Introduction
Introduction
This project aims to summarize
measurements of sentiments, emotions, and reception and model potential
levels of user-generated content on Twitter about the LIV Golf
professional golf series between November 8th, 2022, and November 30th,
2022. Since the creation of a well-funded competing golf league to the
PGA Tour, the golfing world has been split between two camps, those
supportive of the newly established professional golf league and those
strongly opposed to the league. While the reasons for resistance to the
disruptive golf league vary, the 645 Twitter observations creating the
dataset used in this study were gathered exclusively in the English
language, represent a sampling of users distributed globally, and aim to
represent users’ opinions on both sides of the contentious topic. This
study takes place after the first year of the competing golf league and
captures conversations about the golf league on the Twitter platform for
analysis. Through exploratory and descriptive analytics conducted in
“R,” the research will better understand the LIV Golf brand perception,
followers of LIV Golf, and the reception of the competing golf league
into the sporting world. At the time of this paper’s writing, there are
many editorial articles about the LIV Golf vs. PGA Tour debate and the
future of golf globally; however, this study is the only known analysis
of sentiment, opinion, and prediction of LIV Golf through Twitter
analysis.
The LIV Golf field is currently comprised of a global representation of players of varying abilities at different points in their sporting careers. Notable participants in the golf league are Major Championship winners and relatively well-known names in the golfing world, like Sergio Garcia, Dustin Johnson, Brooks Koepka, Phil Mickelson, Bryson DeChambeau, and the current number one player in the world, Cam Smith. LIV Golf has spent exorbitant money to create the league and to gain the consideration of these accomplished players to depart from their current careers on the PGA Tour. It is reported that the total salaries of the players, as mentioned earlier, are estimated at greater than 750 million US dollars. While only some participants on the 54-player roster have received the same compensation as some of the most popular players in the sport, the contract is guaranteed income and is not dependent on performance. Additionally, the total purse of all LIV Golf events is 225 million dollars (Camenker, 2022). Even when excluding the lesser-known players and operating expenses, over one billion dollars has been spent by organizers to create the roster and establish LIV Golf.
The sovereign wealth fund of Saudi Arabia finances LIV Golf. A sovereign wealth fund is a state-owned investment that invests in real and financial assets such as stocks, bonds, real estate, private equity funds, and hedge funds. A persistent narrative in media reporting on the challenging golf series is that creating the golf league is an effort by the Saudi Arabian government to conduct large-scale sports washing. The term sports washing describes efforts by individuals, corporations, or governments leveraging the appeal of sports to improve sentiment toward their socially unacceptable practices (Wojtowicz, 2022). The most famous example of sports washing occurred in the 1936 Olympic Games hosted by Germany and the Nazi regime. The German government made a significant effort to convey a modern, civil, and ethical society to the rest of the world. Other instances of sports washing occurred in the Moscow Olympics in 1980, Beijing Olympics in 2008, Sochi Olympics in 2014, The World Cup in Italy in 1934, The World Cup in Argentina in 1978, and The World Cup in Qatar in 2022. Each event allows a global observer to become less focused on social issues that would tarnish their reputation and become more focused on the positives of sport.
This study is not the first research project that has sought to extrapolate conclusions of a population based on a sample created by the opinions found on Twitter. Studies from the Pew Research Center have found that 80% of the American population accesses the internet daily and approximately 15% of those users frequent Twitter (Smith, 2013). The age, gender, and income levels have been established to be well distributed which aid in eliminating specific sociological factors (Smith, 2013). Through accessing Twitter data about a topic, broad generalizations and conclusions can be formed. Unfortunately, because downloads represent all social groups, it can be difficult to extract meaningful conclusions about any social group. For that reason, the target in this study does not focus on age, gender, or income level and focuses conclusions exclusively on the entire population.
Like other studies, specific focus was placed on removing spam, advertisements, and bots from the data set. Previous studies have found that as much as 10% of the data on Twitter is spam (Chowdury, 2012). The prevalence of spam can be problematic from a research perspective because of how it can lead to faulty conclusions and inaccurate analysis. To create accuracy of analysis, this study took painstaking effort and exhausted all efforts to remove advertisements, bots, spam, and other irrelevant data for the study.
During this study, established Natural Language Processing (NLP) methods have been used at the word, and word pair level to extract meaning and opinions from Twitter observations. Studies by Argawal, Xie, Vovsha, Rambow, and Possenneau in 2011 conduct further analysis into emoticons to deduce meaning from Tweets that have increased accuracy over the state-of-the art methods used in this study. Unfortunately, the research in this study does not leverage those methods but continues with the established state-of-the-art baseline methods.
