Fig 1.
Circle flowchart for operationalizing a sample of political influencers and their generated content.
Factors relevant before data collection are depicted in light gray, whereas post-data-collection factors are depicted in dark gray.
Fig 2.
Content produced by political influencers on Instagram.
Percentages denote the share of posts for each influencer in our sample coded as belonging to either category. Example interpretation: The median account includes advertisements in about 31% of all posts.
Fig 3.
Weekly percentages for content types of posts by political influencers.
Points denote average weekly values, and bands denote loess-smoothed uncertainty estimates. Example interpretation: In week 26, about 29% of posts included political content and advertisements compared to about 18%, including support or disapproval of political entities and events.
Fig 4.
Socio-demographics and item responses of survey participants and their association with being aware of or following influencers.
Points depicted in the forest plot denote logit coefficients from two logistic regression models (one for each dependent variable). Bars denote 95% confidence intervals.
Fig 5.
Helpfulness of influencer content for voting decisions of survey respondents.
Panel A. Various items for how helpful factors were in determining their voting decisions for the federal election. The Likert scales range from 1 - not helpful to 7 - very helpful, with 4 representing the neutral category. Numbers indicate the cumulative percentages for neutral, helpful, and unhelpful indicators. Example interpretation: 6.4 percent of respondents indicated that influencers’ content helped determine their voting decision, whereas 38.1 percent found television news helpful. Panel B. Violin plots for respondent age by sex and respondents’ dichotomized answer to how useful influencer content is for their voting decisions. Example interpretation: the median age of male respondents who do not find influencer content helpful for their voting decisions is 56, while the median age of male respondents who find this content helpful is 35.