Flint Water Crisis
This page contains additional material for the letter to the editor of the
Significance
magazine regarding the "The murky tale of Flint's deceptive water data" article.
Background
Blog post discussing the 90th percentile computation:
Beware the Argument: The Flint Water Crisis and Quantiles
Beyond Counting in the Sample
useR2017 lightning talk
(5 minute) about the effect of investigating compliance with the 15 ppb threshold based on a hypothesis test.
Blog post about the details for computing confidence intervals for quantiles:
Better Confidence Intervals for Quantiles
Code
R Code used in the letter to the editor
.
R Code computing type 1-9 for the Flint data as well as counter examples
for each of type 2-9 violating the duality between quantile and proportion.