This dataset was created because the United States Census Bureau
modifies the geographic boundaries in each Decennial Census to keep
the number of people and households per block group and tract
relatively constant over time. However, such spatial adjustments make
direct decade-to-decade comparisons of places challenging.
The neighborhood file provided by MOIT
(https://data.baltimorecity.gov/Neighborhoods/Neighborhoods-Shape/ysi8-7icr)
contains socioeconomic and demographic variables for year 2010 that
were the result of a custom re-tabulation from the United States
Census Bureau. This served as quality check.
Decadal tract data from 1930 to 2010 were acquired from the National
Historical Geographic Information System (NHGIS, https://nhgis.org,
Manson et al 2017). Model Builder was used in ArcGIS 10.2.2 to
Intersect, Dissolve, Join, and calculate % owner occupied homes, %
vacant homes, and % African American per year 2010 neighborhood (as
defined by MOIT) from the NHGIS-provided Census Tract data, based on
the proportion of the tract that fell within the neighborhood
boundaries.
The proportional weighting system uses polygon area for proportionally
adjusting both numerator and denominator values while calculating
percentages such as percent owner occupied housing units. For example,
if a tract was 50% within a neighborhood and contained 1,000 housing
units with 500 of them owner occupied, that portion of the
neighborhood would be assigned 250 owner occupied housing units and
500 housing units (500 x .5=250, 1000 x .5=500). All portions of
intersecting tracts within a neighborhood are then pooled to create
neighborhood-level estimates.
This approach implicitly assumes that both the numerator and
denominator values are homogeneously distributed within each Census
tract, which is known to be unlikely. This approach also implicitly
assumes that residential lands are homogeneously distributed within
each tract, which is also unlikely. Nevertheless our comparisons for
year 2010 between our proportional weighting system and the Census
Bureau's re-tabulation (the benchmark standard) demonstrate that
despite these assumption violations, the proportional system
faithfully transfers historic demographic and socioeconomic data into
common boundaries of relevance and importance today.
The 1937 Home Owners' Loan Corporation (HOLC) were acquired from the
Lord and Nordquist (2010). The year 2017 Housing Market Typology (HMT)
from Baltimore City, Department of Planning
(https://planning.baltimorecity.gov/maps-data/housing-market-typology).
The 2017 Housing Market Typology was created by the Baltimore City
Planning Department, Department of Housing & Community
Development, and Reinvestment Fund. HOLC and HMT polygons were
intersected with the neighborhood file and a majority-in rule was used
to assign which HOLC and HMT boundaries each neighborhood falls
within.
Finally, tree canopy change data were accessed from TreeBaltimore
(http://treebaltimore.org/) here:
http://baltimore.maps.arcgis.com/apps/webappviewer/index.html?id=b4d5f007c0974e2aa575295654919545
The area of loss, persistence and gain were calculated for each
neighborhood. Then the percent loss, percent gain, and net percent
change were calculated.
Citations
Lord, C., and Norquist, K. (2010) Cities as emergent systems: race as
a rule in organized complexity. Environmental Law, 40(2), pp. 551-597.
Manson, S., Schroeder, J., Riper, D. V., and Ruggles, S. (2017) IPUMS
National Historical Geographic Information System: Version 12.0
[Database]. Minneapolis: University of Minnesota.
http://doi.org/10.18128/D050.V12.0
