Project 1 argues that “gentrification” is only one form of neighborhood change (contingent on a rapid increase in a neighborhood’s socioeconomic status). Therefore, to understand gentrification, it needs to be stripped of all its politically-charged attributes. I argue that gentrification - at its most rudimentary form - is a rapid increase in socioeconomic status. All other indicators normally touted to be a part of gentrification (i.e. displacement, increase in white residents, higher rents, positive changes to the built infrastructure [investment from public and private entities], etc.) become secondary attributes that will be readdressed later in the analysis. With this in mind, we first need a uniform measure of socioeconomic status (SES) using publicly available data from the U.S. Census (that - in theory - can accurately capture neighborhood SES across all U.S. cities [i.e. be able to be comparative]).
I began with a reflective factor analysis using neighborhood-level measures of SES: (1) Median Household Income, (2) Occupational Prestige, (3) Median Home Value, (4) Educational Attainment. I captured these measures across five consecutive samples of census tracts within the cities of Philadelphia and Boston from 2014 to 2018. While I originally began with all of the tracts within these two cities, tracts with populations less than 500 residents or with over one-third of the population living in group quartering were excluded from the analysis (the U.S. Census tracks population counts for various communal living facilities like collegiate dormitories, military barracks, rehabilitation or medical care centers, and correctional facilities. The bureau identifies these facilities as “group quarters.”). These tracts contain “specialized” populations whose socioeconomic status does not directly reflect the socioeconomic status of the residents who have “chosen” to live in the neighborhood. The factor analysis used maximum likelihood estimates and oblimin rotation. It was tested for normality using Shapiro-Wilk test and all variables were highly correlated at 0.001 significance before running the factor analysis. All of the loadings were greater than the accepted threshold of 0.40, and all measures of fit and reliability were favorable - except the empirical chi-square tests and the Root Mean Square Error of Approximation (RMSEA). See Table 1 (Microsoft Word doc) to see the exact loadings.
The factor loadings were extracted to create a series of SES scores (i.e. an SES index) for each tract from 2014 to 2018. The individual SES scores for each census tract during a given year were compared to the mean SES score of each city at the same given year. Each score was then marked as “High” if it was one standard deviation above the city’s mean SES score that year, “Medium” if it equal to or greater than the city’s mean SES score that year, and “Low” if it was below the city’s mean SES score that year. Below are the maps of Boston’s Neighborhood SES Scores from 2014 to 2018 (they are interactive so you can zoom in and out to specific neighborhoods).
## OGR data source with driver: ESRI Shapefile
## Source: "E:\6th_Year_Fall\Dissertation\Project 1\Shapefiles\Tracts_Boston_BARI", layer: "Tracts_Boston BARI"
## with 178 features
## It has 16 fields
## Integer64 fields read as strings: ALAND10 AWATER10 BRA_PD_ID City_Counc WARD Police_Dis Fire_Distr## OGR data source with driver: ESRI Shapefile
## Source: "E:\6th_Year_Fall\Dissertation\Project 1\Shapefiles\Tracts_Boston_BARI", layer: "Tracts_Boston BARI"
## with 178 features
## It has 16 fields
## Integer64 fields read as strings: ALAND10 AWATER10 BRA_PD_ID City_Counc WARD Police_Dis Fire_Distr## OGR data source with driver: ESRI Shapefile
## Source: "E:\6th_Year_Fall\Dissertation\Project 1\Shapefiles\Tracts_Boston_BARI", layer: "Tracts_Boston BARI"
## with 178 features
## It has 16 fields
## Integer64 fields read as strings: ALAND10 AWATER10 BRA_PD_ID City_Counc WARD Police_Dis Fire_Distr## OGR data source with driver: ESRI Shapefile
## Source: "E:\6th_Year_Fall\Dissertation\Project 1\Shapefiles\Tracts_Boston_BARI", layer: "Tracts_Boston BARI"
## with 178 features
## It has 16 fields
## Integer64 fields read as strings: ALAND10 AWATER10 BRA_PD_ID City_Counc WARD Police_Dis Fire_Distr## OGR data source with driver: ESRI Shapefile
## Source: "E:\6th_Year_Fall\Dissertation\Project 1\Shapefiles\Tracts_Boston_BARI", layer: "Tracts_Boston BARI"
## with 178 features
## It has 16 fields
## Integer64 fields read as strings: ALAND10 AWATER10 BRA_PD_ID City_Counc WARD Police_Dis Fire_DistrDuring the five-year period, each census tract (i.e. neighborhood) witnessed a change to their SES score over the five-year period. While most remained unchanged in their ranking as “Low,” “Medium,” or “High,” all experienced a numeric difference. These differences were the result of phenomenon occurring within the tract itself that affected the tract’s SES score (e.g. new homes were built for wealthier families increasing the entire neighborhood’s home values, employment levels in the neighborhood dropped decreasing the median household income, a group of well-educated individuals moved to the neighborhood increasing the overall occupational prestige, etc.). Therefore, it can be argued that each tract witnessed its own trajectory of SES score during the five-year-period that was the result of individual neighborhood-level variations.
For this reason, a mixed effects model was used to extract the “random effects” coefficients (i.e. the slopes) of each census tract during the five-year period. These slopes represent the individual trajectories of each of the tracts’ SES scores during the five-year period (this variability would be lost in a simple linear regression model since the individual “effect” of the neighborhood would be unaccounted for in the total trajectory of the combined SES scores of all the neighborhoods).
The following (interactive) map shows the breakdown of each census’ slope z-scored over the five-year period. The mean SES rise for the city, as a whole, is “0” (this does not mean the city as a whole saw no increase or decrease to its SES score; rather it is the comparative slope for all neighborhoods). Therefore, any tract with a negative slope saw a below-average increase or a decrease in SES over the five-year period. Any tract with a positive slope saw an increase in SES above the city’s mean. For context, a summary of the values are listed in the output.
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## -2.96612 -0.55344 -0.08958 0.00000 0.39068 7.51861## OGR data source with driver: ESRI Shapefile
## Source: "E:\6th_Year_Fall\Dissertation\Project 1\Shapefiles\Tracts_Boston_BARI", layer: "Tracts_Boston BARI"
## with 178 features
## It has 16 fields
## Integer64 fields read as strings: ALAND10 AWATER10 BRA_PD_ID City_Counc WARD Police_Dis Fire_Distr