Tract level data from the US Census 2017 American Community Survey 5-Year Estimates (2013-2017), table C16001, was obtained using the American Fact Finder (www.factfinder.census.gov) on March 19, 2019. This data was geographically limited to all Census Tracts within Pennsylvania (n=3,218) and contained population estimates for Pennsylvania residents (5 years and over) who speak one of the 12 most common languages/language groups (L/LG) in Pennsylvania and their corresponding English Proficiency. Population counts displayed in the map (excluding English-only, HD01_VD02 & HD02_VD02) for each language/language group (e.g. HD01_VD03) and corresponding high (e.g. HD01_VD04 “very well”) and low (e.g. HD01_VD05 “less than ‘very well’”) English Proficiency. Percentage of Census Tract population for each L/LG was calculated by dividing the estimated tract population count for each language by the estimated total population count (HD01_VD01) of each tract (e.g. (HD01_VD03 ÷ HD01_VD01) x 100 ). Tract specific percentages for high and low English Proficiency were calculated for each L/LG by dividing the estimated high or low proficiency population count by the corresponding total L/LG population count for that tract (e.g. (HD01_VD04 ÷ HD01_VD03) x 100 ). Using ArcGIS Pro, the data was spatially joined in to the Census Tract boundary shapefile (see data source #2) for naming and labeling purposes. The resultant data was copied to 26 individual visualized layers which created a L/LG population percentage layer and low English Proficiency layer for each L/LG and additional statewide aggregate layers of all L/LGs. Population count layers were visually represented using a varying number of classes for each layer and manual interval breaks; the number of classes and interval breaks were determined by the map author by comparing the maximum and minimum tract population percentage for each L/LG and use the fewest number of classes while maintaining logical and useful interval breaks. Low English Proficiency (LEP) layers were visualized for each L/LG to highlight only Census Tracts where the LEP percentage was higher than 30%. Highlighting was achieved by using a display filter with a contrasting border and opaque fill color. The map was then exported from ArcGIS Pro to ArcGIS Online where pop-ups were configured for each L/LG layer to provide tract-specific information about population counts, percentages, and LEP. Once this was completed, the map was turned into an interactive Web App using the ArcGIS Web AppBuilder and published.
