Presidential Popularity From Truman To Johnson1

Downloaded from https://www.cambridge.org/core.
Durham University Library, on 17 Jan 2021 at 15:51:27, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.2307/1955610
PRESIDENTIAL POPULARITY 1FROM

TRUMAN TO JOHNSON
J O H N E. MUELLER
University oj Rochester
I think [my grandchildren] will be proud of two Four variables are used as predictors of a
things. What I did for the Negro and seeing it President's popularity. These include a measure
through in Vietnam for all of Asia. The Negro of the length of time the incumbent has been in
cost me 15 points in the polls and Vietnam cost office as well as variables which attempt to esti-
me 20. mate the influence on his rating of major inter-
Lyndon B. Johnson' national events, economic slump and war. To as-
sess the independent impact of each of these
With tenacious regularity over the last two variables as they interact in association with
and a half decades the Gallup Poll has posed to its Presidential popularity, multiple regression anal-
cross-section samples of the American public the ysis is used as the basic analytic technique.
following query, "Do you approve or disapprove
of the way (the incumbent) is handling his job I. T H E DEPENDENT VARIABLE: PRESIDENTIAL
as President?" The responses to this curious POPULARITY
question form an index known as "Presidential The Presidential popularity question taps a
popularity." According to Richard Neustadt, the general impression about the way the incumbent
index is "widely taken to approximate reality" seems to be handling his job at the present mo-
in Washington and reports about its behavior ment. As Neustadt notes, the response, like the
are "very widely read"3 there, including, the question, is "unfocused,"5 unrelated to specific
quotation above would suggest, the highest cir- issues or electoral outcomes. The respondent is
cles. asked to "approve" or "disapprove" and if he
Plotted over time, the index forms probably has "no opinion," he must volunteer that re-
the longest continuous trend line in polling his- sponse himself. He has infrequently been asked
tory. This study seeks to analyze the behavior why he feels that way—and many respondents
of this line for the period from the beginning of when asked are able only vaguely to rationalize
the Truman administration in 1945 to the end their position.6 And only at times has he been
of the Johnson administration in January 1969 asked to register how strongly he approves or
during which time the popularity question was disapproves.
asked some 300 times.* A disapproving response might be considered
a non-constructive vote of no-confidence: the
'This investigation was supported by a grant respondent registers his discontent, but he does
from the National Science Foundation. At various not need to state who he would prefer in the
stages helpful comments, criticisms, and complaints Presidency. Thus the index is likely to be a very
were lodged by Richard Fenno, Gerald Kramer, imperfect indicator of success or failure for a
Richard Niemi, Peter Ordeshook, Alvin Rabushka, President seeking re-election. While approvers
William Riker, and Andrew Scott. are doubtless more likely than disapprovers to
' Quoted, David Wise, "The Twilight of a Pres- endorse his re-election, on considering the oppo-
ident," New York Times Magazine, November sition some approvers may be attracted into
3,1968, p. 131. voting against the incumbent just as some dis-
'Richard E. Neustadt, Presidential Power: The approvers may be led grudgingly to vote for
7
Politics of Leadership (New York: Wiley, 1960), him.
p. 205n. 5
* A general picture of what this line looks like Op. cit., p. 96.
can be gained from the figure in Robert A. Dahl, 'See for example the breakdowns in Gallup
Pluralist Democracy in the United States (Chica- Opinion Index, March, 1966, p. 4.
go: Rand McNally, 1967), p. 107. The Presidential ' There is also a more technical reason why the
popularity data for the Johnson administration popularity index has little direct relevance to the
have been taken from the Gallup Opinion Index. electoral result: Gallup does not ask the question
All other poll data, unless otherwise indicated, during a President's re-election campaign. Thus
have come from the archives of the Roper Public for the months between early summer and late
Opinion Research Center at Williams College, fall in 1948, 1956, and 1964 no Gallup data on
Williamstown, Massachusetts. Presidential popularity exist. One other technical-
18
Downloaded from https://www.cambridge.org/core. Durham University Library, on 17 Jan 2021 at 15:51:27, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.2307/1955610
1970 PRESIDENTIAL POPULARITY FROM TRUMAN TO JOHNSON 19
Whatever peculiarities there are in the ques- disapproval; the correlation between the two is
tion itself, they are at least constant. Unlike — .98. And, most conveniently, this almost
many questions asked by the polling organiza- means that the President's popularity at a given
tions, wording has not varied from time to time moment can be rendered by a single number:
by whim or fashion. The stimulus has therefore the percentage approving his handling of the
been essentially fixed; only the response has job. The no opinion percentage is almost always
varied. close to 14 percent and the percentage disap-
And the variation has been considerable. proving is, of course, the remainder.
Harry Truman was our most popular President There is, however, one small wrinkle. The
in this period—for a few weeks in 1945 when no opinion percentage does get a bit out of
more than 85 percent of the public expressed hand, quite understandably, in the early weeks
approval—and our least popular—from early of the Kennedy and Eisenhower administrations
1951 until March 1952 when less than 30 per- as substantial numbers of respondents felt in-
cent were usually found to be favorably inclined to withhold judgment on these new men.
clined. Other Presidents have stayed within This inordinate withholding of opinion declined
these limits with Lyndon Johnson most nearly in the first weeks to more "normal" levels with
approaching the Truman extremes. President the result that both the level of approval and
Eisenhower's popularity was never higher than disapproval tended to increase.9
79 percent, but it never dropped below 49 per- Since one of the propositions to be tested in
cent either. President Kennedy also maintained this study proposes that there exists a general
a rather high level of popularity but was in no- downward trend in each President's popularity,
ticeable decline at the time of his death. this initial rating situation causes something of a
The proportion of respondents selecting the problem. If the disapproval score is used as the
"no opinion" option, averaging 14 percent, re- dependent variable there will be a slight bias in
mained strikingly constant throughout the favor of the proposition. It seems preferable to
period.8 This is a little surprising since it might load things against the proposition; hence for
be expected that when opinion changes, say, the purposes of this study the dependent vari-
from approval to disapproval of a President, the able is the percentage approving the way the in-
move would be seen first in a decrease in the cumbent is handling his job as President.10 The
support figure with an increase in the no opinion average approval rating for the entire twenty-
percentage, followed in a later survey by an in- four year period is 58 percent.
crease in the disapproval column with a decrease
in the no opinion portion. There are a few occa- II. THE INDEPENDENT VARIABLES
sions in which the no opinion percentage seems If one stares at Presidential popularity trend
to rise and fall in this manner, one occurring in lines long enough, one soon comes to imagine
the early weeks of the Korean War, but by and one is seeing things. If the things imagined seem
large it would appear that if movements into also to be mentioned in the literature about the
the no opinion column do occur they are com- way Presidential popularity should or does be-
pensated for by movements out of it. have, one begins to take the visions seriously
This means therefore that the trend in ap- and to move to test them.
proval is largely a mirror image of the trend in "In the case of President Eisenhower, the no
opinion response actually rose a bit before it
ity is worth mention. There ia a slight underrep- began to descend, reaching the highest level re-
resentation of data points in Truman'a first years. corded for any President in the period in March
By 1950, except for the election year phenomenon, 1953 when 28 percent had no opinion. (President
the Gallup organization was asking the question Nixon has proved to be the greatest mystery of
on virtually every survey conducted—some dozen all: fully 36 percent registered no opinion after
or sixteen per year. Before that time the question his inauguration.)
10
was posed on the average only about half as It is argued by some that percentages should
frequently. Neither of these technical problems, not be used in their pure state as variables, but
however, is likely to bias the results in any im- rather should be transformed into logits: Y* =
portant way, especially since so much of the log. [Y/(l—Y)]. The transformation was tried in
analysis allows each administration a fair amount the analysis, but it made little difference. Therefore
of distinctiveness. the more easily communicated percentage version
* The standard deviation for the no opinion re- haa been kept. In any event the dependent vari-
sponse is 2.93. By contrast the comparable statistic able rarely takes extreme values. It rises to 80
is 14.8 for the approve response and 14.5 for the percent only three or four times and never dips
disapprove response. below 23 percent.
20 THE AMERICAN POLITICAL SCIENCE REVIEW VOL. 64
In this manner were formulated four basic steel mills in 1952 made neither labor nor man-
"independent" variables, predictor variables of agement (nor the Supreme Court, for that mat-
Presidential popularity. They are: 1) a "coalition ter) happy. For the mayor of New York, situa-
of minorities" variable that suggests the overall tions like this seem to arise weekly.
trend in a President's popularity will be down- There are other, only vaguely related, reasons
ward; 2) a "rally round the flag" variable which to expect an overall decline in popularity. One
