Cola Wars Redux

Monday, June 8th, 02009 at 01:10 UTC

It has been almost 20 years since the Cola Wars raged on our TVs. My ill-be-gotten youth was spent watching TV cola commercials and seeing the casualties of each side pile-up night after night. In hindsight, it seems as if no one challenged any of the outstanding claims which were being made by either side—if they did, the general public never heard them.

I have always sided with Coca-cola, never been much of a Pepsi drinker, but then again, I’m just a fare-weather fan. I never picked-up arms for one side or another. I consider myself more of a Switzerland of Cola refuge, supporting the little guys as well as the knock-off brands. In fact, while I have never created or tasted OpenCola, I strongly support the idea.

So I wondered, all those “sip-tests” over the years that showed that people enjoyed the taste of Pepsi over Coke, were those right? Could I personally tell a difference? In Malcolm Gladwell’s book Blink, he supposes that the sweater, more sugary taste of Pepsi was what swayed people’s decisions at that moment. Therefore, selecting Pepsi over Coke based on a single sip.

This is where the conspiracy theories start to emerge. Coke was losing to Pepsi in these tests, so they needed a way to win against this very specific test criteria. To retaliate, they unveiled New Coke, which was even sweater than Pepsi. People took one sip and loved it. The problem is that we don’t ever just take one sip of cola, we usually drink the whole glass, and New Coke was so sweat it was undrinkable. It passed the very specific sip test, while ruining the rest of the experience. On paper, New Coke was a hit, in the sip test it out performed Pepsi, but sales plummeted, the general public hated it and demanded their original Coke back.

Now, some think that this was a ploy by Coca-cola all along—most intelligent people disagree. Some argue that Coke needed to change its formula. Sugar cane was becoming too expensive, therefore they wanted to switch to the cheaper corn syrup. The problem, as the conspiracy theorists see it in the eye of Coke management, is that they could not change the secret recipe mid-production, too many people would taste the difference. This is where New Coke comes in, it was a chance to take Coke off the market, change the formula, make everyone mad at New Coke, so they could return looking humbled to their customers while secretly switching the formula. This myth has been since debunked and proven incorrect.

Recently, several instances of similar swaps and tests peaked my interest and rekindled memories of the Cola Wars I survived. Pepsi logosPepsi is attempting to create a new look for their brand, which hasn’t gone so well for them. There was a leaked memo called “The Pepsi Gravitational Field” which is an fascinating look into the thought process of a multi-million dollar branding exercise. Along with that, I was reading a book entitled Elephants on Acid: And Other Bizarre Experiments and one experiment that crossed my eyes stuck. A researcher served two wines to wine experts, one white wine and one red wine. He simply asked the participants to write down their thoughts about each wine. The experts wrote down fairly generic terms you would usually attribute to white wines, then a second set of distinct terms you would use to describe a common red wine. The trick was that the red wine was in fact the same white wine with a bit of food coloring. The experts were fooled into smelling and tasting different experiences based purely on the color of the liquid!

This made me wonder. I fancy myself a Coke drinker and always have because I was never fond of Pepsi’s more sugary taste, but now I wonder if I could pass the sip test. Would I select Coke over Pepsi and could even tell the difference‽

Luckily, this is a pretty cheap and easy experiment to conduct. I obtained 8 plastic cups, a 2 liter bottle of Coke and a 2 litter bottle of Pepsi. The next step was to design a double blind taste system which would allow me and a few friends to independently propose our answers without the examine curator biasing the experiment. We selected the following system:

Myself and friends would participate in the tasting. A fifth friend volunteered to poor the drinks. We labeled the 8 plastic cups 1-8. Our fifth friend would randomly fill-up 4 cups with Coke and 4 cups with Pepsi.Pepsi and Coke Bottles Each numbered cup to cola type was recorded out of the view of the 4 of us tasters. The cups were then presented and we each randomly select 2 of the 8 to drink from. This created a double blind in which the fifth friend was unaware of what was being served to us. We recorded which cups we selected and our guesses at which drink was inside.

