Introduction

The data set consists of 271 homes sampled with three water lead contaminant values at designated time points. The lead content is in parts per billion (ppb). Additionally, some location data is given about each home.

To get started, read in the flint.csv file using function read_csv().

library(tidyverse)

flint <- read_csv("flint.csv")

Let’s preview the data with function glimpse()

glimpse(flint)

Observations: 813
Variables: 5
$ id   <dbl> 1, 2, 4, 5, 6, 7, 8, 9, 12, 13, 15, 16, 17, 18, 19, 20, 21,…
$ zip  <dbl> 48504, 48507, 48504, 48507, 48505, 48507, 48507, 48503, 485…
$ ward <dbl> 6, 9, 1, 8, 3, 9, 9, 5, 9, 3, 9, 5, 2, 7, 9, 9, 5, 6, 2, 6,…
$ draw <chr> "first", "first", "first", "first", "first", "first", "firs…
$ lead <dbl> 0.344, 8.133, 1.111, 8.007, 1.951, 7.200, 40.630, 1.100, 10…

id: sample id number
zip: zip code in Flint as to the water sample’s location
ward: ward in Flint as to the water sample’s location
draw: water sample at one of three time points
lead: lead content in parts per billion

Analysis

Summary data

Part I

Let’s see how many samples were taken from each zip code.

flint %>%               # data
  group_by(zip) %>%     # perform a grouping by zip code
  count()               # count occurrences

Which zip code had the most samples drawn?

Part II

Next, let’s look at the mean and median lead contaminant values for each zip code and draw combination. We have eight zip codes and samples taken at three time points. How many combinations do we have?

flint %>% 
  group_by(zip, draw) %>% 
  summarise(mean_pb = mean(lead))

flint %>% 
  group_by(zip, draw) %>% 
  summarise(median_pb = median(lead))

How many rows are in each of two above data frames?

Part III

Modify the code below to compute the mean and median lead contaminant values for zip code 48503 at the first draw. What should you put in for draw == "-----"? Don’t forget to uncomment the second line of code.

flint %>% 
  # filter(zip == 48503, draw == "-----") %>% 
  summarise(mean_pb = mean(lead),
            median_pb = median(lead))

Visualizations

Let’s make some plots, where we will focus on zip codes 48503, 48504, 48505, 48506, and 48507. We will restrict our attention to samples with lead values less than 1,000 ppb.

flint_focus <- flint %>% 
  filter(zip %in% 48503:48507, lead < 1000)

Below are side-by-side box plots for the three flushing times in each of the five zip codes considered. Add x and y labels; add a title by inserting title = "title_name" inside the labs() function.

ggplot(data = flint_focus, aes(x = factor(zip), y = lead)) +
  geom_boxplot(aes(fill = factor(draw))) +
  labs(x = "--------", y = "--------", fill = "Flushing time") +
  scale_fill_discrete(breaks = c("first", "second", "third"),
                      labels = c("0 (sec)", "45 (sec)", "120 (sec)")) +
  coord_flip() +
  theme_bw()

Add labels for x, y, a title, and subtitle to the code below to update the corresponding plot.

ggplot(data = flint_focus, aes(x = factor(zip), y = lead)) +
  geom_boxplot(aes(fill = factor(draw))) + 
  labs(x = "--------", y = "--------", fill = "Flushing time",
       subtitle = "--------") +
  scale_fill_discrete(breaks = c("first", "second", "third"),
                      labels = c("0 (sec)", "45 (sec)", "120 (sec)")) +
  coord_flip(ylim = c(0, 50)) +
  theme_bw()

What is the difference between the two plots?

References

Langkjaer-Bain, R. (2017), The murky tale of Flint’s deceptive water data. Significance, 14: 16-21.

Flint Water Crisis