Content-Type: text/html

Late-night Comedy in Election 2000:The Direct Effects of Exposure on Candidate Trait Ratings and the Moderating Role of Political Knowledge



Running Head: Late-night Comedy in Election 2000: The Direct Effects of Exposure on Candidate Trait Ratings and the Moderating Role of Political Knowledge

Late-night Comedy in Election 2000:
The Direct Effects of Exposure on Candidate Trait Ratings
and the Moderating role of Political Knowledge

Dannagal E. Goldthwaite
Annenberg School for Communication
University of Pennsylvania







Dannagal Goldthwaite is a Ph.D. candidate at the University of Pennsylvania. The Electronic Dialogue 2000 Project, from which individual level data was obtained, was supported by grants to Dr. Joseph Cappella and Dr. Vincent Price from the Pew Charitable Trusts and the Annenberg Public Policy Center of the University of Pennsylvania.  Data for use in the late-night joke content analysis was obtained from the Center for Media and Public Affairs (CMPA).  The views expressed here are those of the author alone and do not reflect the views of the funding agencies or of the organizations from which data was obtained.  The author wishes to express thanks to CMPA's Research Director, Dan Amundson, for the use of CMPA's late-night joke database, as well as Joseph Cappella and Vincent Price for their guidance, and Lela Jacobsohn for assistance in the content analysis.

Dannagal E. Goldthwaite, Annenberg School for Communication, University of Pennsylvania, 3620 Walnut Street, Philadelphia, PA 19104. E-mail: <a href="/cgi-bin/wa?LOGON=A3%3Dind0209A%26L%3DAEJMC%26E%3D8bit%26P%3D4928796%26B%3D--%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D%253D_2906021%253D%253D_%26T%3Dtext%252Fplain%3B%2520charset%3Diso-8859-1" >[log in to unmask]</a>
 Late-night Comedy in Election 2000: The Direct Effects of Exposure on Candidate Trait Ratings and the Moderating role of Political Knowledge


Throughout the 2000 Presidential election, there was much discussion among journalists, politicians and comedians alike, about the role of late-night comedy in the formation of public opinion.  In a New York Times article entitled "The Stiff Guy vs. the Dumb Guy," Marshall Sella suggested that  "_part of what turns random episodes like the RATS controversy[1] into icons, what inflates them into pivotal campaign events, is late-night comedy" (Sella, 2000, p. 72).  Mandy Grunwald, Clinton media advisor in his 1992 campaign, and media advisor to Hillary Clinton during her 2000 Senate campaign, stated, "If Leno or Imus or Dennis Miller are making jokes about you, you have a serious problem.  Whatever take they have on you is likely to stick much more solidly than what is in the political ads in papers like the Washington Post" (Grenier, 1999, p. 103).
Meanwhile, political comedians tended to downplay the effects of jokes when discussing the issue publicly.  While speaking to Al Gore's class at the Columbia Journalism School, David Letterman, host of CBS's The Late Show with David Letterman, stated that "_he doubted that he had any impact on the presidential election" (Berner, 2001):
I would guess that very few votes were cast based on a joke that either I or Jay Leno made. (David Letterman addressing Al Gore's class, as reported in Berner, 2001)