anticipates that international crises and similar would be disillusionment. In the process of being
phenomena will give a President a short-term elected, the President invariably says or implies
boost in popularity; 3) an "economic slump" he will do more than he can do and some disaf-
variable associating recessions with decreased fection of once bemused supporters is all but in-
popularity; and 4) a "war" variable predicting evitable. A most notable example would be the
a decrease in popularity under the conditions of case of those who supported President Johnson
the Korean and Vietnam wars. in 1964 because he seemed opposed to escalation
1. The "coalition of minorities" variable. In a in Vietnam. Furthermore initial popularity rat-
somewhat different context Anthony Downs has ings are puffed up by a variety of weak follow-
suggested the possibility that an administration, ers. These might include leering opposition par-
even if it always acts with majority support on tisans looking for the first excuse to join the ag-
each issue, can gradually alienate enough minor- grieved, excitable types who soon became bored
ities to be defeated. This could occur when the by the humdrum of post-election existence, and
minority on each issue feels so intensely about bandwagon riders whose fair weather support
its loss that it is unable to be placated by ad- dissolves with the first sprinkle.15 As Burns
ministration support on other policies it favors. Roper notes, "In a sense, Presidential elections
A clever opposition, under appropriate circum- are quadriennial myth builders which every four
stances, could therefore forge a coalition of these years make voters believe some man is better
intense minorities until it had enough votes to than he is. The President takes office with most
overthrow the incumbent.11 of the nation on his side, but this artificial
Transposed to Presidential popularity, this 'unity' soon begins to evaporate."16
concept might inspire the expectation that a For these reasons the coalition of minorities
President's popularity would show a general variable, as it is dubbed here, predicts decline.
downward trend as he is forced on a variety of "Love," said Machiavelli, "is held by a chain of
issues to act and thus create intense, unforgiv- obligation which, men being selfish, is broken
ing opponents of former supporters. It is quite whenever it serves their purpose."17
easy to point to cases where this may have oc- The coalition of minorities variable is mea-
curred. President Kennedy's rather dramatic ef- sured simply by the length of time, in years,
forts to force back a steel price rise in 1962, since the incumbent was inaugurated (for first
while supported by most Americans, tended to terms) or re-elected (for second terms). It var-
alienate many in the business community.12 Ad- ies then from zero to about four and should be
ministration enforcement of the Supreme negatively correlated with popularity: the lon-
Court's school desegregation order tended to ger the man has been in office, the lower his
create intense opposition among white Southern- popularity. It is; the simple r is —.47. The de-
ers even if the Presidential moves had passive cline is assumed to start over again for second
majority support in most of the country.13 terms because the President is expected to have
Realistically, the concept can be extended spent the campaign rebuilding his popular coali-
somewhat. From time to time there arise exqui- tion by soothing the disaffected, re-deluding the
site dilemmas in which the President must act disillusioned, and putting on a show for the
and in which he will tend to alienate both sides bored. If he is unable to do this, he will not be
no matter what he does, a phenomenon related re-elected, something which has not happened in
to what Aaron Wildavsky has called a "minus the post-war era although twice Presidents have
sum" game.14 President Truman's seizure of the declined to make the effort.
n
An Economic Theory of Democracy (New
The analysis will assume a linear decline in
York: Harper and Row, 1957), pp. 55-60.
popularity. That is, a President's popularity is
u
See the data in H. G. Erskine, "The Polls," 28
assumed to decline at an even rate for all four
15
Public Opinion Quarterly, 341 and 338 (Summer On the bandwagon effect among nonvoters, see
1964). Angus Campbell, Philip E. Converse, Warren E.
"See John M. Fenton, In Your Opinion (Bos- Miller, and Donald E. Stokes, The American
ton: Little, Brown, 1960), p. 146. Voter (New York: Wiley, 1960), pp. 110-115.
M
"The Empty-head Blues: Black Rebellion and "Burns Roper, "The Public Looks at Presi-
White Reaction." The Public Interest, Spring 1968, dents," The Public Pulse, January 1969.
11
pp. 3-16. The Prince, ch. XVII.
1970 PRESIDENTIAL POPULARITY FROM TBUMAN TO JOHNSON 21
years of his term: if a decline of 6 percentage bumps associated with some are considerably
points per year is indicated, he will be down 6 more obvious than others.
points at the end of his first year, 12 at the end In general, a rally point must be associated
of the second, 18 at the end of the third, and 24 with an event which 1) is international and 2)
after four years. There is nothing in the justifi- involves the United States and particularly the
cation for the coalition of minorities variable President directly; and it must be 3) specific,
which demands that the decline must occur with dramatic, and sharply focused.
such tedious regularity but, when curvilinear It must be international because only devel-
variants were experimented with, little or no im- opments confronting the nation as a whole are
provement was found. Hence the reliance in this likely to generate a rally round the flag effect.
study on the linear version which has the ad- Major domestic events—riots, scandals, strikes.
vantage of simplicity and ease of communica- —are at least as likely to exacerbate internal di-
tion. visions as they are to soothe them.
2. The "rally round the flag" variable. This To qualify as a rally point an international
variable seeks to bring into the analysis a phe- event is required to involve the United States
nomenon often noted by students of the Presi- and the President directly because major con-
dency and of public opinion: certain intense in- flicts between other powers are likely to engen-
ternational events generate a "rally round the der split loyalties and are less likely to seem rel-
flag" effect which tends to give a boost to the evant to the average American.
President's popularity rating. As Kenneth Waltz Finally the event must be specific, dramatic
has observed, "In the face of such an event, the and sharply focused in order to assure public at-
people rally behind their chief executive."18 Tom tention and interest. As part of this, events
Wicker: "Simply being President through a which transpire gradually, no matter how im-
great crisis or a big event . . . draws Americans portant, are excluded from consideration be-
together in his support."19 Richard Neustadt cause their impact on public attitudes is likely
notes "the correspondence between popularity to be diffused. Thus sudden changes in the
and happenings,"20 Burns Roper finds "approval bombing levels in Vietnam are expected to cre-
has usually risen during international crises,"21 ate a reaction while the gradual increase of
and Nelson Polsby observes, "Invariably, the American troops is not.
popular response to a President during interna- Errors in this process could occur by includ-
tional crisis is favorable, regardless of the wis- ing events whose importance is only obvious in
dom of the policies he pursues."22 retrospect or by ignoring events like the Geneva
The difficulty with this concept is in opera- summit of 1955 which may seem minor in his-
tionalizing it. There is a terrible temptation to torical perspective but were held significant at
find a bump on a popularity plot, then to scurry the time. For this reason more reliance has been
to historical records to find an international put on indexes of newspaper content than on
"rally point" to associate with it. This process broad, historical accounts of the period.23 In
all but guarantees that the variable will prove general if there has been a bias in selecting rally
significant. points it has been in the direction of excluding
The strategy adopted here to identify rally border-line cases. This was done in profound re-
points was somewhat different and hopefully spect for the lack of public interest and knowl-
more objective. A definition of what a rally edge on most items of international affairs.
point should look like was created largely on a At that, some 34 rally points were designated.
priori grounds and then a search of historical In general they can be said to fall into six cate-
records was made to find events which fit the gories. First, there are the four instances of sud-
definition. Most of the points so identified are den American military intervention: Korea, Le-
associated with bumps on the plot—that after banon, the Bay of Pigs, and the Dominican Re-
all was how the concept was thought of in the public. A second closely related category encom-
first place—but quite a few are not and the passes major military developments in ongoing
u wars: in Korea, the Inchon landing and the
Kenneth N. Waltz, "Electoral Punishment and Chinese intervention; in Vietnam, the Tonkin
Foreign Policy Crisis" in James N. Rosenau, Do-
23
mestic Sources oj Foreign Policy (New York: Especially valuable was Eric V. Nordheim and
Free Press, 1967), p. 272. Pamela B. Wilcox, "Major Events of the Nuclear
19
Tom Wicker, "In the Nation: Peace, It's Age: A Chronology to Assist in the Analysis of
Wonderful," New York Times, July 4, 1967, p. 18. American Public Opinion," Oak Ridge National
M
Op. cit., p. 100. Laboratory, Oak Ridge, Tennessee, August 1967.
a
Op. cit. Other sources often consulted included the New
22
Congress and the Presidency (Englewood York Times Index and the Chronology section of
Cliffs, N. J.: Prentice-Hall, 1964), p. 25. the World Almanac.
Bay episode, the beginning of bombing of North separated into two groups: "good" rally points
Vietnam, the major extension of this bombing, (e.g., the Cuban missile crisis) in which the last-
and the Tet offensive. Third are the major dip- ing effect on opinion was likely to be favorable
