Examining the impact of stay at home orders on neighboring states’ air pollution levels

A case study involving California and Arizona

​

Sophia Anderson, Oana Enache, Elizabeth Haderer & Hayley Nemeth

Background, Hypotheses & Data used

Background

• Fine Particulate Matter 2.5 is an environmental pollutant associated with hazy skies and adverse health effects
• Harvard has shown elevated PM 2.5 levels and COVID19 Deaths are associated, reasoning lower PM 2.5 could improve outcomes
• States with stay-at-home orders could potentially see a short term environmental benefit

Hypothesis

• PM 2.5 levels will change significantly in both California and Arizona after California’s stay-at-home order was implemented on March 19th, due to weather movement between the states.

​

Data

• Data on PM 2.5 levels were downloaded from the EPA’s Air Now Daily Outdoor Data reports for all reporting sites in Arizona and California between January 1st, 2020 and April 15th, 2020
• Dates of note:
• March 19th, 2020: California’s stay at home order begins
• March 30th, 2020: Arizona’s stay at home order begins. (Note: non-essential businesses in Arizona can still remain open as of 4/21/20)

Results

Overview of PM 2.5 levels so far in 2020. Especially beginning in mid-March, PM 2.5 levels were lower in both Arizona and California. Note: 8 significant outlier (>50) PM 2.5 values excluded from visualization so that trends could be seen more clearly.

​

​

​

Analysis

• California saw a significant difference in PM 2.5 levels pre & post stay-at-home order (March 15th, 2020)
• Mean difference of 1.84 (95% CI: 1.65, 2.04), p-value < 2.2 e-16
• Arizona also saw difference in PM 2.5 levels, but it is not statistically significant (AZ stay at home order: March 31st, 2020)
• Mean difference of 0.33 (95% CI: -0.03, 0.68), p-value = 0.07092

Discussion & Next Steps

• Conclusions
• Analysis indicated a statistically significant difference in California PM2.5 levels
• Arizona saw a difference in PM2.5 levels, difference was not statistically significant
• Other ideas for follow up include:
• Date/time considered: We analyzed levels based on CA’s stay at home order date, but:
• Many people were social distancing in weeks prior to stay at home order
• In graph, PM2.5 begins to decrease at beginning of March
• Would be interesting to adjust hypothesis to include weeks prior to stay at home order
• Exploring the limitations of PM 2.5 as a metric and/or introducing other air quality measures
• Expanding the length of time examined (eg. is 2 weeks/1 month enough to see a difference?)
• Especially since Arizona has yet to close non-essential businesses, do people actually drive less there? Other data set/s may clarify this.
• On the whole California has more data than Arizona and covers a much larger geographic space-- maybe follow up could focus on just Southern California?