Jon Stewart, host of Comedy Central's satirical program, The Daily Show, argued that it was unlikely that people were influenced by satirical programs since "[writers and comedians] need [viewers] to know something before they even make a joke about it"  (Nightline, September 18, 2000).  Similarly, Jay Leno, host of NBC's The Tonight Show stated, "we (writers and comedians) reinforce what people already believe" (Shaap, 2000, p. 75).
        Empirical research to date regarding the effects of political satire on public opinion is scarce.  Examinations of the effects of exposure to editorial political cartoons on political opinions has been inconclusive (e.g.: Brinkman, 1968; Carl, 1968).  In an experiment assessing the direct effects of textual satire on character ratings of Nixon, Gruner (1971) found minimal evidence of a persuasion effect, but concluded that higher doses over a longer period of time may produce a statistically significant effect. Concluding a content analysis and examination of third person effects of late-night in the 2000 campaign, Verena Hess argues that late-night jokes act as a "rhetorical device" (Hess, 2001, p. 18) and an illustration of how the media "construct our political world"  (p. 18).
This analysis is an attempt to assess the over-time effects of late-night comedy exposure on people's perceptions of those candidate attributes caricatured in late-night content.  These analyses were completed using survey data gathered throughout the course of the campaign, thus avoiding the problem of short-term exposure faced in experimental analyses such as Gruner's (1971).  The first question to be addressed here concerns the extent to which high exposure to late-night comedy programming led to more negative ratings of the candidates on the particular attributes being caricatured in the content of late-night throughout the 2000 campaign.  The second question concerns whether or not the effects of late-night exposure on candidate attribute ratings over time varied with the political knowledge of the audience.
Late-night as a non-traditional source of political information.
According to a report published by Pew's Research Center for the People and the Press (2000), while newspapers, network and local news have all declined in prominence as people's primary source of campaign information over the past four years, the number of people who report learning something about the campaign from late-night comedy programs, such as those hosted by Jay Leno and David Letterman, is up slightly from four years ago (Pew Research Center, 2000).  America's youngest eligible voters (Age 18-29), who happen to report the lowest use of newspapers and network news, also report receiving more campaign information from late-night programs than any other age group with almost half reporting learning something about the campaign sometimes or regularly from late-night programming, a rate about twice that of Americans over age thirty (Pew Research Center, 2000).
The Pew Report also suggests differences in sources of campaign information by political knowledge group.  When broken down into three levels of political knowledge, people low in political knowledge report regularly learning about the campaign from daily newspapers, news magazines, and the Internet at about half the rate of individuals high in political knowledge.  Meanwhile, almost half (45%) of those low in political knowledge report learning about the campaign regularly from late night, compared to only 20% of individuals with high political knowledge.  (Pew Research Center, 2000).  Pew's report suggests that not only do people use late-night comedy programs as a source of political information, but among certain populations (namely those low in political knowledge and low in exposure to traditional political information) late-night content may play a particularly important role.
Getting the joke.
Late-night comedy is not simply an alternative source of political information.  It is political information delivered in the form of jokes based on caricatures, the understanding of which involves certain cognitive processes which have important implications.  Humor theorist, Arthur Koestler, discusses humorous caricatures in his 1964 work, The Act of Creation, in which he argues that humor occurs at the intersection of two incompatible scripts which must be reconciled to "see the joke" (Koestler, 1964, p. 85).  To understand the joke, the listener must "bridge _ [the] logical gap by inserting the missing links" (p. 84), a process he calls interpolation.  In doing so, audience members are active participants in the creation of a humorous construct, repeatedly accessing caricatured information about the candidates to get the joke.  By basing jokes upon what is familiar or easily accessible in the mind of the listener, comedians and joke writers facilitate the resolution of the inc
ongruity in the joke. As Jon Stewart, host of Comedy Central's The Daily Show, points out, "_if [The Daily Show] made jokes about Gore being clumsy, nobody would laugh because it's not in the consciousness" (Nightline 9/18/00). It is for this reason that archetypal representations such as stereotypes and caricatures make efficient material for jokes. Koestler writes:
A caricature is comic only if we know something of the victim, if we have a mental image, however vague of the person, or type of person, at which it is aimed- even if it is an Eskimo, a cave-man, or a Martian robot.  The unknown cannot be distorted or misrepresented. (Koestler, p. 71)

Because candidate caricatures in late-night joke content most often involve negative attributes of a candidate (as we will see in our content analysis), and since the same attributes are caricatured repeatedly throughout the course of the campaign, high exposure to late-night content is expected to cause individuals to have more negative perceptions of the candidates on those attributes that are caricatured most often.

Hypothesis 1:  Individuals with high exposure to late-night comedy programming will rate the candidates lower over time on the attributes being caricatured than will individuals with lower exposure to late-night programs (controlling for possible confounders).

The role of political knowledge.
As discussed earlier, research suggests that individuals with low political knowledge report obtaining less information from traditional political sources, and more information through late-night comedy than individuals with moderate or high political knowledge (Pew Research Center, 2000).  This is an important finding, not only because it helps us identify which subgroups might receive the most political information from late-night, but also because individuals low in political information might be more susceptible to the effects of certain kinds of messages than those high in information.  In their discussion of the consequences of political knowledge, for example, Delli Carpini and Keeter (1996) address the notion that political knowledge can lead to susceptibility to persuasion in some cases and resistance to persuasion in others.
On the one hand, better-informed citizens should be more likely to change their attitudes in response to certain kinds of critical information.  On the other hand, they should be less susceptible to efforts to manipulate them, less vulnerable to propaganda, less affected by particular events or messages that are irrelevant, and less likely to manifest "response effects" in surveys (Delli Carpini & Keeter, 1996, p. 234).

In other words, better-informed citizens have a greater capacity to evaluate political messages, either to put the information in their cap if it is important or to throw it out if it is not.  In John Zaller's  (1992) reception-acceptance model, he outlines three types of resistance that "politically aware" individuals are likely to exercise more than the politically unaware when encountering political messages: partisan resistance as a function of their political values, inertial resistance as a function of their larger stores of preexisting considerations, and countervalent resistance as a function of their exposure to communications which run counter to the dominant message in the political environment (Zaller, 1992, p. 167).   Each of these forms of resistance is relevant in our discussion of the effects of late-night political jokes.  For instance, individuals with less political information may be less adept at "recognizing messages that are inconsistent with their values,"
less able to use their "informational ballast to counteract the effects of new messages," and less likely to have exposure to "countervalent communications" (in this case non-caricature-based candidate information) to dilute the effects of the dominant messages (Zaller, 1992, p. 167). Based on these arguments, the effects of late-night exposure on attribute ratings of the candidates are expected to be highest among individuals with low political knowledge.