lomatic developments of the period: crises over to the President and "bad" one (e.g., the U-2
Cuban missiles, the U-2 and atomic testing, the crisis, the Bay of Pigs) in which the initial fa-
enunciation of the "Truman doctrine" with its vorable surge could be expected to be rather
offer of aid to Greece and Turkey, the beginning transitory. Two separate rally round the flag
of, and major changes in, the peace talks in Korea variables were then created with the anticipa-
and Vietnam, and the several crises in Berlin. tion that they would generate somewhat differ-
Fourth are the two dramatic technological de- ent regression coefficients. The differences how-
velopments: Sputnik and the announcement of ever were small and inconsistent. The public
the first Soviet atomic test. The fifth category seems to react to "good" and "bad" interna-
includes the meetings between the President and tional events in about the same way. Thus, to
the head of the Soviet Union at Potsdam in this limited extent, the equal weighting of rally
1945, Geneva in 1955, Camp David in 1959, points seems justified.
Paris in 1960, Vienna in 1961, and Glassboro in In tandem, the concepts underlying the coali-
1967. While these events are rarely spectacular tion of minorities and rally round the flag vari-
they, like crisis, do generate a let's-get-behind- ables predict that the President's popularity will
the-President effect. Because they are far less continually decline over time and that interna-
dramatic—even if sometimes more important— tional crises and similar events will explain short
Presidential conferences with other powers (e.g., term bumps and wiggles in this otherwise inex-
the British at Nassau) are excluded as are orable descent.24
American meetings with the Soviet Union at the 3. The "economic slump" variable. There is a
foreign minister level. goodly amount of evidence, and an even goodlier
Sixth and finally, as an analytic convenience amount of speculation, suggesting a relationship
the start of each Presidential term is rather ar- between economic conditions and electoral be-
bitrarily designated as a rally point. Presidents havior. The extension of such thinking to Presi-
Truman and Johnson came in under circum- dential popularity is both natural and prece-
stances which could justifiably be classified dented. Neustadt, for example, concludes the
under the "rally round the flag" rubric although recession in 1958 caused a drop in President Ei-
the crisis was a domestic one. The other points senhower's popularity.25
all involve elections or re-elections which per- The economic indicator used here will be the
haps might also be viewed as a somewhat unify- unemployment rate. The statistic recommends
ing and cathartic experience. itself because it is available for the entire period
These then are the events chosen to be asso- and is reported on a monthly basis.20 It is used
ciated with the rally round the flag variable. No as a general indicator of economic health or
listing will satisfy everyone's perspective about malaise and is not taken simply as a comment
what has or has not been important to Ameri- about the employed. It is assumed that the indi-
cans in this 24 year period. However, in the vidual respondent, in allowing economic percep-
analysis the variable has proven to be a rather tions to influence him, essentially does so by
hardy one. Experimentation with it suggests the comparing how things are now with how they
addition or subtraction of a few rally points is were when the incumbent began his present
likely to make little difference. term of office. If conditions are worse, he is in-
The rally round the flag variable is measured
21
by the length of time, in years, since the last Roper notes that President Kennedy's highest
rally point. It varies then from zero to a theo- point of popularity occurred after the Bay of
retical maximum of about four or an empirical Pigs invasion and concludes this fact says some-
one of 1.9. Like the coalition of minorities vari- thing special about that crisis event (op. cit.)
able, it should be negatively correlated with But this phenomenon is due to two effects—the
popularity: the longer it has been since the last rally round the flag effect and the fact that the
rally round the flag event, the lower the popu- event occurred very early in Kennedy's admini-
larity of the incumbent. It is; the simple r is stration when the value for the coalition of minor-
— .11. Some experiments with curvilinear trans- ities variable was yet very low.
23
formations of the variable were attempted but, Op. cit.,-p. 97ff.
26
since improvement again was marginal at best, Data were gathered from Geoffrey H. Moore
the variable has been left in linear form. (ed.), Business Cycle Indicators, Vol. II (Prince-
Each rally point is given the same weighting ton: Princeton University Press, 1961), p. 122;
in the analysis. One effort to soften this rather and, for more recent data, issues of the Monthly
crude policy was made. The rally points were Labor Review.
1970 PRESIDENTIAL POPULARITY FBOM TRUMAN TO JOHNSON 23
clined to disapprove the President's handling of Harvey Segal notes, "What was important was
his job, if things are better he is inclined to ap- the vague but pervasive feeling of dissatisfaction
prove. The economic variable therefore becomes with the performance of the economy, the pain
the unemployment rate at the time the incum- that made the public receptive to JFK's ap-
bent's term began subtracted from the rate at peals."28 In 1968, representing administrations
the time of the poll.27 It is positive when things that had presided over an unprecedented period
are worse and negative when things are better. of boom, Vice President Humphrey never
It should be negatively correlated with popular- seemed able to turn this fact to his advantage.
ity. But it isn't. It is important to note that in practice this
Unemployment reached some of its highest variable, which will be called the "economic
points during the recessions under the Eisen- slump" variable because of its inability to credit
hower administration. The problem, to be exam- boom, takes on a non-zero value only during the
ined more fully in Section VI below, is that Ei- Eisenhower administration and during the un-
senhower was a generally popular President. employment rise of 1949-50. In symbolic form
Thus even though his popularity seemed to dip the variable's peculiarities can be expressed in
during the recessions, high unemployment comes the following; the units are the percentage of
to be associated with a relatively popular Presi- unemployed:
dent. This problem can be handled rather easily
E = U,-U,0 if U, - Uh > 0
within regression analysis by assigning to each
of the Presidential administrations a "dummy" =0 if U, -Uk< 0
variable, the care and feeding of which will be where
discussed more fully in Section III below. Ut= Unemployment rate at the time of the
However, even when this circumstance is survey
taken into account, the correlation coefficient and
and the regression coefficient for the economic £/(„= Unemployment rate at the beginning of
variable remain positive. This seems to be the incumbent's present term
largely due to the fact that both unemployment
4. The "war" variable. It is widely held that
and the popularity of the incumbent President
the unpopular, puzzling, indecisive wars in
were in general decline between 1961 and 1968.
Korea and Vietnam severely hurt the popularity
The correlation for the period is .77.
of Presidents Truman and Johnson.29 As noted
Therefore a final alteration administered to
in the quotation that heads this report, Presi-
the economic variable was to set it equal to zero
dent Johnson himself apportions 20 percentage
whenever the unemployment rate was lower at
points of his drop in popularity to the Vietnam
the time of the survey than it had been at the
War.
start of the incumbent's present term. This al-
This notion seems highly plausible. The popu-
teration is a substantive one and is executed as
larity of Presidents Truman and Johnson was in
the only way the data can be made to come out
steady decline as the wars progressed with rec-
"right." In essence it suggests that an economy
ord lows occurring during each President's last
in slump harms a President's popularity, but an
year in office at points when the wars seemed
economy which is improving does not seem to
most hopeless and meaningless. The wars un-
help his rating. Bust is bad for him but boom is
questionably contributed in a major way to
not particularly good. There is punishment but
their decisions not to seek third terms and then,
never reward.
when they had stepped aside, the wars proved to
Perhaps this can be seen in a comparison of
be major liabilities for their party's candidates
the 1960 and the 1968 campaigns. In 1960, as
in the next elections. Overall, the correlation
" One wrinkle, which is intuitively comfortable between Presidential popularity and the pres-
but makes little difference in the actual results, ence of war is — .66.
was to do something about the unemployment There are problems with this analysis, how-
rates at the start of the first terms of Presidents ever. The coalition of minorities concept argues
Truman and Eisenhower when unemployment was that decline is a natural phenomenon and, in-
"artifically" depressed due to ongoing wars. Pre- deed, a glance at a plot of Presidential popular-
sumably the public would be understanding about ity clearly shows Truman and Johnson in de-
the immediate postwar rise in unemployment. cline before the wars started. Furthermore, as
Therefore for these two terms the initial unem-
28
ployment level was taken to be that level which "The Pain Threshold of Economics in an
held six months after the war ended while the Election Year," New York Times, July 15, 1968.
29
economic variable for the few months of the war Waltz, op. cit., pp. 273ff, 288; Neustadt, op. cit.,
and the six month period was set equal to zero. pp. 97-99; Wicker, op. cit.; Roper, op. cit.
will be seen, both men experienced noticeable which involved the examination of hundreds of
declines during their first terms when they had regression equations—a search was made for
no war to contend with. The real question is, other variables which could profitably be added
then, did the war somehow add to the decline of to the predictor set.
popularity beyond that which might be expected International developments are reasonably
