Professional Documents
Culture Documents
Abstract
Twitter has been on the rise in the sport world, and this study hoped to see how it is being used
and interacted with. There is a lot know about the use and interaction with Twitter within the
realm of sport, yet there remains much to explore. Using the Minnesota Timberwolves and
further generalizing, this case study looked at the ways teams, athletes, and reporters post on
Twitter and how fans engaged with these tweets. What are they saying? When are fans
engaging? 891 tweets were coded based on their content into specific categories developed by
Hambrick and Geurin-Eagleman. Statistical methods were used to analyze the tweets
collectively. The study concluded that there are significant differences in what is tweeted by
teams, athletes, and reporters and that timing is a very relevant factor when it comes to fan
engagement.
Literature Review
Since the birth of social media, sport fans of all kinds have been daily bombarded by information
through twitter. On average, 500 million small sets of words, known as “tweets,” were posted by
Twitter’s users in 2019. Obviously, not all of these came from sports teams, athletes, and
reporters. However, a very large number of them were coming from sport related entities. All of
the internet has become more and more over-saturated with information. Twitter has not been an
exception. Sport fans, however, have chosen their own fate, showing that they enjoyed this
bombardment, as it allowed them to easily and more thoroughly keep up with their teams and
sports of choice.
From these sport related accounts that fans have subscribed to, different forms of content
have been sent out to them. However, there have been so many tweets, that the only way in
which to study them should be in terms of the conglomerate, as literally everything the human
mind can fathom has probably already been tweeted. Generally, tweets have contained either
text-based or visual content. Different people, such as Hambrick and Geurin-Eagleman, have
created subcategories within textual (Hambrick 2010) and visual content (Geurin-Eagleman
2016) in which to further categorize these tweets in order to systematically make sense of them
Without proceeding any further, some unclear terms need to be defined. Starting at the
most basic level, Twitter is a social media platform in which users post small bits of words or
pictures. Twitter has developed into the dominant source for the spread of sport-related
information on the internet. A tweet, to reiterate, is the thing that is posted on Twitter, whether it
contains a link, a picture, or just words. Fan engagement, or just engagement for short, is the
total of likes, comments, and retweets of a tweet. To engage with a tweet would be to either like
WHAT ARE THEY SAYING? 4
a tweet, comment on a tweet, or retweet a tweet. A beat reporter, often referred to later as just
‘reporter,’ is the main, or leading reporter for a specific team or city. In this case, Jon
Krawcynski is the beat reporter who’s tweets we are studying, although he reports on both the
Ann Pegoraro, in her case study, studied the tweets of just athletes. She came to the
conclusion that some athletes have, in fact, taken to Twitter in order to increase pubic visibility
through engagement with fans and fellow athletes. She also noted that this growth in the use of
Twitter by athletes should spark new opportunities for the communication between athletes and
fans and the study thereof (Pegoraro 2010). The present study has continued this research in
seeking to discover if there exists a difference in how teams, athletes, and reporters tweet. The
present study has also drawn upon Hambrick and Geurin-Eagleman’s conclusions in order to
detect what these differences are, both textually and visually, should they exist. More
specifically, do the types of textual tweets of athletes, teams, and reporters significantly differ
from one another? And if so, how do they differ? Similarly, do the types of visual tweets of the
three account categories differ from one another? And if they do, how do these differ?
Upon reception of these millions of sport related tweets, sport fans have been left with a
choice. Just the same with any common tweet, a sports-related tweet can be engaged with or just
skimmed over. From the corporate side of sports, engagement has proven to be a very important
marker to keep track of. The maximizing of engagement allows sports teams to get more
advertising material past their followers faces, making them more money. At the lowest level,
this is simply game promotions and ticket sales but can be extrapolated further to sponsorship
The emergence of social media platforms in the US came with fear from sports teams on
the corporate level, thinking that these medias would take away from the tv and radio traffic
sports teams got. Furthermore, the reproduction of any sport related scene or picture became
illegal without the expressed, written consent of the league therein. King, however, discussed
the thoughts of Adam Ostrow, who believed this type of censoring to be counterproductive for
the teams (King 2009). In lieu of these somewhat uncommon opinions, social media has been
realized as a platform in which more money can be made by teams and athletes.
