Our methodology and analysis heavily relied on census tract level data and was thus greatly impacted by the scale and aggregation effects of the modifiable areal unit problem (MAUP). The census tract scale of our data may have been insufficient in conducting an analysis on gentrification, as the impacts and influences of gentrification occur at the neighbourhood scale or even at smaller scales. The census tract data set displayed a more aggregated representation of metropolitan Atlanta and may have been insufficiently refined to accentuate the factors influencing gentrification. However, given the availability of data and scope of our project, the usage of data pertaining to the census tract scale data proved to be the most beneficial and informative to our analysis, despite potential impacts on our overall analysis.

Our analysis was also limited by variable selection and availability. The variables utilised in our analysis were chosen based on common factors identified during our preliminary research and literature review of gentrification studies. As such, the variables used were by no means comprehensive. Rather, the selected variables simply represented a subset of many that could have been considered in the evaluation of gentrification, as there is no consensus as to how gentrification should be measured and what variables should be considered. The time-scale of the data also influenced our analysis, as our chosen change interval of 2011-2019 was chosen due to the lack of available data variables from the United States Census Bureau beyond those years, and the results thus did not account for more recent changes, which could have been included had more recent data been available.

Uncertainty was also introduced in the processing necessary to turn disperse points into census tract-level values. Some subjectivity was necessary for the rescaling of kernel density results into scores at a level of magnitude comparable to other explanatory variable layers. The data source for amenities may also have not been comprehensive, therefore leaving out potential points that may have changed the analysis.