The LIV Golf field is currently comprised of a global representation of players of varying abilities at different points in their sporting careers. Notable participants in the golf league are Major Championship winners and relatively well-known names in the golfing world, like Sergio Garcia, Dustin Johnson, Brooks Koepka, Phil Mickelson, Bryson DeChambeau, and the current number one player in the world, Cam Smith. LIV Golf has spent exorbitant money to create the league and to gain the consideration of these accomplished players to depart from their current careers on the PGA Tour. It is reported that the total salaries of the players, as mentioned earlier, are estimated at greater than 750 million US dollars. While only some participants on the 54-player roster have received the same compensation as some of the most popular players in the sport, the contract is guaranteed income and is not dependent on performance. Additionally, the total purse of all LIV Golf events is 225 million dollars (Camenker, 2022). Even when excluding the lesser-known players and operating expenses, over one billion dollars has been spent by organizers to create the roster and establish LIV Golf.
The sovereign wealth fund of Saudi Arabia finances LIV Golf. A sovereign wealth fund is a state-owned investment that invests in real and financial assets such as stocks, bonds, real estate, private equity funds, and hedge funds. A persistent narrative in media reporting on the challenging golf series is that creating the golf league is an effort by the Saudi Arabian government to conduct large-scale sports washing. The term sports washing describes efforts by individuals, corporations, or governments leveraging the appeal of sports to improve sentiment toward their socially unacceptable practices (Wojtowicz, 2022). The most famous example of sports washing occurred in the 1936 Olympic Games hosted by Germany and the Nazi regime. The German government made a significant effort to convey a modern, civil, and ethical society to the rest of the world. Other instances of sports washing occurred in the Moscow Olympics in 1980, Beijing Olympics in 2008, Sochi Olympics in 2014, The World Cup in Italy in 1934, The World Cup in Argentina in 1978, and The World Cup in Qatar in 2022. Each event allows a global observer to become less focused on social issues that would tarnish their reputation and become more focused on the positives of sport.
This study is not the first research project that has sought to extrapolate conclusions of a population based on a sample created by the opinions found on Twitter. Studies from the Pew Research Center have found that 80% of the American population accesses the internet daily and approximately 15% of those users frequent Twitter (Smith, 2013). The age, gender, and income levels have been established to be well distributed which aid in eliminating specific sociological factors (Smith, 2013). Through accessing Twitter data about a topic, broad generalizations and conclusions can be formed. Unfortunately, because downloads represent all social groups, it can be difficult to extract meaningful conclusions about any social group. For that reason, the target in this study does not focus on age, gender, or income level and focuses conclusions exclusively on the entire population.
Like other studies, specific focus was placed on removing spam, advertisements, and bots from the data set. Previous studies have found that as much as 10% of the data on Twitter is spam (Chowdury, 2012). The prevalence of spam can be problematic from a research perspective because of how it can lead to faulty conclusions and inaccurate analysis. To create accuracy of analysis, this study took painstaking effort and exhausted all efforts to remove advertisements, bots, spam, and other irrelevant data for the study.
During this study, established Natural Language Processing (NLP) methods have been used at the word, and word pair level to extract meaning and opinions from Twitter observations. Studies by Argawal, Xie, Vovsha, Rambow, and Possenneau in 2011 conduct further analysis into emoticons to deduce meaning from Tweets that have increased accuracy over the state-of-the art methods used in this study. Unfortunately, the research in this study does not leverage those methods but continues with the established state-of-the-art baseline methods.
Frequency
Column
Frequency Analysis
A crucial step in the exploratory analysis
phase of research was to build upon previous actions identifying the
primary source of LIV Golf content. During this step of the analysis,
the study sought to explore the frequency at which Twitter users created
content associated with LIV Golf and when LIV Golf made content each day
during the observation period. Extracted from this research step are the
number of times LIV Golf authored content and the number of times users
contributed to the overall conversation. As one research question in
this study aims to understand the level at which LIV Golf motivates
users to engage with content and add to the conversation, frequency
analysis establishes the foundation for future linear regression
modeling in this study.
Immediate conclusions from this research step are that the media group was most active during the observation period between the 13th and 17th of November, and the most active date from all groups occurred on the 29th of November. Additionally, LIV Golf didn’t create content daily, and the frequency of tweets is not normally distributed.
The cumulative relative frequency analysis shows that the media’s most active date relative to the whole creates close to 25% of all conversations on Twitter about LIV Golf on several dates during the observation period. The study also identified the user group as the most active group during the observation period by a significant margin. Finally, LIV Golf created the least amount of conversation about the league. Research conducted through frequency analysis is the foundation for future predictions of user activity. Conclusions formed from this research phase contribute to understanding if user activity levels are dependent on activity levels from LIV Golf and assist in determining interest in the league among Twitter users.
