Date of Completion

2020

Document Type

Honors College Thesis

Department

Mathematics and Statistics

Thesis Type

Honors College

First Advisor

Christopher M Danforth

Second Advisor

Peter Sheridan Dodds

Third Advisor

Matthew Price

Keywords

mental health, stigma, natural language processing

Abstract

Major depression is a serious health issue afflicting hundreds of millions of people each year, with many going untreated due to the intense stigma surrounding mental illness. In this project, we explore perceptions of mental health on social media, attempting to quantify the level of stigma present on Twitter and track how it has changed in the past decade. To explore trends in the appearance of various words and phrases, we collect roughly 10% of all tweets starting in 2008, process English tweets into 1-, 2-, and 3-grams, and determine their usage frequency and rank. Using these values, we can examine how often the topic of `mental health' is discussed on Twitter, and we find that the phrase has increased in rank by an order of magnitude since 2013. We attempt to disentangle the components of this rise in prevalence, determining how much of the rise is explained by decreased stigma and how much is explained by a convergence in linguistics. We look at messages containing `mental health' posted in 2012 and 2018, as these years are before and after the drastic increase in rank of this phrase, and examine the divergence of the language in both subsets. In further efforts to measure stigma, we compile a list of negative labels commonly used in stigmatizing language and track the rank and frequency throughout the past decade, and we find that many of these labels have decreased in rank in recent years. We also identify statements of self-disclosures of Twitter users and examine how many appear over the years in a subset of tweets specifically about depression. These results all provide valuable insight into how the discussion around mental health has shifted over time.

Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 License.

Share

COinS