By Robert D’Onofrio, Facebook Data Editor
Facebook is where real people connect with their family, friends, and the public voices they care about to discuss (and sometimes debate) the topics and moments that matter to them most. Many people experience big events across news, entertainment, and sports in large part through this global social conversation on Facebook, from the initial anticipatory buzz through to the sustained “water cooler” discussion people engage in afterwards. The Facebook data team often provides an insights-driven lens into these conversations by highlighting their overall size and geographic spread, the most popular topics discussed, top social moments, and relevant demographic trends. Today we’re excited to announce a new way of looking at social conversation on Facebook.
NFL Football is one of the most popular topics of conversation on Facebook, particularly during cornerstone events like the NFL Draft and the Super Bowl. Following last season’s thrilling contest between the Patriots and the Seahawks, the data team found that 65 million people had 265 million Facebook interactions related to the big game. We discovered that New Hampshire was where conversation was most concentrated, and that Tom Brady and Marshawn Lynch were the most talked-about players. We also shared a video highlighting the number of people-per-minute (PPM) engaged in the social conversation during the game, peaking with the final whistle. Given the debate sparked by the goal-line interception that sealed the game, it would have also been fascinating to better understand the Super Bowl Facebook audience of 65 million in terms of sentiment; was the conversation more positive for one team or another, and how did that excitement shift during critical plays for each team?
Introducing Sentiment Data for Sports
To celebrate the 2015 NFL regular season kickoff between the Patriots and Steelers, Facebook is taking a closer look at sports conversation and sentiment for the very first time. Sports talk by its very nature is ideally suited for sentiment analysis, since most fans either root for or against various teams and/or players. When it comes to favorites they cheer on the most, hopes rise and fall on a regular basis, and fans take to Facebook to express their opinions, sometimes positive and negative, about the latest breaking sports news. Fans love to talk about sports on Facebook, the largest community of sports fans in the world with 650 million people who like at least one sports Page.
This global sports fan community on Facebook includes nearly 50 million people who like an official NFL team Page. Since the first training camp started, many of these fans have been closely following their teams and engaging with other fans on Facebook about the latest injuries, depth chart battles, and preseason games. The scale of this conversation, combined with the large number of NFL fans on Facebook, gives us the unique opportunity to take a detailed look at what NFL fans are saying about their favorite teams.
Specifically, we took a look at all Facebook posts and comments NFL fans made during the preseason (July 22 – September 7) about the teams they Like on Facebook, and classified them according to how positive or negative they were. This enabled us to compare and rank all 32 NFL teams by how positive the overall conversation among each fan base was during the preseason (Figure 1), as well as look more closely at trends over time for specific teams. As always, the insights presented here are based on depersonalized, aggregated data.
Heading into the 2015 NFL season, the most optimistic fan base belongs to the Kansas City Chiefs – perhaps not too surprising, considering the Chiefs were the only team to finish the preseason undefeated (4-0). Likewise, the Vikings (ranked #3) finished the preseason 4-1, have Adrian Peterson returning, and Teddy Bridgewater has looked solid as he begins his second season. The Titans (ranked #5) had the second overall pick in the most recent NFL Draft, and their selection of Marcus Mariota has clearly boosted fan spirits. Other teams in the Top 10 share similar characteristics, related to high-promise rookies, coaching changes, and solid preseason game performances.
On the flip side, until very recently, Patriots fans were bracing for four games without Tom Brady, so it’s not too surprising to see them rank among the least positive fan bases since July 22. QB turmoil also impacted the Bills, Redskins, and Jets this preseason, leaving their fans less enthusiastic about the pending season compared with other fans. Steelers fans were disappointed to find out star receiver Martavis Bryant was going to miss four games, and Chargers fans spent half of the preseason worrying about the Rivers contract situation. I’ll leave additional editorializing to the reader – fans of each team can likely think of a number of possible factors impacting their team’s Facebook fan sentiment.
Beyond this overall snapshot of the 32 NFL teams, we were curious to see how sentiment played out on a daily basis as team-specific news broke on Facebook. The NFC East was rife with fascinating storylines this summer, and is home to some of the most noted rivalries in football. Figure 2 below displays how positive fan sentiment changed each day for these four teams:
Fan sentiment on Facebook moves as you might expect it to – down with negative team news, and up with positive news. Eagles fans peaked this preseason with the long-awaited debut of Sam Bradford, while Cowboys fans fretted most when one of their starting cornerbacks was lost for the season. Similarly, Redskins fan sentiment seems to perfectly reflect all the drama around RGIII and the team ultimately deciding to move on from their one-time franchise quarterback. It will be interesting to see how these team trends continue into the regular season and beyond.
More About Sentiment Data
After the 2012 U.S. presidential election, it was clear that voters were increasingly taking to Facebook to share their thoughts about candidates, issues, and politics in general. Social conversation throughout the 2014 U.S. midterm election culminated with 27 million people having 85 million Facebook interactions on election day alone. With so many people sharing their thoughts about politics, and a historic presidential election ahead of us 2016, the Facebook data team was curious about how people were feeling about potential 2016 candidates and likely key issues. We collaborated internally to develop a robust, privacy-safe system that classifies political posts and comments according to their sentiment (positive, negative, or neutral), which rolled out on Election Day 2014. All of the data was aggregated and depersonalized, meaning no sentiment classifications were associated with any specific individuals. Sports seemed a natural extension of this analytic framework, and we will continue to explore other areas of meaningful social conversation on Facebook that might lend themselves to a similar sentiment analysis in the future.