X and Online News Moved in Different Directions by Time of Day During the Istanbul Race
In Türkiye’s polarized media environment, the relationship between X and online news was anything but uniform. A large election-period analysis found that agenda leadership changed with the clock, with morning online news more often preceding later same-day X issue salience, while evening X issue salience more often preceded the next morning’s online news agenda. The pattern was not mirrored cleanly at the attribute level, where cross-platform relationships were weaker and less stable.
The study examined the 31 days leading up to and including the 31 March 2024 Istanbul local election, a politically charged contest in a city with outsized national significance. Researchers analyzed 448,664 X posts and 10,357 articles from 19 mainstream online news outlets, then split the data into 12-hour AM/PM windows. Across 122 cross-lagged panel analyses, the direction of agenda flow depended on both timing and agenda level, revealing a media system that behaves more like a shifting relay than a fixed hierarchy.
Issue Agendas Aligned Far More Closely Than Attribute Agendas
The first-level issue agendas were strongly synchronized, with correlations often approaching or exceeding r = .90. In practical terms, the platforms were tracking the same broad topics with striking regularity. Election news, campaign conflict, political maneuvering, and other issue-level signals moved in near lockstep far more often than the finer-grained language used to describe those issues.
Attribute agendas told a different story. Second-level relationships were weaker and less consistent, suggesting that the two arenas can agree on what deserves attention without converging as tightly on how that attention is framed. In a polarized pluralist system, that split makes sense: newsrooms and social platforms may share exposure to the same campaign events, yet their editorial routines, ideological filters, and audience expectations still shape the descriptors attached to those events.
The difference is not trivial. Issue salience can be synchronized by the same breaking story, the same candidate appearance, or the same institutional announcement. Attribute salience requires another layer of alignment, because it reflects which characteristics, tones, or evaluative cues rise to prominence around the issue. When those features remain loose, the media agenda may look shared at the surface while still diverging in texture and emphasis.
Morning News Led Same-Day X Attention, While Evening X Often Led the Next Day
The most consequential pattern emerged in the timing of cross-lagged effects. During same-day transitions, morning online news issue salience more often preceded later X issue salience. During overnight transitions, evening X issue salience more often preceded the following morning’s online news issue salience. A single day could therefore contain two different leadership structures, each pointing in a different direction.
This temporal asymmetry matters because it suggests that agenda leadership is not a stable property owned permanently by either platform. It shifts with the news cycle, the pace of electoral developments, and the hours in which audiences and editors are most active. In the morning, newsroom production may still set the day’s issue frame. By evening, the accelerated circulation of posts on X may push particular topics into the next editorial cycle.
The pattern fits a hybrid media environment in which journalism and social media continuously react to one another. X is not simply a mirror of mainstream coverage, and mainstream coverage is not merely a delayed replay of social media chatter. Each platform can temporarily assume the lead depending on when the issue is sampled and where the cycle sits in the day.
Cross-Lagged Associations Outpaced the Rozelle–Campbell Baseline More Often at the First Level
Most first-level cross-lagged correlations exceeded their Rozelle–Campbell baseline, strengthening the case that the observed temporal ordering was more than simple contemporaneous synchronization. At the issue level, then, the timing signal carried enough structure to suggest a directional relationship beyond shared exposure to the same campaign environment. The analytical contrast with the baseline sharpened the interpretation without turning the findings into a claim of causality.
Second-level coefficients crossed that baseline less consistently. In other words, the more granular attribute agenda was less likely to show a temporal pattern robust enough to stand apart from synchronized movement. That result is consistent with the idea that attribute transfer requires stronger editorial resemblance than issue selection does. It also implies that the descriptive layer of media content is more resistant to cross-platform ordering than the agenda of topics themselves.
The authors were careful not to overread the statistics. These were temporally ordered associations, not proof that one platform directly caused changes in the other. Even so, the pattern is difficult to dismiss as noise when it repeats across 122 panels and holds more strongly for issues than for attributes.
Türkiye’s Polarized Media System Shaped the Tempo of Agenda Leadership
The setting matters. Türkiye’s media system combines extreme political polarization, concentrated ownership, and strong political parallelism, creating a high-pressure environment for agenda competition. Istanbul, as the country’s largest electorate and a major financial center, concentrates political attention in ways that intensify both newsroom vigilance and social media response. During an election campaign, those pressures can compress the distance between what is trending online and what is published in news outlets.
This study also fills a gap in the Türkiye IMAS literature, where empirical work has been comparatively scarce and where first-level and second-level agenda-setting have rarely been analyzed in the same integrated design. By combining both levels, the analysis shows that broad issue attention and finer descriptive framing do not travel across platforms with the same speed or reliability. The result is a more segmented picture of intermedia influence, one in which timing and content type both shape the relationship.
The scale of the dataset strengthens that reading. With 448,664 X posts and 10,357 online news articles, the analysis captured a dense slice of campaign communication rather than a narrow anecdotal sample. The use of 12-hour windows also allowed the study to detect intraday differences that a coarser daily design could easily miss.
What the Election Cycle Reveals About Hybrid Agenda Power
The study’s central contribution is a simpler claim than it first appears to be: agenda leadership in a hybrid media system is temporally contingent and level-specific. Morning news can lead afternoon social attention. Evening social attention can lead the next morning’s news. Issue agendas can move together with striking strength, while attribute agendas remain more uneven and fragmented.
That pattern has direct implications for how political communication is studied in high-velocity media settings. If researchers collapse the day into a single unit, they may miss the alternating leadership that emerges within it. If they treat issue salience and attribute salience as interchangeable, they may overstate the extent to which platforms share the same agenda-making power.
The final caution is built into the findings themselves: these are associations, not causal demonstrations. The data show order, not direct influence. Any interpretation beyond that would go past what the analysis can support, and the distinction is especially important in a campaign environment where both platforms respond to the same political shocks, not just to one another.