.png){kind=icon}
.png){kind=icon}
Return
Testing forecast visualization designs for improving agricultural decision support products
.png){kind=icon}
PRE
AI
J
S
J
M
DOI：10.1088/1748-9326/ae5aba.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Changing climate, soil, and water conditions continue to shape how agricultural decisions are made and how crop yields are managed. The Dashboard for agricultural water use and nutrient management (DAWN) provides forecasts and risk information to support the decisions of farmers and water managers in the U.S. Corn Belt region (14 states). DAWN is an online toolkit that generates and communicates agriculture-focused downscaled forecasts and historical data through various interactive graphics. As some DAWN probabilistic forecast visualizations are relatively new, cross-functional and cross-regional teams of DAWN researchers determined that testing and evaluating these visualizations using social scientific methods before their public release or re-release would be helpful, especially given their potential to impact decision-making. This empirical study ensures that DAWN forecast visualization products align with best practices and evidence from social science and human-centered design research. Using mixed methods frameworks applied to previous forecast visualization testing studies, we conducted an online survey-experiment to test the current DAWN visualizations (tercile) against alternative probabilistic graphics (shaded arrays and sina plots) designed using evidence-based recommendations from multidisciplinary literature. We found that: (a) both the treatment graphics improved participants' understanding (accuracy); (b) participants in the shaded arrays group had higher efficacy, while Sina plots were most well-understood; (c) subjective likeability of graphics did not match participants' objective accuracy on interpretation tasks. We present and discuss the results of our multiple linear regression analyses, proposing evidence-based recommendations to improve the design and impacts of DAWN's probabilistic forecast visualizations, advancing impact-driven decision support research, and facilitating operational implementation of research findings in product designs and interfaces.
Keywords:
decision-making
data visualization
uncertainty
human-centered design
climate change
Journal
IF:
5.6
Papers:
9.5K
Citations:
5.0W
Organization
Cited Papers