Immediate conclusions from this research step are that the media group was most active during the observation period between the 13th and 17th of November, and the most active date from all groups occurred on the 29th of November. Additionally, LIV Golf didn’t create content daily, and the frequency of tweets is not normally distributed.
The cumulative relative frequency analysis shows that the media’s most active date relative to the whole creates close to 25% of all conversations on Twitter about LIV Golf on several dates during the observation period. The study also identified the user group as the most active group during the observation period by a significant margin. Finally, LIV Golf created the least amount of conversation about the league. Research conducted through frequency analysis is the foundation for future predictions of user activity. Conclusions formed from this research phase contribute to understanding if user activity levels are dependent on activity levels from LIV Golf and assist in determining interest in the league among Twitter users.
Column
Histogram
Cumulative Relative Frequency
Reception
Column
Reception Analysis
To understand the reception of LIV Golf
content among Twitter users, reception metrics needed to be created.
Nested within Twitter analytics are metrics that allow users to
understand how their content resonates with users on the platform.
Unfortunately, while many of the metrics are valuable to the user, they
provide little help for the research as they are generated based on a
user’s total amount of content. For example, if a user has created 1,000
tweets since their account creation, all analyses that Twitter provides
measure engagements against the totality of a user’s activity and not
specific content.
For this research project, one focus area is the reception of LIV Golf-generated content only during the observation period and therefore needed to create its own means of measuring reception. The reception score used in this study measures user engagement through retweets and favorites. By creating this metric, the analysis can now establish how willing Twitter users were to engage with content, regardless of their opinion about LIV Golf. A Twitter user can retweet content for numerous reasons. The research in this study does not attempt to correlate why the content was shared but only measures the amount users were motivated to share and engage with content.
The formula to generate the reception score (RS) required gathering the ratios measured against the total number of followers for an account. The purpose of the reception score is to determine the amount that content motivated viewers of the content to engage through likes and retweets on the platform. For this research, the reception score was created first by understanding that every follower has the potential to retweet content with each of their friends and followers. Therefore, the potential number of users that could be exposed to content is elevated based on the number of followers they have. For example, a user that creates content for a follower base of 15 people has significantly less reach than a user like LIV Golf, with followers totaling more than 200,000 people. Therefore, it was determined that considering potential reach to generate the retweet ratio was important in measuring engagement.
The formula used to create the retweet ratio (RR) for every tweet is the number of retweets times two divided by the number of author account followers times 100. The multiplication of the number of retweets accounts for the possibility that every follower can share LIV Golf content with their followers. A retweet amplifies content through sharing and when assessing reception, the retweet ratio is more important that the favorite ratio. If content is shared and amplified, it is indicative of content that resonates more with the user. The formula used to create the favorite ratio (FR) is the number of times a tweet was favorited by users divided by the number of followers of the author’s account times 100. Finally, the mean value of each reception score was created by a mean of RR and FR and attached to each tweet in the data set to finalize the reception score.
For this research project, one focus area is the reception of LIV Golf-generated content only during the observation period and therefore needed to create its own means of measuring reception. The reception score used in this study measures user engagement through retweets and favorites. By creating this metric, the analysis can now establish how willing Twitter users were to engage with content, regardless of their opinion about LIV Golf. A Twitter user can retweet content for numerous reasons. The research in this study does not attempt to correlate why the content was shared but only measures the amount users were motivated to share and engage with content.
The formula to generate the reception score (RS) required gathering the ratios measured against the total number of followers for an account. The purpose of the reception score is to determine the amount that content motivated viewers of the content to engage through likes and retweets on the platform. For this research, the reception score was created first by understanding that every follower has the potential to retweet content with each of their friends and followers. Therefore, the potential number of users that could be exposed to content is elevated based on the number of followers they have. For example, a user that creates content for a follower base of 15 people has significantly less reach than a user like LIV Golf, with followers totaling more than 200,000 people. Therefore, it was determined that considering potential reach to generate the retweet ratio was important in measuring engagement.
The formula used to create the retweet ratio (RR) for every tweet is the number of retweets times two divided by the number of author account followers times 100. The multiplication of the number of retweets accounts for the possibility that every follower can share LIV Golf content with their followers. A retweet amplifies content through sharing and when assessing reception, the retweet ratio is more important that the favorite ratio. If content is shared and amplified, it is indicative of content that resonates more with the user. The formula used to create the favorite ratio (FR) is the number of times a tweet was favorited by users divided by the number of followers of the author’s account times 100. Finally, the mean value of each reception score was created by a mean of RR and FR and attached to each tweet in the data set to finalize the reception score.
