Data and code for "Spatiotemporal trends in self-reported mask-wearing behavior in the United States: Analysis of a large cross-sectional survey

This repository provides the data and source code for the following study: Juliana C Taube, Zachary Susswein, Shweta Bansal. "Spatiotemporal trends in self-reported mask-wearing behavior in the United States: Analysis of a large cross-sectional survey." JMIR Public Health and Surveillance. https://doi.org/10.2196/42128.

How to use this resource

To rerun the models and reproduce the figures, start by opening covid_masking.Rproj. From here, run the files in numerical order, starting with 06GENERATE-BIAS.r if you don't have access to the individual data (see Individual Data below). Code will not run correctly if files are sourced out of order. All necessary input data files should be in the repository, with the exception of a large CDC table with daily county mask mandate data which can be downloaded by users (see data/).

Estimates (data/estimates/)

Raw survey responses aggregated to the county-month level and model estimates at the county-month level for both the CTIS (fb) and ONM (onm) surveys.

• fb_estimates_binomreg_FINAL.csv contains self-reported mask-wearing estimates (p_est) from binomial regression model only
• fb_estimates_binomreg_rake_FINAL.csv contains self-reported mask-wearing estimates (p_est) from binomial regression model with raked and resampled individual responses
• fb_estimates_binomreg_rake_debias_FINAL.csv contains self-reported mask-wearing estimates (unbiased_p) from binomial regression model with raked and resampled individual responses and an offset for bias
• fb_estimates_binomreg_rake_debias_FINAL_excl.csv contains self-reported mask-wearing estimates (unbiased_p) from binomial regression model with raked and resampled individual responses and an offset for bias where influential fips (pareto k $\ge$ 0.7 in the initial model) are excluded
• fb_comm_estimates_binomreg_rake_FINAL.csv contains community-reported mask-wearing estimates (comm_p_est) from binomial regression model with raked and resampled individual responses
• fb_comm_estimates_binomreg_rake_FINAL_excl.csv contains community-reported mask-wearing estimates (comm_p_est) from binomial regression model with raked and resampled individual responses where influential fips (pareto k $\ge$ 0.7 in the initial model) are excluded
• onm_estimates_binomreg_FINAL.csv contains self-reported mask-wearing estimates (onm_est) for the grocery store setting in the Outbreaks Near Me survey from the binomial regression model only
• fb_processed.csv contains raw self-reported mask-wearing data (mask_prop_most) aggregated to the county-month level
• onm_processed.csv contains raw self-reported mask-wearing data from the ONM survet (mask_grocery_very_prop) aggregated to the county-month level

Data (data/)

Reference files that may be required to run the code, including fips and zipcode crosswalk files, urban/rural classifications, etc.

• acs_target_data_for_fb_raking.csv contains age, sex, education, and some race/ethnicity targets for raking from the 2019 ACS
• archived_vaccination_data/data_master_county_20210823.csv contains weekly vaccination coverage data for each fips code as recorded at https://vaccinetracking.us
• Average_Household_Size_and_Population_Density_-_County.csv contains population density for each fips code (from https://covid19.census.gov/datasets/USCensus::average-household-size-and-population-density-county/explore?location=4.945434%2C0.315550%2C1.99&showTable=true)
• county_neighbors_fips.txt contains a fips code neighbor list
• COUNTY_ZIP_062021.xlsx is the file for crosswalking fips code to zip code (from https://www.huduser.gov/portal/datasets/usps_crosswalk.html)
• ZIP_COUNTY_062021.xlsx is the file for crosswalking zip code to fips code (from https://www.huduser.gov/portal/datasets/usps_crosswalk.html)
• fb_resampled_comm_data_from_raking_age_sex_educ_FINAL_corrected.csv contains county-month level resampled data on community-reported masking following raking on age, sex, and education, and with correction of weights in nonconvergent county-months
• fb_resampled_data_from_raking_age_sex_educ_FINAL_corrected.csv contains county-month level resampled data on self-reported masking following raking on age, sex, and education, and with correction of weights in nonconvergent county-months
• fb_resampled_vax_data_from_raking_age_sex_educ_FINAL_corrected_aprmay.csv contains county-month level resampled data on vaccination following raking on age, sex, and education, and with correction of weights in nonconvergent county-months
• NCHSURCodes2013.xlsx contains urban-rural classifications for each fips code (from https://www.cdc.gov/nchs/data_access/urban_rural.htm)
• nyt-us-rolling-avg.csv contains COVID-19 case data from the New York Times (from https://github.com/nytimes/covid-19-data/blob/master/rolling-averages/us.csv)
• state_fips.csv contains each state's fips code
• U.S._State_and_Territorial_Public_Mask_Mandates_From_April_8__2020_through_August_15__2021_by_State_by_Day.csv contains mask mandate data at the state-level (from https://data.cdc.gov/Policy-Surveillance/U-S-State-and-Territorial-Public-Mask-Mandates-Fro/62d6-pm5i)
• U.S._State_and_Territorial_Public_Mask_Mandates_From_April_10__2020_through_August_15__2021_by_County_by_Day.csv contains mask mandate data at the county-level (from https://data.cdc.gov/Policy-Surveillance/U-S-State-and-Territorial-Public-Mask-Mandates-Fro/62d6-pm5i)
  • This file is large and will be uploaded in the future. In the meantime, users can download the csv from the above link.
• vax_data_for_time_series.RDS contains cumulative proportions of all U.S. residents vaccinated by week (from vaccinetracking.us, code to produce these estimates is commented out in 13FIGURES.csv)
• worry_df_for_time_series.RDS contains aggregated weekly data on proportion of respondents very or somewhat worried about severe COVID-19 from the CTIS

Code (scripts/)

Scripts to clean data, rake survey responses, resampled weighted survey responses, run binomial regression models, and reproduce figures. Scripts for analyzing individual responses are provided for reproducibility but will not run without the original individual-level data (see Individual Data section below). File names briefly describe the purpose of each script (where COMM stands for community-reported and ONM for the Outbreaks Near Me dataset).

Individual Data

Individual CTIS survey responses cannot be shared by the authors, but researchers can visit https://cmu-delphi.github.io/delphi-epidata/symptom-survey/data-access.html if they would like to enter an agreement for data usage with CMU Delphi. Individual ONM responses also cannot be shared by the authors, but researchers can contact the OutbreaksNearMe team at Boston Children's Hospital and Momentive to inquire about access to these data.

If users have access to the individual data, they will be able to run files 01 through 05.