Hypothesis 2:  The effects of late-night exposure on candidate attribute ratings will be higher among those low in political knowledge than among those high in political knowledge
Methods
Design
This analysis was part of a larger study on the role of electronic discussion among citizens during the 2000 presidential election, under the supervision of Vincent Price and Joseph Cappella.   The larger project was designed specifically to assess the effects of group deliberation on people's engagement with and opinions on political issues. In February 2000, a random sample of American citizens age 18 and older (N=3967) was drawn from a nationally representative panel of survey respondents maintained by Knowledge Networks, Inc. of Menlo Park, California.  The Knowledge Networks panel included a large number of households that had agreed to accept free WebTV equipment and service in exchange for completing periodic surveys on line.  Fifty-one percent of those recruited through the February recruitment survey agreed to participate in the project.  Participants were invited either to 1. discuss politics on-line once per month and complete monthly surveys (discussants), 2. complete
the monthly surveys but not discuss politics on-line (control group), or 3. complete the baseline and final surveys only (set aside).  Analysis of group characteristics (demographics, age, race, gender, political interest, ideology, and party leanings) confirmed that the randomization was successful.  The discussion group consisted of 906 individuals, the control consisted of 139, and the set-aside (first and last surveys only) consisted of 969.  The survey which included the late-night comedy exposure items was issued in September to discussants and control group members only.  Late-night exposure measures were obtained for 64% of discussants (N = 580) and 63% of the control group (N = 88) due to panel attrition over the months of the project.

Measures
Late-night comedy exposure  was obtained through self report on the September survey (N = 668).  Two separate items were issued to participants in randomized order, asking if they happened to watch Jay Leno and if they happened to watch Dave Letterman with three possible responses, "Yes, much of the time," "Yes, but not very often," and "no."  Based on the content analysis of Leno and Letterman jokes, the main candidate caricatures mentioned on the two programs were quite similar with the following exceptions: Leno made more frequent mention of Bush's "drug habit" (10% of Leno jokes) than did Letterman (1% of his jokes); Letterman made a few more references to Gore reinventing himself than did Leno (5.4% v. 1.8%) and Leno made a few more references to Gore being an exaggerator than did Letterman (7.2% v. 3.2%).  As the content of the two programs is basically the same except for these few exceptions, the two measures of exposure (to Leno and to Letterman) were collapsed into one s
ummarized late-night exposure measure to increase the predictive power of the variable.  The summarized late-night exposure measure which ranges from 0 to 2, has a mean of .3265, a standard deviation  of .4151, and kurtosis of 2.249.  The exact question wording, and the calculation of the numerical values attributed to these responses can be found in Appendix A.  A means comparison of the exposure measure by discussant v. control group assignment shows no significant differences in exposure between the two groups[2].

Political Knowledge was obtained on the baseline survey and was calculated as the average of 24 items in three knowledge batteries (For a complete list of items, see Appendix B). Correct answers were coded 1 and incorrect and "don't know" answers were coded 0.  The first knowledge battery consisted of 7 questions regarding the candidates' biographies (4 about democrats and three about republicans).  The second battery consisted of 7 questions about the candidates' issue positions (4 about democrats and three about republicans ). The third battery consisted of 10 questions of general civics knowledge.  The three batteries were combined into a full knowledge scale, with values ranging from 0-1. Alpha for the full knowledge scale is .82.

A Content Analysis of Late-night Jokes was completed using data from the Center for Media and Public Affairs, a non-profit organization that tracks aspects of media coverage of politics and social affairs. The content analysis was used to determine what constituted the most frequently caricatured traits of the candidates in late-night joke content, and hence, which trait ratings we would expect to vary as a function of exposure to late-night comedy. The content analysis was completed on a database containing texts of 879 late-night joke made by Dave Letterman and Jay Leno from January 1, through November 30, 2000.  Although some episodes were re-run, thus duplicating certain jokes, each individual joke was only coded once. The coding scheme consisted of 12 possible caricature traits specifically for Bush and 12 specifically for Gore, as well as one category for jokes focusing on general disillusionment with the campaign, without mention of any caricature traits.  For details of th
e coding scheme see Appendix C. The reliability of the coding scheme was tested between the author and a second coder, obtaining a Kappa of .81 for the Bush caricatures and .93 for the Gore caricatures.   Throughout the period from January 1 to November 30, 2000, 494 jokes mentioned G.W. Bush, 359 jokes mentioned Al Gore, and 26 focused on general disillusionment with the campaign. Based on Table 1, the most salient caricature traits in late-night are Bush as unintelligent, Gore as stiff and dull, and Gore as an exaggerator and a liar.
[Note to reviewers: Examples of late-night jokes were excluded to cut down on length, but will be included in the presentation to illustrate the various caricature categories.]

Evaluations of candidate traits were obtained from a series of closed-ended questions issued three times: March, July and September.  Respondents were asked how well certain traits described each of the candidates.  The traits (inspiring, knowledgeable, and honest) were issued in randomized order.  Response options were:  Extremely well, quite well, not too well, and not well at all, scaled from 0 to 1 where 1 is a more favorable rating.  For our analyses, we will be associating Gore's inspiring rating with Gore's late-night caricature as stiff and boring, Bush's knowledgeable rating with Bush's caricature as unintelligent, and Gore's honest rating with Gore's caricature as an exaggerator and a liar.

