Realpolitik

I have yet to see anyone get to the end of the game without having to tell a few lies. I’m not sure it can be done. Distasteful as it may be, that’s just part of playing the game in Survivor. And many of the most successful players are those who can lie convincingly, and who know when its time to betray an alliance. You can win playing a relatively ethical game (see seasons 10 and 13 for example), but I haven’t seen anybody win while being totally honest. As the great poets of Depeche Mode put it:

You better learn your lesson well:
hide what you have to hide
and tell what you have to tell.

You’ll see your problems multiplied
if you continually decide
to faithfully pursue
the policy of truth

That being said, I’ve seen juries punish players who go out of their way to be duplicitous for no reason, and who try to be everything to everyone. If you’re the swing voter, don’t promise each side that you’re rock solid; try to hedge your bets. There’s nothing wrong with saying “I’m still thinking about who I’m going to vote for”; players should use that approach more. Then you get less blow back after you have to vote against people. You can avoid outright lies and still avoid “faithfully pursuing the policy of truth.”

The data bears this out. Look how many positions ahead the Machiavellian players are! About 6 places ahead on average. The box widths in the boxplots² above are proportional to the number of observations within them, so you can see that players are generally pretty honest.³ Perhaps too honest.

Find Someone to Trust

I haven’t been tracking this, but players in Survivor don’t usually make it far without an alliance, and an alliance usually has two or three people at its core. If you can find someone to trust (preferable two), you’re much more likely to go far.

Play Nice

So often, people are voted out simply because the other players find them obnoxious or abrasive. And even if the annoying players make it to the final two or three, often the jury votes pettily and will punish them simply because they don’t like them. Play nice and don’t be defensive and you’re more likely to go far.

So far the data does not really support this hunch. Looking at the violin plot⁴ above, you’ll see that abrasive players are going slightly further than the pleasant ones.⁵ Abrasive players last on average an extra position than pleasant ones. We’ll see if this trend holds up, doesn’t make much sense to me. Generally, most players are pleasant; the width of the violin plot figures are proportional to the number of observations within them.

Limit Your Visibility

The worst thing you can do in Survivor, besides annoying people, is to be considered a threat. You have to mitigate your threat level. You do that by not being an obvious leader, and in challenges–avoid taking a prominent role. Even if you are a leader, make it seem to your group like you’re all co-equal partners. Leaders often get picked off.

Once again, the data so far does not really bear my supposition out when you look at the ridgeline plot above. Leaders generally last an extra position compared to followers.⁶ We’ll see if that’s just small sample size theater. The distributions are slightly different too. Leaders are more likely to be voted out earlier, but if they pass that hurdle, generally manage to get pretty far, further than the average follower.

Work Hard

Players who are lazy around camp generate a lot of resentment.⁷ Often, especially in the early stages of the game, the vote can be heavily influenced by voting out those who don’t contribute. Gather some wood, go fishing!

Sigh. Looking at the density plot⁸ above, it’s clear that once again the data says the exact opposite of what I’m preaching. But I think I could plausibly explain this one away. One common strategy is to vote out the “threats” at the beginning of the game; these threats tend to be the athletic men, who also tend to be the ones who have the energy and strength to help around camp more. So maybe the hard-working ones are getting voted out earlier not because they’re hard-working but because they happen to be threats too? That’s what I’m going with.

When I showed these results to my brother, he pointed out something that may help explain the counter-intuitive findings. I’m not really accounting here for the fact that often weaker survivor players last precisely because they are weak. That is, stronger players want to bring weaker players to the final rounds with them because they know they will win. So, when the data shows that more abrasive and lazy players are going further in the game, that is not because it is a winning strategy, but because they’re being brought along to the final rounds as goats to the slaughter.

Anyway, so much for my theories; not let’s look a demographic factors. I have no theoretical expectations for these, just curious to see what the data shows.

Gender

Only a very slight difference between men and women in terms of likelihood of advancing in the show according to the plots shown above. Women last about 0.3 positions longer on average than men, but also win the whole show less (at least according to the seasons I have watched, where all the winners have been men). You can see this tendency in the graphic; look at the bulge in the women’s boxplot between positions 15-17. That’s the most likely spot for women to reach, and they get there more than men, but don’t go on to positions 18-20. I put this down to the physical nature of the immunity challenges. In the seasons I have watched, the final immunity challenge usually is a physical task that is geared more heavily towards men (because it prioritizes physical strength or athleticism), and thus the final immunity challenge is always won by men. These men then go on to pick the person they are most likely to beat in a jury vote, and so they typically win.

