A Statistical Test of DeMause's Hypothesis
by Ted Goertzel
Published in Political Psychology 14: 711-723, 1993.
ABSTRACT: A content analysis of imagery in editorial cartoons
published from 1989 to 1991 suggests that the primary emotional function
of these cartoons is the ritual humiliation of leaders through shame and
ridicule. Indulgence and fear are also frequent themes of the cartoons,
as are dangerous men, enemies and death. Sexuality, birth and children
appear infrequently. Contrary to DeMause's hypothesis, there is no unusual
change in the imagery during the period leading up to the Gulf War. Saddam
Hussein may have served the American psyche more as a target for externalizing
guilt feelings than as a feared enemy.
Lloyd DeMause (1991) advanced the provocative hypothesis that United States participation in the Gulf War was the result of a temporary mental disorder. This hypothesis was consistent with his general theory of war as a "group-psychotic episode, with patterns of thinking, levels of imagery, and degrees of splitting and projection that are usually only found in the limited psychotic episodes of individuals" (DeMause, 1982: 92).
DeMause documented his argument with illustrations of emotional imagery from editorial cartoons published in newspapers and magazines prior to and during the War. His focus was not on the manifest topic of the cartoons, but on the underlying emotions expressed by the cartoons' symbolism. This is an appropriate focus in analyzing editorial cartoons, since the medium seems better designed to evoke feelings than to communicate rational arguments. Indeed, a study by Carl (1968) found that 70% of newspaper readers were unable to correctly identify the meaning which the cartoonist intended to communicate in an editorial cartoon. Furthermore, trends in the political position taken by the cartoonist do not seem to vary in keeping with trends in public opinion on war issues. An analysis (Gamson and Stuart, 1992) of the manifest content of editorial cartoons during the Cold War era found that dovish themes predominated over hawkish ones, regardless of the clearly hawkish trend of public opinion during much of this time.
The chief weakness of DeMause's methodology is the lack of systematic statistical data. Instead, DeMause and others working within his tradition present selected illustrations which support their conclusions. This leaves them open to the possibility that they may be selectively attentive to cartoons which fit their theory and inattentive to those that do not.
Methods. This research note reports on a preliminary attempt at a quantitative test of the DeMause hypothesis. A sample of 1272 cartoons were selected from microfilm copies of newspapers published in 1989 to 1991. This time period was chosen because it corresponded to DeMause's argument and illustrations. The cartoons were selected from three sources. Each Sunday, the editors of the New York Times and the Philadelphia Inquirer publish a selection of "Cartoons of the Week" taken from newspapers around the country. All cartoons published in these sections in both newspapers during the three years were included in the sample. These reflect the editors' judgment of the cartoons which best captured the mood of the country during the week. In addition to providing a sample of cartoons from papers around the country, the use of the weekly selections enabled the coder to find four or five cartoons as quickly as finding one in a daily newspaper file.
Use of these editors' selections may, of course, have introduced their biases into the sampling. To diversify the sample, the Sunday cartoons from each issue of the Courier-Post, a daily newspaper published in Cherry Hill, New Jersey, were also included in the sample. The Courier-Post was chosen because a microfilm file was available in our library. It is a typical suburban daily newspaper.
The cartoons were coded for the presence of the key symbols used in DeMause's analysis. In DeMause's theory, these symbols indicate the presence of conscious or unconscious emotional content in the cartoon. The coding guide included thirteen variables, defined as follows:
CHILD - Does a child or any symbolism related to children or parenting appear in the cartoon? (Does not include birth imagery).
DEATH - Does death or the fear or threat of death appear in any way in the cartoon?
FEAR - Is fear conveyed in any way by the cartoon?
MOUTH - Does any kind of prominent, exaggerated or stylized mouth appear in the cartoon?
DANGEROUS MAN - Does a dangerous man of any kind appear in the cartoon?
DANGEROUS WOMAN - Does a dangerous woman of any kind appear in the cartoon?
SUICIDE - Does suicide or the possibility of suicide appear in the cartoon?
ENEMY - Does an enemy or any kind of enemy imagery appear in the cartoon?
INDULGENCE - Does indulgence or decadence appear as a theme in the cartoon?
SEX - Does sex or sexuality appear as a theme in the cartoon?
SHAME - Does shame, ridicule or humiliation appear in any way as a theme in the cartoon?
GUILT - Does guilt appear in any way as a theme in the cartoon? (Guilt is feeling that one has violated one's own conscience or internal moral code.)
BIRTH - Does a theme or imagery of birth or rebirth appear in any way in the cartoon?