Column
LIV Reception Plot
User Reception Plot
Sentiment
Column
Sentiment Analysis
At the onset of the study, it was
understood that sentiment scores are not a reflection of opinion either
advocating for LIV Golf or expressing disapproval of the emerging
league. Sentiment values are derived through a numeric value attached to
each word in the tweet based on a comparison of used words from a known
dictionary, then a mean of the scores of the words creates the sentiment
score. The sentiment scores equate to the tone of conversation through
the language used in the observation. To accomplish thorough
understanding about the tone of conversation about LIV Golf, the study
extracted sentiment scores for the entire data set. Then the data set
was subset to create data frames of only user-generated and LIV
Golf-generated content.
The purpose of one research question aims to understand the tone of conversation among users about LIV Golf. Therefore, the study only explored the sentiment of user-generated content and excluded LIV Golf-authored content in sentiment analysis. Aside from isolating the author of the content, additional separation was accomplished by grouping sentiment scores chronologically during the observation period. The user sentiment scatter plot illustrates each observation of user-generated content for each day the tweet was authored. By conducting this exploratory process, the research aims to understand if negative, positive, or neutral tweets dominate the conversation, and if so, what dates does this occur?
The boxplot illuminates several critical aspects of the user sentiment over time in conjunction with establishing mean sentiment values per day of observation. Evident in the boxplot is the number of outliers that reside significantly away from the mean. For example, on November 22nd, there were four total outliers; three scored in the negative range, and one observation was significantly higher than the mean sentiment score. In addition, analysis shows that mean sentiment scores are neutral or positive even though some sentiments are very positive or negative. The boxplot creates additional insight into the tone of the conversation, but this study aims to understand trends among Twitter users about the LIV Golf topic.
The violin chart contains additional evidence of the varying degree of user sentiment throughout the observation period and creates more refined conculsions from the data set. On November 25th, user content created observations with sentiment scores varying the least amount compared to any other date. Conversely, by isolating November 21st, there is evidence to state that user-generated content was the most inconsistent throughout the observation period creating the tallest violin in Figure 5.5. Regardless, analysis can answer the second research question about the sentiment of user content during the observation period. Through exploratory analysis, the tone of language used in content associated with LIV Golf is mainly positive, regardless of outliers.
The conclusion that the overall tone of conversation is largely neutral is supported by evidence from the mean user sentiment chart. However, it is important to understand that sentiment does vary among users, and varies between content created by LIV Golf and original user generated content.
The purpose of one research question aims to understand the tone of conversation among users about LIV Golf. Therefore, the study only explored the sentiment of user-generated content and excluded LIV Golf-authored content in sentiment analysis. Aside from isolating the author of the content, additional separation was accomplished by grouping sentiment scores chronologically during the observation period. The user sentiment scatter plot illustrates each observation of user-generated content for each day the tweet was authored. By conducting this exploratory process, the research aims to understand if negative, positive, or neutral tweets dominate the conversation, and if so, what dates does this occur?
The boxplot illuminates several critical aspects of the user sentiment over time in conjunction with establishing mean sentiment values per day of observation. Evident in the boxplot is the number of outliers that reside significantly away from the mean. For example, on November 22nd, there were four total outliers; three scored in the negative range, and one observation was significantly higher than the mean sentiment score. In addition, analysis shows that mean sentiment scores are neutral or positive even though some sentiments are very positive or negative. The boxplot creates additional insight into the tone of the conversation, but this study aims to understand trends among Twitter users about the LIV Golf topic.
The violin chart contains additional evidence of the varying degree of user sentiment throughout the observation period and creates more refined conculsions from the data set. On November 25th, user content created observations with sentiment scores varying the least amount compared to any other date. Conversely, by isolating November 21st, there is evidence to state that user-generated content was the most inconsistent throughout the observation period creating the tallest violin in Figure 5.5. Regardless, analysis can answer the second research question about the sentiment of user content during the observation period. Through exploratory analysis, the tone of language used in content associated with LIV Golf is mainly positive, regardless of outliers.
The conclusion that the overall tone of conversation is largely neutral is supported by evidence from the mean user sentiment chart. However, it is important to understand that sentiment does vary among users, and varies between content created by LIV Golf and original user generated content.
