That time when we published our research on the design of an AQI scale in effecting face mask-wearing behavior
Four UW HCDE graduate students
UW Professor and Peers
Time Frame & Constraints
January 2020 through March 2020
In 2015, air pollution caused 8.8 million premature deaths worldwide and had reduced life expectancy by 2.9 years. Today, wildfires in the United States burn, on average, 78 days longer and cover twice the area compared to 50 years ago. The increasing impact of wildfires and other emissions will lead to more days and locations with poor air quality. In 2018, the California wildfires caused California cities to rank as the world’s most polluted cities over places in India and China.
Previously, researchers have studied the effect of Air Quality Index (AQI) alerts on actions, primarily focusing on avoiding the outdoors and physical activity. As air quality continues to decrease worldwide, the importance of understanding AQI alerts on resulting behavior change becomes more critical.
Figure 1: Typically, an Air Quality Index (AQI) alert contains the air quality rating number, a one-word description (Good, Bad, Hazardous, etc.), a color indicator (green to red/purple), and a recommended health action.
Our team constructed a research problem and specific research question based on relevant literature examining how visual anchoring on an Air Quality Index scale effects intended behavior change to protect against adverse health outcomes (i.e., staying home, changing location to be inside, reducing time outside, wearing a face mask).
To learn about the effect of visual anchoring on intended behavior change, such as wearing a face mask, we created hypotheses for our online research experiment, identified relevant variables, weighed ethical concerns, and designed experimental AQI instruments (Figure 2). Group #1 participants were exposed to the “Single Number” AQI scale (Figure 2: left), while participant Group #2 was exposed to the “Slider Scale” AQI visual (Figure 2: right). Our team collaboratively wrote the research survey and then collected participants’ survey data using the software tools SurveyMonkey and Mechanical Turk.
After implementing our SurveyMonkey study onto Mechanical Turk, we selected and applied appropriate data analysis methods using R to evaluate our experimental research study’s validity and uncover users’ insights and tendencies.
Figure 2: The AQI scale visuals were designed in black and white to control the impact of color on the scale severity perception.
Figure 3: Our team collected survey data using SurveyMonkey and Mechanical Turk.
Our research study’s most salient benefit indicates that people’s past behavior and the visual information displayed on the AQI scale can help people potentially increase their intent to wear a face mask in the future. As health hazards continue to increase, it is necessary to understand how to communicate the most critical information that can influence protective health behaviors.
This study was inconclusive on anchoring visual design in AQI alerts, indicating that more research is necessary, which could, in part, be due to the social stigma of wearing air masks in the United States. However, since this study was conducted, America’s climate regarding wearing a face mask has changed dramatically since this study took place in the early stages of COVID19.
Our research experiment could be expanded upon in future research examining anchoring visuals to increase trust in AQI alerts since mask-wearing has become normalized worldwide. The potential impact of predicting this type of intended behavior change for decreasing air quality and the COVID-19 pandemics is crucial.
I am very grateful to have suggested that our team research Americans’ attitude on mask-wearing behaviors as we gathered our research findings two weeks before our state went into lockdown due to COVID-19. If we were to conduct this research today, our results would be dramatically different since the public attitude about mask-wearing behavior has shifted for public health and political reasons.