This morning I rolled out two new graphs for PoliticsEQ. On the keyword pages you can now see a breakdown of average keyword sentiment by publisher and by detractors, which I've defined as the 10 writers who have written at least 10 articles on a particular keyword whose sentiment is the most negative.
This data appears to confirm what we already knew but had no measurable proof of until now. That is, that different publishers treat presidential candidates differently.
The graphic below is a comparison of negativity around the keywords Donald Trump and Hillary Clinton by publisher. What we see is a clear representation of bias on the part of some media outlets. Politico and Huffington Post are way harder on Trump than FoxNews ,for instance. And the converse is true as well. When it comes to Hillary Clinton, FoxNews leads the pack on negative sentiment while Politico is the most forgiving.
There are also some anomalies here as well. I would have expected Breitbart to be among the softest on Donald Trump but in fact, while they are definitely not as hard on him as Politico, they aren't as easy on him as FoxNews. This is likely NOT because Breitbart is a fair and impartial source of journalism, but rather an overall editorial tone affecting the coverage of all subjects. It might be fun to come up with an adjustment for "house" tone. I suspect this is likely affecting Vox.com's sentiment score as well given their tone is substantially less emotional than other sources.
It is also worth noting that Hillary Clinton's keyword sentiment scores are considerable, as well as consistently, more negative than Trump. What could cause this isn't clear. Perhaps there's a general negativity around Clinton? Or perhaps there is sarcasm aimed at Trump that the Tone Analysis is improperly scoring. Or perhaps there are some gender-related effects in play. In the future it would be fun to compare to other politicians. Right now we simply don't have the resources.