These categories do not capture the full subtlety and complexity of DeMause's theory. In particular, we were unable to operationalize his theory of the four stages of the political cycle which parallel the stages of fetal life - strong, cracking, collapse and upheaval (DeMause, 1982: 245-246). We attempted to develop categories which could be coded as objectively and unambiguously as possible. We also coded the topic of each cartoon, and the date on which the cartoon appeared.
Even with our simplification, the categories were subtle and difficult to code. Different coders may disagree as to whether guilt or birth imagery, for example, appear in a cartoon. For this study, all cartoons were coded by the same individual in order to eliminate inconsistency between raters. It was not possible to check this coder's reliability by having him recode a portion of the cartoons, however, since he was certain that he would remember how he had coded many of them.
As an alternative method of checking reliability, 80 cartoons were recoded by the principal investigator who had closely reviewed the coding done by the original rater. The following reliabilities coefficients were obtained using Scott's (Scott, 1955; Holsti, 1969: 140) and the more familiar Pearson's r which gave very similar results in all but one case:
Pearson's r Scott's
CHILD .86 .83
DEATH .65 .63
FEAR .54 .55
MOUTH .73 .57
DANGEROUS MAN .60 .61
DANGEROUS WOMAN .49 .49
SUICIDE .50 .58
ENEMY .64 .67
INDULGENCE .37 .42
SEX 1.00 1.00
SHAME .47 .45
GUILT .47 .53
BIRTH .49 .57
Except for Child and Sex, these reliabilities are weak. While there is no objective standard for reliability scores, Krippendorf (1980: 147) suggests as a rule of thumb that full confidence should be given only for reliabilities above .8 while those between .67 and .8 should be used only for "drawing highly tentative and cautious conclusions." Krippendorf (1980: 147) also observed that "if it is an exploratory study without serious consequences [the] level of reliability may be relaxed considerably, but it should not be so low that the findings can no longer be taken seriously." By these standards, the findings reported here (except with regard to the Child and Sex variables) must be viewed as tentative and exploratory. They should, however, be taken at least as seriously as those from DeMause's studies which offer no attempt at quantification. The judgments required in coding cartoons for their emotional content are quite subtle, so the difficulty in reaching agreement in judgment between raters is inherent in the task.
It should be emphasized, however, that these inter-rater reliabilities do not measure the consistency of the work done by the single individual who did the actual coding. Given the highly interpretative nature of this coding, it seems likely that one person is more consistent than two people would be. Coding by a single individual may also be less valid, however, since idiosyncratic interpretations may be introduced. Again, this is also true of the original work done by DeMause.
Findings. The topical content of the cartoons is shown in figure 1. As expected, the topics reflected the issues of the day. The most prominent topics were the collapse of Communism, the Iraqi War, Presidential appointments, abuse of power by government officials, and the doings of famous people. In addition to the manifest topical content, most of the cartoons clearly showed a strong emotional content. Indeed, only four percent of the cartoons had none of the themes or symbols on our coding list. The average cartoon had 2.12 of the thirteen symbols or themes.
Figure 2 shows the percentage of the cartoons which included each of
the thirteen symbols or themes. The most prominent theme was shame, ridicule
or humiliation which was a feature in just over half the cartoons. These
cartoons showed people in embarrassing situations, such as President Bush
all tangled up in fishing wire which he had used to "catch" President Noriega
of Panama, or California drivers stuck in a traffic jam because they never
learned how to drive in the rain, or businessmen throwing darts at a poster
of Japanese businessmen who have outperformed them. Shame was much more
common than guilt, perhaps reflecting greater preoccupation with shame
than guilt in American culture. On the other hand, this may simply reflect
the nature of the editorial cartoon as a medium, since we coded all cartons
which "made fun" of anyone as showing shame or humiliation. "Making fun"
of the powerful appears to be a central emotional function of editorial
Figure 2 had the following data:
|Emotional Theme or Symbolism||Percent of Cartoons|
Indulgence and fear were the second most common themes, and these may indirectly reflect guilty feelings. People may be afraid because they have been doing something wrong for which they may be punished, or they may feel guilty because they indulged themselves. Dangerous men were much more common than dangerous women. Also, almost all of the enemy images were of men, suggesting that men are much more feared or resented than women. In general, images of men were much more common than those of women, reflecting the fact that most positions of power in the world are held by men.
Death and suicide appeared more frequently than sex, birth, or children. We coded suicide fairly liberally, including not only blatantly suicidal images such as Shamir and Begin driving their bus off a cliff or Andy Rooney with a severed head, but also instances of self-destructive behavior such as members of minority groups preparing to slaughter each other or justice department investigators hiding under a vault which is about to explode. The incidence of exaggerated or stylized mouths may not seem large compared to the other categories, but these were coded only when the mouth was clearly exaggerated or stylized as a focus of the cartoon.
