Methods: Aggregate Data on Race, Gender, and Placement Through 30 Seasons
In order to assess whether or not certain racial or gender categories have immediate advantages or disadvantages, I compiled the name, race, gender, and “boot number” (the place in which they were voted out) of all 535 contestant to play in the first 30 seasons. These data did not exist prior to my research, so I obtained a list of every contestant, and then sorted it by race and gender. While I did so as meticulously as possible, there is some potential for error in this method. There are some people whose ethnicity is ambiguous, and for whom I used my own judgment based on researching the contestant as well as any experiences they have shared on the show that reveal their ethnicity. There were not so many cases of this that the margin of error would likely impact the data in any meaningful way.
For all of the following data I have left contestants who quit or were medically evacuated in the data, because social factors often lead to quits. For example, it is more likely that a contestant will want to quit if he or she feels unwelcome or disliked by the tribe, or if they feel they have no chance of winning, which makes it possible that quits are influenced by the same factors that cause a contestant to be voted out. Quits and medevacs are such a small portion of overall eliminations that this makes very little difference in the results. I counted these contestants’ exits as if they were voted out normally in the place in which they left. Finally, I have also counted each time a contestant plays as entirely separate. If a player has played on three different seasons, I count each season as completely individual. While a player’s returning status does influence his or her game, there is no reason to believe that playing before would influence the relevant factors. Overall, the 535 total refers to games played, as opposed to individual contestants.
Data: Placement in the Game by Race and Gender
The categories are divided by race, and then further broken down into gender as a subset of race. However, due to severe lack of representation for Hispanic/Latinx and Asian contestants, those categories are not broken down by gender. This is because breaking them down by gender would provide a uselessly small dataset for some categories. For example, there are only 8 identified Hispanic/Latinx male contestants, and three of them are the same man who played thrice. Therefore, these categories are taken as a whole to show a more accurate depiction of how their race influences their games, as breaking down by gender would not give a useful representation of how those groups fare.
Table A: Average Place in Which Contestants are Eliminated
|Demographic||Average boot number|
|Women of Color||7.87|
|Men of Color||9.85|
|People of Color||8.75|
From Table A, it appears that race and gender do not significantly determine a contestant’s overall performance in the game. Overall, every group performs within two places of the expected value of 9.52. However, the group that has the lowest average placement is black women, followed by women of color. This table also provides insight into the importance of the intersection of race and gender. While black men and men of color actually perform slightly higher than the average, people of color and black people as a whole are taken down an entire place number due to the lower success rate of black women and women of color. Women as a whole do very moderately worse than average, but white women fare the best out of all women. The success of black men/men of color and white women as compared to black women shows that obstacles in Survivor are not tied to race or gender, but rather the intersection of race and gender, which creates unique challenges for women of color in the game. Table B below further illustrates these challenges faced most by women of color.
Table B: Contestants Who Placed in the Bottom 5
|Demographic||% Voted off within the first five boots||Number voted off within first five boots||Number of contestants overall|
|Women of Color||45%||31||69|
|Men of Color||26%||14||54|
|People of Color||37%||45||123|
Table B shows the number and percentage of contestants belonging to different racial and gender categories that were voted off within the first five boots. I chose five somewhat arbitrarily, however, it is an appropriate number because the first five boots are still within the beginning phase of the game, when the merge and jury phase are still far enough away that most votes, overall, are likely based on personalities and challenge performance. Due to personality and ability being the most common reasons for elimination, it is reasonable to conclude that this is the phase of the game in which people are being voted out for who they are, rather than for how they play, much more so than later phases of the game where the votes are based mostly off of strategy. Studying the first five boots most closely reflects my interest in determining whether or not race and gender can give an inherent and immediate disadvantage in the game. The expected value for any given group should be close to 28% – out of the 535 contestants, 150 have been in this group. It is important to note the groups that fall near or below this average: Men overall, black men, white women, white men, white people overall, and men of color all fall very close to and below this value, and women as a whole are above but still close to this average at 32%. This demonstrates that all categories of men, and all categories of white people are most likely to survive the first five eliminations. However, while women are only slightly more likely to be eliminated in this phase than would be expected, they are still 9% less likely to survive this phase than men are. An interesting observation I found during my research, however, is that while women comprise 57% of all first five boots, when the data are reduced to the first four boots, women jump to 65%, which is more indicative of women’s immediate disadvantage than this table shows. Women were the only group in which the difference between first four and five boots was noteworthy.
The groups who fall distinctly above the average in terms of their likelihood to be eliminated in this phase are black people, Hispanic/Latinx people, women of color, people of color, and most significantly, black women, who are exactly twice as likely to be eliminated in the first five Tribal Councils than the expected value of 28%. This demonstrates that all people of color are less likely to survive the first five tribal councils, with the exception of Asians. Even more telling, this means that black women make up 12% of all first five boots, despite being only 6% of all contestants. Similarly, women of color comprise 20% of all first five boots, while only making up 13% of contestants.
