Download Female Judges' Impact on Contestant Survival in MasterChef Junior: Gender & Race Analysis and more Exercises Economics in PDF only on Docsity!
Whip It Like A (Wo)Man: Survival Analysis of
Minority Contestants in Masterchef Junior
December 19, 2018
1 Introduction
Masterchef Junior is a popular American cooking show featuring children ages eight to thirteen. It is based on a similar show, Junior Masterchef, a U.K.- based children’s cooking show. In the show, a cohort of children (between 15 and 25 in a typical season) compete for the coveted Masterchef Junior trophy and $100,000. Celebrity judges such as Gordon Ramsay, an international cook- ing sensation, Joe Bastianich, a successful restaurateur, Graham Elliot, a three Michelin star recipient, and Christina Tosi, a renowned pastry chef and two-time James Beard Award winner, evaluate the contestants’ performances in weekly challenges and ultimately eliminate at least two people each week until only the top two remain. The finale is an elaborate display of culinary technique and finesse as the finalists prepare a self-designed three course meal for the judges.
As much fun as the show is to watch, it is not immune to biases we hold as to what a chef “looks like.” Subtle comments are often made towards white male contestants that they remind the judges of themselves as young people. In fact, the white male contestants reflect the industry as a whole. A 2015 study by the Restaurant Opportunities Centers United (ROC United) found that “81 per- cent of management positions in 133 fine-dining restaurants were held by white employees” [5], many of whom are male. Women of color often are questioned and challenged when they rise to positions of authority in a professional kitchen. “[Jamie] Suddoth, 37, works at a market-cafe hybrid called Earl’s Grocery and runs Jamie’s Cakes and Classes, both in Charlotte, but she has dabbled in other specialties. No matter what part of the kitchen she walks into, she feels a need to prove her talent as a woman of color. ‘It’s a male-dominated field,’ Suddoth added. “I don’t get respect right off the bat” [5]. To date, there have been only five black Beard Foundation Best Chef Award nominees. Discrimination against women and people of color manifests itself in wage discrepancies as well. According to the ROC United study on restaurants in California, white female restaurant employees make 80 percent of what white male restaurant employ- ees earn on average. That gap is consistent for other minority groups as well. Male employees of color earn 82 percent while women of color earn 71 percent
of what their white male counterparts earn. Some of this is explained by a lack of minorities in the highest paying positions which is another cause for concern [6]. Aggressive behavior in the kitchen, often coming from well-known male chefs, is regarded as powerful and captivating [1]. The “bad boy” attitude has kept many women out. Gordon Ramsay, one of the show’s star judges, is one of the best examples of an explosive temper that undoubtedly draws in fans. In 2009, when ten female chefs won Michelin stars in the U.K., The Telegraph announced “It could be the beginning of the end for the foul-mouthed, macho, and defiantly male master chef“ [2]. While managers may claim that the biases we hold as to how chefs look and behave does not adversely affect hiring and promotion outcomes, a 2009 study found that innate biases were predictive of ”socially and organizationally significant behaviors“ [4].
Innate biases that the chefs have about who is a chef often plays out in their commentary on different contestants. In a team challenge in season 2, the judges had to eliminate two of the three members of the losing team. They chose to keep Samuel, a white boy, over Sean, an Asian-American boy, because Samuel ”looked most comfortable“ even though Sean far out-performed him. Sean’s steak had ”blown the other team’s out of the water“ while Samuel ”botched his fish and failed to bring his team together“ [3]. Female contestants are often asked questions like ”Could you do that again, or is this luck?“ and ”Is this a fluke?“ while male contestants are asked how much they would charge for their dish at their future restaurant, even when they perform at a similar level. In part responding to this criticism, in its third season MasterChef Junior introduced a new female judge, Christina Tosi. In this paper we will rigorously investigate whether or not the introduction of a female judge changes the trajectory of minority contestants, both women and people of color. We hypothesize that the presence of a female judge will increase the survival of minority contestants. In section 2 we will discuss our data and methods, in section 3 we will discuss the results of our analysis, in section 4 we will discuss these results in context, and finally in section 5 we will provide some concluding remarks.
2 Data and Methods
To obtain our data, we copied tables we found on Wikipedia [7, 8, 9, 10, 11, 12] for each season that contained detailed information about the contestants and their statuses at the end of each episode. We then cleaned the data and added new columns that would be relevant to our analysis. One of these was a column for race, which we determined using photographs of the children and onscreen comments they made about their heritage and how it influenced their cooking. Our race values are either white or non-white for simplicity, and we classified races from the Mediterranean region as non-white.
