KmB8yvR8

14. DiscussionThe present meta-analysis provides the first comprehensive quantitative evaluation of sex differences in humor production ability. We sought to collect all available studies assessing new humor outputs (mostly verbal) rated for funniness by independent judges blind to any characteristic of the producer. We believe that these types of humor production tasks and the method by which they are evaluated reflect the most objective measure of true humor abilities. Results reveal a small to moderate effect, with men scoring higher than women.We took several steps to address and minimize the possibility of publication and other potential biases. First, we used random-effects models and weighted means to estimate the combined effect size, a more conservative approach for analyzing the data (Borenstein et al., 2009; Card, 2015). Second, our statistical analyses and tests, together with visual inspections of funnel and Normal Q-Q plots, showed no indication of publication bias. Overall, more than 60% of the data (20 out of 36 samples) were unpublished, either from unpublished datasets, theses and dissertations, or based on peer-reviewed publications where the data did not appear in the paper. The latter consists of 12 
232samples (44% of the published data), and included studies that did not focus on sex differences in HPA. Thus, it is unlikely that the presence or absence of sex differences in HPA played any role in the decision to publish the study. In fact, the combined effect size for peer-reviewed publications was almost identical to that of not peer-reviewed studies (.333 and .280 respectively). Moreover, given the debate surrounding sex differences in HPA, it is unclear what type of bias might have existed. Results suggesting that men and women do not differ in their HPA are as important and informative as data indicating that sex differences in HPA do exist. Indeed, roughly half of the peer-reviewed publications revealed no significant sex differences, while the other half showed men having higher HPA. Thus, it is unlikely that the reason an unpublished manuscript was not published had anything to do with the presence or absence of sex differences in HPA. In sum, all indications are that there is little evidence for publication bias, and our view is that such bias is unlikely.Third, to address the possibility of additional biases, we included in our meta-analysis a large number of moderators that could potentially influence the results. One aim was to test whether there is any biasstemming from the authors’ or judges’ sex. We found no evidence for such biases. The sex of the first author, and whether a single or mixed sex team conducted the study, had no influence on the results. In addition, despite variation in the male to female judge ratioacross studies, the overall number of male and female judges was nearly identical, 171 and 169 respectively (based on all studies that reported the figure, 26 out of 36 samples). A moderator test revealed no effect for this variable on 
333the combined effect; thus, it seems unlikely that judges’ sex had any impact on the results.Only one of the variables yielded a significant effect on the estimated combined effect size. This moderator, number of responses per task suggests that asking participants to produce one humorous output may result in higher sex differences in HPA for men, compared with tasks that allow for multiple humour outputs. This finding may indicate that men might have a further advantage when asked to produce their best humor once, while women may be funnier when given the chance to create multiple responses. Still, it is important to remember that in either scenario, men still scored significantly higher than women.Though our moderator analyses did not reveal other statistically significant moderators’ effects, some results may still illuminate important trends in the sources of sex differences in HPA. In particular, an important aspect of all humor creation tasks is whether the participants produced humor under time constraints or not. The results showed that having unlimited time to complete the task was associated with larger sex differences in HPA, compared to tasks restricted in time. Having unlimited time to create humor may have reduce the stress involved in creating humor, and allow participants more time to think about the task, something that benefited men more than women. However, it is also important to note that for all studies where participants were limited in the amount of time to produce humor, more time was not associated with larger sex 
434differences. In other words, just knowing that the time is limited seemed to have some effect on the magnitude of sex differences.It was also interesting to find some null moderator effects. For example, sex differences remained similar in more than 40 years of research, though there are relatively few studies prior to 2000 for which the data were available. Similarly, the results were surprisingly similar across cultures and samples, with college students and non-college students showing almost identical sex differences. Nonetheless, these results are provisional, as researchers may find different results in samples that are more diverse in the future.4.1. Explaining Sex Differences in Higher Humor Production AbilitySex differences in HPA found in this study may reflect evolved sex differences in mating preferences and strategies that were shaped by sexual selection theory (Darwin, 1871). Women’s higher parental investment and larger reproduction costs than men make them choosier when selecting a mate, and more attentive to traits that could result in higher fitness (Buss, 2016; Trivers, 1972). HPA could be one such trait, as people vary in their ability to produce humor, and it is a reliable, hard to fake signal of intelligence, a highly desirable trait that increases fitness and serves as a mental fitness indicator to attract mates (Christensen et al., 2018; Greengross & Miller, 2011; Howrigan & MacDonald, 2008; Kaufman, DeYoung, Reis, & Gray, 2011; Miller, 2000a, 2000c). As women are choosier than men, we would expect women to be more sensitive and attuned to men’s display of high HPA. As a consequence, a stronger intra-sexual competition among men 
