Summary of Research Findings

As I have repeatedly mentioned in my blog posts, this summer has been one for the books. I have come out knowing more about myself as a person and as an academic. For starters, I need to constantly be setting deadlines and challenging myself in order to actually get work done. Working in as much interpersonal interaction into a workday will also aid in motivation. As an academic I know how I face trials and tribulations such as a challenging code segment day in and day out.

In the realm of academics, I’ve learned unquantifiable amounts regarding the research process. Although I started researching my intended paradigm last November, it wasn’t until truly starting the study that I understood where my problems were and how to go about intense research. The research process, once an abstract to do list, is now familiar (albeit still daunting).

My summer has been structured into several parts:

Getting to know Twitter API and basic python functions. Throughout this period I read countless articles on the best way to analyze tweets. I investigated the “OAuthHandler” setup with Tweepy and have a registered consumer code/access token which allow me to stream tweets. In Python I got to know Tweepy and basic functions such as how to load a file/have Python create a written text file. Once I had baseline understandings of Twitter data and Python, I went on to merge the two. The next chunk of my time was getting to know Twitter through Python; seeing what information I could summon and how to translate it into .txt files and graphs. This was by far the most challenging part of the summer. Heading into the summer with no knowledge of Python, I emerged from this stunt as still amateur but with a proficiency in the program.

For the next few weeks I manually scrubbed data myself, collecting images/screengrabs and inputting statistics such as the number of tweets during a specific time frame, the average number of responses to a tweet, the content of said tweet (whether an image or a video). I focused on gubernatorial elections from 2016 and 2017. VA was the most statistically interesting by far (although a lot of information lies in the lack of tweets in 2016, I believe). Based on what I have found: Democrats tend to tweet far more than republicans, and to use more negative-press tactics. Republicans tend to use Twitter for policy. Most of the tweets I gathered this summer focused on healthcare.

Looking towards the future, I hope to take the data that I collected and my knowledge of python on towards an honors thesis. Many of the screen grabs i have obtained could be adapted for an experiment, and I am interested in submitting an outline to the Omnibus survey. I intend to get started on that proposal in the next few weeks.

I look forward to presenting the fruits of my labor in more detail this fall at the Charles Center showcase.

Speak Your Mind