Tweaking my proposal after diving into the data

And so it begins. I hit the ground running on my research project this summer, wasting no time diving right into the data. After spending the past few months outlining a research project informed by an in-depth literature review of my topic and my proposed data sources, my first task was determining which parts of my proposal were actually feasible and subsequently tweaking my project. Over the course of my first week, I livestreamed current Tweets, ‘hydrated’ historical Tweets, recorded historical Twitter data by tediously scrolling through athletes’ public profiles, downloaded Google Trends data, and requested access to survey data to obtain a better sense of my options. Then I examined the accessibility and feasibility of using these kinds of data to answer four key questions: 1) which athletes are speaking out about politics, 2) what are they saying, 3) who is listening, and 4) how are people responding. I quickly discovered that many of the methods that I believed I could use in November were infeasible, while many methods that I had not previously considered provided opportunities for additional analysis. At the end of the week, I finalized a new project outline that I have included below which is more comprehensive and feasible than my original proposal.

Question 1: Which athletes speak out about politics?
Hypothesis: Three factors make athletes more likely to speak out about politics: minority status, intrinsic factors (age, college degree, degree of outspokenness), and external characteristics (all-star status, followers, high salaries, contract security, and liberal fans)
Method: I will download tweets from every American starting NBA basketball player between 8/1/17 and 10/31/17. Then I will determine which athletes mention political key words. Using a probit model, I will regress being a political tweeter on the following variables: salary, age, number of followers, number of tweets, all-star status (current NBA all-star), team’s state Trump vote percentage, college grad, minority status, and years left on their contract. This will allow me to determine which factors cause a statistically significant increase on likelihood of tweeting about politics.

Question 2: What are these athletes saying?
Hypothesis: NBA players are most likely to address racial injustice and police brutality
Method: I will create a word cloud and determine which key words and topics occur the most among NBA players’ political tweets

Question 3: Which of these athletes’ tweets have the most influence? Which topics are the most popular?
Hypothesis: Tweets from athletes with all-star status, more followers, and higher salaries will be favorited, retweeted, and commented on more. Tweets addressing police brutality or racial injustice will also be favorited, retweeted, and commented on more.
Method: I will regress the number of favorites, retweets, and comments on the political tweets I collected earlier on whether the tweets mentioned different keywords (Trump, Obama, President, America, White House Invite, Kneeling, Charlottesville, Racism, Police), the number of followers the tweeter has, and whether the tweeter has all-star status.

Question 4: Can athletes’ offline protests dominate national conversation over Twitter?
Hypothesis: The most popular keywords and hashtags in tweets mentioning the word “White House” the day that the Philadelphia Eagles were disinvited from the White House will reference the Eagles’ protest. The most retweeted users will also include many sports accounts. These trends will differ dramatically from tweets mentioning ‘White House’ a week later.
Method: I livestreamed tweets mentioning the keyword ‘White House’ the day that the Eagles were disinvited from the White House and a week after. I will create a code in python that will allow me to compare the top keywords, top hashtags, and most retweeted users between these days. Note: this is a case study.

Question 5: Do athletes’ offline protests cause people to research political issues?
Hypothesis: Following every athlete protest covered by ESPN since 2012, the number of Google searches for racial injustice and police brutality will increase.
Method: I will collect Google Trends Data and determine whether increases in searches for the keywords ‘racial injustice’ and ‘police brutality’ are more frequent following athlete protests.

Question 6: Who listens to athletes? Who is more likely to support their protests?
Hypothesis: Minorities, young people, college grads, and democrats will be more likely to support athlete protests than whites, older people, high-school grads, and republicans, however these will be more likely to support athlete protests than the protests of regular Americans.
Method: I will consult a survey released by CNN following Colin Kaepernick’s protest to determine which factors make individuals more likely to support athlete protests and protests from regular citizens.

Question 7: Do athlete protests change the way people talk about political issues?
Hypothesis: People will be more critical of racial injustice and police brutality when they are directly responding to athlete protests.
Method: I will compare the sentiment of tweets using the hashtag #blacklivesmatter that mention keywords related to athlete protests to tweets that do not mention keywords related to athlete protests.

Speak Your Mind