to occur on other grounds ? well incorporated into the analysis with a spe-
An answer can be approached through multi- cific variable included for war and another for
ple regression analysis. After allowing for a gen- major crisis-like activities. Domestically, how-
eral pattern of decline under the coalition of mi- ever, there is only the half-time variable for
norities variable, the additional impact of a economic slump and the important but very un-
variable chosen to represent war can be as- specific coalition of minorities variable.
sessed. It is also possible in this manner to Accordingly it would be valuable to generate
compare the two wars to see if their association some sort of domestic equivalent to the rally
with Presidential popularity differed. round the flag variable to assess more precisely
The presence of war is incorporated in the how major domestic events affect Presidential
analysis simply by a dummy variable that takes popularity. Operationally, however, this is a
on a value of one when a war is on and remains difficult task. First, while it is a justifiable asser-
zero otherwise. The beginning of the Vietnam tion that international crises will redound in the
War was taken to be June 1965 with the begin- short term to a President's benefit it is by no
nings of the major US troop involvement. At means clear how a domestic crisis, whether riot,
that point it became an American war for the strike, or scandal, should affect his popularity.
public; before that ignorance of the war was Furthermore major domestic concerns have var-
considerable: as late as mid-1964, 25 percent of ied quite widely not only in intensity and dura-
the public admitted it had never heard of the tion, but also in nature. Labor relations, which
fighting in Vietnam.30 rarely made big news in the mid-1960s, were of
Other war measures of a more complex na- profound concern in the middle and late 1940s
ture were experimented with. They increase in as a multitude of major strikes threatened to
magnitude as the war progresses and thus cripple the nation and the adventures of John L.
should be able to tap a wearying effect as the Lewis and the Taft-Hartley bill dominated the
years go by and should be negatively associated headlines. In the 1950s, however, labor broke
with popularity. These measures, however, are into the news only with an occasional steel or
very highly correlated with the coalition of mi- auto strike or with the labor racketeering scan-
norities variable for the two relevant Presiden- dals in the last years of the decade. On the
tial terms and thus are all but useless in the other hand, race relations, of extreme impor-
analysis. The simple dummy variable suffers this tance in the 1960s, made, except for the Little
defect in lesser measure, although it is far from Rock crisis of 1957 and an occasional election-
immune, and thus, despite its crudities, has been time outburst, little claim to public attention
used.31 before that time. From the late 1940s into the
5. Other variables. The analysis of Presi- mid-1950s sundry spy and Communist hunts
dential popularity will apply in various ways were of concern, but the issue fairly well fizzled
only the four variables discussed above—a after that. Other issues which might be men-
rather austere representation of a presumably tioned had even briefer or more erratic days in
complex process. As will be seen, it is quite pos- the sun: the food shortage of 1947, the MacAr-
sible to get a sound fit with these four variables, thur hearing of 1951, various space flights. Simi-
but at various stages in the investigation— larly, personal crisis for the Presidents such as
heart attacks and major surgery for Presidents
30
A. T. Steele, The American People and China Eisenhower and Johnson and the attempted as-
(New York: McGraw-Hill, 1966), p. 294. In May sassination of President Truman could not
1964 Gallup found almost two-thirds of the popu- readily be fashioned into a predictor variable. In
lation said they paid little or no attention to de- any event, these events seem to have far more
velopments in South Vietnam. Lloyd A. Free and impact on the stock market than on popularity
Hadley Cantril, The Political Beliefs of Americans ratings.
(New York: Clarion, 1968), pp. 59-60. Scandal is a recurring feature of public
31
Although the Korean War continued into Pres- awareness and thus is more promising as a po-
ident Eisenhower's administration, he is not tential variable in the analysis. Besides the scan-
"blamed" for the war in the analysis since of dals associated with alleged spies and Commu-
course he was elected partly because of discontent nists in the government during the McCarthy
over the war. Accordingly the war variable is set era and those associated with labor in the late
at zero for this period. 1950s, Americans, with greatly varying degrees
of pain, have suffered through the five percenter In this section the relation of the first three
scandal of 1949-50; charges of corruption in variables to Presidential popularity will be as-
the RFC in 1951, in the Justice Department in sessed. In the next section the war variable will
1952, and in the FHA in 1954; and scandals over be added to the analysis.
Sherman Adams in 1958, over television quiz The association between the first three vari-
shows in 1959, over industry "payola" in the ables and Presidential popularity is given in its
late 1950s, over Billie Sol Estes in 1962, and baldest form in equation 1 in Table I.32 The
over Bobby Baker in 1963. While scandal is equation explains a respectable, if not sensa-
never worked into the regression analysis some tional, 22 percent of the variance. The coalition
preliminary suggestions as to its relevance to a of minorities variable shows, in conformity with
"moral crisis" phenomenon which may in turn the speculation above, a significant negative re-
affect Presidential popularity are developed in lationship. The equation suggests that, in gen-
Section VI below. eral, a President's popularity rating starts at 69
Some thought was given to including a "lame percent and declines at a rate of about six per-
duck" variable when it was observed that the centage points per year.
popularity of Presidents Truman and Johnson However, while the coefficients for the rally
rose noticeably after they decided not to seek round the flag and economic slump variables are
third terms. The trouble is, however, that Presi- in the expected direction, they are not signifi-
dent Eisenhower was a lame duck for his entire cant either in a statistical or a substantive
second term and it was found easier to ignore sense. The trouble with the economic slump
the whole idea than to decide what to do about variable was anticipated in the discussion about
this uncomfortable fact. it in Section II: the economic decline occurred
One domestic variable which did show some during the relatively popular reign of President
very minor promise was a dummy variable for Eisenhower; while the slump seems to have hurt
the presence of a major strike. The variable his popularity, even with the decline he re-
takes on a zero value almost everywhere except mained popular compared to other Presidents;
in parts of President Truman's first term. After hence what is needed is a variable to take into
that time major strikes were rather unusual account this peculiar "Eisenhower effect."
and, when they did occur, usually lasted for To account for this phenomenon, equation 2
such a short time that there was barely time to mixes into the analysis a dummy variable for
have a public opinion survey conducted to test each of the Presidents. This formulation insists
their effects. Despite these peculiarities, the that all Presidents must decline (or increase) in
variable did show statistical significance, though popularity at the same rate but, unlike equation
only after the Korean War dummy had been in- 1, it allows each President to begin at his own
corporated in the equation to allow for a major particular level. Thus peculiar effects of person-
peculiarity of President Truman's second term. ality, style, and party and of differences in the
Substantively the variable suggests a popularity conditions under which the President came into
drop of less than three percentage points when a office can be taken into account.33
major strike is on and, as such a minor contrib- The addition improves things considerably.
utor, it is not included in the discussion below. 82
Its small success, however, may suggest that Each equation is displayed vertically. The de-
further experimentation with the effects of spe- pendent variable, the percentage approving the
cific domestic events could prove profitable. way the President is handling his job, has a mean
of 57.5 and a standard deviation of 14.8. The num-
III. RESULTS WITHOUT T H E WAR VARIABLE ber of cases is 292. The figures in parentheses are
In summary the expected behavior of Presi- the standard errors for the respective partial re-
dential popularity is as follows. It is anticipated gression coefficients. To be regarded statistically
1) that each President will experience in each significant a regression coefficient should be, con-
term a general decline of popularity; S) that ventionally, at least twice its standard error. All
this decline will be interrupted from time to equations reported in this study are significant (F
time with temporary upsurges associated with test) at well beyond the .01 level. The Durbin-
international crises and similar events; 8) that Watson d is an indicator of serial correlation which
the decline will be accelerated in direct relation suggests decreasing positive serial correlation as
to increases in unemployment rates over those the statistic approaches the value of 2.0. All equa-
prevailing when the President began his term, tions in this study exhibit a statistically signifi-
but that improvement in unemployment rates cant amount of positive serial correlation.
33
will not affect his popularity one way or the The dummy variables formalize the sort of
other; and 4) that the President will experience discussion found in Neustadt, op. dt., p. 98. They
an additional loss of popularity if a war is on. account for what a singer might call tessitura.
TABLE 1. REGRESSION RESULTS INCLUDING ADMINISTRATION EFFECTS
Equations
(1) (2) (3) (4)
Intercept 69.37 54.51 68.15 71.52
Independent variables
Coalitions of minorities (in years) -6.14 -5.12