In general, our sampling confirms DeMause's observation that emotional
symbolism is rampant in editorial cartoons. It is this content, rather
than the political ideology or position of the cartoonist, which accounts
for much of the importance of editorial cartoons. This imagery also seemed
to be more or less independent of the subject matter of the cartoon, as
DeMause has observed. Table One shows the average number of emotional symbols
found in cartoons in each topical category. Emotional symbolism was frequent
in cartoons dealing with Panama and Noriega and with arms sales, but not
particularly frequent in those dealing with the Iraqi-Arab war or the Gulf
War. There was also a high frequency of emotional symbolism in cartoons
dealing with educational issues.
Central to DeMause's argument is the hypothesis that there are trends over time in the frequency of certain kinds of emotional imagery. DeMause suggests, for example, that the imminence of war could be predicted by the increasing frequency of certain images in the editorial cartoons. With our sample of 1272 cartoons over a three year period, we have an average of 35 to 45 cartoons per month, which is barely adequate for an analysis of trends by month. Breaking things down by week, for a finer measure of change over time, would leave too few cases in each category.
Figure 3 shows the percent of enemy imagery by month over the three year period of our study. Statistically, this figure does seem to show more than random fluctuation (the chisquare is significant at p = .035, but with 4 of 72 cells having fewer than five expected cases. The 72 cells were obtained by cross-tabulating the 36 months with the presence or absence of enemy imagery). There is a high percent of enemy images in the fall of 1990 and in early 1991, but a lull in November of 1990. The highest percent of enemy images is in August of 1992. There are considerable fluctuations from month to month, however, and one might even detect a cyclical pattern, with the frequency of enemy imagery peaking every six to eight months, then subsiding appreciably for a month or two. More data are needed, however, before this pattern can be verified.
Figure 4 shows the trends in Death Imagery. This figure also shows variation over time (the chisquare is significant at p=.007 with 4 of 72 cells having fewer than five expected cases). The sharpest peak is during January and February of 1991, the war months. There is, however, no pattern of increase in death imagery prior to the war which would have permitted one to predict it. Death imagery was most frequent in the cartoons which dealt with certain topics: human rights, arms sales, and gun control. Perhaps surprisingly, it was not particularly strong in cartoons dealing with Gulf War topics.
The theme of Indulgence showed a strong statistical correlation with
time (p=.00000 by chisquare test with 3 of 72 cells having fewer than 5
expected cases). There is a marked variation over time in the presence
of this theme, as shown in figure 5, but it is difficult to associate this
variation with historical events such as the Gulf War. Again, there is
an apparent cyclical variation over a 6 to 8 month time period, but more
cases would be needed to establish this with certainty.
Table below is frequency of Indulgence Imagery.
There is also some fluctuation in the total number of emotional symbols per cartoon over time, as shown in figure 6, but the variations are not strong enough to verify any kind of cyclical trend, F(34,1231)=1.279, (p - .132). (The F-test was used because data in this table are mean scores). Variations over time of the other variables were less pronounced than those presented in the figures.
Summary and Conclusions
There is evidence of a high frequency of emotional symbolism, as defined by DeMause's theory, in editorial cartoons. Even if we question the one study (Carl, 1968) which showed that readers had difficulty grasping the ideological message of editorial cartoons, there is little doubt that emotional symbolism is what gives editorial cartoons much of their importance. Further research might test this hypothesis by showing cartoons to respondents and asking them to report on their emotional and cognitive responses.
The themes of shame, indulgence, and fear are prominent in the cartoons' emotional symbolism. Dangerous men and enemies appear frequently, while feminine imagery, children and birth are less frequent. Suicide, death and self-destructive behavior are surprisingly frequent. There is surprisingly little sexual imagery, in contrast to the very frequent use of sexual symbolism in advertising and entertainment media.
The evidence on time trends in symbolism is ambiguous. There does seem to be some statistical evidence of fluctuations over time in certain symbols, but this should be confirmed with larger samples and other raters. Although our total sample seems large, the sample of cartoons from any given month is small. There is a suggestion of a six to eight month cycle in the amount of emotional symbolism in editorial cartoons, particularly with regard to indulgence, death and enemy imagery. This cyclical pattern was unanticipated, and should be tested over a longer time period as well as with more cases.
