Humor Preference Changes with Age

Dec. 8, 2007


Summary: Two Internet surveys were used in order to test the validity of predicted changes of humor preference with age and gender. The results show clearly a strong agreement with predicted changes of age preferences but no significant evidence of gender preference.

Introduction: There are numerous models of humor (1) and of animal behavior, yet there are very few serious attempts at linking (2) the two fields of study, perhaps because humor, like language, is recognized by some influential thinkers as purely human activity and "There is nothing useful to be said about behavior or thought at the level of abstraction at which animal and human communication fall together." (3) Here is a hypothesis of humor types based on categorization of animal aggression with specific age preference predictions. An Internet survey of 16 jokes tested the hypothesis under easy commercial conditions and collected email addresses. The subjects who gave email addresses were invited to participate in a stricter 40 jokes scientific survey. Data collection lasted over one year; histogram results are shown for sample sizes of 2703 subjects for the commercial survey and 145 subjects for the scientific survey. The scientific survey also included 50-item Big Five personality assay. A secondary hypothesis of gender humor preference was also tested on the same subjects.

The background to the hypothesis was described in technical details by the author (4) in 1988. The following is a shorter description.

The Hypothesis: For the purpose of this work, humor is defined as a sudden falsification of a perceived threat. Depending on cognitive factors, threat induces either fear or aggression, types of humor follow the classification of four types of animal aggression as suggested by Paul Brain: self-defensive, predatory, social and parental.(5) The fifth type, infanticidal aggression, is ignored as being of no significant parallel with human behavior. These types correspond respectively to the humor types of Emotional, Feeding, SocioSexual and Parenting. Notice that these are motivational, not cognitive categories, all of which are divided cognitively into externalizing and internalizing, which basically describes the difference between fearful and aggressive behaviors in the context of the four motivational types. Emotional humor is self-assuring, Feeding mocks threat to turf and resources, SocioSexual humor falsifies threats to the social message and its contents, and Parenting mocks threat to offspring, or is self defeating or lies in the wider context of altruistic behavior.

Animals usually display aggression for a reason, I observe the type of aggression with the most utility in young people is to protect their feeding turf and resources in order to secure their needs to grow, and when adult is to protect their mates and social group and when mature their offspring. People also protect their own self-existence independently of age and body needs constantly throughout their lives. Translated into humor preference, I predict that "Feeding" type of humor is highest when young and declines with age, "SocioSexual" rises until adulthood then starts falling down and "Parenting" keeps rising with age, but "Emotional" humor is age insensitive. The secondary somewhat intuitive hypothesis of gender preference predicts that females internalize while males externalize.

The use of the above categorization in computer personality simulators is protected by U.S. patent pending #11557013.

The Experiment: In order to test the humor preference hypothesis, two surveys of scoring jokes were conducted online.

The Commercial Humor Preference Survey: Subjects were solicited by search engine advertising over a period of one year. Has scores for 16 jokes, 4 per motivational type, 2 per cognitive category. Responses were collected in database, including IP address, date and time of response, geographical location of subject, age group divided into six cells, gender and scores of joke preference from 1 to 5. A question was asked if the subject was willing to participate later on in a more scientific survey, if yes then he/she was asked to leave email address. The survey provided instant score reading and a comparison with the average scores of the other participants. Commercial grade self-auditing was applied, which basically meant insuring that each participant has a unique IP address and the responses are non-trivial.

The Scientific Humor Preference Survey: Randomly selected email respondents from the commercial survey were contacted and invited to take the scientific survey. The selection of participants was intended to reflect the age and gender distribution of the world population (data stratification) but this was abandoned since only 145 out of more than 700 emails responded in time. The scientific survey has scores for 40 jokes, 10 per motivational type, 5 per cognitive category. It also included the 50-item Big Five IPIP questionnaire . In addition to the ranking responses, the questionnaire asks for contact information and whether or not the subject is willing to participate in additional academic or commercial surveys. Repeated responses and subjects with trivial scores were removed before the analysis.

The Analysis: In preparation for analysis, contact information was removed and the responses for both surveys were saved in .csv format. The analysis was performed using R Software . First, the rankings were added for each type and category, then Z-scored and the scores averaged. The raw .csv data files can be viewed and analyzed for the commercial and scientific surveys.

The Results: The raw .csv data were used in linear regression analysis against age groups. The results are attached for the commercial and scientific surveys. Comparing standard deviation estimates between raw data and linearly fitted shows significant reduction of uncertainty in almost all cases. Tables 1 and 2 below show the values of averaged Z-scores, figures 1 and 3 are the histograms. We can identify one freak cell in the commercial survey results and 2 or maybe 3 in the scientific. The freak result of the commercial survey may be explained by double interpretation, which I would like to call the Dangerfield phenomenon, after Rodney Dangerfield’s famous joke: I was born so ugly the Doctor slapped my mother, which showed old age preference as well as young and a dip in the middle. The reduction of uncertainty is particularly poignant at the SocioSexual humor type commercial data when using two lines for regression; considering that the standard deviations for each line are estimated with only one degree of freedom yet the reduction of uncertainty is better than 50%.

