You're probably peeling bananas the wrong way

Using data to measure the current international state of banana peeling.

Posted by Alaska Bunch on August 23rd, 2016

If you haven’t heard already, there’s a right and wrong way to eat a banana, and you’re probably doing it wrong.

See for yourself - a quick Google search shows a near unanimous agreement that the non-stem end is the correct end to open the banana from, with a list of reasons why:

  • It’s less likely to mush or bruise while opening
  • You can use the stem as a handle while you eat
  • It’s the way that monkeys peel their bananas

We can save the discussion of whether we should be modeling our behavior after monkeys for another time, but our curiosity was piqued, and we wanted to know whether human behavior reflected the advice of the experts. How many people actually peel bananas from the non-stem end, and who are they? Our guess was not many.

Collecting the Data

Testing this hypothesis happened to be a good way to showcase the simplest use of our Create a Poll tool. All it required was a 2-picture poll, and Alaska Bunch poll screenshot since we were interested in everyone's responses, there was no need to exclude any demographics when picking our audience for the poll. The screenshot to the right shows what the poll looked like to voters on our mobile app.

Getting the data is simple enough, but just as important is making sure that data is good. To do this, each poll has a minimum time to vote to mitigate accidental clicks, and the order of the pictures is randomized for each voter to minimize the effect of Order Bias, which is low already for a 2-choice poll. These two measures alone have given us greater than 98% accuracy on our control questions (e.g. which device are you using?). This number will only improve as our behavior-tracking algorithm weeds out more bad clicks and bad voters.

The total cost to collect 1,000 votes for this study on Alaska Bunch was $21.

The Results

Now for the fun part. As we expected, the vast majority of people are ignoring the advice of the experts (or have never heard it), and are peeling from the stem-end of the banana. Beyond this, we were interested to see how the split would look (pun intended) for each demographic segment.

For now, we limit the demographic information we collect from voters to gender, age, self-reported location, platform, and device, all of which are anonymized. With this, we were able to measure a stronger preference in stem-end peeling for girls over guys, and older people over younger people. iPhone and Android users were tied.

We did not get enough data from many of the smaller U.S. states to be meaningful, but of the states with a statistically significant number of responses, California had the highest rate of non-stem peelers, and Michigan came in with the lowest. Our international voters (8% of responses) had only 14% non-stem peelers, lower than any region in the U.S.

Our Analysis

Opening a banana from the non-stem end is not intuitive - for human beings, at least. If you're a kid figuring out how to open a banana on your own for the first time, you'll probably pull at the stem a bit and get the banana to open up without too many issues. We came into this as skeptics to the concept of opening a banana any other way, but while doing this poll we couldn't help but try it out. Admittedly, opening from the non-stem end has proven to be a very slight improvement on our lives.

This puts this information in an interesting category of importance - it's just mildly interesting enough to be posted online and shared with friends, but not nearly important enough that anyone needs to know it. We suspect that the internet is the driving force behind a lot of the people opening bananas this way (directly or indirectly), and that the websites posting about this happen to have greater readership among guys and younger people. This is our going theory at least - more research is needed.

As for the results in California - we just can't seem to do anything normal.

You can view the raw data for this poll by downloading this CSV.