Pegoraro also concludes that “although fans follow these athletes in large numbers,
athletes are not capitalizing on this audience of consumers. Although Twitter is offering fans
unprecedented access to athletes, its potential as a marketing tool is not being realized by
athletes,” (Pegoraro 2010). This conclusion can easily be extrapolated to opportunity for team
accounts as well. However, the key to getting paid, from the team’s perspective, has proven to
There are many factors that contribute to fan engagement with tweets. Jan Boehmer
discovered some of these factors in relation to characteristics of fans, team entities, and the
messages themselves. Her study found that “one’s level of interest in a tweet’s exact topic, the
perceived relevance of the tweet, how similar the information contained in a tweet is to the user’s
personal opinion, and one’s perception of how a tweet would affect one’s followers are the user
characteristics that affect intentions to retweet,” (Boehmer 2015). These findings suggested that
dominating Twitter requires relevance and ‘correct’ opinions, but never mentions anything
relating to timing. The present study planned on continuing this discussion by studying the
timing of tweets. How does engagement change from non-gameday to gameday? Furthermore,
does the time of day at which a message is tweeted affect the amount of engagement?
WHAT ARE THEY SAYING? 6
Methods
This study carefully followed and coded tweets from thirteen accounts relating to the Minnesota
Timberwolves of the National Basketball Association (NBA) for thirty days. The dates studied
were October 16 through November 14. These dates were chosen, due to the timing of the study,
in order to maximize the amount of days during the regular season of the NBA. The NBA
regular season started on October 22, hence twenty-four of the days of the study were within the
regular season. The accounts studied were the official team account @Timberwolves, the beat
reporter @ JonKrawcynski, and all of the rostered athletes who were active on twitter during the
@Holla_At_Rob3). For the purpose of the study, the eleven athletes were grouped together, so
In total, 891 tweets were studied in the 30-day period. Six pieces of data were
individually collected for each tweet. The first piece of data for each tweet was whether or not it
was tweeted on gameday or not. The second, contingent upon the first, was whether their team
won or lost on a gameday. Thirdly, the time of day the tweet was posted was noted. Next, the
total comments were noted. Following that, the added sum of likes and retweets was noted.
Lastly, the tweet was coded based upon its content, using the coding scheme of Hambrick for
textual content and the coding scheme of Geurin-Eagleman for visual content.
Hambrick’s scheme for coding text-based content is broken up into six categories based
purely upon what is said in each tweet. Within this scheme, a text-based tweet is grouped into
Promotional. Interactivity was the account’s direct communication with other athlete’s and fans.
WHAT ARE THEY SAYING? 7
An example of this from the current study from Jordan McLaughlin (@itsmclaughlin11) was
tweeted on October 31: “Happy Halloween everyone be safe tonight.” This tweet is purely an
athlete communicating with his fans. The second category was Diversion. Tweets of this
category were non-sports related, discussing friends and family or personal interests, such as
video games, music, or fashion. An example of this from Karl-Anthony Towns (@KarlTowns)
on October 24 was “If you want to have a chance to win a copy of Call of Duty: Modern
Warfare Operator Edition, follow me on Twitter, IG, and join my discord community. I will be
Thirdly was the code of Information sharing. This is essentially any insight into the team
or athlete being studied, such as details about practices or recent competitive results. An
example of this, tweeted on November 13 by the beat reporter (@JonKrawcynski) was “Teague
finishes with 18 points and 6 assists in 23 minutes. Towns: 28, 11r, 4 a. Wolves cruising to the
end here.” Content, a somewhat confusing title, was defined as any sort of link to pictures,
videos, and other websites. Another tweet by the beat reporter reads “The Timberwolves are
starting to solve the math problem that has plagued them for years.” This tweet was simply the
title of an article linked within the tweet on theathletic.com. This is an example of content as
defined by Hambrick.