535ensues, resulting in an overall higher average HPA for men (Bressler & Balshine, 2006; Bressler et al., 2006; Miller, 2000a, 2000b). Based on this logic, when selecting a mate, men should use humor more often and more creatively to attract women and signal their mate value, while women should be more sensitive to men producing high quality humor. Various research supports this theory, and the view that HPA is valued differently and divulges disparate information for men and women. Compared to men, choosier women value humor as a more important trait when selecting a mate, while men make more effort to impress women and advertise their humor ability, including in real ecological settings, such as dyadic conversations and on dating sites (Lippa, 2007; Provine, 1993; Sprecher & Regan, 2002; Todosijević et al., 2003; Wilbur & Campbell, 2011). Women also prefer a man with higher HPA, while men are more attracted to a woman that laughs at their humor, rather than a woman with high HPA, as smiles and laughter signal the woman may have a romantic interest in them (Bressler & Balshine, 2006; Bressler et al., 2006; Hone, Hurwitz, & Lieberman, 2015). Viewing HPA as a mental fitness indicator relies on the connection between HPA and intelligence (Miller, 2000a, 2000c), and numerous studies have shown positive correlations between the two attributes (Christensen et al., 2018; Greengross & Miller, 2011; Howrigan & MacDonald, 2008; Kellner & Benedek, 2017).Nevertheless, while a universal phenomenon, humor varies across cultures and reflects societal norms (T. Jiang, Li, & Hou, 2019). Little is known about sex differences in any facet of humor among non-Western populations, thus, the universality of sex differences in HPA found in our meta-analysis should be taken with caution. Most of our data come 
636from WEIRD (Western, Educated, Industrialized, Rich, and Democratic) countries, which may delineate only a fraction of all human populations (Henrich, Heine, & Norenzayan, 2010b). Often, findings that are true for WEIRD samples do not replicate in non-WEIRD populations (Gurven, Von Rueden, Massenkoff, Kaplan, & Lero Vie, 2013; Henrich, Heine, & Norenzayan, 2010a; Henrich et al., 2010b). It is possible that non-WEIRD countries use and experience humor differently, which could influence the direction and magnitude of sex differences in HPA. If we want to draw conclusions that will be applicable to all humans, as the evolutionary explanation suggests, we need information about more diverse populations. For example, research suggests that East Asian people, such as from China and Taiwan, laugh less, and view themselves as less funny, compared to Western cultures such the United States and Canada (F. Jiang, Yue, & Lu, 2011). Western countries generally tend to value humor more, perceive humor as a more socially desirable trait, and view humorous people more positively, compared to Eastern countries (Yue, Jiang, Lu, & Hiranandani, 2016).It is possible that the observed difference between men’s and women’s HPA is an artifact of the fact that certain types of humor, the ones that are considered of high quality, are more freely expressed by men than by women. For example, men may feel no restrictions in telling sexual and aggressive jokes, while women may be more inhibited in the use of these types of humor. However, some research suggests that while women are less likely than men to tell jokes in general, when they do, women are just as likely as men to use sexual and aggressive themes (Johnson, 1991). Mickes et al. (2012) compared the themes that men and women used when producing humor using the cartoon captioning task. 
737Their results showed that when generating the humorous captions, men produced significantly more sexual humor and used more profanity than women, though the overall usage of such humor was low (4.30% for men, 1.95% for women). More importantly, the use of sexual humor and profanity did not give men any advantage and did not contribute to their total higher HPA ratings compared to those of women. Also, neither men nor women judges rated these types of humor as funnier. The authors concluded that the higher humor ability of men could not be attributed to the use of sexual humor and profanity, but to other factors.It is also important to note that despite the belief held by many that women enjoy sexual and aggressive humor less than men, reviews of the literature show mixed support to such claims (Lampert & Ervin-Tripp, 1998; Martin, 2014). Earlier research found women to enjoy sexual and aggressive humor less than men (Lampert & Ervin-Tripp, 1998; Martin, 2014), however, many of the jokes and cartoons used in these studies portray women as the target of the jokes, and the jokes tended to be sexist. When the targets of the jokes are men, or the jokes are not sexist or have neutral themes, women and men express similar levels of appreciation to the humor (Lampert & Ervin-Tripp, 1998). In our analysis there is little evidence to suggest that men and women judges evaluated the humor produced by either sex differently. Five studies in our meta-analysis tested for sex differences between judges’ ratings, and only one found a significant difference (Mickes et al., 2012). Mickes et al. (2012) reported that both male and female raters judged men’s HPA as higher than women’s, but male raters gave male participants slightly higher ratings than female raters. In contrast, four other studies that tested for sex differences in judges’ rating did 
838not find any significant differences (Brodzinsky & Rubien, 1976; Greengross, Jones, & Sanoudaki, 2017; Greengross, Martin, & Miller, 2012; Greengross & Miller, 2011). Given the small number of studies to date, the role of judges clearly deserves more attention in future research.There is some evidence that sexual and aggressive stimuli used to elicit the humor production may have an effect on the overall magnitude of sex differences in HPA. Brodzinsky and Rubien (1976) asked participants to produce spontaneous new humor in response to captionless cartoons that contained either sexual, aggressive, or neutral themes. Sex differences in HPA were found for sexual and aggressive themes, but not for the neutral cartoons. With the exception of Brodzinsky and Rubien (1976), no other study in our analysis reported whether the stimuli were sexual or aggressive in nature.Another possible explanation for the observed difference lies in the nature of the task itself. The typical HPA task requires a crisp, focused response. Some research suggests that men are more likely to tell jokes, while women prefer telling funny stories and anecdotes (Crawford & Gressley, 1991). Although most tasks included in the meta-analysis do not exactly imitate a traditional canned joke structure, they are fairly constrained and do not afford longer, narrative-oriented responses. In addition, joke-telling is relatively a small fraction of everyday use of humor, where most humor arises during spontaneous social interactions (Martin & Kuiper, 1999; Provine, 2000). Varying the types of contexts of HPA tasks seems like a particularly promising direction for future work.