Other researchers have noticed cyclical trends in political attitudes. Klingberg (1952) and Holmes (1985) have presented evidence for the alternation of introverted and extroverted periods in American foreign policy attitudes. These swings in the national mood are commonly observed, and President Bush himself confirmed the relevance of this phenomenon when he stated on numerous occasions that one of the benefits of the Gulf War was overcoming the "Vietnam syndrome" and moving the country towards a more activist role in foreign affairs. Of course, these cycles last for decades, not months. However, there is no reason why long term cycles may not coexist with shorter ones in political psychology as they do in economics.
On a deeper level, Lasswell (1932: 538) argued that cycles in political moods reflect the alternating dominance of superego and id in the collective psyche: "prolonged ego and superego indulgence produces redefinitions in directions gratifying to the id; prolonged ego and id indulgence produces redefinitions in directions gratifying to the super-ego." Freud's theory of humor (Freud, 1960, 1961) suggests that cartoons and other humorous media may play a role in this process by helping readers to relate preconscious thoughts to their unconscious motivations. Freud (1960: 166) thought that jokes worked when "a preconscious thought is given over for a moment to unconscious revision and the outcome of this is at once grasped by conscious perception."
In totalitarian countries, political jokes (Cuban American National Foundation, 1989) play a key role in articulating discontent which people are afraid to express in "serious" conversations, let alone in newspapers. In democratic societies, political cartoons and other humorous media, such as monologues on late night television, may play a similar role in expressing hostile feelings toward authority figures which are not yet openly acknowledged by more serious commentators in the same media. There are clearly cyclical trends in the popularity of Presidents as measured by public opinion polls, and it would not be surprising if these same trends were reflected in cartoon imagery.
The Gulf War itself does not seem to have had a marked impact on the frequency of emotional symbolism. Nor could one have predicted the war by looking at these time series data, despite DeMause's claims to have been able to do so. At the same time, it may be that there is a peak in emotional intensity every six to eight months at which time the population would be most receptive to an aggressive initiative by their government. It may be that political leaders are intuitively sensitive to these trends and respond to them in scheduling military actions.
DeMause (1991) attempted a psychiatric diagnosis of the American public on the basis of his analysis of visual material selected from over a hundred magazines and newspapers. He suggested that (DeMause, 1991: 8) "if a patient were to walk into a psychiatric clinic suffering from intrusive images of terrifying figures torturing children, severe depression unrelated to current life events and suicidal wishes, a post-traumatic stress disorder would likely be suspected." To support his diagnosis, DeMause offered a small number of cartoons which vividly portrayed themes such as terrifying parents, hurt children, enemies, the ritual humiliation of the leader, the triumph of good over evil, and the celebration of the rebirth of life. Our sampling does not find all of these themes in great number. There are very few images of terrifying parents and hurt children in our sample. There are a few cartoons showing Saddam Hussein holding children hostage, which fit DeMause's analysis well, but these occurred after the conflict had begun at a time when Hussein was in fact holding children and adults hostage.
One of DeMause's themes, does occur with great frequency in our sample: the ritual humiliation of leaders. This is not surprising, since ridicule and humiliation of powerful leaders have been a persistent feature of political cartoons since the eighteenth century (Morris, 1992). As Press (1981: 11) observed "a political cartoon is worth looking at just because it is enjoyable to stick pins into fools and villains or to watch others do it."
If one were to attempt a speculative diagnosis of the American psyche on the basis of this sample of cartoons, it would differ in emphasis from DeMause's, although not disagreeing with it on all points. To the extent that these cartoons express Americans' inner feelings, Americans appear to be preoccupied with humiliation and shame, perhaps reflecting basic insecurity feelings. These feelings may be rooted in guilt over indulgence and insecurity about not deserving the nation's affluent and wasteful life style. Editorial cartoons meet these emotional needs by ridiculing powerful leaders and authority figures from around the world. These men are offered as scapegoats and targets of externalization for the self-doubts of the newspaper readers.
This analysis suggests that Saddam Hussein may have served the American psyche as a target for externalizing guilt feelings. He was an ungracious upstart who was ungrateful for all the help America had given him, and who overstepped the bounds of his authority and abilities. He had to be put in his place. The cartoons did not typically portray him as threatening, as they did America's enemies in World War II (Kean, 1986). Instead, they portrayed him as a petty and offensive bully. He was scorned and belittled more than he was feared. This may explain the surprising infrequency of death imagery in the Gulf War cartoons.
Even if we should agree on this interpretation of the cartoons, however, we would be left with the question of how well the cartoons reflect the American psyche as a whole. One might obtain a quite different impression from an analysis of other cultural media such as advertising, television programs, or the comics on the funny pages. These media often stress sexuality, romance, family relationships and adventure, and may be in closer touch with the preoccupations of the majority of the population. Readers of newspaper editorial pages are largely middle class, middle aged and well educated. The editorial cartoons may reflect the concerns of this segment of the population more than they reflect those of the country in general.