Fanship, as defined in this study, were the accounts discussing sports other than their
November 12, was “Good win @OrgeonMBB,” as he is referencing the Oregon men’s college
basketball team. The final category was Promotional, defined as publicity regarding
sponsorships, upcoming games, and any other promotions, such as giveaways or discounted
WHAT ARE THEY SAYING? 8
tickets. Josh Okogie (@CallMe_NonStop) tweeted “Call these the nonstops,” with a link to the
In this study, many primarily visual tweets were also studied. These were coded
similarly but with a different scheme. This one was developed by Geurin-Eagleman and codes
tweets of visual content into one of seven categories, purely off of what is shown in the image or
video. The categories within this scheme were Related to personal life, Related to business,
Related to athlete/team sport, Related to other sport of athlete, reposted photo, pop-culture of
landmark, and Meme/screen capture. The first category, Related to personal life, was defined as
visual content referencing lives, routine, weather, parties, charity, family, or friends. An
example of this is a video @KarlTowns posted on October 29 of someone dressed as Joker for
some sort of costume party. Secondly, related to business, was defined as photos or videos
referencing their business/work, including training, travel, games. In the case of this study, this
is everything basketball that is related to the Timberwolves. An example of this was the short
video posted by the official team account on November 13, showing a dunk in the fourth quarter.
Related to athlete/team sport was defined as photos referencing basketball, but not the
Timberwolves, such as any college or high school basketball reference or other NBA teams. An
example was the official Timberwolves account tweeting pictures of Paul Pierce in his Kansas
uniform. Related to other sport or athlete was defined photos focusing on another sport. An
example of this, again from @Timberwolves, was the picture posted on November 13 of Kyle
Rudolph, a Minnesota Vikings player. Reposted photo was simply defined as a photo posted
that is originally from someone else’s account. An example of this was Jaylen Nowell
(@JaylenNowell) reposting a photo that the Timberwolves team account had posted a few days
Pop-culture or landmark tweets were defined as visual content that focus on movies, tv,
or landmarks/locations, and stem from shared cultural knowledge. An example was Karl-
Anthony Towns tweeting a video in which him and his teammates are running around
Philadelphia searching for cheesesteak before their game. Lastly, meme/screen capture, was
defined as any visual content that is not a photo or video. An example of this was the meme
posted by the Timberwolves of Kendrick Perkins turning up the thermostat on November 14.
Results
As aforementioned, thirteen accounts and a total of 891 tweets were studied. Of these tweets,
252 (28.28%) were tweeted on non-game days, as opposed to the 639 (71.72%) tweeted on game
days. 454 (50.95%) of them were tweeted by the official Timberwolves account, 89 (9.99%)
were tweeted by the athletes, and 348 (39.06%) were tweeted by the beat reporter of the
Timberwolves.
Research question 1A asked if the textual tweets of athletes, teams, and reporters differ
from one another based on their code in Hambrick’s coding scheme. A chi-square test was
performed to examine the relation between type of account and type of textual content. The
content was significantly different based on the type of the poster (X 2(10, 559) = 150.47, p < .00).
The two most common content types for the team to post were information sharing (n = 92,
62.2%) and interactive (n = 20, 13.5%), while the least common was content (n = 1, 0.7%). For
the athletes, the most common was interactive (n = 33, 46.5%), while the least common was
promotional (n = 3, 4.2%). For the beat reporter, the most common was information sharing
(n = 216, 63.5%), while the other categories each made up less than 12% within the reporter’s
tweets. The team and reporter included more information sharing than the athletes.
WHAT ARE THEY SAYING? 10
Research question 1B asked if the visual content of tweets of athletes, teams, and
reporters differ from one another based on their code in Geurin-Eagleman’s coding scheme. A
chi-square test was performed to examine the relation between type of account poster and type
of visual content. The content was significantly different based on the type of the poster (X 2(12,
332) = 137.06, p < .00). For the team, the most common type of visual tweet was related to
business (n = 178, 58.2%), while every other category contained less than 15% of the team’s
tweets. For the reporter, the most common type was a repost (n = 6, 75%), with only 8 visual
based tweets in total. For the athletes, the most common type was a repost (n = 12, 66.7%). The
team included more visual content in total than the other types of accounts. The team included
more visual tweets related to business than the reporter and athletes, with 96.7% of them.