Party-ideology index, TV news exposure, Race, Gender, Age, Income were included as controls in all the models.  For a description of these control variables, see Appendix D.


Table 1

Results of joke content analysis in order of frequency
Bush
Total
Percent
Gore
Total
Percent
Intelligence
142
28.74
Stiff/boring/robotic
71
19.78
Death penalty
82
16.60
Exaggerator
49
13.65
Drugs
60
12.15
Other
45
12.53
Alcohol
48
9.72
DNC kiss
40
11.10
Other
47
9.50
Failing campaign
29
8.08
Swearing
35
7.09
Reinvents himself
25
6.96
Foreign affairs
27
5.47
Illegal fundraising
24
6.69
Malapropisms
22
4.45
Makeup
23
6.40
Failing campaign
12
2.43
Florida
21
5.85
Silver spoon
10
2.02
Aggressive/rude
13
3.62
Smirk
4
.81
Lieberman
8
2.22
Friend of the rich
3
.61
Environment
7
1.95
Florida
2
.40
Silver spoon
4
1.11
TOTAL
494
100.00
TOTAL
359
100.00
Note. For descriptions of caricature categories see Appendix C

 Analyses

To look for direct effects on candidate trait ratings as a function of late-night exposure, the relationship between amount of late-night viewing and net change over time in candidate attribute rating was tested.  Changes in participants' attribute ratings of the candidates from early March to early July, and from early July to early September were analyzed.  Net change in an attribute rating from March to July was obtained by subtracting the March attribute rating from the July attribute rating[3]; similarly for change from July to September (September attribute rating minus July attribute rating).
If a direct persuasion effect were taking place, exposure to late-night would be a significant predictor of net change in the caricature attribute ratings.  For instance, if exposure to late-night from March to July actually persuaded people that Al Gore was uninspiring, a negative relationship between late-night exposure and net change on Gore's inspiring rating from March to July should be observed.  As the independent variables of theoretical interest to us here are late-night exposure and political knowledge, the following discussion will focus on the effects of those variables.  In a preliminary set of regression analyses, younger people and people with high television news exposure were found to be higher viewers of late-night, so age and television news exposure were included as controls in the models.  Also included as controls because of their possible relationship to candidate attribute ratings were political interest, an indexed measure of party affiliation and ideology
, income, race, and gender.  For a description of these control variables, see Appendix D
Equation 1.
Net change in candidate caricature attribute rating = b0 + b1 *exposure +  (b2 * controls).

Twelve regressions were run, using late-night exposure and controls[4] to predict net change in "knowledgeable," "inspiring," and "honest" for Bush and for Gore, from early March to early July, and from early July to early September. In none of the twelve regressions was late-night exposure found to be a statistically significant (p<.05) or marginally significant (p<.1) predictor of net change in any candidate attribute ratings.
To test the hypothesis that late-night exposure interacts with political knowledge in the effects process, we tested a model with the interaction of late-night exposure and political knowledge to predict net change in candidate attribute ratings over time. If the extent to which late-night exposure influences one's candidate attribute ratings varies as a function of one's political knowledge, then in equation 2, b3 would be significantly different from zero.
Equation 2. Interaction of Late-night exposure and Political Knowledge

Net change in candidate caricature attribute rating = b0 + b1 * late-night exposure +  b2* political knowledge + (b3* late-night exposure * political knowledge)+ (b4 * controls)

To test equation 2, twelve regressions were run[5]. The regression coefficients of these variables in predicting net change in candidate attribute ratings over time are shown in Table 2[6].  To explore the interaction, the sample was first split into three political knowledge groups; low, medium, and high.  The mean knowledge score for each group was substituted into the model to represent the three different knowledge groups (.42, .65, .82 respectively) and five different levels of late-night comedy exposure (0, .3, .6, 1, 1.3) were substituted. Figures 1 through 5 illustrate the predicted values of net change in attribute ratings based upon political knowledge and late-night exposure. The lines within each graph represent the three different knowledge groups, with the low knowledge group represented by the dashed line.
In the case of net change in Bush knowledgeability and in Gore knowledgeability from March to July, the interaction term of late-night and political knowledge approached statistical significance where the less politically aware show more critical ratings of Bush's knowledgeability at high levels of late-night exposure, and less critical ratings of Gore's knowledgeability at high levels of late-night exposure.  Meanwhile, net change in Bush inspiring from March to July, and net change in Gore inspiring and Gore honest from July to September, show statistically significant trends in which the low knowledge individuals showed a decrease in the candidate attribute rating over time as a function of late-night exposure (See Figures 2-5).
Figure 2 suggests that as late-night exposure increased, those people with low political knowledge reported more critical ratings of Bush's knowledgeability over time and those with high political knowledge showed more positive perception of Bush's knowledgeability over time.  Although this relationship is not quite statistically significant, in a follow-up analysis where OLS regressions were run separately for low and high knowledge portions of our sample, the coefficient of late-night exposure in predicting net change in Bush knowledgeability from March to July is positive for high knowledge individuals, and negative for low knowledge individuals, suggesting that the role played by late-night exposure in influencing impressions of Bush's knowledgeability may well have been different for individuals with different levels of political knowledge.


