RUNNING IN

COLOR

Analyzing the difference in men's and women's running shoe colors.

By Lindsey Poulter & Hesham Eissa

If you had to guess, would you say that men and women have similar or different choices when it comes to running shoe colors? Do you think running shoe colors conform to typical gender stereotypes, with more blue shoes for men and more pink shoes for women? We asked ourselves these questions based on comparing our personal experiences and came up with a theory that there was a difference in both color availability and shade.

To prove our theory, with data, we analyzed the running shoes available on RunRepeat, a website that aggregates running shoes from 213 retailers and writes in-depth reviews. For all shoes available in the United States, we collected the gender the shoe was marketed for, the color category, the brand, and the model name.

To better understand the running shoe market, let's start by looking at the color and gender classifications of the 8,884 available running shoes.

Men's vs. Women's Shoes by Color

pink

purple

beige

silver

white

blue

grey

orange

black

green

brown

red

yellow

gold

100% MoreMen

67%

33%

67%

100% MoreWomen

After we found out there was a clear contrast between men and women's color choices, we wanted to know how the shades and lightness differed between genders. For example, blue, a stereotypical color for men, was pretty popular for women too. However, did this mean that both men and women had navy blue, royal blue, and light blue to choose from?

To answer this question, we looked at all non-neutral colors and identified the most dominant color of each shoe. Then, to allow for easy comparison between genders, we used each shoe's hue, saturation, and lightness to cluster the data into similar groups.

Now, let's explore the spectrum of colors available through three different lenses: the cluster colors, the actual shoe colors, and lightness. Each available shoe is represented below as one circle.

Scroll Down

Blue

Color Shoes By

Cluster

Actual

Lightness

Total Shoes

Cluster

Actual

Lightness

Top Brands

0% Women

100% Women

Men's Shoes

Women's Shoes

Men's Shoes

Women's Shoes

Men's Shoes

Women's Shoes

Men's Shoes

Women's Shoes

Men's Shoes

Women's Shoes

Men's Shoes

Women's Shoes

Men's Shoes

Women's Shoes

Men's Shoes

Women's Shoes

We collected data on all running shoes available in the U.S. from RunRepeat , on October 25, 2020. RunRepeat is a website that aggregates running shoes across all brands (Adidas, Nike, etc.) and retailers (Amazon, JackRabbit, etc.) and provides in-depth reviews with the ability to purchase through affiliate links. We used RunRepeat over-collecting from each brand individually to allow for standardized classification and collection methods.

For every shoe, RunRepeat classifies it into one (and only one) of fourteen color categories and as for men or women. We used these classifications throughout our entire visualization and did not modify any values. We used Python and BeautifulSoup library to gather this information for each shoe as well as the model name, brand name, and the preview image. We only used data and images that appeared on the results page and showed as being available to purchase by at least one retailer.

Once we had our main dataset, we wanted to look beyond the fourteen color categories and understand the different tints and shades that made up each category. We were only interested in the non-neutral colors: blue, purple, pink, red, orange, yellow, and green. We removed the white background from all images and then used Color Thief module to determine the dominant color of each image. Since we knew what color category RunRepeat classified each shoe as, we verified that the dominant color was within that color category. If it was not, we manually checked the shoe on RunRepeat and used our best judgment to correct the dominant color. Occasionally, if we felt RunRepeat did not have an up-to-date image, we would Google the model and color to verify it was a real option. If that failed, we would remove the shoe.

Once we had one hex color for each shoe, we converted it to the HSL color space. Then, for each gender and color category, we ran k-means clustering on the hue, saturation, and lightness of the shoes within that subset. This created between 2-5 colors to represent each color/gender. We ran these colors through Name that Color Javascript library to determine the closest matching color name.