Column
LIV Mean Sentiment
User Sentiment Scatter Plot
User Mean Sentiment Plot
User Sentiment Boxplot
User Sentiment Violin Chart
Emotion
Column
Emotion Analysis
While it is meaningful to understand the
tone of the conversation about LIV Golf, research required additional
analysis to gain further insight into the opinions expressed during the
observation period. The study conducted emotion analysis to answer
research questions seeking to understand feelings expressed in Twitter
content about LIV Golf on data frames of LIV Golf content and
user-generated content. The decision to conduct research in isolated
data frames was to create more refined conclusions. The conclusion from
this executing emotion analysis in this manner is that the study can
understand the differences between the emotions used in LIV Golf content
and Twitter user content. Had the data emotion analysis been conducted
on the entire data set, some of the LIV Golf emotions would have
impacted conclusions about user emotions.
Like the sentiment analysis process of scoring words used in a Tweet against a known dictionary, emotion analysis leverages similar techniques. However, words are binned and cataloged based on the emotion expressed instead of collecting tonal information about the lexicon. This process is used to understand the frequency of words used across ten distinct feelings: anger, anticipation, disgust, fear, joy, negative, positive, sadness, surprise, and trust. Emotion analysis gathers each word in a tweet throughout the collection period and stores them in each emotion bin.
The LIV Golf Emotion chart presents a histogram of the words used in LIV Golf authored content. The x-axis in the LIV Golf Emotion Chart is each bin of emotions, and the y-axis is the number of times the words conveying those emotions were counted from the data set of LIV Golf content. This analysis shows that LIV golf generally conveys a positive message with words used to express anticipation potentially about future events, singings, or initiatives. Negative word usage from the bins of disgust, fear, negative, and sadness is the least used in LIV Golf content.
The User Emotion Chart graphically illustrates the results of the analysis of user-generated content among the ten established emotions. Evident within the results is that user-generated content remains primarily positive. The feelings of anticipation, trust, and positivity are three of the four highest aggregate emotions. However, there is a notable increase in negative emotions from user-generated content. The feelings of sadness and anger occur at the same frequency as surprise. Words conveying fear and negativity present themselves at a far higher frequency than observed in LIV Golf content. The conclusion drawn from this research step illuminates that even though sentiment about LIV Golf is mainly positive, the expressed feelings convey mixed emotions among Twitter users contributing to the conversation. Additionally, users convey negative emotions far more regularly than LIV Golf.
An important consideration when interpreting the results of emotion analysis is that the frequency of emotion does not necessarily reflect advocacy or disdain for LIV Golf as an organization but reflects the feelings used in the lexicon of observations. The research does not aim to understand the allegiance of Twitter users between the PGA Tour and LIV Golf and is instead focused on capturing the emotion of content. Therefore, it is reasonable to conclude that just as it was observed in sentiment analysis, there is not a dominant emotion displayed in Twitter content attached to LIV Golf during the observation period. Users show positivity and negativity just as they express positive and negative sentiments. After additional analysis and informed by an understanding of the emotions and sentiments of Twitter users, the study will present potential future challenges to LIV Golf.
Like the sentiment analysis process of scoring words used in a Tweet against a known dictionary, emotion analysis leverages similar techniques. However, words are binned and cataloged based on the emotion expressed instead of collecting tonal information about the lexicon. This process is used to understand the frequency of words used across ten distinct feelings: anger, anticipation, disgust, fear, joy, negative, positive, sadness, surprise, and trust. Emotion analysis gathers each word in a tweet throughout the collection period and stores them in each emotion bin.
The LIV Golf Emotion chart presents a histogram of the words used in LIV Golf authored content. The x-axis in the LIV Golf Emotion Chart is each bin of emotions, and the y-axis is the number of times the words conveying those emotions were counted from the data set of LIV Golf content. This analysis shows that LIV golf generally conveys a positive message with words used to express anticipation potentially about future events, singings, or initiatives. Negative word usage from the bins of disgust, fear, negative, and sadness is the least used in LIV Golf content.
The User Emotion Chart graphically illustrates the results of the analysis of user-generated content among the ten established emotions. Evident within the results is that user-generated content remains primarily positive. The feelings of anticipation, trust, and positivity are three of the four highest aggregate emotions. However, there is a notable increase in negative emotions from user-generated content. The feelings of sadness and anger occur at the same frequency as surprise. Words conveying fear and negativity present themselves at a far higher frequency than observed in LIV Golf content. The conclusion drawn from this research step illuminates that even though sentiment about LIV Golf is mainly positive, the expressed feelings convey mixed emotions among Twitter users contributing to the conversation. Additionally, users convey negative emotions far more regularly than LIV Golf.