Table 2
Predictors of Net candidate attribute rating change from early March to early July, and from early July to early September

Net Change from early March to early July
Al Gore
G.W. Bush
Insp
Know
Hon
Insp
Know
Hon
Predictor
B
(SE B)
B
(SE B)
B
(SE B)
B
(SE B)
B
(SE B)
B
(SE B)
Late-night Exposure
0.005
(.03)
0.03
(.031)
0.03
(.03)
-0.05
(.03)
-0.01
(.03)
-0.03
(.029)
Political Knowledge
0.09
(.08)
0.02
(.08)
0.13#
(.07)
0.16*
(.08)
0.008
(.07)
0.12
(.07)
Political Knowledge X
Late-night
-0.06
(.17)
-0.29#
(.16)
-0.02
(.16)
0.34*
(.167)
0.28#
(.146)
0.24
(.16)
R2
N
.01
494
.019
495
.015
486
.021
499
.028
499
.02
496

Net Change from early July to early September
Al Gore
G.W. Bush
Insp
Know
Hon
Insp
Know
Hon
Predictor
B
(SE B)
B
(SE B)
B
(SE B)
B
(SE B)
B
(SE B)
B
(SE B)
Late-night Exposure
.02
(.032)
-5.9E-5
(.032)
-.06
(.031)
-.03
(.032)
-8.1E-5
(.03)
-.01
(.030)
Political Knowledge
.06
(.076)
-.09
(.078)
.06
(.076)
.13#
(.078)
.14
(.074)
.04
(.074)
Political Knowledge X
Late-night
.33*
(.167)
.175
(.17)
.36*
(.168)
.018
(.167)
.057
(.158)
-.05
(.159)
R2
N
.04
455
.015
456
.022
456
.022
454
.029
453
.014
453
Note. Numbers in parentheses are standard errors of the regression unstandardized coefficients. Control variables in the models include Party-ideology index, Average Interest, TV news exposure, Race, Gender, Age, Income and, in the July-Sept models only, overall favorability of that candidate.
#p<.1  *p<.05  **p<.01 ***p<.001




Figure 1.  Net change in Gore knowledgeability rating from March to July (July minus March) as a function of late-night exposure, by knowledge.
Note.  This relationship was not highly statistically significant (p<.1).








Figure 2.  Net change in Bush knowledgeability rating from March to July (July minus March) as a function of late-night exposure, by knowledge group.
Note.  This relationship was not highly statistically significant (p<.1).
                                .


Figure 3.  Net change in Bush inspiring rating from March to July (July minus March) as a function of late-night exposure, by knowledge
Note.  This relationship was statistically significant (p<.05).




Figure 4. Net change in Gore inspiring rating from July to September (September minus July) as a function of late-night exposure, by knowledge group
Note.  This relationship was statistically significant (p<.05).




Figure 5. Net change in Gore honesty rating from July to September (Sept minus July) as
a function of late-night exposure, by knowledge group.
Note.  This relationship was statistically significant (p<.05).


Gore's inspiring rating was found to vary (from July to September) as a function of late-night exposure and political knowledge (Figure 4). Increased exposure to late-night was related to lower ratings on Gore's inspiringness among those people with lower knowledge and higher ratings on Gore's inspiringness among those people with higher knowledge.  The fact that people with high political knowledge might have given Gore higher inspiring ratings and Bush higher knowledgeability ratings with increased exposure to late-night is a curious finding. One possible explanation is that individuals with higher political knowledge are more critical viewers of late-night, so rather than being influenced by the political jokes, they see it as political information from a source that lacks credibility, and became more critical of late-night's message as a result.  This would be consistent with Zaller's (1992) reception-acceptance model, though there was no way to empirically examine this mechanism with variables available here.
The third caricature attribute of interest was that of Gore's honesty.  Here, again, it was found that people with lower political knowledge became more critical of Gore's honesty at higher levels of late-night exposure (Figure 5).  The obvious question following each of these findings concerns how the content of the late-night jokes made during that time might explain these effects.  This question was pursued in the context of the changes in Gore's honesty rating. To explore the relationship between joke content and net change trait ratings, three categories of Gore jokes concerning his honesty were collapsed into one "honesty" category:  Exaggerator/liar, reinvents himself/strategic, and Illegal campaign fundraising.
The frequency of Gore "Stiff/dull" jokes, the most frequent caricature overall, was compared to the frequency of Gore "honesty" jokes by Month from March to September (Figure 6).  Although Gore's caricature as stiff and boring dominated the jokes being made from March through May, by August and September Gore's caricature as dishonest was the more frequent of the two (Figure 6). This might explain the effect found in changes in Gore's honesty ratings from July to September as a function of late-night exposure and political knowledge.  It is curious that while there were direct effects on Gore's inspiring rating from July to September, no similar effects were observed from March to July. It may be the case that from March through July although late-night did not affect Gore's inspiring rating itself, the repeat activation of the Gore caricature as "stiff and dull" throughout that time period contributed to the priming of that caricature trait, an effect which was explored in a sepa
rate series of analyses.