Or, more simply, the cartoons may simply reflect the biases and preoccupations of the cartoonists themselves. Gamson and Stuart (1992) found that the ideological perspective of the cartoons was predominantly liberal and dovish, even though the newspapers they published in were often more hawkish and conservative. When Rothman and Lichter (1982) administered projective tests to a sample of dovish activists, they found them to be characterized by weakened self-esteem, injured narcissism, and paranoid tendencies. Biographies of eminent doves (Goertzel, 1992) suggest that they often project their hostile feelings onto the authorities and institutions of their own societies so as to avoid expressing them toward their nation's enemies. The imagery in editorial cartoons may be more reflective of the concerns of this group than of the society at large, which may be why they are of so much interest to psychohistorians who share many of the same concerns.
In discussing an early version of this paper with Lloyd DeMause and several of his colleagues at the 1992 meetings of the International Psychohistory Association, DeMause objected to our using the weekly reviews as a source of cartoons. He claimed that certain prominent cartoonists were never listed in these reviews because their work was too expensive, and that these cartoonists were in better touch with the American psyche. One participant objected vehemently that quantitative sampling was a waste of time, since one cartoonist who is in touch with the mood of the masses is more meaningful than several hundred who are not, just as in group psychotherapy one remark may be particularly meaningful after a long period of superficial conversation. If either of these objections is valid, then some other means of assessing the true mood of the masses must be found. Otherwise, there is no way to know whether the cartoonists selected reflect the national psyche or just the preconceptions of the analyst.
Even if the cartoons reflect the concerns of only a segment of the population, however, it is certainly a segment with significant influence on public affairs. DeMause is correct in drawing our attention to the large amount of emotional symbolism in the cartoons, and especially in highlighting their importance in the ritual humiliation of leaders. Further research should compare this symbolism to that in other media, including news magazine covers (Goldman, 1991) and comedy monologues, as well as using interviews, projective tests and other methods such as Q factor analysis (Kinsey and Taylor, 1982) to verify the extent to which the media actually reflect the mass consciousness.
Carl, L. (1968). Editorial cartoons fail to reach many readers. Journalism Quarterly, 45, 533-535.
Cuban American National Foundation. (1989). Chistes: political humor in Cuba," Washington: CANF.
Demause, L. (1992). The gulf war as a mental disorder. Journal of Psychohistory, 19, 1-20.
DeMause, L. (1982). Foundations of psychohistory. New York: Creative Roots.
Freud, S. (1960). Jokes and their relation to the unconscious. New York: Norton.
Freud, S. (1961). Humor. In The standard edition of the complete psychological works of Sigmund Freud (pp. 159-166). London: Hogarth.
Gamson, W., & Stuart, D. (1992). Media discourse as a symbolic contest: the bomb in political cartoons. Sociological Forum, 7, 55-86.
Goertzel, T. (1992). Turncoats and true believers: the dynamics of political belief and disillusionment. Buffalo, N.Y.: Prometheus Press.
Goldman, I. (1991). Narcissism, social character, and communication: a Q-methodological perspective. Psychological Record, 41, 343-360.
Holmes, J.E. (1985). The mood/interest theory of American foreign policy. Lexington: The University Press of Kentucky.
Holsti, O. R. (1969). Content analysis for the social sciences and humanities. Reading, Mass: Addison-Wesley.
Kean, S. (1986). Faces of the Enemy. San Francisco: Harper and Row.
Kinsey, D & Taylor, W. (1982). Some meanings of political cartoons. Operant Subjectivity, 5, 107-114.
Klingberg, F.L. (1952). The historical alternation of moods in American foreign policy. World Politics, 4, 239-73.
Krippendorf, K. (1980). Content analysis: an introduction to its methodology. Beverly Hills: Sage.
Morris, R. (1992). Cartoons and the political system. Canadian journal of communication, 17, 253-258.
Press, C. (1981). The political cartoon. Rutherford, NJ: Fairleigh Dickinson University Press.
Rothman, S., & and Lichter, R. (1982). Roots of radicalism: Jews, Christians and the new left. New York: Oxford University Press.
Scott, W.A. (1955). Reliability of content analysis: the case of nominal
scale coding. Public Opinion Quarterly, 19, 321-325.
Following are some sample cartoons, for illustrative
and instructional purposes. The reader is invited to try coding them
both for manifest content and for latent emotional imagery.