Research has shown that voters have partisan perceptual filters when engaging with political information, which leads them to seek confirmatory information, and to treat opposing information with a high degree of scepticism. The United States of America has a particularly polarized electorate, which should lead to a strongly observable trend showing differing responses to negative shocks in an election campaign across partisan lines. This paper applies methods from the computer science field of sentiment analysis in a new way to a dataset of over 57 million tweets gathered during the 2016 US presidential election campaign. Using these techniques, I investigate the extent of partisan biases in reaction to the email scandals which affected Hillary Clinton’s campaign, and the sexual harassment allegations which affected Donald Trump’s campaign, as well as tracking the polarization of opinion following the official election debates. By looking at the differing shifts in Twitter sentiment between the states which Trump and Clinton won on election day, over the course of the campaign, I test theories of motivated scepticism, and partisan bias. While broadly finding results which confirm the existence of partisan bias, I also find some unexpected results surrounding reaction to events affecting Trump negatively in Republican states, suggesting that there may be multiple effects in play beyond perceptual biases. Following the discussion of this paper’s results, I raise new areas of research, which follow from the findings and methodologies of this paper.