Table C: Contestants Who Placed in the Top 5
|Demographic||% Who made it into the top 5||Number who made the top 5||Number of contestants overall|
|Women of Color||22%||15||69|
|Men of Color||30%||16||54|
|People of Color||25%||31||123|
Table C demonstrates the percentage of those who made it to the top five in their respective seasons from the same demographics. Although contestants who made it to the finals were never actually eliminated, I counted their placement as if they were voted out in that spot. For example, if a contestant came in first on a 16-person season, I counted them as the 16th boot. Because this affects the data in that in 20-contestant seasons the average placement is 10, whereas in a 16-contestant season it is 8, I have averaged all of the seasons to find that the overall average placement of all contestants is 9.52, against which I will compare the average placement of each group. Once again, the expected value is 28%. Again, while most groups hover near the average with no notable deviation, black women and women of color fall 9% and 8% below the average respectively, demonstrating the barriers to women of color faring as well as a group as their fellow competitors. One of the most notable changes between tables B and C is how the gap between Hispanic/Latinx contestants and Asian contestants completely closes by the end of the game, showing that the exact same proportions make it to the end of the game. One thing the table doesn’t show, however, is one of the few times in which counting all returning players as individual games clouds some facts. Specifically, of the 8 Hispanic/Latinx contestants who made it to the top 5, two of them are two-time winner Sandra Diaz-Twine, and two of them are Ozzy Lusth, who advanced far in two seasons. This means that Asian and Hispanic/Latinx contestants may not actually fare equally in Survivor if there were more data, but rather that Diaz-Twine and Lusth are simply unusually good at the game. As far as I can discern, this is the only instance in which counting returning players as entirely separate makes a noteworthy difference in the data.
Table C also shows a somewhat surprising development: Black men and men of color actually fare the best (though by an extremely slim margin) in making the top 5. However, these data do not tell the whole story. A significant proportion of black people and people of color come from two unique seasons, Survivor: Cook Islands and Survivor: Fiji, which each have an unusual relationship with race. Cook Islands, the show’s thirteenth season, introduced a twist in which players were initially divided into four tribes, based on their race. The tribes, divided into White, Latinx/Hispanic, Asian, and Black, were dissolved rather quickly and mixed into two tribes, and the most prominent lasting impact of the racial divide was the equal representation it provided, with five contestants belonging to each category represented. This was the first time the show has come remotely close to equal representation. Fiji, the season immediately following Cook Islands, also featured very close to equal representation. Based on the racial composition of the cast, it is likely that Fiji was also meant to be a season initially divided by race, but a woman quit immediately before filming and it was too late to replace her. This left the season with only 19 people, and therefore dividing into even tribes based on race was impossible, as one tribe would have been at a numbers disadvantage, so the race twist was not carried out again, but the near-equal representation it provided remained. Without Cook Islands and Fiji, the number of top 5 contestants who are black drops from 14 to 10, the number of top 5 contestants of color drops from 29 to 21, and the number of men of color winners drops to zero, as these are the only two seasons to be won by men of color. Most significantly, the number of each Asian and Hispanic/Latinx contestants without these two seasons drops from 28 to 18, meaning that over one third of all Asian and of all Hispanic/Latinx contestants are in just two seasons, showing the extreme difference between these two seasons and all others.
There is, of course, no evidence that Yul Kwon, the winner of Cook Islands, who is Asian, and Earl Cole, the winner of Fiji, who is black, would not have won on any other season they played. However, given that they are the only two men of color winners on the only two seasons with equal or near-equal representation, it stands to reason that these seasons are more conducive to people of color faring well, potentially for reasons more complicated than the obvious fact that more people of color on a season means a higher chance of a person of color winning, which I will explain later in this paper. However, it is interesting that despite Cook Islands and Fiji benefitting men of color in terms of overall placement and winners, the equal representation did not lift black women and women of color in making it to the top five in the same way, as they still remain the two lowest groups. This further demonstrates that while the evidence does not support that Survivor has a significant race or gender problem, it likely has a problem when it comes to people who face challenges based on race and gender. Despite this, disparities between all groups are smaller in Table C than they are in Table B, demonstrating that the most significant disadvantages women of color face at the beginning of the game. The effects of Cook Islands and Fiji will be further discussed later in this paper.
Another explanation for the convergence of many categories in the top 5 is that after the merge, the people who are more likely to win are more likely to be voted out. Those who are the most likeable, athletic, and respected are the biggest threats to win the game, and therefore are targeted during the individual portion of the game in a way that they were not before the merge when they were assets to the tribe. The much more even distribution at the end of the game than at the beginning demonstrates that the positive traits that allow contestants to easily survive the pre-merge, make them targets at the end. However, some groups, such as black women, still have low rates due to the overwhelming number of black women voted out in the pre-merge, denying them the opportunity to do well in the post-merge where they are less likely to be targeted.
These tables also show how many people from each demographic have been represented on Survivor. White people make up the overwhelming majority of contestants, representing over 75% of contestants, with white men comprising 40% of overall contestants, despite being 31% of the US population, demonstrating an overrepresentation of that group. Black men average one per season, while Hispanic/Latinx and Asian contestants of each average one per season of any gender. These data do not specifically describe winners of color, since there are too few to permit statistical conclusions, however they are certainly important to discuss. Six seasons have been won by people of color, two of them by the same woman, Sandra Diaz-Twine, a Hispanic woman who is the only contestant to ever win the game twice. Apart from Diaz-Twine, there have been two female winners of color, Vecepia Towry, a black woman, and Natalie Anderson, originally from Sri Lanka. As previously discussed, there are two men of color winners, and Kwon and Cole’s wins are unique in that they were during equal-representation seasons. One theory is that contestants’ ethnicities may be proportional to the viewership of Survivor, as casting may wish to reflect its audience. However, there are no public, accessible data on the race or gender breakdown of the show’s audience, though such data likely exists.