Sunday, February 23, 2020

Laboratory: Women avoid competing with men; men seem to anticipate the lower competitiveness of female opponents, as evidenced by their greater tendency to compete against women

Gender and Willingness to Compete for High Stakes. Dennie van Dolder, Martijn J. van den Assem and Thomas Buser. Tinbergen Institute Discussion Paper TI 2020-011/I, Feb 2020. https://papers.tinbergen.nl/20011.pdf

Abstract: We examine gender differences in competitiveness, using a TV game show where the winner of an elimination competition plays a game of chance worth hundreds of thousands of euros. At several stages of the competition, contestants face a choice between continuing to compete and opting out in exchange for a comparatively modest prize. When strategic considerations are absent, we observe the well-known pattern that women are less likely to compete than men, but this difference derives entirely from women avoiding competition against men. When the decision to compete is strategic and contestants should factor in the competitiveness of others, women again avoid competing against men. Men, in turn, seem to anticipate the lower competitiveness of female opponents, as evidenced by their greater tendency to compete against women. Ability differences are unlikely to explain these results. The findings underline the importance of the gender of competitors for the analysis of differences in willingness to compete, and shed new light on the persistent gender gap at the male-dominated higher rungs of the career ladder.

JEL: D91, J16
Keywords: gender differences, competitiveness, willingness to compete, game show

