Data Visualization Debrief: Sex and Data at the BET Awards

General

Just over a month ago, we set out to compare male and female nominees for the category of Best New Artist at the BET Awards in a way that was objective, accessible, easy to navigate and visually appealing. In addition to making an objective prediction for who the winner should be based on publicly available data, we analyzed the gender discrepancy among both nominees and winners for the category since the award show’s inception.

We found that in the BET Awards’ 17-year history, 34% of nominees and 19% of winners of the Best New Artist category had been women, and in both 2004 and 2012, there had been no female nominees for the category whatsoever.

To make our objective prediction in the lead-up to this year’s show, we gathered information from a mix of data sources, including streaming numbers from Spotify, social engagement metrics from Instagram, and critical reviews aggregated by albumoftheyear.org -- sources that we selected based on their influence in their respective categories and the quality of their data. Our conclusion based on this information was that SZA was the winner, which, indeed, also ended up being the case at the BETs.

You can view the data visualization here.
 

Design

Colour-scheme: We chose the colours based on what we believed was evocative of BET branding, while still adding our own twist by going slightly bolder. In addition, we used duotones in the images to grab readers’ attention.

Font: Work Sans font was selected in order to satisfy the criteria of being both bold and modern while still being legible and flexible.

Animation: Using animation was a means of both grabbing readers’ attention and setting the project apart from typical, static data visualizations. We opted to use CSS-only animations, which included the use of cubic-bezier transitions, animation delays and child-div hover states on parents divs.  You can see an example of this in practice, here.

A unique challenge we faced in the design and UX of the data viz project was the list-styling for cross-referencing. To create this effect, we needed to assign unique classes of “winner” and “female” to the table cell, and then toggle booleans using Crossfilter in order to change the styling of a table cell reflective of whether or not it contained these classes. You can see an example of this in practice, here.

Coding

To collect initial information about who the winners and the nominees were for the category of Best New Artist each year, we crawled Wikipedia’s BET pages. We wrote scripts to capture the results year over year, and then processed and analyzed the data in order to create something more interactive and accessible.

We created the visualization by using crossfilter, DC.js and some custom D3.js. Crossfilter was used to make it interactive, but we needed to write our own methods to filter the names below the table which went beyond what crossfilter and DC.js typically allow you to do. The primary concern here was to create the best UX possible.

Some of the challenges of this process had to do with the tedious nature of crawling Wikipedia pages, as each page is structured differently. It required that we carefully write scripts that were generic enough to work on most pages, while also being specific enough to be able to determine features like the sex of the artist based on the contents of the article.

Wavo Native Advertising

We spent approximately $140 and served the ad to selected lists of people, reaching 996 people and logging 3,149 impressions.

Our primary takeaway from our personal experience using our native advertising tool was that the creative would have worked better as a Facebook and/or Instagram in-newsfeed ad, or as a boosted post. Instead, we shared it through Instagram stories where it felt ad-like, which may have inadvertently led to slowed and/or reduced engagement, as compared to content that feels more native to the platform.

Marketing

We put together a press kit, which included a press release, logos and information about the company and cold-emailed reporters at select publications who we believed would be most interested in the information being presented in our data visualization based on their previous work. Still, in spite of initial interest from a variety of reporters, our main issue was the lack of turnaround time we allowed for by contacting outlets the week of the awards show. Additionally, part of the reason it might not have been picked up by press may have had to do with the promotional blurb for Wavo at the bottom of the data visualization as well as the volume of text, which may have prevented the visualization from speaking for itself.

Conclusion and Main Takeaways

All in all, the data visualization we put together representing sex discrepancy at the BET Awards and the objective prediction we made about who the winner would be are an example both of the democratization and increased accessibility of data, and of the ways we can work to fight biases in award shows, for example, by reducing the need for subjective opinion.

Moreover, humans are creating data at a dizzying pace, but only a fraction of this data is put to use, and an even smaller fraction is made accessible through data visualizations. By putting together projects such as this, Wavo is trailblazing efforts to demystify data and make it accessible for all.