New York Times – Political Candidates

produced using: R (packages – wordcloud, stringr, tm, snowballC, lsa)

Description

This project looks at a total of 550 NYTimes editorials from 5 notable columnists that appeared in print from January 2016 to early 2017. This text analysis focuses on comparing the linguistic complexity and general attitudes between columnists and the candidates.

Most Used Words

As seen in the chart below, it appears that the most common word used between all 5 authors is “Trump” (which was used almost 3,000 times). The word “people” is the second most used word while “president” and “Clinton” are the 14th and 15th most frequently used words.

Linguistic Complexity

The graph below shows the Linguistic Complexity (which corresponds to the “grade-level” of the writing) of each author through time. In general, most of the author’s complexity grade levels stayed consistent from April 2016 to March 2017. However, both Brooks and Dowd are exceptions since their Linguistic Complexity Grades have increased starting around April 2016.

This also compares the Linguistic Complexity of each author by using boxplots. Kristof appears to have the widest range of Linguistic Complexity in his articles.

Sentiment Analysis

This displays the range of sentiment used when writing about each presidential candidate. For Trump, the range is extremely wide with some articles writing very positively about him while other articles were extremely negatively. Clinton had both negative and positive sentiments, but never reached the same range as Trump.

This chart shows the change in sentiment over time by each author for the candidates. Brooks initially produced positive content about Clinton but then the sentiment experiences a dip around April 2016 (perhaps this is due to Bernie losing the primary?). Dowd had positive content about Trump around January 2017 (around inauguration day) but since then her content has grown more negative.