 Figure 6. Frequencies of Gore "dishonest" and "stiff/dull" jokes per month
Note. Numbers in parentheses represent the total number of candidate jokes made that month.
Two other curious results were obtained from March to July.  The first concerns the differential changes in Gore's knowledgeability rating as a function of late-night exposure and political knowledge (Figure 1), where people with low knowledge and high exposure gave Gore higher knowledgeable ratings with increased exposure to late-night.  Perhaps as people with low political knowledge and high exposure found Bush less intelligent over time (Figure 2), Gore seemed more intelligent, suggesting a contrast effect occurring between Gore and Bush.   However, because these effects on both Gore and Bush's knowledgeable ratings were only approaching significance, it would be a mistake to over interpret the effect.
The second curious finding is in the case of Bush's inspiring rating (Figure 3). In this case, it is important to consider how late-night's effects on candidate attribute ratings might translate to other attributes of the candidates.  After all, neither Leno nor Letterman made enough jokes specifically about Bush being uninspiring enough to justify a direct effect on Bush's inspiring rating, though one could argue that lack of intelligence might translate into uninspiringness. Since honesty, intelligence and inspiringness are all fundamentally leadership qualities, they may function interchangeably when an individual is called upon to judge the candidate on any one individual trait.  This question merits the attention of future research.
It is also plausible that if late-night exposure affects certain people's perceptions of candidate attributes, so too might it directly affect their overall perception of that candidate. To explore this secondary hypothesis, a follow-up analysis was completed in which the effects of late-night exposure*political knowledge on net change in overall candidate favorability ratings from early March to early August was tested.  In these analyses, neither the main effects of late-night, nor the interaction term with political knowledge was a significant predictor of change in overall favorability of Bush or of Gore.
Unfortunately, the limitations of this study are numerous, and include the small sample size, the very low R-square in the models, the minimal number of points in time at which data was available, and the attrition of people high in political knowledge over the course of the project. The latter of these "limitations" may actually lend support to our findings, as the effects of late-night that were witnessed here all operate as a function of political knowledge. Since the knowledge distribution in our sample tends to under-represent low-knowledge individuals, the role of political knowledge in susceptibility to the effects of late-night may be even greater in reality than what we have witnessed here. Another concern to be raised is what other aspects of the late-night content, other than the jokes, may have played a role in these effects. For instance, both candidates appeared on Leno and on Letterman in the months before the election, and although neither of their appearances coin
cided with the data gathering for these analyses, there was discussion by the late-night hosts about the candidates' upcoming visits.

Conclusion
This study is a first step in the empirical analysis of the effects of political satire on public opinion. The magnitude of these findings is not huge, as is illustrated by the R-Square of the models and the small predicted variation in the dependent variables. However, what has been shown is that late-night exposure may have different effects on people high in political knowledge than on people low in political knowledge. Since four of the five relationships indicate that individuals with low knowledge gave lower attribute ratings at high levels of exposure, these findings may be particularly important in terms of how campaigns choose to deal with late-night joke content, and in how scholars choose to approach late-night jokes as an object of research. An important question left to be explored concerns whether or not late-night exposure was related to changes in behavior regarding the candidates (i.e.: voting). It would also be of value to explore effects of late-night exposure o
n other dimensions of political engagement such as cynicism, interest in the campaign, or trust in institutions.
        In light of these findings, it is again interesting to note that late-night hosts publicly state that their jokes do not influence public opinion. If people with low knowledge watch the most late-night, report obtaining information about the campaign from late-night, and receive less traditional political information from the news than high knowledge people (Pew Research Center, 2000), and if these low-knowledge individuals are also more likely to become more critical of the caricatured traits of public officials portrayed on late-night, then perhaps late-night writers and hosts ought to consider the possibility that their jokes play an important role in the political climate, particularly among the less-informed.
These findings lend support to the notion that certain entertainment media, such as late-night, may play a role in informing and even shaping the political views of certain kinds of viewers who may not be receiving more traditional information elsewhere. As candidates and political officials show up on the couches of people like Dave Letterman, Jay Leno, or Jon Stewart in the months and years to come, and as politics continue to be a key ingredient for late-night jokes, the questions of how people use these programs, what they learn, and what associations they make become increasingly important. Just because it is called "comedy" does not mean that it should not be taken seriously.
 Appendix A
Late-night exposure measure:  "Do you happen to watch any of the following entertainment programs on television?" followed by a list of several television programs including Jay Leno and Dave Letterman.  48% reported watching neither program, 18% reported watching one of the two, but not very often, 20% reported watching both, but not very often, 6% reported watching one of the programs much of the time; 4.6% reported watching one program much of the time and the other not very often and 1.2% reported watching both programs much of the time.
        To determine quantitative equivalents of the three possible responses, an informal survey was sent out to approximately 100 graduate students (of whom 57 responded), asking them how many days of exposure per week (0-5 days) they would infer from each of the three possible responses (no; yes, but not very often; yes, much of the time).  The mean of these responses was as follows:  Yes, much of the time = 3.76 days; Yes, but not very often = 1.06 days; and No = 0 days.  Each of the two late-night exposure responses (to Jay Leno and to Dave Letterman) was recoded on a scale of zero to one based on the results of this survey, resulting in the following numerical equivalents: Yes, much of the time = 1; Yes, but not very often = .28; No = 0.  The total late-night exposure measure was calculated as the sum of these two measures.
This summarized late-night exposure measure has a mean of .3265, a standard deviation  of .4151, and kurtosis of 2.249.