Results from a t-test indicate that there is a significant difference between gameday and non-
gameday in the amount of engagement with tweets. Non-gameday tweets received more
comments (x = 30.53, SD = 253.35) and likes and retweets (x = 1619.8, SD = 13621.76) than
gameday tweets comments (x = 8.28, SD = 14.5) and likes and retweets (x = 380.48, SD =
791.09), respectively (t889 = 2.21, p < .03), (t889 = 2.29, p < .03). Non-gameday tweets received
Research question 2B asked how the time of day affects the amount of engagement with
tweets. There is a weak, negative correlation between comments and hour of the day, (r = -.12, p
< .01). There is a weak, negative correlation between likes and retweets and hour of the day, (r
= -.12, p < .00). There is slightly more engagement with tweets earlier in the day.
Discussion
WHAT ARE THEY SAYING? 11
This study desired to uncover if teams, athletes, and reporters tweet differently and if the timing
of tweets affect their engagement. To summarize, teams, athletes, and reporters do, in fact,
tweet differently, both textually and visually. On the text-based side, teams and reporters
dominate the information sharing category, while athletes were more likely to produce
interactive tweets than any other category. This did not come as a surprise at all, as one would
expect teams and reporters to be sharing information on Twitter, as it is essentially the reason
they exist. This research agreed with Pegoraro’s research, as the opportunity for athletes to
interact more with fans was still a wide-open door (Pegoraro 2010). Interactivity was the
category most tweeted in for athletes, yet was still small, with only 33 tweets. Not only that, but
in the 30-day window, for all 11 athletes studied, only 3 total tweets were categorized as
promotional. This is not necessarily a bad thing, but it suggest that these athletes in particular do
not care all that much about media attention. This also is not surprising, as these athletes play
for Minnesota, and upon viewing their gameplay, one can easily see that these athletes are not
out here trying to flex when they can just get it done in the right way. This brings up a limitation
of the research.
The Minnesota Timberwolves are not the Chicago Bulls or the LA Lakers. With zero
NBA finals experience and few historical superstars, it only takes one logical step to assume that
generally speaking, Minnesota’s players are not attention seekers on the whole. This is
definitely a limitation of this research. This is not to say that it would be inherently better to
study players with more media-attention, but that studying more than one team or choosing a
team that has more of a balance of players could benefit this research. Another possible, much
larger, future study could study players and teams league-wide, as opposed to simply on one
team, in order to get a more accurate balance on who the archetype NBA player is.
WHAT ARE THEY SAYING? 12
Returning to the above thought, there exists such an opportunity for athletes to promote
themselves on Twitter for any reason, whether money or influence or a voice in the public. This
study once again pokes at that hole, wondering why athletes have not done so yet with all the
On the visual side of the coding process, the team account massively outweighed both the
reporter and the athletes in all categories, but specifically in related to business. This makes
complete and total sense. The team entity of the Timberwolves hires design professionals, social
media experts, and many others taking part in the tweets produced by the official team account.
They produce original content with the goal of having fans enjoy it. Not to mention how easy it
is to produce a five to ten second video of a play that just happened and in turn transforming the
Twitter feed into a highlight real. In the research phase of this study, the researchers noticed that
regardless of the day, the official was tweeting, and it was tweeting original visual content daily.
This could be just something that is a big part of the Timberwolves vision for their Twitter
account or it could be a reflection of a common league-wide practice. Within the current study
there is no way to tell. Once again, a limiting factor of this research is the fact that only one
team was studied, and a future study could include multiple teams.