6. Conclusion and discussion

The present paper examined gender differences in willingness to compete for high stakes, using a TV
game show where the winner of an elimination competition plays a game of chance worth hundreds
of thousands of euros. The amounts that are at stake in this real-life setting are much closer to the
sums involved in promotion competitions at the top of the labor market than the financial rewards
that are commonly employed in laboratory experiments.
We focused on two different games. In the first, contestants choose between competing in the next
elimination game and opting out for a prize. This decision resembles the tournament-entry decision
that subjects typically make in lab experiments. In line with the picture that emerges from the
experimental literature, we find that women are twice as likely as men to avoid the competition. The
comparatively high opt-out rate of women, however, derives entirely from situations where they will
face predominantly male opponents. This suggests that women have a particular dislike of competing
against men, rather than a more general dislike of competition. In this non-strategic game, men do not
appear to condition their behavior on the gender of their opponents.
In the second game one question determines the last elimination, unless one of the two remaining
contestants voluntarily accepts an opt-out prize. This head-to-head game is relatively complex,
because the optimal choice depends on the anticipated behavior of the opponent. The results show
that both genders avoid competing against a male opponent in this strategic setting. The same pattern
arises in the strategic game analyzed in the Appendix. This pattern confirms the result for the first
game that women avoid competing against men, and suggests that in strategic interactions men
anticipate a lower willingness to compete among their female opponents. In line with the results for
the first game, there is no evidence of a main effect of gender on competitiveness.
Altogether, the results confirm that gender differences in willingness to compete also occur in
situations where the stakes are very high, but they also indicate that the stereotypical image and
widely-held belief that women are less competitive than men is too simplistic. The findings show the
importance of the gender of competitors and of the presence of strategic interaction: women appear
to dislike competing against men, and men appear to exploit this when there is strategic interaction.
Hitherto, the literature on gender differences in willingness to compete has largely ignored these two
factors. Exceptions are Booth and Nolen (2012) and Geraldes (2018), who similarly find that women
avoid competing against men. Such a dislike is in line with evidence that women perform worse when
they compete against men (Gneezy, Niederle and Rustichini, 2003; Günther et al., 2010; Backus et al.,
2016; de Sousa and Hollard, 2016; Booth, Cardona-Sosa and Nolen, 2018; Booth and Yamamura, 2018).
Strategic interaction is typically ruled out by design in willingness-to-compete experiments.
Research somewhat removed from the willingness-to-compete literature indirectly supports our
conclusion about the importance of opponent gender in strategic interactions. Babcock et al. (2017)
show that women are more likely than men to volunteer for an undesirable task when interacting in
mixed-gender groups, but not when interacting in single-gender groups. Booth and Yamamura (2018)
study Japanese speedboat races and find that men adopt a more aggressive racing style in mixedgender races than in single-gender races, whereas women act less aggressively in mixed-gender races
than in single-gender races. Hernandez-Arenaz and Iriberri (2018) investigate bargaining behavior in a
Spanish TV game show and find that women ask less from men than from women.
Our setting differs in a number of ways from that of a traditional laboratory experiment. In addition to
the markedly higher stakes, the subject pool that we observe is much more diverse, and because of
the presence of a large audience and cameras, our contestants’ decisions are observed by many and
subject to considerable public scrutiny. A benefit of the diverse pool of contestants is that we can also
explore how competitiveness varies with age. We consistently find a u-shaped relation, with people
who are in their forties displaying the lowest opt-out propensity. Mayr et al. (2012) find a similar
relation between age and competitiveness among subjects in an incentivized experiment. Flory et al.
(2018) and Buser, Niederle and Oosterbeek (2020), however, report different age patterns.
A benefit of the public nature of the competition, is that this feature is shared by many consequential
real world competitions. In the political arena and upper echelons of the corporate world, for example,
people face considerable public scrutiny when they compete for top positions. This is another
argument for why our naturally occurring setting is much more similar to such real-world situations
than anonymous, low-stakes laboratory experiments. Because opting out of the competition ends
public observability and choosing to compete extends it, a difference in the attitudes of men and
women towards public observability would obfuscate the relation between gender and
competitiveness. To the best of our knowledge, the only paper that studies the impact of public
observability on competitiveness is Buser, Ranehill and van Veldhuizen (2019). They find suggestive
evidence that public observability increases men’s willingness to compete, but conclude that public
observability does not alter the magnitude of the gender gap in willingness to compete in an
economically or statistically significant way. This possible confound also occurs in many real-life
settings, but cannot explain why women display a dislike of competing against men and why men are
more competitive when they face women in a strategic setting.
Similar to the tasks that are typically used in willingness-to-compete experiments, the questions that
are central in the elimination competition are often numerical or arithmetic in nature. This similarity
facilitates the comparison of our results with those of previous studies. Research on competitiveness
often intentionally uses such tasks because mathematics is a stereotypically male area, which brings
the research closer to competitive situations in male-dominated workplaces or male-connotated areas
such as management (Niederle and Vesterlund, 2011). Although we cannot entirely rule out that our
results are partly driven by a gender difference in the ability to answer the questions, none of the
additional analyses of our data provide evidence that the quiz questions actually favor men. Second,
based on research into math performance and research into recall of factual information, there is little
reason to believe that such a difference holds true for the general population (Herlitz et al., 1997;
Nilsson, 2003; Hyde et al., 2008; Lindberg et al., 2010). Last, even if any gender difference would exist
among the initial pool of contestants, it would be relatively small among those we study because
weaker contestants are less likely to reach the stages of interest.
Of course, contestants may have held non-rational expectations about their own ability and that of
others, possibly inspired by stereotypical beliefs about the performance of men and women on math
questions. Indeed, in experimental work that uses a neutral or stereotypically female type of task,
competitiveness differences between men and women are sometimes, but not always, weak or absent
(Große and Riener, 2010; Shurchkov, 2012; Dreber, von Essen and Ranehill, 2014; Wozniak, Harbaugh
and Mayr, 2014). Stereotype-biased beliefs about performance may explain why women display a
relatively low propensity to compete against men in all the games that we study, and why men display
a relatively high propensity to compete against women when the game is strategic. This explanation,
however, is not supported by the behavior of men in the first game: there is no evidence that men are
especially eager to compete against women. If anything, men are more likely to opt out against women
than against men in this game.
A possible concern about our findings is that their generalizability might be negatively affected by
selection effects. Selection procedures are inevitable in any lab experiment or field setting, and could
potentially bias comparisons of the behavior of men and women (Larkin and Pines, 2003; Reback and
Stowe, 2011; Hogarth, Karelaia and Trujillo, 2012). Unlike contestants in most other game shows,
however, contestants in our elimination competition do not need to self-select into auditions and are
not screened and then selected by producers prior to their participation. All have won their ticket
through the popular Dutch Postcode Lottery. Even for competition-averse individuals, using this ticket
is attractive: in addition to the lucrative possibility of becoming the finalist, contestants can win many
other large prizes. Nevertheless, subjects in our study are not selected perfectly at random: all are
lottery players who were able to attend the recording, and couples might send the best or most
competitive of the two of them. Still, as a group the subjects in our study do resemble a cross-section
of the general population much more closely than subjects in most lab experiments and other field
studies do. More importantly, it is not clear how selection mechanisms could explain that the
competitiveness of men and women depends on the gender of their opponent.
Our study finds that women avoid competition against men, but remains silent on the underlying
causes of this behavior. According to the literature, differences in reluctance to compete can be driven
by differences in risk attitudes, (over)confidence, and intrinsic attitudes towards competition.
However, even in controlled experiments, and even without strategic interaction and consideration of
the role of the gender of opponents, careful disentangling of the possible determinants of
competitiveness has proven to be methodologically challenging (van Veldhuizen, 2018; Gillen,
Snowberg and Yariv, 2019). Our field data is not rich enough for delving into the possible explanations,
but it seems implausible that risk preferences depend on the gender of the opponents. This leaves
opponent-dependent confidence in performance and intrinsic opponent-dependent attitudes towards
competition as the most likely drivers: women could be less confident about their performance in a
competition against men, or have a more deeply rooted, intrinsic aversion to competing against men.
Such a specific response to the opponent’s gender can, in turn, be partly determined by the culture in
which people grow up (Gneezy, Leonard and List, 2009; Andersen et al., 2013; Booth et al., 2019;
Zhang, 2019).
Research into the effect of social class and ethnicity on competitiveness suggest that gender
differences in competitiveness—including those identified in the present study—may derive from a
more general phenomenon, where willingness to compete relates to social power and status. Almås
et al. (2016b) find that gender differences in competitiveness arise between boys and girls from
families with a high socioeconomic status only, and not between those who are from families with a
low socioeconomic status. Siddique and Vlassopoulos (2019) report that people from an ethnic
minority group are more likely to compete when their competitors are all co-ethnic than when their
competitors are predominantly from the majority group.
Regardless of the possible psychological mechanisms, the finding that women avoid competing against
men has the important implication that male dominance in a professional environment becomes selfperpetuating. This is especially the case if there is strategic interaction, where those who compete are
better off when others abstain from competing. In such a setting, women can expect more pushback
from both male and female competitors. At the higher rungs of the career ladder, where
overrepresentation of men and strategic interaction are both ubiquitous, affirmative action may be
necessary to alter the status quo.

No comments:

Post a Comment