Appendix B
Political knowledge items: Each question was followed by randomly ordered responses (Al Gore and Bill Bradley in the case of the Democrats, and George W. Bush and John McCain in the case of the Republicans).

I.  Knowledge of Candidate Biographies obtained on first baseline survey.
A.      Thinking about the Democrats , to the best of your knowledge who a)     Was a professional basketball player (Bradley); b) Is the son of a former United States Senator (Gore); c) Voted for tax cuts proposed by President Reagan in 1981 (Bradley); d) Served in the United States Senate (Both)
B.      Thinking about the Republicans (John McCain and George W. Bush), to the best of your knowledge, who a) Is a state governor (Bush); b) Is a United States Senator (McCain); c) Was a prisoner of war in Vietnam (McCain).

II. Knowledge of Candidate Issue Positions obtained on first baseline survey
A.      Thinking about Democrats, to the best of your knowledge, who a) Supports a universal health care program (Bradley); b) Favors increased government funding of political campaigns (Bradley); c) Favors giving patients the right to sue their HMO (Both); d) Favors tax-free savings accounts to help parents pay for college (Gore).
B.      Thinking about Republicans, to the best of your knowledge, who a) Supports giving tax credits or vouchers to people who send their children to private schools (both); b) Has pledged to cut federal income taxes by over $1 trillion in ten years (Bush); c) Supports a ban on soft money campaign contributions (McCain).

III.  General Political Knowledge obtained from second baseline survey
A.      Which one of the parties is more conservative than the other at the national level?
[Randomly ordered response options:] Democrats, Republicans, Don't know
B.      Which one of the parties has the most members in the House of Representatives in Washington? [Randomly ordered response options:] Democrats, Republicans, Don't know.
C.      Which one of the parties has the most members in the U.S. Senate? [Randomly ordered response options:] Democrats, Republicans, Don't know
D.      Who has the final responsibility to decide if a law is Constitutional or not?  [Randomly ordered response options:] President, Congress, Supreme Court, Don't know
E.      Which one of the following is the main duty of Congress? [Randomly ordered response options:]  Write legislation; Administer the President's policies; Watch over the governments of each state; Don't know
F.      Whose responsibility is it to nominate judges to the Federal Courts?  [Randomly ordered response options:] President, Congress, Supreme Court, Don't know
G.      How much of a majority is needed for the U.S. Senate and House to override a presidential veto?  [Randomly ordered response options:] Bare majority (one more than half the votes), Two-thirds majority, Three-fourths majority, Don't know
H.      Do you happen to know what job or political office is currently held by Al Gore? [Randomly ordered response options:] U.S. Senator, U.S. Vice President, Governor of Tennessee, Don't know
I.      What job or political office is currently held by Trent Lott? [Randomly ordered response options:] U.S. Senator, U.S. Ambassador to the United Nations, Chief Justice of the U.S. Supreme Court, Don't know
J.      What job or political office is currently held by William Rehnquist? [Randomly ordered response options:] U.S. Senator, U.S. Ambassador to the United Nations, Chief Justice of the U.S. Supreme Court, Don't know

Appendix C
Content analysis for late-night jokes:
Each joke is coded for the caricature trait(s) central to the joke and can be given one code per candidate. A joke containing two caricature traits about the same candidate is given one code for the caricature trait mentioned first.  If no category is applicable to either candidate, both candidates are coded as 0.  If the joke is only applicable to one candidate the other receives a 0.
Bush Codes:
Malapropisms: references to Bush's mispronunciations or misuse of words.
Foreign affairs: references to Bush's lack of knowledge of international affairs, mainly geography and names of leaders.
Intelligence: references to Bush's intellectual capacity which are not about mispronunciations or foreign affairs.
Alcohol: references to Bush having had a drinking problem.
Drugs: references to questions of Bush's drug history or lack of memory of college years.
Smirk: references to his smirk.
Friend of the rich: references to Bush's upper-class friendly policies helping only the wealthy.
Silver spoon: references to Bush's wealthy upbringing or use of his father's name to get power
Failing campaign: references to Bush's likely defeat or failing campaign.
Death penalty: references to Bush in terms of Texas exacutions or position on the death penalty.
Swearing: references to Bush's incident calling Adam Clymer of the New York Times a "major league asshole" when he thought the microphone was off.
FLORIDA: references to Bush acting as president before the race is called on Dec 12.