The table below contains each of the variables we added to our data, as well as an interpretation and range of values for each variable:
judge. Additionally, we tested the overall differences in survival rates using log-rank tests.
3 Results
First, we wanted to investigate survival rates by sex before and after the in- troduction of a female judge to the panel. We estimated survival rates non- parametrically with a Kaplan-Meier curve. In Figure 2, we can see that before the introduction of a female judge, at almost every time-point in the season, male contestants have a higher predicted survival rate than female contestants. At the points where about 15 percent of the season and about 60 percent of the season have passed, the curves are overlapping and we see little difference in survival. In seasons 4 to 6, though, it appears that female contestants have a higher predicted survival rate at most time-points. We also ran log-rank tests to ask whether or not there is a significant difference in survival for male and female contestants overall with and without a female judge. Without a female judge on the panel, there is not a significant difference at α < 0 .05 because p = 0.149 with only 47 observations. Additionally, with a female judge there is also not a significant difference in survival rates for male and female contes- tants at α < 0 .05 because p = 0.068 with only 68 observations. If we had more observations from an additional season (season 7 is expected to air in spring of 2019), we may see more significant results.
Figure 2: Survival (measured by percent of season) of Contestants by Sex
(a) Survival in seasons 1 to 3, before the addition of a female judge
(b) Survival in seasons 4 to 6, after the addition of a female judge
Next, we used the same non-parametric methods to assess the relationship be- tween survival and race before and after the introduction of a female judge. Figure 3 shows that in seasons 1 to 3, white contestants generally have a higher predicted survival rate than non-white contestants. This trend is the most dras- tic in the first 50 percent of weeks passed, where there is a large gap between the predicted survival curves. In seasons 4 to 6, the survival curves are more similar to each other, with non-white contestants now having a higher overall
predicted survival rate. We also wanted to know whether or not the overall survival rates are significantly different. To measure this we used log-rank tests. We found that before the introduction of a female judge there is not a significant difference in survival at α < 0 .05 because p = 0.529 with only 47 observations. After the introduction of a female judge there is also not a significant difference in survival rates at α < 0 .05 because p = 0.903 with only 68 observations.
Figure 3: Survival (measured by percent of season) of Contestants by Race
(a) Survival in seasons 1 to 3, before the addition of a female judge
(b) Survival in seasons 4 to 6, after the addition of a female judge
Based on these plots, we decided to use accelerated failure time models with an interaction term between female judge and race as well as between female judge and gender to pick up on the relationship between those variables. We hypothesized that the interaction between female judge and gender would be significant and that it would predict higher survival for female contestants with a female judge present. There are clear patterns in these Kaplan-Meier curves, but AFT and Cox Proportional Hazards models would give a better way to quantify and assess them.
Table 1 below displays our results based on three different models for contestant survival by sex. These models include a controlled value for age as well as an interaction term between sex and whether or not a female judge was present in the season. Significance levels of coefficients are indicated by number of aster- isks.
of episodes per season tends to increase for later seasons.
Figure 4: Survival Curves from Weibull AFT Model for Gender
In Table 2 below, we used the same models as in the analysis of survival by sex to investigate survival by race. Similarly, these models include a controlled value for age and an interaction term between race and presence of a female judge.
Table 2: Modeling Survival Time by Race
Dependent variable: log(Percent of Season Survived) Weibull AFT Log-Normal AFT Cox Proportional Hazards (1) (2) (3)
Age 0. 076 ∗∗^0. 121 ∗∗∗^ − 0. 167 ∗∗ (0.035) (0.046) (0.072)
White 0. 102 0. 273 − 0. 214 (0.161) (0.204) (0.325)
Female judge present 0. 176 0. 452 ∗∗^ − 0. 341 (0.163) (0.206) (0.330)
White:Female judge present − 0. 095 − 0. 361 0. 147 (0.202) (0.258) (0.409)
Intercept 3. 255 ∗∗∗^2. 316 ∗∗∗ (0.412) (0.516)
Observations 115 115 115 R^2 0. Max. Possible R^2 0. Log Likelihood -525.652 -531.856 -424. χ^2 (df = 4) 5.183 10.403∗∗ Wald Test 5.760 (df = 4) LR Test 5.704 (df = 4) Score (Logrank) Test 5.780 (df = 4)
Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.
