“…Moreover, using the language of the case syllabus, a shortened description of the case often relied upon in the coding of Supreme Court cases (Spaeth et al, 2012, e.g.,), is exceedingly unlikely to provide an accurate signal of the opinion's sentiment. The latter option, training a classifier based on previously utilized measures such as voting patterns (e.g., Corley et al, 2010;Corley, Steigerwalt and Ward, 2013) or dissenting and concurring behavior (e.g., Walker, Epstein and Dixon, 1988;Caldeira and Zorn, 1998;Hendershot et al, 2013), carries perhaps greater appeal. These previously utilized metrics are, certainly, reflective of divisions of opinion on the Court.…”