Fig. 4: Distribution of keywords of the replies in a textual scatterplot.
From: Different honesty conceptions align across US politicians' tweets and public replies
Every term is a dot with two coordinates associated with the user ideology score (x-axis) and keyness of a word in determining the honesty component (y-axis). Each axis represents a Scaled F-Score (SFS) ranging from −1 to 1 (see “Methods” for further details). The word color signals how representative the term is for the two ends of the ideological spectrum (i.e., red = more conservative; blue = more liberal). We only show word labels where SFS > 0.5 or SFS < −0.5 for readability reasons.