An important consideration when interpreting the results of emotion analysis is that the frequency of emotion does not necessarily reflect advocacy or disdain for LIV Golf as an organization but reflects the feelings used in the lexicon of observations. The research does not aim to understand the allegiance of Twitter users between the PGA Tour and LIV Golf and is instead focused on capturing the emotion of content. Therefore, it is reasonable to conclude that just as it was observed in sentiment analysis, there is not a dominant emotion displayed in Twitter content attached to LIV Golf during the observation period. Users show positivity and negativity just as they express positive and negative sentiments. After additional analysis and informed by an understanding of the emotions and sentiments of Twitter users, the study will present potential future challenges to LIV Golf.
Column
LIV Golf Emotion Chart
User Emotion Chart
Language
Column
Text Analysis
By executing textual analysis of the
content in isolated data frames, this study can conclude the main ideas
and topics of conversation based on the frequency of their use in LIV
Golf authored content and user-generated content. The output of the
analysis was the graphical representation of word usage illustrated in
word clouds and through a bigram chart showing joined words, further
illuminating the ideas and opinions of users. The LIV GOlf Worldcloud
illustrates keywords measured by their frequency of use in LIV Golf
authored content. Captured within the word cloud is the number of times
that a specific word was used in all LIV Golf tweets. The larger the
word in the illustration, the more frequently it was used; therefore,
the study can deduce that those words express the dominant ideas or
topics of conversation.
From the LIV Golf Wordcloud, it is evident that LIV Golf communicates to Twitter users content advertising the competing league, expressing opportunities to attend events, and results from past events. Highlighted in the word cloud are announcements of the Australian city of Adelaide recently announced to host an additional event to the LIV Golf season. Expressed through the words “read,” “prices,” “tickets,” and “click” are efforts to engage with Twitter users to motivate them to attend events, visit sites that report on LIV Golf, and increase overall interest in the league. The word cloud of LIV Golf authored content illuminates some aspects of the conversation about the league but identified in frequency analysis; LIV Golf content accounts for roughly five percent of the total conversation. Therefore, the study can create more refined conclusions about topics and opinions about the conversations through textual analysis of user-generated content.
The User Wordcloud illustrates the frequency of word usage from Twitter users in tweets associated with the LIV Golf topic from November 8th to November 29th, 2022. Compared to the word choices used by LIV Golf, the conversation about the emerging golf league is somewhat divided. In all tweets authored by LIV Golf, the words PGA and PGA Tour were never used; only users reference the established golf league. The word cloud does not denote allegiance, either supporting or opposing either league, but it is assumed that users are comparing LIV Golf to the PGA Tour.
Twitter users also frequently used the word “players.” In the LIV Golf content, the organization mentions players by their names to announce signings or results of a tournament. Still, in user-generated content, users are binning players as a group. It is assumed that users are talking about players on the PGA Tour or LIV Tour as a whole and expressing their opinions about the tour through opinions about players. Finally, within user-generated content, there is a significant usage of the words “Saudi,” “Saudis,” and “money .” The word usage implies that there remains a small but significant amount of content about the LIV Golf tour and its connection to the financial backing by the sovereign wealth fund from the Kingdom of Saudi Arabia. The word cloud and textual analysis do not convey whether users express positivity or negativity from their word choices. Therefore, further research is required to deduce user opinions about LIV Golf relating to its financial connection with the Kingdom of Saudi Arabia.
From the LIV Golf Wordcloud, it is evident that LIV Golf communicates to Twitter users content advertising the competing league, expressing opportunities to attend events, and results from past events. Highlighted in the word cloud are announcements of the Australian city of Adelaide recently announced to host an additional event to the LIV Golf season. Expressed through the words “read,” “prices,” “tickets,” and “click” are efforts to engage with Twitter users to motivate them to attend events, visit sites that report on LIV Golf, and increase overall interest in the league. The word cloud of LIV Golf authored content illuminates some aspects of the conversation about the league but identified in frequency analysis; LIV Golf content accounts for roughly five percent of the total conversation. Therefore, the study can create more refined conclusions about topics and opinions about the conversations through textual analysis of user-generated content.
The User Wordcloud illustrates the frequency of word usage from Twitter users in tweets associated with the LIV Golf topic from November 8th to November 29th, 2022. Compared to the word choices used by LIV Golf, the conversation about the emerging golf league is somewhat divided. In all tweets authored by LIV Golf, the words PGA and PGA Tour were never used; only users reference the established golf league. The word cloud does not denote allegiance, either supporting or opposing either league, but it is assumed that users are comparing LIV Golf to the PGA Tour.
