This section gives a description of the manual coding process. The coding was done using the online tool Angrist (Wettstein, 2016).
All material was double coded and inconsistencies between coders were resolved using consensus discussions. All coder pairs were frequently rotated over the material as well as the pairs itself to avoid systematic errors within and between coder pairs.
The coding process consisted of two major components:
In a first step coders were asked to identify and mark all actors, positions and justifications within the news item (coding step 2 to 4 in the codebook) using the coding tool Angrist. Afterwards they discussed their markings with their assigned coding partner adjudicating them via consensus discussion. All news items with final markings were submitted into a pool from which they were randomly assigned to another coder pair for the second coding step.
In a second step coders coded the already marked units of analyses, following the instruction described in step 5 to 10 in the codebook. After their initial coding they again discussed their codings with their assigned coding partner to dissolve all differences via consensus discussion. The Final codings were submitted through Angrist.
Note. News items available in none plain text format (pdf) were marked manually and the markings copied into Angrist, because direct markings in Angrist were not possible. The second step (Coding) was done completely with the assistance of Angrist in the same way as the news items available in plain text format.
Wettstein, M. (2016). Angrist 1.21. Retrieved from https://www.ikmz.uzh.ch/dam/jcr:daf64af8-9ec5-4fdc-a4ef-90cf7db10824/ANGRIST_1-2-en.pdf