As was true in the models for survival by sex, we see that as age of the con- testants increases, so does their survival. However, we do not see the same significance in the interaction term for survival by race in the presence of a female judge as we do for survival by gender in the presence of a female judge. The survival curves pertaining to a female judge being on the panel (indicated by solid lines) in Figure 5 are consistent with this finding, as these curves are essentially identical for both white and non-white contestants. We will address why we hypothesize these results are not surprising in the next section of the paper.
for promoting more diverse mentors in the culinary industry at-large. In the future we would advocate for including more non-white judges on MasterChef Junior. While for one season there were occasionally guest judges of color, they only appeared for an episode or two at a time.
Our results are in line with qualitative studies of discrimination in the restau- rant industry. They are also in line with qualitative discussions specifically of MasterChef Junior. In our introduction we discussed issues of explosive tem- pers, lack of mentors, and questioning of abilities as well as a variety of other mechanisms that push minorities in the culinary industry out of positions of power. There are also claims that these cultural norms in the culinary industry are reflected in the way the judges on MasterChef Junior communicate with the young contestants. We have added to this conversation by using survival anal- ysis methods to back up popular claims. Previously, much of the quantitative analysis included only raw counts of salaries and employees.
We think there are a few reasons that we saw these particular results of our anal- ysis. Namely, as the show aired seasons 2 and 3, the producers started to receive more comments in popular media about the lack of female contestants reach- ing the final weeks of the competition. In part responding to these criticisms, Christina Tosi was hired to the panel of judges, and the initial pool of accepted contestants diversified. With the addition of Tosi to the panel, minority con- testants arguably had a more relatable mentor. With a more diverse contestant pool, minority contestants likely felt less singled-out and had a stronger sense of belonging. In fact, the breakdown of contestants on the show is quite a bit more diverse than the demographic breakdown of the industry as a whole, a good sign of changes to come.
5 Conclusion
In this paper we have shown that a female judge on MasterChef Junior pos- itively impacts the probability of survival for female contestants. It also pos- itively impacts the probability of survival for contestants of color to a lesser degree, largely because of issues with sample size. We found these results while controlling for age as a proxy for cooking experience.
If we continued this research we would think of a way to calculate some sort of ”power ranking“ for each contestant. This would give us a way to assess whether certain contestants win because they are just better cooks, not because of biases. One way to explain the results we got is that the white male con- testants may genuinely be better cooks for whatever reason. A power ranking would give us a way to control for this and ask a more refined question: at similar skill levels, do minority contestants perform better or worse than we would expect holding age constant? In a way, our model controls for skill by using age but that doesn’t completely answer the question we are asking. For
example, there are eight-year-olds who have been cooking since they were four and thirteen-year-olds who learned to cook when they were twelve. In cases like that, the advantage of age may be nullified or significantly decreased.
We also acknowledge that there are numerous limitations in answering our re- search questions. First, the number of weeks and contestants (correlated) is not the same for each season. The seasons tend to get longer as time goes on and the last season is about twice as long as the first. This makes it difficult to do calculations by week so we instead chose to use percent of season. Another limitation is that we had to guess the race of contestants, which is prone to assumptions and errors. We also don’t have data for the first couple of episodes for many of the seasons. Additionally, we don’t have data on people who au- ditioned, only those who were accepted. This means we don’t know how many minority contestants were accepted to begin with relative to the larger pool of applicants. Thus, our analysis reflects a bias towards people who survived long enough to get to the point where data is collected.
With these limitations in mind, we still think that our conclusions speak to an important point about visibility, diversity, and mentorship. Minority contes- tants on the show have measurably better outcomes with a more diverse judging panel. We also think this is a particularly interesting result because many of the children on the show have not yet been exposed to the biases in the culinary in- dustry the same way that the judges have. In this way the judges are even more responsible for modeling good behavior in the kitchen while the children look to them for advice and development of skills. It further demonstrates that while the contestants are developing cooking skills on the show, they are also devel- oping interpersonal skills that can later help them navigate the diversity issues they encounter. We hope our analysis can be a starting point for discussions like this in the future.
References
[1] Jen Agg. Sexism in the kitchen. Oct. 2012. url: https://www.nytimes. com/2012/10/20/opinion/sexism-in-the-kitchen.html. [2] Charlotte Druckman. Why Are There No Great Women Chefs? 2010. url: https://gastronomica.org/2010/02/05/why-are-there-no-great- women-chefs/. [3] Caroline Framke. ‘On MasterChef Junior, innate biases are hard to beat’. In: The Atlantic (Feb. 2015). url: https : / / www. theatlantic. com / entertainment/archive/2012/02/masterchef-junior/382079/. [4] John T Jost et al. The existence of implicit bias is beyond reasonable doubt: A refutation of ideological and methodological objections and exec- utive summary of ten studies that no manager should ignore. Nov. 2009.
