search

Carceral-Ecologies / Carceral-Proximity-Analysis

Working version can be found at...
http://critical-data-analysis.org/shiny/proximity/proximity-app/
GNU General Public License v3.0
1 stars 1 forks source link

Statistical Significance Tests #23

Open lindsaypoirier opened 4 years ago

lindsaypoirier commented 4 years ago

@shapironick asked on Slack: One stray idea: for the spatial analysis could we also include some t-test p-values in the dashboard? For example even if we don’t have a threshold for when a number of brown fields is toxic and how many facilities are at or above that threshold, we could test if we can say if carceral facilities are statistically significantly more likely to be in a CD with multiple brownfields than if they were placed randomly and how many times higher the likelihood is that they are than if they were placed randomly. Does something like that seem like a good idea?

Reading up, this is definitely something that we can do. Right now, I think that this involves creating a census tract dataframe with all census tracts across the US as observations and variables has_prison and has_brownfield set to TRUE or FALSE, permuting the values in these two columns to randomize the association between the two, and then devising a statistics test to compare the randomized association to the observed values.

Most of this can be done in a few lines with the infer package.

shapironick commented 4 years ago

Amazing! I just sent this issue to an epidemiologist friend to get her insights.