood of an incumbent party defeat. As is
demonstrated in Tables 2 and 3, taken together, the
thirteen keys correctly classified all 31 elections from
1860 to 1980 into Class I or Class C according to the
decision rule that the incumbent party prevails if D b 6
and the challenging party prevails if D 6. Thus, the
party in power is predicted to lose the popular vote if
and only if it loses 6 or more of the thirteen keys.
No question was included in the analysis unless
it distinguished between incumbent and challenger
victories according to the criterion P(i/C) P(i/I) N 0.1.
In addition, each question must be part of a set that
correctly distinguished incumbent and challenging

Table 4
Keys To The White House: Advance Predictions, 1984-2004
YEAR

K1

K2

K3

K4

K5

K6

K7

K8

K9

K 10

K 11

K 12

K13

CLASS

1984
1988
1992
1996
2000
2004

0
0
1
1
0
0

0
0
0
0
0
0

0
1
0
0
1
0

0
0
1
1
0
0

0
0
1
0
0
0

1
0
1
0
0
1

0
1
1
1
1
1

0
0
0
0
0
0

0
0
0
0
1
0

0
0
0
0
0
1

1
0
0
1
1
0

0
1
1
1
1
1

0
0
0
0
0
0

2
3
6
5
5
4

I
I
C
I
I
I

I = Incumbent win in popular vote, C = Challenger win. The popular vote and the Electoral College vote diverged.

A.J. Lichtman / International Journal of Forecasting 24 (2008) 301309

Table 5
Keys To The White House: Timing of Predictions
ELECTION
YEAR

DATE OF
PREDICTION

SOURCE

1984

APRIL 1982

1988

MAY 1988

1992

SEPTEMBER
1992

1996

OCTOBER
1996

2000

NOVEMBER
1999

2004

APRIL 2003

How to Bet in '84.

Washingtonian Magazine,
April 1982, 147-149.
How to Bet in November.
Washingtonian Magazine,
May 1988, 115-124.
The Keys to the White House.
Montgomery Journal,
September 14, 1992, 12.
The Keys to the White House:
Who Will Be the Next American
President?
Social Education,
October 1996, 358-360.
The Keys to Election 2000.
Social Education,
November/December 1999,
422-424.
The Keys to the White House.
Montgomery Gazette,
April 25, 2003, 4.

predictions did not change a single key in any of the six

elections.
3. Judgment and the keys
Answers to some of the questions posed in the Keys
require the kind of informed evaluations that historians
invariably rely on in drawing conclusions about past
events. Two constraints distinguish these assessments
from the ad hoc judgments offered by conventional
political commentators. First, all judgment calls are
made consistently across elections; the threshold standards established in the study of previous elections
must be applied to future contests as well. Second, each
Key has an explicit definition that is briefly summarized in Table 1. As indicated by these definitions, Keys
1-3 and 5-6 require little or no judgment. Key 4 requires
an assessment of the prospects of a third party
candidate. The rule of thumb is that the percentage
vote polled by insurgent candidates usually equals
about half of their peak standing in the polls. For
additional details on defining and turning the keys, see
Lichtman (2005, pp.1-48) and Lichtman (2006).
Keys 7-13 require more considerable judgment,
but the calls on these keys are not difficult, given their