Race/Ethnicity

In the density plots above, I’m mixing up race and ethnicity just for the sake of convenience. Obviously a person can be Black and Latino at the same time in real life (or White and Latino), but not in my graphic! Contestants are settled into discrete categories purely based on my perception. And based on those categories, Asians are killing it in Survivor. In the seasons I have watched, no Asian player has even been voted out earlier than sixth. If you’re wondering what’s happening with the Latino density plot, that smaller second peak represents an outlier by the name of…well, I don’t want to give away spoilers, but let’s just say his name is geographic. A legendary player.

Here is how far (i.e. what position they are voted out) a generic player of each race/ethnicity lasts:

Yep, the Asian players are about 4 positions ahead of the next group. The dots are weighted by the number of players. Latinos might be doing better if they let more of them onto the show!

Physical Attractiveness

Like with anything in life, it helps to be good looking in Survivor. You can see it’s pretty rare for a plain Jane or Shane to make it past round 12.

Obviously this is a highly subjective score; one derived by combining the assessment of my wife’s score with my own of each Survivor contestant. But at least there is both male and female input on the contestants, right?

Age

Age is just a number in Survivor. Not much of a relationship. You can see that by the relatively flat prediction line. Note that in the graphic above I’ve switched the “rounds lasted” to the y-axis to make it easier to interpret the regression line. To be more precise with what the regression line is showing, if a contestant is one year older than another contestant, they are only likely to lose 2 percent of a place in terms of positions. So not a lot. It does make a modest difference in the extremes of age though. A 70 year old will place one position lower on average than a 20 year old. At least according to the data I have now; it’s being heavily skewed by Scout, an older player who went far. We’ll have to see if it holds up if I get more geriatrics in my sample. You would think much older players would do worse because of the physical challenges; I expect we will see that. So far though, no difference when it comes to being 20 or 40, for example.

Occupation

I’ve arranged these in descending order of winningness. Clearly it pays to be in construction! And it hurts you to be an athlete? A bit strange that. How to explain…hmm, maybe the construction workers are just tougher and more helpful around camp? People trust them more? The athletes are over-confident? We’ll need more data…two brilliant construction workers are all it takes to mess this data up.

Full Model

So which factors matter? Below, you can see the output of my kitchen sink regression model. Just throwing all the variables together into an ordinary least squares regression model (although with robust standard errors to guard against violating assumptions of homoskedasticity).⁹ Not the type of model I’d use with my real work; I prefer using computational methods like matching with observed data (which I’m not doing here because of a small sample size and, frankly, convenience), but it is useful just to see what stands out.

To interpret the table, just look at the size of the coefficients in the “estimate” column. These are the number of positions moved based one each variable. Most of them are binary variables, so it’s just if you meet that qualification or not. For example, if you are Machiavellian, you can expect to place 7.58 positions ahead of someone who is not. Attractiveness is on a 10 point scale. Being 1 point more attractive than a competitor gives you a 0.71 position edge. Not huge, but obviously being 5 points more attractive is going to be 5*0.71, which is 3.55 positions ahead, which is a decent boost. The smallest is age; 1 additional year only gives you a 0.02 position boost. It may be easier to see graphically:

These are arranged from the largest effect size to the smallest. Machiavellianism is clearly the key trait; both substantively and significantly (it’s the only one where the confidence interval does not include 0, meaning we can be confident that there is actually an effect beyond mere chance). The variable that matters least is age.

So far, my analysis has been focused on merely advancing in the game, not winning per se. It may be useful to consider the statistical profile of the actual winners.

Think of these as percentages. So, of the four seasons included in this analysis, the winners are: never women, usually white (75% of the time), always pleasant, never lazy, of above average personal attractiveness, somewhat Machiavellian (half of winners qualify), and generally leaders (75% of the time). See, being pleasant and hard working does matter!

Now onto some other thoughts about Survivor.

Season 10: Vanuatu

Lots of great story lines in this season: one tribe losing so many immunity challenges, the relationship arc between Ian and Tom, two people quitting, and a very worthy winner.¹⁰

Season 28: Cagayan

This was the first full season I watched after Season 1, after a decades-long hiatus. My sister-in-law made us watch it while visiting us in Morocco. Skeptical at first, I was quickly sucked in to the action. So many annoying characters to root against, and Tony in the middle master-minding it all but also growing increasingly paranoid episode after episode.