(0.71) (0.48)
Rally round the flag (in years) -0.31 -2.15 -1.87 -2.63
(1.95) (1.35) (1.07) (1.07)
Economic slump (in % unemployed) -0.09 -3.18 -5.30 -5.86

(0.96) (0.75) (0.60) (0.60)
Dummy variables for administrations
Eisenhower 24.08 0.51

(1.42) (2.29)
Kennedy 23.87 11.33

(1.82) (2.96)
Johnson 10.46 4.26

(1.54) (2.48)
Coalition of minorities variable for

administrations (in years)
Truman -11.44 -12.54

(0.84) (0.50)
Eisenhower 0.83 0.13

(0.57) (0.47)
Kennedy -5.96 -1.58
(1.36) (0.82)
Johnson -9.12 -8.68

(0.67) (0.50)
d .13 .30 .46 .44
Standard error of estimate 13.16 8.88 6.86 7.06
fi2 .22 .65 .79 .78
The fit is much better and the rally round the centage points per year but that each has done
flag and economic slump variables attain re- so at his own particular level. President Tru-
spectable magnitudes in the predicted direction, man's decline is measured from a starting point
although the rally round the flag variable does of 54.51 percent (when the dummy variables for
not quite reach statistical significance. the Eisenhower, Kennedy, and Johnson adminis-
The equation suggests that the Presidents trations are all zero). President Eisenhower de-
have declined at an overall rate of overfiveper- clines from a much higher level, about 79 per-
cent (54.51 + 24.08), President Kennedy from In equation 3, Presidents who served two
78 percent, and President Johnson from 65 per- terms were required to begin each term at the
cent. same level and their rate of decline or increase
The importance of these dummy variables also had to be the same in each term. Liberation
clearly demonstrates that an analysis of Presi- from these restrictions is gained in the rather
dential popularity cannot rely entirely on the cluttered equation 5 of Table 2 which is like
variables discussed in Section II, but must also equation 3 except that it affords a term by
incorporate parameters designed to allow for the term, rather than simply an administration by
special character of each administration. To an administration comparison. As can be seen Pres-
extent this is unfortunate. The beauty of equa- ident Eisenhower managed a statistically signifi-
tion 1 is that it affords a prediction of a Presi- cant increase of popularity of some two and a
dent's popular rating simply by measuring how half percentage points per year in his first term.
long he has been in office, how long it has been His second term ratings showed a more human,
since the last rally point, and how many people but very minor and statistically non-significant
are unemployed. Such predictions, however, decline.35
would be quite inaccurate because the fit of the No important differences emerge in the Ei-
equation is rather poor. Instead one must in- senhower phenomenon when the economic slump
clude the administration variables, the mag- variable, which functions mainly during the Ei-
nitudes of which cannot be known until the senhower years, is dropped from the equation.
President's term is over. So much for beauty. No matter how the data are looked at then
In equation 3 administration effects are in- the conclusion remains the same. President Ei-
corporated in a different manner, greatly im- senhower's ability to maintain his popularity,
proving fit and reducing serial correlation. In especially during his first term, is striking and
this formulation each President is allowed to unparallelled among the postwar Presidents. An
begin at his own level of popularity as in equa- examination of some of the possible reasons for
tion 2, but in addition each may decline (or in- this phenomenon is conducted in Section VI
crease) at his own rate: for each administration below.
there is a different coefficient for the coalition of The rally round the flag and the economic
minorities variable. Three of the four values so slump variables emerge alive and well in equa-
generated are strongly significant while the mag- tions 3, 4, and 5 (and 6). Both are usually sta-
nitudes of the administration dummies drop tistically significant but their substantive impor-
greatly. When the administration dummies are tance varies as one moves from an administra-
dropped entirely from consideration, as in equation by administration formulation of the coali-
tion 4, the regression coefficients mostly remain tion of minorities variable (equations 3 and 4)
firm and the fit of the equation is scarcely weak- to the term by term formulation in Table 2.
ened. It is clear that the important differences Specifically, the rally round the flag variable
between administrations do not lie so much in gets stronger while the economic slump variable
different overall levels of popularity, but rather weakens.
in the widely differing rates at which the coali- The rally round the flag variable is very
tion of minorities variable takes effect. much a parasite—it is designed to explain
The popular decline of Presidents Truman bumps and wiggles on a pattern measured
and Johnson has been almost precipitous. Presi- mainly by the other variables. Consequently the
dent Truman's rating fell off at some 11 or 12 rally round the flag variable does very poorly on
percentage points per year while President its own and only begins to shine when the over-
Johnson declined at a rate of around 9 points a all trends become well determined by the rest of
year. President Kennedy was noticeably more
successful at holding on to his supporters. Then see if this peculiarity had any major impact, equa-
there is the Eisenhower phenomenon: in spite of tions 3 and 4 were recalculated using the per-
all the rationalizations for the coalitions of mi- centage disapproving as the dependent variable.
norities concept tediously arrayed in Section II, This manipulation causes no fundamental differ-
President Eisenhower's rating uncooperatively ences, although President Eisenhower's rating be-
refuses to decline at all.34 haves a little less outrageously.
34 85
It was noted in Section I that some minor bias If the term dummies are dropped from the
in these results is introduced by an embellished equation to attain a version comparable to equa-
rate of "no opinion" in the first weeks of the tion 4, the Eisenhower phenomenon holds except
Kennedy and first Eisenhower terms. As this rate that his first term increase drops to 2.00 (still sig-
declined, there was some tendency for the Presi- nificant) and his second term decrease is a slightly
dents' approval and disapproval rates to rise. To steeper —0.36 (still not significant).
TABLE 2. BEGRESSION RESULTS INCLUDING TEBM The declining fortunes of the economic slump
EFFECTS AND THE WAB VARIABLES variable suggest that the variable in equations 3
and 4 was partly covering for the differences be-
Equations tween the two Eisenhower terms: the first term
was associated with increasing popularity and a
(S) (6) smaller recession, the second with somewhat de-
Intercept 72.00 72.38 clining popularity and a larger recession. With
the Eisenhower terms more thoroughly differen-
Independent variables tiated in equation 5, the variable is reduced to a
Rally round the flag (in years) -4.88 -6.15
more purely economic function. The magnitude
(1.04) (1.05) of the coefficient of the economic slump variable
Economic slump (in % unemployed) -2.67 -3.72 in this equation suggests a decline of popularity
(0.65) (0.65) of about three percentage points jor every per-
centage point rise in the unemployment rate
Dummy variables for terms over the level holding when the President began
Truman—second -15.25 -12.41 his present term. Since the unemployment rate
(3.69) (3.57) has varied in the postwar period only from
Eisenhower—first -3.17 -2.41 about 3 to 7 percent, the substantive impact of
(3.15) (3.02) the economic slump variable on Presidential
Eisenhower—second -5.30 -4.35 popularity is somewhat limited.
(3.07) (2.94)
Kennedy 7.53 7.18 IV. RESULTS WITH THE WAR VARIABLE ADDED
(3.29) (3.14)
The variable designed to tap the impact on
Johnson—first 7.14 6.77
(5.50) (5.26) Presidential popularity of the wars in Korea and
Johnson—second — 1.15 -0.79
Vietnam was applied with no great confidence
(3.06) (3.25) that it would prove to have an independent,
added effect when the coalition of minorities had
Coalition of minorities variable already been incorporated into the equation
for terms (in years)
especially given the problem of multicollinearity.
Truman—first -9.21 -8.93 It is obvious from a perusal of a plot that, as
(1.41) (1.35) noted in Section II and as demonstrated in
Truman—second -7.98 -2.83 equation 5, Presidents Truman and Johnson
(1.00) (1.37) were in popular decline during their warless first
Eisenho wer—first 2.45 2.58 terms.36 Furthermore each was in clear decline
(0.85) (0.82) in the first part of his second term before the
Eisenhower—second -0.07 0.22 wars started and it is not at all obvious that
(0.65) (0.62) this trend altered when the wars began.
Kennedy -5.11 -4.76 The equations suggest otherwise, however.
(1.21) (1.16)
When a war dummy was appended to the equa-
Johnson—first -4.98 -3.71
(14.01) (13.39) tions already discussed, it emerged significant
-8.15 -8.13
and suggested that the presence of war de-
Johnson—second
(0.66) (0.80) pressed the popularity of Presidents Truman
and Johnson by over seven percentage points.
Dummy variables for wars The next step, obviously, was to set up a sep-
Korea -18.19
arate dummy variable for each war. This
(3.43) brought forth the incredible result documented
Vietnam -0.28 in equation 6: the Korean War had a large, sig-
(2.79) nificant, independent negative impact on Presi-
dent Truman's popularity of some 18 percent-
d .57 .67 age points, but the Vietnam War had no inde-
Standard error of estimate 6.07 5.80
.84 .86 pendent impact on President Johnson's popular-
ity at all.
Confronted with a result like this, one's first
the equation. In the end, the rally round the flag 38
variable suggests a popularity decline of around Regression statistics relating to President
five or six percentage points for every year since Johnson's first term are very unreliable, as the
the last rally point—about the same magnitude size of the standard errors suggests, because the
as the coalition of minorities variable in its gen- popularity question was posed so few times dur-
eral state as in equations 1 and 2. ing this brief period.
impulse is to do something to make it go away. plaints from the right during Vietnam that
This impulse was fully indulged. Variables were President Johnson had adopted a "no win" pol-
transformed and transmuted, sections of the icy there, these were continually being undercut
analysis were reformed or removed, potentially by public statements from General William
biassing data were sectioned out. But nothing Westmoreland—a man highly respected by the
seemed to work. The relationship persisted. In right—insisting that he was receiving all the
fact under some manipulations the relationship support he needed from the President and was
became stronger. getting it as fast as he needed it.
One's second impulse, then, is to attempt to If these observations are sound, the single
explain the result. One speculates. event which best differentiates the impact of the
The wars in Korea and Vietnam differed Korean and Vietnam wars on Presidential popu-
from each other in many respects, of course, but larity was President Truman's dismissal of Gen-
it seems unlikely that these differences can be eral Douglas MacArthur. That move was a
used in any simple manner to explain the cu- major factor in the politicization of the war as
rious regression finding. This is the case because, Republicans took the General's side and echoed
as one study has indicated, public response to his complaints that it was the President's med-
the wars themselves was much the same. Sup- dling in policy that was keeping the war from
port for each war, high at first, declined as a being won.39
logarithmic function of American casualties— The differing impact of the wars on Presiden-
quickly at first, then more slowly. The functions tial popularity therefore may be due to the fact
for each of the wars for comparable periods were that Korea became "Truman's war" while Viet-
quite similar. Furthermore both wars inspired nam never in the same sense really became
support and opposition from much the same seg- "Johnson's war."40
ments of the population.37 39
See John W. Spanier, The Truman-MacArthur
Therefore it is probably a sounder approach
Controversy and the Korean War (Cambridge,
in seeking to explain the regression finding to
Mass.: Belknap, 1959); also Neustadt, op. dt.,
look specifically at popular attitudes toward the
passim; and Trumbull Higgins, Korea and the
President's relation to the war, rather than to
Fall oj MacArthur (New York: Oxford, 1960).
perceptions of the war itself. A comment by
See also the data in George Belknap and Angus
Richard Neustadt seems strikingly relevant in
Campbell, "Political Party Identification and At-
this respect. "Truman," he observes, "seems to
titudes Toward Foreign Policy," 15 Public Opin-