305

definitions and the history of calls made in the 37 elections from 1860 to 2004. To secure Policy Change Key 7,
an administration must either fundamentally redirect the
course of American government, as in the administrations
of Warren Harding or Ronald Reagan, or achieve major
innovations of policy, as in the first administrations of
Woodrow Wilson or Franklin Roosevelt. Particular
legislative achievements such as the tax cuts of George
W. Bush's first term are not sufficient to secure this key.
Social Unrest Key 8 favors the party controlling the
White House unless unrest manifests itself in sustained
violent challenges to authority, as in the Reconstruction
Era or in the tumultuous 1960s. Isolated incidents such
as the race riots in Miami in 1980 or in Los Angeles in
1992 do not topple this key.
To reach the threshold for turning Scandal Key 9
against an administration, serious wrongdoing must be
widely recognized as directly implicating the president or
producing extensive patterns of illicit activity on the part
of administration officials. Examples include the Watergate Scandals, the impeachment of President Clinton, and
the so-called Great Barbecue of the second administration of Ulysses S. Grant, in which members of his
cabinet and staff had enriched themselves through
bribery, kickbacks, and the outright sale of government
contracts.
Given that the public pays only sporadic attention
to international matters, Foreign/Military Failure Key 10
is turned against the executive party only by a major
setback such as the attack on of Pearl Harbor or the Iran
hostage crisis of the Carter administration. Failed diplomatic initiatives such as Dwight Eisenhower's inability
to gain a nuclear test-ban treaty with the Soviets or Bill
Clinton's inability to secure a peace settlement between
the Israelis and Palestinians do not count against the party
in power.
As with the policy-change key, major successful
initiatives by an administration are necessary to
prevent the Foreign/Military Failure Key from being
turned against the party in power. For example, the
INF Treaty and the thawing of U.S.-Soviet relations
during Ronald Reagan's second term fit both criteria,
but the first-term intervention in the tiny Caribbean
island of Grenada, which was a minor splash, neither
involved critical American interests nor significantly
advanced the nation's standing in the world.
Of all of the keys, Incumbent and Challenger
Charisma/Hero Keys 12 and 13 are the likeliest to be

306

A.J. Lichtman / International Journal of Forecasting 24 (2008) 301309

influenced by the election campaign itself. To meet the

threshold for heroic stature, a candidate's achievements,
like those of Ulysses Grant and Dwight Eisenhower,
must be critical to the nation's success in an important
endeavor. In contrast, George McGovern's heroism as a
bomber pilot in World War II did not turn Key 13 against
the party in power in 1972. Very few candidates have
demonstrated truly inspirational qualities. Examples
include Democrats William Jennings Bryan and John F.
Kennedy, and Republicans Theodore Roosevelt and
Ronald Reagan. In each case their charisma was widely
recognized in press commentary at the time. No sitting
president who was charismatic or a national hero has
ever been challenged within his own party, faced a
significant third-party campaign or a charismatic
opponent, or been defeated for reelection.
4. Subsets of the keys
Some subsets of the thirteen keys also correctly
predict the outcome of all elections since 1860. Why not,
then, use a smaller set of keys? First, the full set of thirteen
keys provides the greatest separation between incumbent
wins and losses over the entire sample of elections.
Second, given the limited sample of prior elections, the
thirteen-key system has the greatest capacity to capture
future variations in political circumstances, which smaller
subsystems might miss. For example, at the time we
originally developed the system in 1981, the combination
of only four keyscontest key 2, third-party key 4, shortterm economy key 5, and policy-change key 7
correctly predicted every election between 1860 and
1980 (the party in power wins unless two keys are turned
against it). However, this four-key system would have
miscalled President Clinton's reelection in 1996 (the
party in power lost Keys 4 and 7), which the thirteen- key
system correctly predicted. Third, the greater stability and
range of the full system also provides the ability to
conduct long-term forecasts of a future election. Finally,
all of the thirteen keys have strong theoretical justification
for their inclusion. This criterion is consistent with the
dual objective of the Keys to both predict and explain
the outcomes of American presidential elections.
5. Extensions of the keys
We used more advanced pattern recognition
systems to modify our simple binary system in two

