A framework measures ownership patterns in news
Measuring Organizational Influence on Digital News: An Information-Theoretic Framework for Media Pluralism and Editorial Independence · arXiv:2610.40172
In plain words
Carragher and Carley introduce a way to measure how ownership relates to news outlets’ choices, such as which events they cover and which stories they feature. It separates overlap linked specifically to an owner from overlap that is common across the wider news field.
In digital news data, pairs of outlets with the same owner shared more information about their coverage choices than a comparison with owner labels shuffled. This shows a measurable pattern, but does not prove that owners caused it.
The research
News outlets often make similar editorial choices, but how can we tell whether that similarity reflects their common owners or a news agenda shared across the wider field?
Without a comparable measure that separates those patterns, it is difficult to assess how ownership relates to editorial independence.
The authors separate owner-specific information in outlets’ coverage from information shared across the wider news field, then apply that measure to digital news.
They apply the measure to events covered, sources cited, and stories featured on outlet homepages.
Same-owner pairs shared more information across all three choices than pairs in a comparison with owner labels shuffled, while a separate model linked more event-selection variation to owners than to network affiliates.
These comparisons show that ownership patterns can be measured apart from the wider news agenda. The authors also find acquired outlets experience sustained independence losses sooner than matched never-acquired outlets.
These are observational patterns, so they do not establish that owners caused outlets to make similar choices or that acquisitions caused the measured changes. The event filter can miss local, investigative, or niche reporting, and crawler blocks create gaps in the data.
What the evidence shows
- A measure for editorial choices
Carragher and Carley introduce an information-theoretic framework to measure how ownership relates to editorial independence and how political-bias groups relate to internal pluralism. The question is difficult because outlets may make similar choices simply because they respond to the same news agenda.
Earlier research often relied on individual cases and different methods, making comparisons across outlets difficult.
- Separate group and field
The framework compares an outlet’s coverage with counts of coverage by other outlets in its group and by outlets outside it. Partial information decomposition separates information uniquely associated with the group from information shared with the wider field.
This helps prevent common coverage, such as a broadly shared news agenda, from automatically counting as owner dependence. Ownership and political-bias groups apply the same basic construction to different questions.
- Compare measured patterns
The authors apply the measure to digital outlets’ event selection, source selection, and front-page placement. Same-owner pairs shared more information than the shuffled-owner comparison across all three decisions.
A separate model also found more variation in event-selection independence among outlets sharing an owner than among outlets sharing a network affiliation. Together, these results show that the framework can distinguish and compare organizational patterns in measured editorial choices.
- Follow change around deals
The authors then follow acquired outlets and matched never-acquired outlets. At six months, acquired outlets had a higher rate of sustained independence loss, using a rule that required the measure to remain more than one standard deviation below its pre-deal level for at least two months.
Targets in 11 blocked or abandoned deals did not close their coverage gap to the would-be acquirer after one year, a contrast consistent with convergence after completed acquisitions but too small and nonrandom to establish causation.
- Bound the conclusion by medium
A comparison of station websites and broadcast captions shows that the pattern depends on the medium: in sampled web data, linguistic redundancy follows ownership, while in sampled broadcast data it follows network affiliation.
The authors’ results therefore support measuring ownership patterns in digital editorial choices, while the acquisition comparisons and media contrast bound what can be inferred from them.
Deep
- Original paper authors
- Peter Carragher, Kathleen M. Carley
- Submitted to arXiv
- License
- CC BY 4.0