The Brains vs Brawn vs Beauty gimmick did not really work, primarily because the Brains tribe was the most incompetent tribe I have ever seen. They confused intelligence with occupation; the Brains contestants had fancy jobs, but showed an incredible lack of common sense or basic competence. It made for good television though. In one memorable episode, the Brains tribe lost a challenge even though the other tribe was intentionally trying to lose.

Season 9: Vanuatu

I love this season though it does have an unworthy winner, which leaves a bad taste in the mouth. The Battle of the Sexes premise made for interesting viewing, especially given that they adjusted the challenges so that they were not purely physical. I wish more seasons had challenges that did not rely so much on physical fitness or the ability to swim fast.

Season 13: Cook Islands

The premise promised to be interesting (Black, White, Latino, & Asian) but did not quite live up to what it could have been. It also featured a lot of personal attacks, which I hate to see. When it gets personal, I don’t enjoy watching.

Tony: Season 28

Tony just made season 28 so interesting with his combination of manipulation and paranoia. He even built a spying shack where he could spy on other contestants! His total amoral play makes me a little nervous to realize his profession is policing, but hopefully it does not extend to real life!

Steph: Season 10

Steph was a compelling figure, not because she was so interesting personally, but because of the incredible adversity she faced on the island. Watching her team (through no fault of her own) lose every single immunity challenge until she was alone on the island was tough. You couldn’t help root for her. She was incredibly resilient and tough. That being said, she is an unusual player for me to like since she really just played a straight game without much strategizing (until forced to). Just mainly got by on her pleasantness, general competence, and authenticity.

Kass: Season 28

“Chaos Kass”, and very aptly named at that. Seemed to take a perverse pleasure in blindsiding her team-mates. Made for very entertaining tribal councils. There is nothing I like better in Survivor than a good blind side.

Twila: Season 9

One thing I really value in Survivor is authenticity, and Twila has that in spades. In fact, she probably should have toned it down somewhat. She also grew as player a lot in the course of the game. She turned from a player other players openly talked about manipulating, to ending up being one of the leaders of the game.

Cao Boi: Season 13

A divisive character, but again, I value authenticity, and for good and ill, Cowboy was authentic. He was authentic to the detriment of his game, but as a viewer I appreciated it. He added some novelty to the game too, as he was doing things like basing strategy on dreams.

He did bring one definite skill – he could cure migraines by rubbing a person’s temple and then sharply pulling the skin between the eyes. It left an angry red welt, but hey, it appeared to cure the migraine.

Adam: Season 13

My main problem with Adam is that he and Candice made the game so personal. Not a lot of other redeeming traits either coming from the screen. Appeared whiny, entitled, and lazy. Had this annoying smirk too.

Candice: Season 13

Again, I don’t like contestants who are so petty and bitter towards the other players. In that way the relationship between Candice and Adam seemed fitting; two very negative people should pair up. Apparently she ends up as a “hero” on a later Survivor Season called Heroes and Villains. Not sure how she earned that designation.

Kim: Season 10

A potent mix of supercilious, lazy, and incompetent. I gave a huge sigh of relief when she was finally voted out. For such a poor player, she got an inordinate amount of screen time. Perhaps the producers knew she was the right kind of player to provoke a reaction out of the audience? It worked. She’d give interviews to the camera where she talked about being the brains of the team, and how it was so unfair that she was being singled out for being lazy when she was “physically incapable” of helping out around camp.

Ami: Season 9

Perhaps it is a bit harsh to put Ami on this list. Unlike Candice, Adam, and Kim, she was an excellent player. A natural leader, crafty, and played with a surprising amount of integrity. My main gripe with her was how condescending I felt she was with the other players.

Spencer: Season 28

Again, this is harsh because Spencer was a good player; I just found him personally annoying. Mostly his arrogance and his gate-keeping. He kept lecturing the other players about what a worthy survivor winner is as if there aren’t multiple ways to play the game.

You’ll notice all of the above pics are taken of the players while they were on the show. That’s another lesson from Survivor: players always look their best after they’ve spent some weeks on the island roughing it. You get tanned and lean, the men grow some rugged facial hair, and the women aren’t caked in make up. It’s ironic, but when the contestants reunite for the final episode in America, where they’ve had a lot of time to prepare their looks, they usually look much worse than the more natural look they sport on the island.