Gore Codes:
Stiff/boring/robotic: references to Gore's boring, uninspiring personality or robotic and stiff appearance.
Reinvents himself: references to Gore's changing personality and the perception that he is willing to do or say anything to get elected.
Exaggerator/liar: references to Gore's exaggerations or lies (e.g.: inventing the internet).
Makeup: references to Gore's use of makeup during the televised debates.
DNC kiss:  references to Tipper and Gore's public kiss at the Democratic National Convention.
Environment: any mention of Gore's reputation as an environmental fanatic.
Aggresssive/rude: references to aggressive behavior during the debates, or elsewhere.
Silver spoon: references to Gore's wealthy upbringing or use of his father's name to get power.
Failing campaign:  references to Gore's likely defeat or failing campaign.
Lieberman: references to Lieberman's religion or Gore's choice of VP as strategic.
Illegal funding: references to Buddhist monk fundraising incident and other illegal funding.
Florida: references to Gore's refusal to concede of the election in the weeks after November 7.

Appendix D
Control Variables included in the models:
Party-ideology index:  scale from -5 to +5 where -5 are strong conservative republicans and +5 are strong liberal democrats.
TV news exposure: average number of days in the past week that the respondents 1. Watched national network news, 2. Watched local TV news, 3. Watched cable news on TV 4. Read a daily newspaper, 5. Listened to political talk radio.   All items loaded on the same factor, explaining 40.41% of the variance.  Alpha for the news media exposure scale is .64.
Interest: average interest in public affairs and how much the respondent cares which party wins the election.  Both items loaded on one factor, explaining 72.64% of the variance.  Alpha = .62.
Race: a dummy variable where 1 = white and 0 = non-white.
Gender: a dummy variable where 1 = male and 0 = female.
Age: the participant's age in years.
Income: the participant's self reported income in thousands of dollars.

References

Berner, K. (2001). Dave Standup Guy in Columbia Gig, New York Daily News, 5 April, p. 3.
Brinkman, Del (1968) Do Editorial Cartoons and Editorials Change Opinion?  Journalism Quarterly 45:4 (Winter 1968) 724-726.
Carl, LeRoy M. (1968) Editorial Cartoons Fail to Reach Many Readers, Journalism Quarterly 54:3 (Autumn 1968): 533-535.
Center for Media and Public Affairs. (2000)  Media Monitor, January/February.
Delli Carpini, M. & Keeter, S. (1996). What Americans Don't Know About Politics and Why it Matters. New Haven: Yale University Press.
Grenier, Cynthia (1999). Late-night Gurus." World and I. 14(8) 103.
Gruner, C. R. (1971). Ad Hominem Satire as Persuader: An Experiment, Journalism Quarterly, 48(1), 28-131.
Hess, V. (2001, September). The Role of Political Comedy in the 2000 Election Campaign: Examing Content and Third-Person Effects. Paper presented at the annual meeting of the American Political Science Association Political Communications Division, San Francisco, CA.
Kleinbaum, D. G., Kupper, L. L., Muller, K. E., & Nizam, A. (1998). Applied Regression analysis and other Multivariable Methods. Pacific Grove, CA: Duxbury Press.
Koestler, A. (1964).  The Act of Creation.  London:  Hutchinson and Co.
Nightline. ABC News. (2000, 18 September).
Pew Research Center For the People and The Press. (2000, February). Audiences Fragmented and Skeptical: The tough Job of Communicating with Voters [On-line Report]. Available: http://www.people-press.org/jan00rpt2.htm
Sella, Marshall. (2000). The Stiff Guy vs. the Dumb Guy.  The New York Times. September 24, 2000.  72.
Shaap, D. (2000, July). Why are we laughing at Bush and Gore. George, 5(6), 72-103
Suls, Jerry (1972).  "A Two-Stage Model for the appreciation of jokes and cartoons:  an information-processing analysis."  Psychology of Humor.  Jeffrey H. Goldstein and Paul E.  McGhee (eds)  New York and London:  Academic Press.
Zaller, John (1992).  The Nature and Origins of Mass Opinion. New York: Cambridge University Press



[1]  The "RATS" controversy refers to a political ad made by the Bush campaign in which the word "RATS" flashed on the screen for a fraction of a second.  Republicans argued it was simply part of the word "DEMOC-RATS," while Democrats argued that the Bush campaign was engaging in subliminal advertising.
[2]  Late-night exposure means comparison.  Discussants (Mean = .33, SD = .42) Control group (Mean = .34, SD = .40). ANOVA: ( F= .187, Sig. = .666)
[3]  E.g.:  Net change in Gore inspiring rating from March to July = (Gore inspiring rating from July survey - Gore inspiring rating from the March survey)
[4]  Control variables included in the model:  Mean interest, Party-ideology index, TV news exposure, income, race, gender, age
[5]  In predicting net change from early July to early September, all the controls included in the above model remained, with the addition of "overall favorability of the candidate."
[6]  To avoid problems posed by multicolinearity, all interaction terms were centered by subtracting their means, while the main effects remain in their original form (Kleinbaum et al., 1998). The lowest tolerance in any of the twelve regressions was .55.