The Bowdoin Orient

Let’s talk about bottled water.

First of all, bottled water is significantly more expensive than tap water. A one-liter bottle of Poland Spring at the C-Store costs $1.31. A one-liter Smartwater costs $3.52, but a liter of tap water is essentially free (or $0.0009 per liter in Brunswick and Topsham, if you want to get technical). 

Second, bottled water is bad for the environment. We have turned parts of the ocean into giant plastic soups by improperly disposing of our plastic products. Students that buy bottled water unavoidably enlarge their carbon footprints, even if they recycle every single bottle they drink.
Tap water is also more likely to be clean and free of contaminants. In some areas of the world, it’s necessary to drink bottled water because the tap water isn’t safe to drink. Brunswick is not one of those areas. 

Tap water is not only as clean as bottled water, but it’s also more stringently regulated. The Food and Drug Administration (FDA) sets safety standards for all bottled water, while the EPA, state, and regional municipalities work together to create tap water quality standards. 

The New York Times reported in 2009 that the FDA does not require bottled water companies to provide consumers information on what contaminants their water contains, where the water was sourced from or how it was treated. By contrast, the EPA mandates that all municipalities report the source of their tap water and any evidence of contaminants to their residents. 

Erin McAuliffe wrote in a 2009 Orient article that though Maine’s water is the nation’s third most contaminated, the tap water supplied by the Brunswick Topsham Water District (BTWD) is comparatively clean. According to the latest annual report by the BTWD, the concentrations of all potentially hazardous materials found in our tap water were all low enough to pose no health risk. 

However, there are differences in the cleanliness of different sources of tap water on campus. Several water fountains—those snazzy zero-touch water bottle fill stations and the water dispensers in the soda machines in the dining halls—have reverse osmosis filtration systems. The bathroom sinks and the other water fountains do not. 

Director of Facilities Operations and Maintenance Ted Stam could not say whether filtered water was significantly safer than non-filtered water, adding that the choice to drink one or the other should depend on taste preference.

“Some people like having the natural minerals and their taste in water,” he said. “Others prefer having the minerals filtered out and the absence of the taste.”

On November 19, members of  the Green Bowdoin Alliance conducted a blind taste test at Smith union to investigate if students could tell the difference between bottled water, filtered water and tap water from the bathroom sink. 

After conducting statistical work with math professor Jack O’Brien, I found that although people could tell the difference between water types, they had no significant preference for any one type.

Assuming our 70 samples were collected randomly, we can generalize that Bowdoin students don’t prefer bottled water to tap water.

Some people actually believe that Smartwater’s electrolytes make you smarter. Smartwater only contains “electrolytes added for taste,” as printed on every single Smartwater bottle. I couldn’t find the concentrations of Smartwater’s electrolytes on the Internet, so I emailed customer relations at Coca Cola, the brand’s owner.

They responded that the “exact amounts of minerals in Smartwater is proprietary information to our company,” and assured me that their products’  electrolyte concentrations “are insignificant from a nutritive value point of view.” It’s not a sports drink—it’s a slightly salty bottled water. 

Obviously, you have the right to buy bottled water. I only want you to justify your decision: Buy that bottle of water at the C-Store if your preference for its subtle taste outweighs its burden to your wallet and to the already dirty planet.

Statistical Analysis

Prepare for scientific speak. Each student sampled bottled, filtered, and unfiltered water kept at equal temperatures. The students were then asked which sample was which type of water. Table 1 breaks down how the subjects faired in the guessing part of the taste test. Students were then asked to rank the taste of the three samples. Table 2 details how the subjects ranked each type of water. Afterwards the identity of each type of water was revealed. Using data on the subjects’ preferences and guesses, I aimed to answer two questions with Professor O’Brien.

Are there a significant number of wrong guesses, suggesting people cannot tell the difference between the types of water? To answer this question, I set up a null and alternate hypothesis for a statistical test. Under the null, students cannot tell the difference between each type of water. Under the alternate, students can tell the difference. To figure out which hypothesis agrees with the data, I used the Fisher’s Exact Test, which is extremely similar to the Chi-Squared Test. A Fisher’s test calculates the expected values of the data under the null, and then calculates the probability that our data would occur under the null. We reject the null if the probability of our data is lower than an arbitrary number. Many scientists, as well my Bio 1109 lab instructor, Pam Bryer, require the p value to be less than 0.05 to reject the null. In other words, Pam would only reject the null if her data occurs 5% of the time under the null. For example, after achieving 95% confidence that the Higgs-Boson particle exists, those scientists poured billions into achieving 5sigma confidence, equivalent to achieving a p-value lower than 2.85 x 10^-7. For the sake of my data, I’ll follow the scientific standard of 0.05.  For the guessing question, the Fisher test spat out a p-value of 0.0002, which means the probability that our data would occur under the null is 0.0002. Given 0.0002 is less than 0.05, I can say with 95% confidence that students can tell the difference between the different types of water.

Table 1. Shows the percentage of people who guessed a type of water after given a type of water. For example, 27% of subjects given bottled water guessed it was filtered water. Subjects guess correctly significantly more often than incorrectly (Fisher, p = 0.0002).

Do students consistently rank one type of water over another? Again, we must set up a null and alternate hypothesis. Under the null, students do not prefer one type of water to another. Under the alternate, students prefer a type of water over another. I used another Fisher Test, which spat out the p-value of 0.2. Though low, this probability is not low enough for Pam’s statistical standards. Since our p-value is greater than 0.05, we cannot reject the null, so we conclude that our subjects have no preference for one type of water over another.

Table 2.  Shows the percentage of subjects who ranked a given type of water as their favorite (1), their least favorite (3), or somewhere in between (2). Some subjects could not rank any type as their favorite or least favorite, so they ranked all types as 2. Subjects did not significantly favor one type of water over another (Fisher, p = 0.2).

A Statistical Generalization. Assuming our samples were collected randomly, which is to say that the people participating in our taste test were essentially drawn from a hat containing every Bowdoin student, the 70 subjects who participated in the blind taste test sufficiently represent the Bowdoin student body. This means that the conclusions from these Fisher Tests can be generalized to the Bowdoin community. In other words, Bowdoin students can tell the difference between these types of water, but they do not prefer one over another.

Michael Butler is a member of the Class of 2017.