Missing and sparse data and the associated uncertainty are inevitable in science, and their accurate portrayal in media is essential for upholding scientific transparency and credibility. Yet, in an era of conflicting information and deceptive sources, revealing uncertainty can be seen as unfavorable, in particular for science engagement media. Our study focused on conveying incomplete data on Venus’s upper atmosphere to an adolescent audience using a scientific visualization designed for a planetarium’s dome. Through a comparative study of visualizations with an unprocessed versus a processed dataset, we found that translating data into a Voronoi diagram can make the concept of sparse data understandable and aesthetically pleasing to a broader audience, yet it can come at costs of lower perceived details and accuracy. Additional results hint that participant’s preference for visualization can differ from their perceptions of clarity and that neither the preference nor the clarity appears to be linked to participants’ science literacy. Finally, we discuss the potential to design immersive science media experiences in planetariums and dome settings.

Science communication predominantly engages individuals who already possess a pre-existing interest in science, thereby limiting its reach to a broader and more diverse audience. Situated approaches that intertwine content with spatial design may overcome interest barriers as it allows for spontaneous encounters with science. This paper describes the design of a multi-modal physicalisation coined Star Listener as the result of a collaboration between scientific experts (in astrophysics), design researchers and media professionals. Star Listener consists of different sensory engagement modalities, including textual information panels, speakers that distribute sonified data of stars, two public video displays with an animated explainer and interactive touchscreens with a mini-game. Star Listener was placed in an entryway of a public library for a period of 2 months. We evaluated this set-up by observing and interviewing passersby, both in an in-the-wild study and as a controlled in-the-wild study. We present design recommendations for the design of situated data representations and narrative physicalisations to support engagement of varied audiences with science.

As data literacy skills are increasingly important in today’s society, scholars have been exploring strategies to engage people with data, for example through storytelling and familiar media such as video. In this paper, we present the design of a video-based data storytelling application that prompts children and their families to explore and interpret historical weather data through a personalised weather forecast. The application was displayed at a 2-month summer exhibition of a popular television channel. In a controlled comparative study, we investigated how the application triggered reflection, as well as emotional and narrative engagement of families at home and at the exhibition. We combined this approach with an in-the-wild study, in which we observed spontaneous interactions of visitors. Our findings indicate that data engagement is encouraged when family interactions occur, which may be facilitated by external environmental conditions and internal story design. Here, we uncover 5 design recommendations for data video storytellers.

Public displays are increasingly deployed to make civic data easily and publicly consumable. While augmenting such public visualizations with a narrative design strategy could be promising to engage a lay audience, they might perform differently on public displays than on common online media because of the more context-sensitive environment. We therefore report on a comparative in-the-wild study of a public display that contrasts an identical public visualization with and without a narrative structure, and unravel how this affects the user engagement and insight creation process. Our findings indicate how a narrative strategy in relation to contextual aspects supports deeper, more personal reflection on data, connects authorship to the surrounding environment, and overcomes comprehension issues. We believe these results are useful for making public visualizations more effective, as well as understanding why and how lay users interact with and learn from narrative data visualization in general.

The in-the-wild methodology involves the evaluation of a functioning prototype in an everyday context, during which the participants are typically left unaware of the actual study context. As the material dimensions of such a prototype imply a preliminary status, the apparent difference between prototype and the final end product might affect the actual ecological validity of the evaluation results. By replicating an in-the-wild study of an identical yet progressed high fidelity prototype versus its research product one year apart, we aim to investigate the impact of material dimensions on user behavior. Our results demonstrate how impermanent material dimensions tend to increase the participation rate and augments reflection on ownership; imperfect dimensions reduce the expectations and contextual appropriation of an installation; and incomplete dimensions imply a relationship with the investigator. We thus claim that material dimensions affect the evaluation outcomes of in-the-wild evaluation studies.

As cycling is increasingly promoted as an environment-friendly, cheap and even fast alternative, there exists an increasing need to civically involve the potentially engaged and opinionated user group of cyclists. Therefore, we designed and evaluated Bicycle Barometer, an interactive bicycle count display that gathers the opinions from cyclists and conveys real-time, multi-dimensional data to them regarding cycling behavior. Our user-centered design process focused on optimizing the user experience by comparing several alternative cyclist-specific interaction designs, which resulted in the combination of a pressure sensitive floor mat, push button and low-resolution LED display. An in-the-wild evaluation study resulted in a set of design recommendations for cyclist-specific interaction, providing concrete insights into how a specifically targeted interaction method for public display is able to afford engagement and enthusiasm from a particular target audience.

In this paper, we investigate the potential of controlled in-the-wild studies as an evaluation methodology that merges the benefits of lab-based and in-the-wild studies. Our exploratory investigation builds upon a comparative, between subject experiment benchmarking different interaction features of a custom public installation that visualized a series of urban datasets. In order to evaluate the usefulness of the in-the-wild versus the controlled in-the-wild methodologies, we compared the resulting findings in terms of participant engagement, insight generation, and social interaction. We propose that a controlled in-the-wild study offers a viable alternative when evaluating more complex interaction methods in public space, hereby potentially reducing the practical efforts of in-the-wild studies to involve participants.