We needed to repeat the experiment a few times to eliminate the possibility of just guessing correctly. If 4 of the 8 glasses were Coke and 4 Pepsi, this gives us an 8 choose 2 combination. To find out how many combinations there are we use the equation n!/k!(n-k)!, substituting 8 nCr 2 we get 8!/2!(8-2)! = 28. There are 28 possible combinations to randomly choose 2 cups. 6 of those combinations will be two Cokes and 6 of those will be two Pepsis. This equates to a 42.8% chance of selecting two of the same drink and a 57.1% chance of selecting two alternative drinks.

We wanted to build in enough rounds of taste tests so that chance of randomly guessing correctly became so small we knew we were selecting based on our personal tastes rather than chance. We had three possible configurations when choosing, 2 cups of Coke, a 21.4% chance, 2 cups of Pepsi another 21.4% chance and 1 cup of each drink with a 57.1% chance.

We decided 5 rounds was enough iterations to remove chance. Each round, if we were to select at random, there are 8 possible cups, each cup has 1 or 2 possible contents. This gives you 16 combinations for the first cup. The second cup has only 7 possibilities and 2 possible contents, for a possible 14 combinations. So for each tasting round there is 224 possible combinations of cups and liquids. Multiply this by the number of rounds, 5 in this case, and you get 1120 possible combinations during the experiment. If we scored well, that would mean our taste buds could truly select Coke over Pepsi rather than us randomly guessing correctly.

One afternoon, we set to task. Here are the results we recorded:

Actual Contents

This is the table of the actual contents of the cups. This was unknown to us to prevent bias between the experimenter and the experimentees.

Round Cup #1 Cup #2 Cup #3 Cup #4 Cup #5 Cup #6 Cup #7 Cup #8
1 Pepsi Coke Coke Pepsi Pepsi Coke Coke Pepsi
2 Coke Pepsi Coke Coke Pepsi Pepsi Coke Pepsi
3 Pepsi Coke Pepsi Coke Coke Pepsi Pepsi Coke
4 Coke Coke Coke Coke Pepsi Pepsi Pepsi Pepsi
5 Coke Coke Pepsi Pepsi Pepsi Coke Pepsi Coke

Contestant #1

Round Cup #1 Cup #1 Contents Cup #2 Cup #2 Contents
1 4 Pepsi 7 Coke
2 2 Coke (x) 5 Coke (x)
3 4 Coke 1 Pepsi
4 6 Coke 7 Pepsi
5 8 Coke 3 Pepsi

Contestant #2

Round Cup #1 Cup #1 Contents Cup #2 Cup #2 Contents
1 8 Pepsi 3 Coke
2 6 Pepsi 8 Coke (x)
3 2 Coke 5 Coke
4 1 Coke 5 Coke (x)
5 7 Pepsi 1 Coke

Contestant #3

Round Cup #1 Cup #1 Contents Cup #2 Cup #2 Contents
1 1 Pepsi 5 Coke (x)
2 3 Coke 4 Coke
3 7 Pepsi 8 Coke
4 2 Coke 8 Pepsi
5 4 Pepsi 6 Coke

Contestant #4

Round Cup #1 Cup #1 Contents Cup #2 Cup #2 Contents
1 2 Coke 6 Coke
2 4 Coke 1 Coke
3 6 Pepsi 3 Pepsi
4 4 Coke 3 Coke
5 5 Pepsi 2 Pepsi (x)

As you can see, we did very well in selecting correctly. At most, 2 mistakes out of 10, that is an 80% correct ratio, much better than guessing.

All of the errors that slipped through were Pepsi being guessed as Coke, except one where Coke was mistaken as Pepsi.

We weren’t asking people about which they preferred, we simply want to determine if it was possible to identify the difference between the two based on the taste rather than preference. I would conclude that we are successfully capable of doing so.