have run afoul of the twin notions that a war-
ion Quarterly 601-23 (Winter 1951-52). Note es-
time Chief Executive ought to be 'above politics'
pecially the strong party polarization on the issue.
and that he ought to help the generals 'win.' " 38
That the public was strongly inclined to support
President Johnson seems to have run consid-
General MacArthur in the dispute can be seen
erably less afoul. In seeking to keep the war
from poll data. The first polls, conducted as the
"above politics," he assiduously cultivated bi-
General was making his triumphal, "old soldiers
partisan support for the war and repeatedly
never die" return to the United States in April
sought to demonstrate that the war effort was
1951, suggest more than twice as many people sup-
simply an extension of the policies and actions
ported the General as supported the President. As
of previous Presidents. He was especially suc-
Neustadt suggests (op. dt., p. 97), emotion on the
cessful at generating public expressions of ap-
issue faded during the Senate Hearings on the is-
proval from the most popular Republican of
sue which lasted until June and this seems to have
them all: General Eisenhower. Vocal opposition
been to the benefit of President Truman's posi-
to the war in Vietnam came either from groups
tion. The Truman point of view received its great-
largely unassociated with either party or from
est support in late June and early July as peace
members of the President's own party. Then,
talks were being begun. As the talks began to
when the latter opposition began to move from
prove unproductive, however, public opinion be-
expressions of misgivings at congressional hear-
gan to revert to its previous support of General
ings to explicit challenges in the primaries, Pres-
MacArthur, until, by the first days of 1952 (when
ident Johnson removed himself from the battle
the polling agencies grew bored with the issue),
precisely, he said, to keep the war "above poli-
the MacArthur position was as strongly approved
tics." And, while there were occasional com-
and President Truman's as strongly rejected as
37
John E. Mueller, "Patterns of Popular Sup- ever.
port for the Wars in Korea and Vietnam," un- " There is evidence which suggests that World
published paper, Department of Political Science, War II, a much more popular (and much larger)
University of Rochester, 1969. war than either Korea or Vietnam, may have
38
Op. dt., p. 97. worked to the distinct benefit of President Roose-
30 THE AMEKICAN POLITICAL SCIENCE REVIEW VOL. 64
One other item of speculation might be put ysis is applied in the Korean period it is found
forth. Domestically, the war in Vietnam was ac- that a variable associated with the Korean War
companied by a profoundly important crisis as does show significance even after other variables
America confronted its long-ignored racial di- have been taken into account. What the regres-
lemma head on. There seems to have been noth- sion analysis shows therefore is that, while the
ing comparable during the Korean War. The Korean War does seem to have had an indepen-
clamor associated with McCarthyism comes to dent, additional impact on President Truman's
mind but many analysts feel that, however im- decline in popularity, the Vietnam War shows
portant to politicians, intellectuals, and journal- no such relation to President Johnson's decline.
ists, McCarthyism was of rather less than major
concern to public opinion.41 Furthermore its V. THE RESIDUALS
dramatic climax, the Army-McCarthy hearings, An analysis of the residuals finds that equa-
took place months after the Korean War had tion 6 predicts worst in President Truman's first
ended and over a year after President Truman term. The President's extremely high initial rat-
left office. ings are not well predicted suggesting that the
It may be, then, that the discontent asso- equation does not adequately account for the
ciated with the racial crisis was enough by itself trauma of President Roosevelt's death42 com-
to cause much of President Johnson's popular bined as it was with the ending of World War
decline and thus that the unhappiness over the II and with important peace conferences. It was
Vietnam War could make little additional in- almost as if Americans were afraid to disap-
road. In the Truman case, there was no pro- prove of President Truman.
found independent domestic source of discon- From these spectacular highs, President Tru-
tent: his second term coalition of minorities de- man plunged to great lows during the labor tur-
cline is usually found as in equation 5 to have moil of 1946. These ratings are also badly speci-
been less than his first term decline and when, fied by the equation. The Truman popularity
as in equation 6, a variable has already ac- rose in early 1947, as the labor situation eased,
counted for the war effect, his decline is quite and then declined for the rest of the term. Thus
moderate. Thus in a sense there was "room" for while President Truman's first term, like the
the war to have an independent impact. other Democratic terms, shows an overall de-
It would be wise in concluding this section to cline of popularity, that decline was considera-
emphasize what has and what has not been said. 42
In late November 1945, over six months after
It has not been argued that the war in Vietnam
President Roosevelt's death, Gallup asked his sam-
had nothing to do with President Johnson's de-
ple, "In your opinion, who is the greatest person
cline in popularity and thus the analysis cannot
living or dead, in world history?" Fully 28 percent
really be used to refute the President's own esti-
proffered Roosevelt's name. Abraham Lincoln was
mation of the impact of Vietnam as indicated in
mentioned by 19 percent, Jesus Christ by 15 per-
the quotation that heads this study. However it
cent, and George Washington by 8 percent. No
is argued that whatever impact the war had was
one else received more than 2 percent. And the
tapped by the other variables in the equation,
aura lasted. A survey conducted in June 1949 in
especially the coalition of minorities variable
the city of Philadelphia (which had voted 59 per-
which is specifically designed to account for gen-
cent for Roosevelt in 1944 as against a national
eral overall decline. When the same sort of anal-
rate of 55 percent) posed this question: "Could
velt. The National Opinion Research Center in you tell us the name of a great person, living or
its post-election poll in 1944 asked Roosevelt sup- dead, whom you admire the most?" The most
porters if they would have voted for Dewey "if commonly mentioned names were Roosevelt with
the war had been over." Enough answered in the 42 percent, Lincoln with 9 percent, and Washing-
affirmative to suggest that Dewey might well have ton with 5 percent. (The absence of Jesus Christ
won in a warless atmosphere. From data supplied on this latter list presumably can be laid to the
by the Inter-University Consortium for Political peculiarities of question wording—or of Philadel-
Research. phians.) Fillmore H. Sanford, "Public Orientation
11
Samuel A. Stouffer, Communism, Conformity, to Roosevelt," 15 Public Opinion Quarterly 190-
and Civil Liberties (Garden City, N.Y.: Double- 91, 200 (Summer 1951). In 1948, Roper found 43
day, 1955), especially ch. 3; Campbell, et al., op. percent of a national sample offering Roosevelt's
cit., pp. 50-51; Nelson W. Polsby, "Toward an name when queried, "Considering all the men in
Explanation of McCarthyism," 8 Political Stud- America who have been prominent in public affairs
ies 250-71 (October 1960); Elmo Roper, You and during the past 50 years, which one or two have
Your Leaders (New York: Morrow, 1957), pp. you admired the most?" Dwight Eisenhower was
250-51. second at 17 percent. E. Roper, op. cit., p. 22.
1970 PRESIDENTIAL POPULARITY PROM TKUMAN TO JOHNSON 31
bly more erratic than the others. The dummy 2. Early in his first term President Eisen-
variable for strikes, discussed briefly in Section hower was able to present to the public one sen-
II, improved matters only slightly. sational achievement: he ended the Korean War
Beyond this, the residuals are reasonably well —or, at any rate, presided over its end. This ac-
behaved. There are small but noticeable effects complishment was seen by the public as he left
from the lame duck phenomenon at the end of office to be a great one45 and was used with
the Truman and Johnson administrations and profit by the Republicans in a Presidential cam-
from the "no opinion" peculiarity of the initial paign a full 15 years after it happened. From
weeks of the Eisenhower and Kennedy adminis- the standpoint of public opinion it may well
trations. And here and there are data points have been the most favorable achievement
whose magnitudes have somehow managed to turned in by any postwar President. As such it
escape specification by the variables in the may have tended to overwhelm the negative im-
regression equation. One can of course gener- pact of anything else the President did, at least
ate a unique explanation for each of these but for the first years of his administration. Some
this procedure clutters the analysis more than it credit for this is given in the regression analysis
is worth. Besides, the laws of sampling insist since the signing of the truce is counted as a
that Gallup must have made some mistakes. rally point, but this may be a totally inadequate
As the magnitude of the Durbin-Watson d recognition.
indicates, serial correlation has by no means There is another aspect of President Eisen-
been eliminated in the regression equations. Al- hower's first term which may not be sufficiently
lowing the coalition of minorities variable to be accounted for in the rally round the flag vari-
specified for each term improved things considerable: the euphoria of the "spirit of Geneva" pe-
ably, but much is left to be desired. riod toward the end of the term when the Presi-
dent's popularity should have been at its lowest
VI. THE EISENHOWER PHENOMENON ebb.
Great noise was made in Section II about the 3. President Eisenhower's amateur status
coalition of minorities variable with its stern may also have worked to his benefit, at least for
prediction that a President's popularity would a while. The public may have been more will-
decline inexorably over his four year term. The ing to grant him the benefit of a doubt, to ex-
noise was not entirely unjustified since the vari- tend the "honeymoon" period, than it would for
able proved to be a hardy and tenacious predic- a President who is a political professional. It is
tor for the postwar Democratic administrations. also easier under these circumstances for the
The variable fails for the Eisenhower admin- President to appear "above the battle" and thus
istration, however, especially for the General's to be blamed only belatedly and indirectly for
first term. The analysis suggests then that if a political mishaps, thereby softening their impact.
President wants to leave office a popular man he 4. President Eisenhower may have been cu-
should either 1) be Dwight David Eisenhower, riously benefited by the fact that, especially on
or 2) resign the day after inauguration. the domestic front, he didn't do anything.48 In-
The Eisenhower phenomenon, noted but left
dangling without explanation or rationalization "De Gaulle and Eisenhower: The Public Image of
in Section III, deserves special examination. the Victorious General," in Angus Campbell, et al.,
Why didn't President Eisenhower decline in Elections and the Political Order (New York:
popularity like everybody else? A number of Wiley, 1966), pp. 292-345.
suggestions can be proffered. 45
In December 1960 the public was asked what
1. To begin with, credit must be given to it felt was Eisenhower's greatest accomplishment.
President Eisenhower's"personalappeal: he was The ending of the Korean War was mentioned by
extremely likeable—a quality very beneficial in 11 percent and a related comment, "he kept us
a popularity contest and one lacked in abun- out of war," was suggested by an additional 32
dance by, say, Lyndon Johnson. As Fillmore percent. No other specific accomplishment was
Sanford has observed, "The American people, in mentioned by more than 5 percent; only 3 percent
reacting to a national leader, put great emphasis mentioned anything having to do with the do-
on his personal warmth"43—a quality projected mestic scene. See also Neustadt, op. cit., p. 98.
to an unusual degree by President Eisenhower. 40
As Irving Kristol argues, ". . . when a con-
As part of this, he was able to project an image servative administration does take office, it pur-
of integrity and sincerity which many found to sues no coherent program but merely takes satis-
be enormously attractive.44 faction in not doing the things that the liberals
41
Op. cit., p. 19S. may be clamoring for. This, in effect, is what hap-
"See Philip E. Converse and Georges Dupeux, pened during the two terms of President Eisen-
deed analysts of the Eisenhower administration TABLE 3 . THE GREENFIELD INDEX

