Study title

Full edition for scientific use. The AUTNES dataset on TV confrontations contains data collected from fifteen debates between pairs of top candidates in the 2013 national election campaign. The debates were televised by the Austrian national broadcasting company ORF between August 29th and September 24th, 2013. All statements spoken during the debates are part of the dataset. Natural sentences are the unit of analysis. The coding procedure applies the AUTNES relational approach of recording subjects, predicates, and objects to TV confrontations. The subject is defined as an actor that positions himself/herself towards an issue in the sentence. There are two types of objects: issues and object actors. The one issue that fits the content of the spoken sentence best is selected by coders from the AUTNES issue coding scheme. The issue predicate numerically records whether the subject’s position towards the issue is one of support, rejection/criticism, or conveys a neutral stance. Up to three object actors are recorded per sentence, each with their name (if an individual is present), organisational affiliation and political relevance, as well as their evaluation by the subject actor (positive, negative or neutral). In addition to the basic subject–predicate–object structure, we code character traits and party records for all subject and object actors as well as justifications for issue statements. Variables: Variables referring to the debate and coded sentence: case-ID; debate-ID; sentence-ID; sender (TV station broadcasting the debate); date of the debate; begin time and end time of the debate; duration of the debate; party affiliation and name of discussant 1 und discussant 2; name of the debate’s moderator; organisational affiliation and name of the sentence’s speaker (various actors besides the participants of the debate, i.e. journalists, experts, etc.); sentence; sentence length in words; sentence length in characters; variables referring to the subject actor:...