Along with implications of the first research question comes the implications of the
second. The fact that engagement is significantly higher on days without games is actually quite
surprising. One may predict that fans are more likely to see and engage with tweets during
games. However, this straight up is not the case. Fans prefer to engage with tweets on non-
gamedays. This could be because of the type of tweets posted on gameday verses non-gameday.
For instance, it was much more likely in this study for the team to post a whole bunch of videos
while they are playing in a game. If a fan is watching the game, they may still see these tweets
WHAT ARE THEY SAYING? 13
and may even very much enjoy their existence, but amidst the game they have no reason to
interact with these videos, as they are simply recaps of what they have already seen. Similarly,
if the fan is not able to watch the game and relies on Twitter to get their share of game action
and highlights, they will most likely simply scroll through them as there is a ton of content
posted on gamedays, not taking the time to stop and retweet or comment or like the tweets.
On the flip side, non-gamedays pose a less busy day for everyone involved. In this ‘free
time,’ without a game that evening, fans can reconcile spending more time on Twitter and
engaging with tweets. This disagrees with King’s article on not allowing the reproduction of
sport-related material on social media being counterproductive (King 2009), as the gameday
engagement levels suggest that this type of censorship would have a small effect.
Another possible reason for this is the massive variety of tweets’ engagement levels. The
researcher team noticed that there were just a few tweets that received thousands of likes, while
many, many more received little to no engagement. These anomaly tweets are outlaw-ish in
their behavior. Every once in a while, one of the accounts would post something spectacular,
generating massive engagement. Every time this happened was on a non-gameday. If these
outlaws are taken out of the equation, there would be much less of a discrepancy between
engagement levels on game day verses non-game day. This poses yet another possible limitation
of the current study. An option for future research fixing this problem would be to study tweets
for a longer period of time, as this would allow the law of large numbers to take affect and
balance out these anomaly tweets from having such a massive effect on the numbers.
Very similar implications can be inferred regarding the negative correlation between time
of day and engagement. NBA games tend to be in the evening, hence, all the above reasons for
fans not engaging during games apply here as well. Along with that, there seemed to be a trend
WHAT ARE THEY SAYING? 14
for 11am being the first tweet of the day from both the team and the beat reporter. Furthermore,
the vast majority of the massively-engaged-with, anomaly tweets mentioned above were tweeted
somewhere between 11am and 1pm. This would therefore shift the engagement numbers to
earlier in the day. In addition, there were a handful of ‘rollover’ tweets that were tweeted after
11:59 pm on gameday, putting them at hour 0 of a non-gameday. It was often these tweets that
received the most engagement for gameday tweets, even thought they were not technically
tweeted on gameday. These tweets could have affected both the gameday/non-gameday
numbers as well as the time of day numbers, as a tweet during hour 0 receive one 1,000 likes
surely helps the likelihood for earlier hours’ tweets to receive high amounts of engagement.
This is a slight problem that could be fixed in future research by simply lying about the time
In this study, the Minnesota Timberwolves, its players, and beat reporter’s tweets were
analyzed. It turns out that the athletes really have not tweeted all that much within the 30-day
period looked at. This implies that the athletes of the Timberwolves are not pursuers of the
Paparazzi. But the lack of athlete’s tweets was made up for by the volume of tweets from the
team and reporter who evidently do their jobs quite well. The Timberwolves official account
posted so much visual content that it is clear that this is what sport fans of today desire to see on
their twitter feed. Likewise, sport fans undoubtedly want to see this content on days that lack a
game for their team, and seemingly would like to view this content earlier rather than later in the
day. But then again, people are crazy and may shift their ideal scenario in just one short second.
WHAT ARE THEY SAYING? 15
References
Boehmer, J., & Tandoc Jr., E. C. (2015). Why We Retweet: Factors Influencing Intentions to
232.
Hambrick, M. E., Simmons, J. M., Greenhalgh, G. P., & Greenwell, T. (2010). Understanding
King, D. (2009). Just Don't Twitter About the Ol' Ball Game. Information Today, 26(9), 39.
Pegoraro, A. (2010). Look Who’s Talking – Athletes on Twitter: A Case Study. International