The data set of the survey comprises content analysis of the entire 24-h programme of the six largest, with respect to audience size, TV stations with nationwide coverage, namely ALPHA, ANT1, MEGA, SKAI, STAR and the public broadcaster ERT. These data were collected from the day after the referendum proclamation, until the day of its conduct, that is between 29 June and 5 July 2015. The project involved 15 coders consisting of academic staf, Ph.D. candidates and graduate and undergraduate students. From the examination and coding of the research material, two distinct data sets were created: in the first one, the media units concerning the referendum were coded, while in the second one, the "talking heads" that appeared in each unit/broadcast constituted the unit of analysis. Each of the 4,917 people (talking heads) who appeared in these segments separately was coded, excluding the host journalists of the shows/news bulletins. The data were then analysed according to the principles of quantitative content analysis. For each unit of analysis, a group of variables was coded. The construction of the variable categories was further derived from a process of qualitative thematic coding. This method was applied to deal with the exceptional characteristics of the referendum campaign.

Total universe/Complete enumeration

Content coding