Saturday, August 14, 2010

Money Melts the Pounds Away -- an in-depth look at The Biggest Loser outcomes.

No, this is not veiled commentary on my social life.
Obesity is a growing issue in the industrialized world generally and the United States in particular. The science behind obesity is still developing, but the popular culture response has already begun. Reality television, which has previously attempted to resolve our lovelessness, joblessness, and lack of fashion has now begun to confront the ‘American lifestyle.’ 

Television network NBC’s hit reality series “The Biggest Loser” takes a group of overweight individuals and sequesters them in a large housing and gym facility. There, under the oversight of expert personal trainers and medical personnel, they attempt to lose weight as quickly as possible. Taking place within a competitive team setting, at the end of each week the people who lost the least weight risk being eliminated. At stake is $250,000 cash for the winning player and a $100,000 prize for the eliminated player losing the most weight by the finale.

The Biggest Loser's game-show world bears only loose relation to the reality of an average person looking to drop a few pounds. However, its dramatic format and inspiring message have proved a global success, with the creation of Biggest Loser UK, Biggest Loser Australia, and Biggest Winner Arab (to name a few). The American version of the show has produced nine seasons since 2004, with a tenth being filmed as of this writing. 
 
With each contestant’s weight loss announced weekly, this television series provides a wealth of data on weight loss under ideal and controlled conditions. With 6 to 8 daily hours of exercise, a rigidly structured diet organized by top-rate personal trainers along with a strong monetary incentive, participants on The Biggest Loser have every advantage in losing large amounts of weight. Some drop over twenty pounds, or over 5% of their body mass, in a single week. The vast majority go on to change their lives by adopting healthier eating habits and frequent exercise. These results demonstrate weight loss at the absolute limit of human capacity. 
 
While every participant on The Biggest Loser puts in a monumental effort to lose weight, there can only be one winner. Contestants experience different outcomes in weight loss in spite of a generally high standard of effort. There have been many debates in the bio-medical field on whether obesity is caused by genetics, culture, age, or something else entirely, and a consensus has yet to emerge. With its wide demographic variety, The Biggest Loser provides an opportunity to test how these differences impact optimal-scenario weight loss. 

 DATA:

The format of the Biggest Loser lends itself to data collection. The first half of each episode follows the contestants through a competitive challenge and their daily exercise. During the latter half, each person is weighed in and their amount is compared to the previous week's measurement. The grand prizes are based on a total percentage of weight lost, so as not to advantage larger contestants. For this study, a similar method was used to balance out starting weight differences between participants. Data was taken from the first nine Biggest Loser U.S. episodes, which featured a total of 158 people. A fuller explanation of the variables continues below. 
 
Sample RWL Calculation for Helen (Season 8)
Dependent Variable: Rate of Weight Loss (RWL) 

In order to compare outcomes across demographics, it was first necessary to condense each person's weight loss up to the finale into a single figure. For each week, every contestant's total weight loss up to that time was recorded. That number was divided by their base weight at the start of the season to arrive at their total percentage lost. Plotting all those points to a graph, then imposing a line of best fit results in a graph for each participant like the one seen here. The slope of the line is the average rate of weight loss (hereafter RWL) for that person across the time they were on The Biggest Loser. In the example above, the slope is 2.309. This is interpreted to mean that Helen (our sample contestant) lost an average of 2.309% of her base weight each week she was competing on The Biggest Loser. She entered weighing 257 pounds, and on average she lost 5.9 pounds each week until the finale. 
 
Independent Variables:

Age: The Biggest Loser draws from every age group, with contestants from the age of eighteen up to over sixty. Age can play two conflicting roles. The first factor is maturity; with age comes experience, determination and mental resilience. A mature person will be more likely to conquer the mental and physical strain of rapidly losing weight. The second factor is infirmity; older people are more likely to enter the contest with pre-existing injuries and are also slower to recover if they are hurt while exercising. While minor injuries are common for nearly all Biggest Loser contestants, being young and healing rapidly is an advantage that plays against maturity. 
 
Base Weight: While all Biggest Loser contestants are “overweight”, both their degree of obesity and general body types vary greatly. Heavier contestants have more weight to lose, but may also be in worse physical shape to start, which slows exercise. It's expected that a higher base weight will have a negative effect on rate of weight loss.