ways. First, we generated multiple distinctive traits

from the learning material in Tables 2 and 3, representing combinations of answers to two or more
questions. Second, we weighted each key according to
its individual predictive power and several other
criteria. The use of combined indicators failed to
improve the results obtained through using Hemming's Distance, whereas the weighting of predictive
factors introduced errors into the classification of
individual elections, given instabilities from one election to the next. However, keys do have trigger effects
on other keys. It may seem counterintuitive, for example, that during the Great Depression, the economy
accounted for the fall of only two keys against Herbert
Hoover's Republicans in 1932. However, the economic collapse also triggered the loss of Mandate Key
1, Social Unrest Key 8, and most likely Challenger
Charisma Key 13, as without the Depression, Franklin
Roosevelt would most likely have followed his plan to
run for president in 1936, not 1932. The Vietnam War
in 1968 only directly counted for one discrepant Key
against the incumbent Democrats, but also triggered
the loss of Mandate Key 1, Party Contest Key 2,
Incumbency Key 3, Social Unrest Key 8, and perhaps
Third Party Key 4.
Given the unlimited number of combinations, it was
possible to mine the data and find a weighted combination of a subset of the Keys that accounts for the outcomes of all presidential elections since 1860. One such
example, suggested by an anonymous reviewer, would
be: Key 1 + (5Key 2) + (Key 5) + (2Key 6) + (2Key
7) +(2Key 8), with the same cut-point rule of six
negative keys predicting the incumbent party's defeat.
However, this model lacks theoretical justification,
depends heavily on the calls on single keys, and, like
the four-key model discussed above, may not be stable
in subsequent elections.
In an extension of the Keys model, regression
methods were used to convert scores from the keys for
the full period from 1860 to 2004 to numerical predictions of the two party presidential vote. The regression
yielded the following result:
V = 36.75 + 1.84 L
V = the percentage of the two-party split going to
the incumbent
L = the number of Keys favoring the incumbent
party

A.J. Lichtman / International Journal of Forecasting 24 (2008) 301309

Table 6
Keys To The White House: Historical Results & Numerical Predictions,
1860-2004
ELECTION

ACTUAL
TWO-PARTY
% FOR
INCUMBENT

PREDICTED
TWO-PARTY
% FOR
INCUMBENT

DIFFERENCE

1860
1864
1868
1872
1876
1880
1884
1888
1892
1896
1900
1904
1908
1912
1916
1920
1924
1928
1932
1936
1940
1944
1948
1952
1956
1960
1964
1968
1972
1976
1980
1984
1988
1992
1996
2000
2004
ABSOLUTE
MEAN DIFF

42.69
55.00
52.70
55.94
48.48
50.05
49.69
50.41
48.31
47.80
53.14
60.00
54.55
35.59
51.63
36.19
65.22
58.79
40.89
62.49
54.97
53.78
52.38
44.52
57.75
49.95
61.35
55.38
61.81
48.93
44.66
59.15
53.88
46.58
54.73
50.26
51.21

47.79
55.15
56.99
55.15
44.11
53.31
47.79
51.47
49.63
45.95
55.15
60.67
55.15
49.63
55.15
45.95
53.31
55.15
45.95
58.83
56.99
56.99
51.47
45.95
58.83
44.11
55.15
45.95
53.31
45.95
45.95
56.99
55.15
49.63
51.47
51.47
53.31

5.10
0.15
4.29
-0.79
-4.37
3.26
-1.90
1.06
1.32
-1.85
2.01
0.67
0.60
14.04
3.52
9.76
-11.91
-3.64
5.06
-3.66
2.02
3.21
-0.91
1.43
1.08
-5.84
-6.20
-9.43
-8.50
-2.98
1.29
-2.16
1.27
3.05
-3.26
1.21
2.10
3.65

This approach was suggested by Armstrong &

Cuzan (2006, p.12).
For all 37 elections, Table 6 reports the actual twoparty percentage for the winning candidate and the
predicted percentage based on the equation outlined
above. The equation correctly predicts the popular vote
winner in each election, with a mean error margin of

307

3.65 percent for the two-party vote percentage. For an

additional test, I applied regression methods to data
from 1860 to 1932. The resulting equation correctly
predicted the winners of all subsequent out-of-sample
elections, with a mean error margin of 3.67 percent.
6. Forecasting 2008
At the time of this writing (November 2007), only
five keys were called in favor of the incumbent
Republican Party. Eight were called against the party
in power, two more than necessary to predict its defeat
(Table 7).
The following five keys favor the incumbent
Republican Party:
Given that New York Mayor Michael Bloomberg
the only third-party contender likely to win at least
5 percent of the popular vote has repeatedly said
that he will not run for president, Third Party Key 4
falls in favor of the party in power.
It does not appear that the economy will suffer a
recession in 2008, so Short-Term Economy Key 5
should fall in line for the GOP.
The absence of social upheavals comparable to the
1960s avoids the loss of Social Unrest Key 8.
The lack of a significant scandal that directly
implicates the president averts the loss of Scandal
Key 9.

Table 7
The 13 Keys To The White House: Current Standings
Key number

Description

Outcome 2008

Key 1
Key 2
Key 3
Key 4
Key 5
Key 6
Key 7
Key 8
Key 9
Key 10
Key 11
Key 12
Key 13

Party mandate
Contest
Incumbency
Third party
Short-term economy
Long-term economy
Policy change
Social unrest
Scandal
Foreign/military failure
Foreign/military success
Incumbent charisma
Challenger charisma

False
False
False
True
True
False
False
True
True
False
False
False
True

True: 5 keys; false: 8 keys.

Prediction: incumbent Republicans lose.

308

A.J. Lichtman / International Journal of Forecasting 24 (2008) 301309

No prospective Democratic challenger thus far

matches the charisma of Franklin D. Roosevelt or
John F. Kennedy, keeping the Challenger Charisma/
Hero Key 13 in line for the incumbents.
Thse following eight keys fall against the incumbent party:
The Democrats won more than enough U.S. House
seats in the 2006 midterm elections to topple Mandate
Key 1.
The Republicans are battling fiercely in choosing a
nominee to replace George W. Bush, turning Contest
Key 2 against them.
Bush's inability to run again in 2008 dooms Incumbency Key 3.
Real per-capita growth so far during the second
Bush administration is slightly below the record
achieved during the last two presidential terms, so
Long-Term Economy Key 6 is counted against the
party in power.
With bitter partisan divisions in Congress, Bush has
failed to achieve the second-term policy revolution
needed to secure Policy Change Key 7.
The war in Iraq is a widely acknowledged failure,
and the administration has achieved no offsetting
triumph in foreign/military affairs, forfeiting Foreign/Military Failure Key 10 and Foreign/Military
Success Key 11.
Of all GOP candidates on the horizon, none appear
to be a Theodore Roosevelt or Ronald Reagan,
toppling Incumbent Charisma/Hero Key 12.
The shift in incumbent party prospects between 2004
and 2008 is most clearly illustrated in the first four keys
the political keys. In 2004, the incumbent Republicans
secured all four of the political keys. For 2008, the
Republicans lose three of the political keys. Setbacks in
foreign policy account for the shift of another key. Thus,
the negative forecast for Republican presidential candidates in 2008 does not depend on the performance of
the economy. The incumbent party would still lose the
presidency in 2008 even if it secured both economic keys.
Although keys do not usually change late in a presidential term, shifting verdicts are at least theoretically
possible. For example, one of the Republican candidates
could sweep to an overwhelming victory in the party
primaries. However, a changed forecast for 2008 re-

quires three of eight negative keys to turn back in favor

of the Republicans, with none shifting to the challenging
party. Barack Obama could emerge as a charismatic
challenging candidate, Bloomberg could still launch a
third-party campaign, the economy could fall into recession during the election year, or the Democratic
Congress could uncover a scandal to pin on the president, not just subordinates. It is also possible, although
unlikely, that the nomination of a breakthrough candidate such as a woman (Hillary Clinton) or an AfricanAmerican (Barack Obama) could break the historical
pattern of election results. The Keys to the White House
firmly predict a Democratic victory in 2008. It is most
improbable that the course of events will swing the Keys
back in line for the GOP.
The regression analysis also provides for a numerical
prediction of the 2008 results. Assuming that the incumbent Republicans retain five keys, their percentage of
the two-party presidential vote is an estimated 45.95
percent.
7. Governing and campaigning
The Keys have implications for governing the country
and conducting presidential campaigns. The keys do not
prove that campaigning is irrelevant to the outcomes of
presidential elections. Primary campaigns can influence
the turning of Contest Key 2, and general election campaigns could produce charismatic candidates, as measured by Keys 12 and 13, although the threshold for
winning these keys is high. However, the keys do suggest
that what mainly counts in presidential elections is governing, as measured by the consequential events of a
presidential term, not packaging, image-making or campaigning. Effective governing keeps incumbent parties in
office and renders conventional campaigning by challengers futile. This relationship between governing and
politics has held true across nearly 150 years of American
history and vast changes in our economy, society, and
politics: suffrage for women and blacks; new immigrants
from Eastern Europe, Asia, and Latin America; the rise of
the corporation; and the advent of polling, television, and
the internet.
It is also plausible that the predictive value of the
Keys depends in part on the fact that historically candidates have followed conventional campaign models. If
so, the candidate predicted to lose by the lineup of the
Keys has an incentive to break the pattern of history by

