VisWorkshops In our workshops, viewers who found the data source of a visualization reputable also tended to trust it.

SciAmMessageAnalysis In visualization captions that included methodological details, such as data collection, these were appreciated for helping coders generate more complete and accurate messages.

CrisisMapSensemaking In comments and interviews on crisis maps, viewers expressed distrust when parts of maps were left blank without explanation, interpreting missing data as a lack of credibility.

SciAmInterviews Data visualizations at the popular science magazine Scientific American typically included references to data sources, distinguishing between peer-reviewed and raw datasets. When combining multiple sources, they aimed to specify which part of the visualization draws from which data. They also aimed to credit all contributors involved in visualization production.

ClimateVisInterviews Both experts and lay viewers criticized the use of complex or abstract language for providing context, such as in captions or annotations. Lay viewers especially struggled with technical terms and high-level descriptions when it came to future scenario descriptions in climate data visualizations.

SciAmMessageAnalysis Our study found that older captions, stemming from the 1980s or so, often used high-level, scientific vocabulary, such as for describing uncertainty or data processing, which coders found harder to distill into messages. Newer captions, such as from 2020 visualizations, used plainer language and helped convey overarching narratives more effectively.

VisProducerInterviews Practitioners stressed the importance of clear language in annotations, labels, and titles and called learning to simplify the most essential advice for newcomers.

SciAmMessageAnalysis Field notes showed that missing context, such as the absence of captions, led to irritation. Then coders often guessed, relied on assumptions, or misinterpreted the visualization. If context was provided, for example through small captions, this fostered message interpretation notably.

SemanticContextExperiment Visual elements like icons helped viewers quickly identify a visualization’s topic. When used meaningfully, such as when directly linked to the data topic or dimension, these elements were perceived as effective “visual anchors”, rather than decorative elements, and said to support understanding.

ClimateVisInterviews Interviewees expressed, that even though visualization design benefits from complexity reduction, it should not come at the cost of transparency. Some found removing uncertainty ranges made visualizations clearer, while others felt it hurt credibility. Accompanying text was suggested to add depth without distracting from the core message.

CrisisMapSensemaking When young, digitally native viewers engaged with crisis maps, we observed that too little context left viewers confused or distrustful. Some viewers expressed a desire for the possibility to access data provenance and other context when needed, possibly through interactive features.

ClimateVisInterviews Experts described interactive techniques as powerful when thoughtfully applied, but warned against adding interaction for its own sake. Lay participants did not report using interactive tools.

VisProducerInterviews Practitioners described using interactive designs that let users click through increasing levels of detail, such as from a simplified overview to a detailed chart and then raw data, tailoring the experience to different viewer needs.