Height: Taller overweight people have two factors working against them. First, having a larger frame increases strain on joints and muscles, increasing the risk of injury. Second, tall people have more bone and muscle mass, making it more difficult to lose a high percentage of weight. Shorter Biggest Loser contestants are expected to have an advantage.

Gender: The effects of male-female differences are difficult to predict, because sex is interconnected with many other demographic variables. Men are on average taller, and generally enter the show with a higher base weight than women. Additionally, many of the male contestants were former athletes (wrestling and football being to most common) so they have more endurance to start. Based on the sample, it's expected that men will have an advantage in rate of weight loss.

Ethnic Background: Contestants were divided between four categories: Black, Hispanic, Asian, or none of the above (“white”). There are limitations to this approach; ethnic differences are clearly more subtle than these categories encompass. Many of the people considered “white” were of Italian origin but American born. The “Asian” category was geographically broad; Heba from Season 5, who was of Egyptian heritage, was classified as Asian. As a result, the results of ethnic background on weight loss are unpredictable. 
 
Marital Status: Contestants with spouses have two contradictory factors at play. Having someone for emotional support during many weeks away from home is a huge advantage. Some seasons of The Biggest Loser were “Couples” themed, and some married couples worked together to lose weight. The majority of married participants were separated from their spouses, however. Missing home hurt the morale of some contestants and made them more willing to leave The Biggest Loser campus. 
 
Kids: Many of the married contestants also had children at home. The number of children is expected to have an ambiguous effect on weight loss. However, it is most likely correlated with age, as older people have had more years to produce children. 
 
Weeks on “Campus”: Perhaps the ultimate advantage in losing weight is more time with the trainers and specialized diet provided to Biggest Loser contestants. More weeks on campus is expected to correlate strongly with a high rate of weight loss. At the very least, people who stay in the longest are also the ones who have the highest, most consistent rates of weight loss.

RESULTS TABLE: 


Irrelevant Variables: 
 
Race: varying ethnic origins did not have a measurable impact on rate of weight loss. The majority of participants were white, and some ethnic groups (e.g. Asians) had very low representation, so statistical significance was hard to achieve.

Northwest, Southwest, Northeast: three of the location dummy variables did not have any statistically significant difference from the base case (the generic “Midwest”). 
 
Season: not discussed much before, the season did not strongly impact weight loss. 
 
Spouse: being married did not make a measurable difference. The contradictory factors at play for a married person canceled out, leaving the variable statistically irrelevant.

Relevant Variables: (in order of greatest statistical significance)

Weeks: as expected, staying on the show longer was a sign of faster weight loss. The extremely high t-stat for this variable shows that more weeks on campus corresponds almost exactly with higher rates of weight loss, other factors being constant.

Male: men had a substantial advantage in losing weight quickly. Anecdotal observation supports this – the majority of seasons were won by men, and most of the finalists were also men. Women did better in the later seasons, but men in general had the advantage.

Southeast: for whatever reason, people hailing from the southeast part of the country had a lower rate of weight loss. None of the other regional variables had a measurable impact, which makes this result especially surprising.

Age, Age2: as anticipated, an older and more mature person has an advantage, but at a certain point, the indignities of age start to impede weight loss. 
 
Base Weight: a higher base weight correlates to lower rates of weight loss, other factors being held constant. While intuitively, it would seem that heavier players have more to lose and should drop it quickly, that was not evident in the data. Being in worse physical shape and having less stamina for sustained exercise was a disadvantage for heavier contestants.

Height: taller participants did not lose weight as quickly as shorter ones. Other factors being constant, a lower stature was an advantage.

CONCLUSION:

Combining all these factors together, we can generate a profile of the “perfect” Biggest Loser participant. Such a person would be: male, middle-aged, below average in height, and lack a Southern accent. Why? I'm not sure, ask a sociologist. My economics work here is done. 

Potential Flaws: 
  • Calculating the RWL requires knowing how long between the last episode of the show and the finale. Contestants are sent home for between two and four months (depending on the season) and the exact amount is not always clear. For three of the seasons, it was necessary to extrapolate the number of weeks before the finale based on the seasons prior and following. 
  • Some participants chose not to attend the finale, so their final weight loss totals are unknown. The number of contestants not appearing was less than five between all nine seasons, but calculating a rate of weight loss based on a different span of time would have introduced serious inconsistencies into the model. To compensate, the weight loss at their last episode in the game was recorded for absentee players as their finale weigh-in total. 
 
Questions, comments, or suggestions for more analysis are welcome.

1 comment: