Source
Barnehl and Schumacher, “Media capture, captured”, Democratization 31(7): 1536–1563 (2024), LIT-1323, read in full in NOTE-1047.
What was actually shown
144,700 articles from the twelve of Nicaragua’s 27 national outlets with an active web presence were modelled with a 30-topic structural topic model, with outlet and date as covariates. Topics relevant to the regime’s legitimacy were chosen in advance. Sentences mentioning Ortega were classified negative, neutral or positive. For each nominally independent outlet (Radio Corporación, Canal 10, Canal 14) and each topic, the score [(d(N, Reg) − d(N, Opp))/d(Opp, Reg) + 1]/2 runs from 0 (as the regime-owned outlets) to 1 (as the opposition outlets). It is weighted across topics by how far apart the poles are, and computed the same way for tone.
All three outlets’ agenda scores fell over 2018–2022. In 2018–2019 they covered the legitimacy topics roughly like the opposition. From the second half of 2019 Canal 10 and Canal 14 moved toward the regime, and Canal 14 did not recover. On tone, the opposition averaged about −.4 and the regime outlets about .2. Canal 14’s tone score fell from about .8 to below .5 by the first quarter of 2022, and Radio Corporación’s stayed close to 1, dipping slightly from 2021.
What could have come out otherwise: the two poles could have failed to separate on the chosen topics or on tone (checks II and III), and the outlets’ agenda and tone could have disagreed about which outlet moved (check V). They did not.
What this does not say
- Not that the outlets were coerced. The paper defines capture as requiring coercion by an external actor, and the scores measure alignment. “it is impossible to definitively reject” that a shift was a free editorial choice. Coercion is argued from the documented harassment in the period.
- Not that the poles were fixed. An opposition outlet that became more critical in exile would lower an unchanged outlet’s score, as the authors note.
- Not with stated uncertainty. No tests or intervals for the scores appear in the main text.
- Not beyond Nicaragua, three outlets, and the researchers’ choice of legitimacy topics.