Gender preference was also tested; the accumulated scores are shown in figures 2 and 4. In both surveys gender preference was similar in slightly preferring externalizing jokes to internalizing.

Estimate of correlation between B5 and humor preference was also sought from the scientific survey, but the results are not shown here; a quick calculation led to insignificant values for most correlations, although some correlation coefficients measured around 0.25. The small count of scientific subjects may be the main cause for not reaching conclusive values.

Under 15 #254

16 to 25 #1129

26 to 35 #430

36 to 45 #310

46 to 55 #294

over 55 #286





























Table 1: Commercial survey age humor type preference in z-score average standard deviations.

Fig. 1: Commercial survey humor preference showing one freak result, Parenting type humor preference for under 15.

Fig. 2: Commercial survey gender humor preference in total score of externalizing and internalizing categories.

Under 15 #12

16 to 25 #33

26 to 35 #27

36 to 45 #22

46 to 55 #23

over 55 #28





























Table 2: Scientific survey humor type preference in z-score average standard deviations.

Fig. 3: Scientific survey humor preference showing two freak results, Feeding and SocioSexual type humor preferences for 36-45.

Fig. 4: Scientific survey gender humor preference in total score of externalizing and internalizing categories.

Past Research: The use of four types of humor is not new; Martin et al.(6) formulated their humor types in a way that easily identifies some as harmful to health while the others as beneficial. The four types are Self-enhancing, Aggressive, Affiliative and Self-defeating, roughly corresponding to Emotional, Feeding, SocioSexual and Parenting respectively. Self-enhancing and Affiliative are identifies as good for health while Aggressive and Self-defeating are not. The article used the Humor Styles Questionnaire to test for age and gender preference. Using linear regression, no significant age correlation was detected. Martin et al’s work is different from this in two important respects: First, no cognitive divide (externalizing/internalizing) is included; a careful examination of the interpretations shows that Aggressive and Self-defeating humor refer only to externalizing categories, which excludes important internalizing contributions to age dependency. Second, there was no definition of hypothesis to test, and so only linear regression was used to fit the data for all age groups. These differences may have prevented the identification of trends in Aggressive and Self-defeating humor types, and the peak in the Affiliative type.

Conclusions: The analysis of the commercial and scientific survey data clearly show agreement between predicted humor preference with age but not with gender. It should be noted that the peak of SocioSexual preference is in the 26-35 year age group and declines from then on in both surveys, this traces the decline of female fertility and may or may not be coincidental, but in my conclusion it warrants more attention in future research. Correlations between B5 factors and humor preference exist, but considering their low value and the small number of participants were thought to be too unreliable to reach concrete conclusions.


(1) Lyttle, J. (2003): Theories of Humor:

(2) Decter, R.R. : Humor in Domestic Dogs (1997):

(3) Chomsky, N. (1968) :Linguistic Contributions to the Study of Mind :

(4) Kadri, F.L. (1989): Animal Drives in Humans. ISBN 0-9694181-0-8.

(5) Brain, P. and Benton, D. (1981): Multidisciplinary approaches to aggression research. Elsevier/North Holland: Amsterdam.

(6) Martin, R.A., Puhlik-Doris, P., Larsen, G., Gray, J., & Weir, K. (2003): Individual differences in uses of humor and their relation to psychological well-being: Development of the Humor Styles Questionnaire. Journal of Research in Personality, 37, 48-75.

About the author: Faisal L. Kadri is an independent researcher not affiliated with any educational institution. His research interest since 1986 is in applying the mathematical tools of nonlinear systems engineering to modeling motivational mechanisms in animals and humans. For more information please visit:


Addendum March 10, 2008: Internal consistency reliability
Cronbach's Alpha coefficient is computed for the commercial 16-Joke and scientific 40-joke tests, here are the values:
Commercial Survey: Alpha: 0.743 Robust Alpha: 0.854, for each category:
Emotion Feeding SocioSex Parenting
0.56 0.50 0.47 0.78

Scientific Survey: Alpha: 0.923 Robust Alpha: 0.957, and for each category:

Emotion Feeding SocioSex Parenting
0.81 0.80 0.83 0.86

I am grateful to Andreas Christmann of the University of Bayreuth for his guidance in calculating the robust Cronbach's alpha.

Copyright © 2007 Faisal L. Kadri, all rights reserved. Reproduction of this article in Internet media is permitted on condition that all links are maintained.