A.J. Lichtman / International Journal of Forecasting 24 (2008) 301309

waging an unconventional, breakthrough campaign.

Such a campaign may not break the historical pattern
of election outcomes, but has other advantages for
the candidate and the nation. In 2004, the Democrats
failed to understand that the historical odds favored the
Republicans. As a result, John Kerry ran a visionless,
consultant-driven general election campaign that failed
to establish a principled opposition, elevate the level of
political debate, inspire grassroots activism, or sustain
his viability for another presidential campaign. In July
2004, Keilis-Borok and I wrote:
Kerry has a choice between following the usual meaningless routine in the hope that setbacks to the administration and the country will elect him in November
or taking a chance on running a new kind of daring,
innovative, and programmatic campaign. With the right
choice, Kerry can achieve an historical breakthrough
that would establish the basis for a principled choice
of our national leader and a grassroots mobilization on
issues that matter to America's future. (Lichtman &
Keilis-Borok, 2004)
In politics, as Kerry learned, the only thing worse
than losing is losing irrelevantly.
Candidates favored by the Keys do not have an incentive to break historical patterns in their campaign.
However, they do have an incentive, within the limits
of conventional politics, to run a substantive campaign
that builds a foundation for governing, thereby
increasing their party's chances of retaining the
White House in the next election. In summary, not

309

only elections, but also election forecasts have

consequences for politics. The models that we use to
predict and understand presidential elections shape the
conduct of campaigns, the relationship between
candidates and the American people, and ultimately
the policies of government.
References
Armstrong, J. S. & Cuzan, A. G. (Feb. 2006). Index Methods for
Forecasting: An Application to the American Presidential Elections. Foresight: The International Journal of Applied Forecasting
(Issue 3), 1013.
Campbell, J. E. & Garand, James C. (Eds.). (2000). Before the Vote:
Forecasting American National Elections. CA: Sage: Thousand
Oaks.
Gvishiani, A. D., Zelevinsky, A. V., Keilis-Borok, V. I., & Kosobokov,
V. I. (1980). Methods and Algorithms for Interpretation of
Seismological DataComputational Semiology. Moscow: Nauka.
Keilis-Borok, V. I. & Lichtman, A. J. (1981). Pattern Recognition
Applied to Presidential Elections in the United States, 1860-1980:
The Role of Integral Social, Economic, and Political Traits. Proceedings of the National Academy of Sciences, 78, 72307234.
Jones, R. J. (2002). Who Will be in the White House? Predicting
Presidential Elections. New York: Longman.
Jones, R. J. (2007). The State of Presidential Election Forecasting in
2004. Paper Presented at the 2007 ISF International Symposium
on Forecasting. New York, NY.
Lichtman, A. J. (2005). The Keys to the White House. Lanham, MD:
Lexington Books.
Lichtman, A. J. (Feb. 2006). The Keys to the White House: Forecast
for 2008. Foresight: The International Journal of Applied
Forecasting(Issue 3), 59.
Lichtman, A. J. & Keilis-Borok, V. I. (July 2004). What Kerry Must
Do to Win (But Probably Won't). Counterpunch. http://www.
counterpunch.org/lichtman07292004.html