often argue that its contribution lies in what it
didn't do. The times called for consolidation, Number of Items under the Heading,
they argue, and President Eisenhower's achieve- "US: Moral Conditions" in
ment was that he neither innovated nor re- Readers' Guide to Periodical Literature,
pealed, but was content to preside over a period 1945-1968, by year
of placidity in which he tacitly gave Republican
respectability to major Democratic innovations 1945 1
of earlier years: the programs of the New Deal 1946 0
domestically and the policies of the Truman 1947 8
Doctrine internationally.47 1948 1
In terms of the justification for the coalition 1949 1
of minorities variable as discussed in Section II, 1950 3
such behavior could have a peculiar result. It 1951 35
was assumed in part that the President would 1952 17
enact programs which, while approved by the 1953 1
majority, would alienate intense minorities 1954 4
which would gradually cumulate to his disad- 1955 2
vantage. But suppose the President doesn't do 1956 0
anything. Those who want no change are happy 1957 7
while, if things are sufficiently ambiguous, those 1958 0
who support change have not really been denied 1959 9
by an explicit decision and can still patiently 1960 32
wait and hope. At some point of course those 1961 23
who want change begin to see that they are 1962 10
never going to get their desires and may become 1963 11
alienated, but this will be a delayed process. At 1964 5
least in moderate, placid times, a conservative 1965 6
policy may dissipate some of the power of the 1966 7
coalition of minorities phenomenon. Were polls 1967 18
available, one might find that President Warren 1968 10
Harding maintained his popularity as strikingly
as President Eisenhower.
5. Although it might be difficult to sort out crisis in the postwar period arose in the early
cause and effect, it is worth noting that Presi- 1950s and was associated with " 'five percenters,'
dent Eisenhower's first term (and most of his deep freezes, mink coats, the Kefauver hearings,
second) coincided with a period of national and a series of basketball fixes," while "the sym-
goodness. In a brilliant article Meg Greenfield bols of our present [1961] decline are Charles
has noted that "moral crises" as appraised and Van Doren, payola, cheating in school, and the
bemoaned by intellectuals seem to follow a decision of Frances Gary Powers not to kill
cyclic pattern: we go through a period in which himself." We never recovered as thoroughly
the popular journals are filled with articles tell- from that crisis as we did from the earlier one
ing us how bad we are after which there is a pe- for, as the crisis showed signs of waning (the
riod of respite.48 success of the Peace Corps began to show how
Miss Greenfield's main indicator of these eth- good we were at heart), new elements—Billie
ical cycles is exquisite: the number of items Sol Estes, Bobby Baker, President Kennedy's
under the heading, "US: Moral Conditions," in assassination, campus revolts, the hippies, and
the Readers' Guide to Periodical Literature. The the city riots—proved once again that we have a
pattern, elaborated and duly pedantified, is "sick society." Our moral crises are regener-
given in Table 3. As she notes, our first moral ated every eight years and seem to coincide with
the end of Presidential administrations.49
hower . . ." "The Old Politics, the New Politics,
49
and the New, New Politics," New York Times It may, or then again may not, be worth not-
Magazine, November 24, 1968, p. 167. ing that Presidential elections in which the incum-
47
See Clinton Rossiter, The American Presi- bent party was removed, 1952, 1960, and 1968, oc-
dency (New York: Harcourt, Brace, 1960), pp. curred during moral crises while the elections in
161-78. which the President was retained, 1948, 1956, and
48
Meg Greenfield, "The Great American Moral- 1964, all took place during times of relative good-
ity Play," The Reporter, June 8, 1961, pp. 13-18. ness.
1970 PRESIDENTIAL POPULARITY FROM TBUMAN TO JOHNSON 33
Of course objective indicators of public mo- The analysis strongly suggests that Presiden-
rality have not been careening in this manner. tial style as well as the ideological and political
Much of the fluctuation in the Greenfield index nature of the administration and the times can
is no doubt due to journalistic fad. A sensational make a sizeable difference in the way popularity
fraud, scandal or disruption causes theologians, ratings behave. A more precise assessment of
journalists, and other intellectuals to sociologize: these relationships would be most desirable.
society is sick. Others pick up the idea and it The study has dealt entirely with general
blossoms into a full moral crisis. In a year or popular approval of the President. Left unexam-
two the theme no longer sells magazines and the ined are the ways population groups differ in
space is filled with other profundities. Fraud, their approach to the President. Supporters of
scandal, and disruption continue, but the moral the President's own party, for example, are
crisis eases. more likely to approve the way he handles his
But—and this is a logical and empirical leap job. Presumably they are also relatively hard to
of some magnitude—to the extent that these alienate, are more likely to be enchanted by his
patterns reflect and influence public attitudes, successes, and are more tolerant of his blunders.
they may be relevant to Presidential popularity. It should also be possible to extend the analy-
The early Eisenhower years are notable for their sis to other bodies of data. Somewhat compara-
absence of moral anguish and they differ from ble data from the Roosevelt administration are
other between-crisis periods in an important re- available. Although the popularity question was
spect: not only were we not demonstrably bad, posed with far less regularity in those days (and
we were positively good for we were undergoing was largely dropped during Warld War II) and
a religious revival. Miss Greenfield looked at the although there are problems with varying ques-
items under the heading, "US: Religious Insti- tion wording, students of President Roosevelt's
tutions." She finds only six items in the popularity ratings emerge with findings which fit
1951-53 period, but 25 in 1953-55 while "in well with those of this study.51 The popularity
the 1955-57 volume, at the height of our virtue ratings of Governors and Senators in states with
. . . the religious listings reached thirty-four active statewide polls can also be analyzed as
with twenty-eight 'see alsos.' " can data on national leaders from such countries
If we were so good ourselves, how could we as Britain, Canada, and France.52
possibly find fault in our leader? spectacular and cumulative international events
VII. FURTHER RESEARCH and shifts in governmental policy—to use the dis-
tinction made by Karl Doutsch and Richard Mer-
This study has been reasonably successful at ritt—differ in impact. "Effects of Events on Na-
generating a regression equation based on only tional and International Images" in Herbert C.
four rather simple variables, which fits quite Kelman (ed.) International Behavior (New York:
well the erratic behavior over 24 years of the Holt, 1965), pp. 132-87.
Presidential popularity index. There is, however, 51
Wesley C. Clark has found some relation be-
much room for improvement and refinement.
tween the Roosevelt popularity and the state of
Little has been done to separate out from the the economy in the 1937-1940 period. He also
coalition of minorities variable the specific and notes a general "downward slant" in the rating
divergent influences of domestic events on Presi- over time and finds a rise of popularity during
dential popularity. There was one variable de- international crises. ("Economic Aspects of a
signed to account in a general way for changes President's Popularity," Ph.D. Dissertation, Uni-
in the economy, some limited analysis was made versity of Pennsylvania, 1943, pp. 41, 28, 35). See
of the relevance of major strikes, and comments also B. Roper, op. cit., and E. Roper, op. cit.,
were interjected about the role of scandal and chapters 2 and 3. And V. 0. Key has observed
"moral crisis." But domestic life is considerably that during 1940 "the popularity of Roosevelt rose
more complicated than this and more precise so- and fell with European crises." Politics, Parties
cial, political, and economic indicators can be and Pressure Groups (New York: Crowell, 1952,
sought. 3rd. ed.), p. 596 (cited in Waltz, op. cit., p. 272.)
It would also be of value to get better esti-
°2 British observers have noted an apparent re-
mates of the impact of different kinds of inter-
lation between unemployment and party prefer-
national events on Presidential popularity—al-
ence in their country: rising unemployment seems
though, as already suggested, such analysis may
to have benefited Labor while declining unemploy-
find that all dramatic international events affect
ment favors the Tories. Henry Durant, "Indirect
popularity in much the same way no matter
Influences on Voting Behavior," 1 Polls 7-11
how they may differ in historical significance.50
(Spring 1965). Extensive data from France on the
x
It may prove valuable to attempt to see how popularity of President De Gaulle have been
VIII. SUMMARY during his second term and actually increased