9394.2. Limitations and Suggestions for Future ResearchAs with any meta-analysis, there is always a possibility that more data exists. In fact, we know with certainty about several studies that matched our inclusion criteria, but for which the data could not be retrieved. Many researchers included measures of HPA in their study but did not report the relevant information, mostly because they did not focus on studying sex differences in humor. Most of these data come from older studies, and the raw data are now lost. Nonetheless, as our analysis on publication bias revealed, adding more data is unlikely to change the overall results. Still, we welcome any new studies on this topic, perhaps with new methodologies for measuring HPA.Our meta-analysis included studies that used verbal humor as their main measure of HPA. It is possible that by focusing on non-verbal humor, the results would have been different. Another limitation is that the measures of HPA included in our meta-analysis are somewhat artificial, and do not represent everyday production of humor. Requesting people to produce humor on demand is challenging, and perhaps disadvantaged women more than men. It also ignores the social context in which most humor is produced (Provine, 2000), context that if taken into account may benefit women more than men. Perhaps sex differences in HPA vary depending on the environment in which it is produced. For example, women may have equal HPA scores as men when interacting with other women. Thus, studying various dyadic interactions of men and women in more ecologically valid situations, such as natural conversations, is crucial for fully understanding when and how sex differences in HPA emerge. Relatively few researchers 
1040conduct these types of studies (Hall, 2015; Provine, 1993, 2000). Still, humor is largely a social phenomenon and most humor is created in a social context while interacting with other people. Studying humorous interactions in the lab (Hall, 2015), or observing them in natural settings (Provine, 1993, 2000) should be a fruitful endeavor that requires more of our effort. Another limitation are the ages of participants included in our meta-analysis. The samples contained somewhat restricted ages, ranging from 15-35, with a median age of 21.7 (see Table 4.). Clearly, such samples are not representative of the whole population, but they do represent individuals at peak reproductive age. At these ages, following sexual maturity, people are at peak fertility. This is the period when the competition over mates is the strongest (Buss, 2016). As a result, and due to women’s choosiness, it might represent the time when men try to impress women with their humor the most, thus resulting in higher HPA than women, as our meta-analysis found. Hence, results might be different for younger or older populations, with different theoretical implications.4.3. ConclusionThe research presented here focused on one specific aspect of humor that is largely under-investigated in humor research, humor production ability. Despite finding men to have higher humor creation abilities than women on verbal humor, this difference should not be seen as representative of other types of humor, including non-verbal humor production ability. In fact, for most aspects of humor, men and women seem to exhibit many similarities, with relatively few differences (Martin, 2014). In regard to humor 
1141production abilities, the topic of sex differences is often reduced to blunt assertions such as that “Women are not funny” (e.g.,Hitchens, 2007). We hope that our meta-analysis will help advance a more nuanced discussion on the topic based on a systematic evaluation of the available scientific data. Examination of such data suggest that regardless of the underlying source of variability, men exhibit higher humor ability than women on the kinds of verbal tasks included in our sample of studies. It is important to remember that though robust, these differences are small to medium in size, and are based on averages. They do not reflect individual abilities, as both men and women vary largely in their abilities to produce humor. We tried to illuminate possible sources for the differences in HPA, what they might mean, theoretical implications, considerations for future research, and limitations. Humor is an important experience for most people, one that is largely unique to humans. We hope that our results will further foster the study of humor, advance theories pertaining to understanding and explaining sex differences in humor and other cognitive abilities, as well as foster research on humor ability.