Twitter users also frequently used the word “players.” In the LIV Golf content, the organization mentions players by their names to announce signings or results of a tournament. Still, in user-generated content, users are binning players as a group. It is assumed that users are talking about players on the PGA Tour or LIV Tour as a whole and expressing their opinions about the tour through opinions about players. Finally, within user-generated content, there is a significant usage of the words “Saudi,” “Saudis,” and “money .” The word usage implies that there remains a small but significant amount of content about the LIV Golf tour and its connection to the financial backing by the sovereign wealth fund from the Kingdom of Saudi Arabia. The word cloud and textual analysis do not convey whether users express positivity or negativity from their word choices. Therefore, further research is required to deduce user opinions about LIV Golf relating to its financial connection with the Kingdom of Saudi Arabia.
Column
LIVGolf Wordcloud
User Wordcloud
Word Network
Column
Word Network Analysis
The bigram count network counted and
ranked the usage of paired words to determine which word pairs are used
most frequently. By conducting this step during research, the study can
determine far more information about tone, emotion, sentiment, and their
connection to LIV Golf. While previous analysis began to form the
understanding of the conversation about LIV Golf on Twitter during the
observation period, the bigram network allows the study to create
additional insight. Figure 5.11 illustrates the bigram network of word
frequency from user-generated content. Displayed in the network are the
most common word pairings from the text variable of user tweets.
The bigram chart illustrates a complex network displaying varying opinions from Twitter users. Some networks do not provide additional insight into user opinions about LIV Golf. For example, the network pair of “best players” does not allow this study to conclude if Twitter users believe the best players reside on the PGA Tour or the LIV Tour or if it represents statements completely unrelated to the two golf leagues. The only conclusion is that when the word best, and the word players were used, they were most frequently used with each other. Word pairings of “join-LIV-will” likely express conversation about additional players joining the golf league. Still, it is impossible to determine if the users are conveying positivity or negativity with the word pair.
While some word pairings are mundane and don’t allow for solid conclusions about user opinion, other word pairings are unmistakably connected with user opinions. For example, in the word pairing “human-rights,” the users that authored content with that word pairing are expressing concern for the Kingdom of Saudi Arabia’s reputation regarding human rights. The topic of human rights is distinct and separate from the typical conversation about a sports league and is indicative of Twitter users confronting the topic of sports washing. Another instance where the connection between LIV Golf and the government of the Kingdom of Saudi Arabia presents itself is from the word pairings “Saudi-country-helped-kill-3000-Americans”. During the observation period, these word pairings were used frequently enough to present themselves as major topics of conversation.
The bigram chart illustrates a complex network displaying varying opinions from Twitter users. Some networks do not provide additional insight into user opinions about LIV Golf. For example, the network pair of “best players” does not allow this study to conclude if Twitter users believe the best players reside on the PGA Tour or the LIV Tour or if it represents statements completely unrelated to the two golf leagues. The only conclusion is that when the word best, and the word players were used, they were most frequently used with each other. Word pairings of “join-LIV-will” likely express conversation about additional players joining the golf league. Still, it is impossible to determine if the users are conveying positivity or negativity with the word pair.
While some word pairings are mundane and don’t allow for solid conclusions about user opinion, other word pairings are unmistakably connected with user opinions. For example, in the word pairing “human-rights,” the users that authored content with that word pairing are expressing concern for the Kingdom of Saudi Arabia’s reputation regarding human rights. The topic of human rights is distinct and separate from the typical conversation about a sports league and is indicative of Twitter users confronting the topic of sports washing. Another instance where the connection between LIV Golf and the government of the Kingdom of Saudi Arabia presents itself is from the word pairings “Saudi-country-helped-kill-3000-Americans”. During the observation period, these word pairings were used frequently enough to present themselves as major topics of conversation.
Column
User Bigram
Linear Regression
Column
Linear Regression Analysis
Research within this study aims to
understand if user activity on Twitter about LIV Golf is the result of
genuine interest in the league or is the response to LIV Golf activity
on the platform. The study asks if LIV Golf were to stop creating
content on Twitter, would user activity, awareness of the league, and
general interest subside, or is there enough interest present in the
league to sustain itself? To answer this question, the study created a
linear regression model measuring user activity levels based on the
amount of activity created from LIV Golf.
Taken from the results of the linear regression model was the intercept of 23.687 and a coefficient of 1.031. To test the model’s accuracy, the study took the residual standard error (RSE), 7.101, divided by the average of user-generated content. The formula used to determine the error rate was RSE/Mean User Content. The model created in this study has an error rate of 28%, which means that 72% of the time, user observations will fit within this model.
The study used the equation of y = a + bx to create future predictions based on the analysis in this study through the “predict” function in R. In the formula, y is the predicted value or the predicted number of user-generated tweets. Next, a is the intercept of the y-axis when the value of y is 0. Next, b is the slope or the amount that changes for every one unit increase in x. Finally, x is the value of the explanatory variable or the number of LIV Golf tweets. During the study, a test was run to predict the number of user-generated tweets based on LIV Golf creating ten tweets in one day. Research can conclude with a 72% level of certainty that if LIV Golf created ten tweets in one day, of the 213,702 followers, 34 tweets would likely be created by users.
