This article examines US congressional candidates’ messaging strategies during the 2018 midterm campaign. Using every candidates’ tweets, in conjunction with an original dataset of 280 million tweets matching a dictionary of election related keywords in the fifty days leading up to the election, we uncover variation in how potential legislators publicly presented themselves and how the electorate responded. Using an affinity model approach of text analysis, we show how candidates’ electoral circumstances, partisanship, and descriptive traits dictated their messaging strategies, such as their willingness to discuss polarizing issues or comment on the partisan issue of the day. In doing so, we assess conventional claims about the importance of Donald Trump, Brett Kavanaugh, health care, and immigration in Democrats’ and Republicans’ respective campaigns. Additionally, we compare these messaging approaches with the political conversations occurring simultaneously in the districts of
candidates using geocoded tweets. Our results highlight the importance of electoral conditions in how campaign appeals are developed and their relationship with election returns.