This investigation has applied multiple during his first term.
regression analysis to the behavior of the re- In considering this Eisenhower phenomenon
sponses to the Gallup Poll's Presidential popu- it is suggested that a combination of several
larity question in the 24 year period from the causes may be relevant: the President's personal
beginning of the Truman administration to the appeal, his ending of the Korean War, his ama-
end of the Johnson administration. Predictor teur status, his domestic conservatism at a time
variables include a measure of the length of when such a policy was acceptable, and his for-
time the incumbent has been in office as well as tune in coming to office at a time of national
variables which attempt to assess the influence goodness.
on his rating of major international events, eco- The second variable, the "rally round the
nomic slump, and war. Depite the austerity of flag" variable, predicts short term boosts in a
this representation of a presumably complex President's popularity whenever there occurs an
process, the fit of the resulting equation was international crisis or a similar event. The vari-
very good: it explained 86 percent of the variable proves to be a sturdy one and suggests a
ance in Presidential popularity. popular decline of about five or six percentage
This degree of fit could only be attained, points for every year since the last "rally
however, by allowing the special character of point."
each Presidential administration to be expressed Economic effects were estimated in the third
in the equation. Thus it does not seem possible variable. The variable could only be made to
to predict a given President's popularity well function if it was assumed that an economy in
simply by taking into account such general phe- slump harms a President's popularity, but an
nomena as the state of the economy or of inter- economy in boom does not help his rating. A de-
national affairs. cline of popularity of about three percentage
The first variable, dubbed the "coalition of points is suggested for every percentage point
minorities" variable, found, as expected, the rise in the unemployment rate over the level
popularity of most Presidents to be in decline holding when the President began his present
during each term. The important differences be- term.
tween administrations do not lie so much in dif- The fourth variable attempted to take into
ferent overall levels of popularity, but rather in account the influence of war on Presidential
the widely differing rates at which this coalition popularity. It was found that the Korean War
of minorities variable takes effect. Specifically, had a large, significant independent negative im-
the popular decline of Presidents Truman and pact on President Truman's popularity of some
Johnson was quite steep while President Ken- 18 percentage points, but that the Vietnam War
nedy seems to have been somewhat better at had no independent impact on President John-
maintaining his popularity. President Eisenhow- son's popularity at all. It is suggested that this
er's popularity did not significantly decline at all difference may be due to the relationship be-
tween the Presidents and the wars: President
Truman was less able than President Johnson to
published: for example, Gallup Opinion Index, keep the war "above" partisan politics and he
March 1968, pp. 27-28. A study by Howard Rosen- seemed to the public to be interfering and re-
thai has investigated regional aspects of the Gen- straining the generals. The absence in the Tru-
eral's popularity. "The Popularity of Charles De man case of a domestic crisis comparable to the
Gaulle: Findings from Archive-based Research," racial turmoil of the Johnson era may also be
31 Public Opinion Quarterly 381-98 (Fall 1967). relevant.

Presidential Popularity From Truman To Johnson1

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Presidential Popularity From Truman To Johnson1

Uploaded by

Copyright:

Available Formats

Downloaded from https://www.cambridge.org/core.

PRESIDENTIAL POPULARITY 1FROM

1970 PRESIDENTIAL POPULARITY FROM TRUMAN TO JOHNSON 19

20 THE AMERICAN POLITICAL SCIENCE REVIEW VOL. 64

1970 PRESIDENTIAL POPULARITY FROM TBUMAN TO JOHNSON 21

22 THE AMERICAN POLITICAL SCIENCE REVIEW VOL. 64

1970 PRESIDENTIAL POPULARITY FBOM TRUMAN TO JOHNSON 23

24 THE AMERICAN POLITICAL SCIENCE REVIEW VOL. 64

1970 PRESIDENTIAL POPULARITY FROM TRUMAN TO JOHNSON 25

26 THE AMERICAN POLITICAL SCIENCE REVIEW VOL. 64

TABLE 1. REGRESSION RESULTS INCLUDING ADMINISTRATION EFFECTS

(1) (2) (3) (4)

Intercept 69.37 54.51 68.15 71.52

Coalitions of minorities (in years) -6.14 -5.12

Economic slump (in % unemployed) -0.09 -3.18 -5.30 -5.86

Dummy variables for administrations

Eisenhower 24.08 0.51

Kennedy 23.87 11.33

Johnson 10.46 4.26

Coalition of minorities variable for

Truman -11.44 -12.54

Eisenhower 0.83 0.13

Johnson -9.12 -8.68

d .13 .30 .46 .44

Standard error of estimate 13.16 8.88 6.86 7.06

fi2 .22 .65 .79 .78

1970 PRESIDENTIAL POPULARITY FROM TRUMAN TO JOHNSON 27

28 THE AMERICAN POLITICAL SCIENCE REVIEW VOL. 64

1970 PRESIDENTIAL POPULARITY FROM TRUMAN TO JOHNSON 29

30 THE AMEKICAN POLITICAL SCIENCE REVIEW VOL. 64

1970 PRESIDENTIAL POPULARITY PROM TKUMAN TO JOHNSON 31

32 THE AMERICAN POLITICAL SCIENCE REVIEW VOL. 64

deed analysts of the Eisenhower administration TABLE 3 . THE GREENFIELD INDEX

1970 PRESIDENTIAL POPULARITY FROM TBUMAN TO JOHNSON 33

34 THE AMERICAN POLITICAL SCIENCE REVIEW VOL. 64

VIII. SUMMARY during his second term and actually increased

You might also like