The Survivor Research Project: Theory and Methods

It's not like anyone needs another theory on the workings of the grandfather of blockbuster reality TV, Survivor. As it is, sociologists, anthropologists, management gurus, strategists, economists and game theorists have lent their expertise to various aspects of the game. There surely is no room for another theory. Well, mine isn't exactly a theory as such. It's more a cobbling together of various observations picked up from following Survivor-related discussion boards since Season 2. And unlike many existing analyses of the show, mine isn't directly about how to win it or what happens in the game.

The actual events of the game are only part of my "theory". The other, and more significant part of it, concerns the signalling by the show's editors. My hypothesis is not so much about who wins, who is booted and why. It's not about the social game of Survivor, but about the TV show, Survivor. It's about what we see as viewers and whether that reflects, post-ante, who wins and who gets booted.

The Hypothesis
How long a contestant lasts in the game = f (Screen time, Impression left on viewers, Likeability).

Screen time is the most important factor, but also the most delicate to model. Early in the season, a lot of screen time could be both good and bad. It could mean that the player is about to leave soon, and is being given air time before he/she gets the boot. Or it could mean that the player has an interesting "growth arc" that will be revealed in later episodes, signalling that he/she will likely go far. A player might also get a lot of screen-time because of a high likeability or attractiveness factor, giving the viewer someone to root for and hence, an incentive to keep watching.

In the middle part of the season, those who get sufficient but not overwhelming screen time are the most likely to progress further. In any individual episode, the person who gets the most air-time is usually the one voted off, or who just narrowly avoids the boot.

Towards the tail end of the season, the ones who appear on screen most often are likely to be the final 3.

Impression Left on Viewers is subjective, but an important complementing factor to Screen time. It's not just about how often you are seen, but how memorable you are to viewers. I think this is equally about the editors as the individual's personalities. After all, even the dullest person in the world would say a few mildly interesting things once in a while. The editors can choose to isolate only these moments for the show. It is in the interest of good TV that the successful players are interesting to viewers, even if it is in the sense of being annoying as hell (Johny Fairplay), a major asshole (Hatch and Heidink) or obnoxious (Jenna M).

Likeability is even more subjective. It works at two levels; within the game, likeable people are likely to go further (although being too likeable can get you voted off if others don't want to come up against you in the Final 2) and outside of the game, in the comfort of our living rooms, we want to root for likeable people. It's no good to have a bunch of people who are interesting in a negative way and are completely unlikeable. It's not pleasant viewing for us. In past seasons, I think the Survivor prdocers have always tried to "redeem" even the most unlikeable finalists with favourable editing, ostensibly to make them more palatable to viewers.


Testing the Hypothesis: The Data
Data and measurement are a big part of my day job. Perhaps it's an occupational hazard, but I am going to try to approach this systematically. What I propose to do is to monitor the three factors of Screen Time, Impression and Likeability for every episode of Survivor this season. At the end of it all, I should have enough data to actually attempt an empirical model of Survivor Success. I will collate data on these three factors for each player, as long as he/she lasts in the game.


Testing the Hypothesis: The Measurement Scale
It would be great to use objective measures, but entirely unpractical. I suppose that the best measure for screen time would be the actual minutes that each contestant is shown on air, but that would involve watching Survivor with a stop watch and a datasheet, and I am not that intense about this project! Therefore, I am going to use Likert Scale measures, based on my perceptions of what I saw. It's not perfect (and would not et through a review process in academic research) but this is Survivor, not the Management Science journal.

For Screen Time, the scale is between 0 to 5, where 0 means no significant screen time. Being shown in the background during the start of a challenge does not count as significant screen time. Neither does voting during tribal council, especially if the player's comments while voting are not aired until the end credits. 5 would mean massive screen-time, both in terms of having personal interviews as well as being shown around camp, during challenges or tribal council.

For Impression, the scale is also between 0 and 5. 0 is awarded when I cannot remember anything a player did or said during the episode.

Likeability is the most subjective of all, as it will be filtered through my personal preferences and prejudices. Again, the scale will be between 0 and 5. When the person is not shown enough for me to have formed any opinion, no score will be awarded. 0 would correspond to hatred levels ala my feelings towards Jenna M and her buddy Heidi. Unlike the other two factors, likeability is not unique to each episode. While a player may be shown very positively during one episode, prior bad behaviour will colour my opinions and therefore affect the score I award. This is a "flow" variable, while Impression and Screen Time are "stock" variables.

So there you have it, the outline of my Survivor Research Project. In the next post, I take on the fun task of reviewing Episodes 1 to 3 and awarding scores based on my measurement framework.