To visually understand the relationship between the independent and dependent variables, a regression line was created and available in the Linear Regression Chart. The y-axis in the chart is the number of user generated tweets, and the x-axis is the number of LIV Golf tweets. The red line in the chart is the regression line created in this study. Shaded in grey in the chart is the standard error or level of certainty based on regression analysis. For prediction purposes, the number of user tweets is expected to fall close to the red line, and there is a reasonable certainty that predictions will fall within the grey-shaded portion of the chart.
Taken from the results of the linear regression model was the intercept of 23.687 and a coefficient of 1.031. To test the model’s accuracy, the study took the residual standard error (RSE), 7.101, divided by the average of user-generated content. The formula used to determine the error rate was RSE/Mean User Content. The model created in this study has an error rate of 28%, which means that 72% of the time, user observations will fit within this model.
The study used the equation of y = a + bx to create future predictions based on the analysis in this study through the “predict” function in R. In the formula, y is the predicted value or the predicted number of user-generated tweets. Next, a is the intercept of the y-axis when the value of y is 0. Next, b is the slope or the amount that changes for every one unit increase in x. Finally, x is the value of the explanatory variable or the number of LIV Golf tweets. During the study, a test was run to predict the number of user-generated tweets based on LIV Golf creating ten tweets in one day. Research can conclude with a 72% level of certainty that if LIV Golf created ten tweets in one day, of the 213,702 followers, 34 tweets would likely be created by users.
To visually understand the relationship between the independent and dependent variables, a regression line was created and available in the Linear Regression Chart. The y-axis in the chart is the number of user generated tweets, and the x-axis is the number of LIV Golf tweets. The red line in the chart is the regression line created in this study. Shaded in grey in the chart is the standard error or level of certainty based on regression analysis. For prediction purposes, the number of user tweets is expected to fall close to the red line, and there is a reasonable certainty that predictions will fall within the grey-shaded portion of the chart.
Column
Linear Regression Chart
Conclusion
Conclusion
This study aimed to understand and measure
sentiment, opinions, and reception, then model the potential amount of
conversation about LIV Golf on Twitter. By conducting a thorough
analysis of the sentiment of content about LIV Golf, the study can
conclude that even though the establishment of the competing golf league
has divided the golfing world, the conversation remains largely
peaceful. The study found evidence of resistance to LIV Golf through
textual analysis because of the connection between the organization and
the government of The Kingdom of Saudi Arabia. While LIV Golf has spent
billions of dollars creating a competing golf league, Twitter users’
conscience and awareness of the practice of sports washing remained a
significant topic of conversation between the 8th to 29th of November
2022. Even though the number of followers increased during the study,
accusations of LIV Golf acting as a mechanism for a foreign government
to improve its global reputation will likely limit its speed of
acceptance.
References
References
Argawal, A., Xie, B., Vovsha, I., Rambow, O., Possonneau, R., (2011). Sentiment Analysis of Twitter Data. Retrieved November 22, 2022, from https://aclanthology.org/W11-0705.pdf
Camenker, C. (2022). What Does LIV Golf Stand For? Explaining the Name, Meaning of the Saudi-Backed Invitational Tour. Sporting News. Retrieved November 24, 2022, from https://www.sportingnews.com/us/golf/news/liv-golf-tour-name-invitational-series-meaning-explained/ietdisstrdtxa8k3u2s5c5kq
Chowdury, A. (2012). The State of Twitter Spam. Retrieved November 26, 2022, from https://blog.twitter.com/official/en_us/a/2010/state-of-twitter-spam.html
Kim, A, Hansen, H., Murphy, J., Richards, A., Duke, J., Allen, J. (December 2013). Methodological Considerations in Analyzing Twitter Data, JNCI Monographs, Volume 2013, Issue 47, Pages 140-146, retrieved November 27, 2022, from https://doi.org/10.1093/jncimonographs/lgt026
Smith, B. (May 31, 2012). Pew Internet & American Life Project. Retrieved November 27, 2022, from http://pewinternet.org/Reports/2012/Twitter-Use-2012.aspx.
Wojtowicz, J., Fruh, K., Archer, A. (2022). Sportswashing: What It Is, Who Does It, and How to Stop It. Liberal Current. Retreived November 25, 2022, from https://www.liberalcurrents.com/sportswashing-what-it-is-who-does-it-and-how-to-stop-it%EF%BF%BC/