The media play a critical role in electoral politics, and recent research has shown that gendered media coverage of politicians can have measurable electoral consequences (Van Remoortere and Vliegenthart 2023; Van Erkel et al. 2020). Building on this, I provide the first systematic test of how well-established categories of media sexism (Van Der Pas and Aaldering 2020; Shim 2025)—e.g., coverage volume, tone, policy categorization, personality traits, gender identifiers, and private-life emphasis—affect the re-election prospects of legislators. While prior studies often document that female legislators receive less media attention, are more often linked to “feminine” traits and issues, and are framed through gendered identifiers, it remains unclear whether different categories of media sexism damage electoral careers to different degrees.
To address this question, I examine two understudied but theoretically illuminating cases: South Korea and Taiwan (2004–2024). The comparison enables a two-dimensional mediation test: (i) the proportion of women in the legislature (Taiwan surpassed the 30% “critical mass” threshold in 2012, while Korea has remained below 20%); and (ii) the electoral tier under mixed electoral systems—both democracies have mixed systems with district and party-list tiers.
I employ a mixed-methods design. Quantitatively, I use regression analyses on an original dataset that links media coverage of all elected legislators in the major national newspapers to their subsequent re-election outcomes. Qualitatively, I conduct 30 elite interviews with legislators from across parties to illuminate how politicians with high degrees of media sexism exposure perceive, internalize, or strategically counteract gendered media portrayals.
The findings demonstrate that not all media sexism categories are electorally equivalent. For instance, policy-based gendering exerts clear negative effects, and these effects are disproportionately harmful for district-tier legislators in South Korea, where media sexism interacts with low female presence and clientelistic expectations. The paper advances research on gendered mediation by disaggregating sexism effects, integrating electoral system logics, and showing how levels of descriptive representation shape the magnitude of media-bias penalties.
