STT 864
Final Project
Spring 2016
Due Date: Wednesday before 5:00pm EST, April 27, in C418 Wells Hall. Late papers
will not be accepted.
Instructions: You must work independently. You may not discuss this project
with anyone except the course instructor.
The project aims to analyze Flint water data sets. The background of the Flint water
crisis was introduced in Lab 2. The data analyzed in this project were downloaded from the
website https://www.michigan.gov/flintwater. One data set is a sentinel sampling results data set (file name: “Sentinel Data Set 1A-B 515890 7”), which were collected from
about 600 sentinel sites in Flint (see Figure 1 for a map of locations). The sentinel sites
were chosen by the environmental protection agency (EPA) and the Michigan Department
of Environmental Quality (MDEQ) according to the Detroit Free Press (in the article “EPA
says Flint water tests show some progress made”). The second data set was the residential water testing samples collected between Sept. 3, 2015 and Mar. 15, 2016 (file name:
“Test Results Flint 513922 7”). More details about these two data sets could be found on
the website https://www.michigan.gov/flintwater. Both data sets are also available
on our class website http://www.stt.msu.edu/users/pszhong/STT-864-2016.html.
You are free to perform any statistical analyses that answer real questions related to
the data sets that you consider to be relevant and informative. Use R to perform calculations and computations. Do not submit pages of raw computer output, but you need
to include important tables or attach graphs to support your conclusion and discussion.
Write concise answers that clearly describe the steps in your analysis and your conclusions.
Remember that there is no absolutely best answer and I expect to receive many different
answers. You will be graded with respect to how well your analyses reveal interesting aspects of the data sets, the interpretation of your results, how well you justify your methods
of analyses, and the overall quality of your writing. The details of the grading rubrics will
be posted on the class page website.
Your report should not exceed eight typed pages, excluding attached graphs or tables,
with the font size and spacing used in these instructions. You may cut out computer
output and simply paste it in appropriate places in your report. You do not have to
type your report. You can submit a handwritten report. The limit on the number of
handwritten pages depends on how large you write, but is limited to what could have fit
on eight typed pages. Your report should including the following (but not limited to):
1. Provide a one paragraph summary of your major findings. This should not contain
any formulas or mathematical symbols. It should be well written so that it could be
easily understood by anybody else who is not a statistician.
1
Flint Water Incident - Sentinel Sampling Results - Lead
W
X
W
X
W
X
W
X
WX
X
WX
W X
XX
W
WX
XX
WX
W W
WW
WX
WX
WX
W X
W X
X
W
W
X
W
X
W
X
W X
X
W
X
W
WW
X X
W
X
W
X
W
X
X
W
48506
W
X
W
X
W
X
48504
W
X
W
X
W
X
W
X
W
X
WW
X
XX
W
W
X
WX
W
XX
W
W
X
W
X
WX
X
W
X
W
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
WX
X
W X
X
W
WX
WX
W
WX
W
W
WX
X
W X
X
W
WX
W X
X
WX
W
X
W
X
W
WX
X
W
WX
X
WX
X
W
X
W X
WX
WX
WX
W
WX
X
WX
WW
X
WX
WW
X
WX
X
WX
X
W
W
X
W
X
W
X
WX
X
W
W
W
X
W
X
WX
X
WX
W
X
W
X
W
X
W
X
WX
X
WX
X
WX
W
X
W
X
W
W
WWX
WX
X
WX
X
W
WX
W
X
WX
W
X
W X
X
W
X
W
W
W
X
W X
X
W
X
W X
WX
W
X
WX
X
W
W
X
W
X
W
WX
X
WX
W
W
X
WX
X
W
W X
X
W X
X
WX
X
W
W
W
X
WX
X
WX
W
W
WX
X
WX
W
X
W
X
WX
W
WX
X
W
X
W
W
X
W X
X
W
W X
X
WX
X
W
XX
W
W W
X
W
X
48532
W
XX
W
X
W
W
X
WX
X
W
WX
X
W
W
WX
X
W
X
W
X
XX
W
W
X
WX
WX
W
W
W
X
W
X
WX
WX
X
W
X
WX
X
W
W X
X
W
X
W
WX
X
W
X
WX
W
W
X
48532
W
X
48505
XX
W
W
W
X
W
X
W
X
W
X
48502
W
X
X
W
X
W
W
X
X
W
W
X
W
X
W
X
XX
W
W
WX
X
W
X
W
W
X
W
X
X
W
W
X
W
WX
X
WW
X
WX
WX
WX
WX
W X
X
W
X
W
X
W
W X
WX
X
W
X
W
X
X
W
W
W X
X
W
WX
X
W
X
WW
W
WX
X
WW
WX
X
W
X
WX
X
WX
X
WX
X
W
W X
X
WX
W
W
X
W
X
WX
WX
X
WX
X
W
WX
X
W
X
W
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
W
X
WX
W
WX
X
W
X
W
X
WX
X
W
W
X
W
X
W
WX
W
X
W
WX
WX
X
W
X
W X
WX
X
WX
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
W
X
X
X W
W
XX
WW
WX
X
W
W
X
W
WX
WW X
X
WX
X
W
X
W
X
XX
W
X W
W
WX
X
W
W
WX
X
W
XX
W
WX
W
X
48503
W
X
W
X
W
X
48507
W
X
W
X
W
X
W
X
W
W
X
WX
X
W
X
X
W
W
X
W
X
W
X
W
XX
W
WX
X
W
X
W
W
X
W
X
W
X
W
W X
W
X
WX
WX
X
W
WX
WX
X
W
X
W
X
W
X
X
W
W
X
W
X
W
X
W
WX
X
W
X
48507
W
X
W
X
W
X
W
X
W
X
W
W
X
W
X
WX
WX
WX
X
WX
X
W
X
WX
X
W
WX
X
W
W
W
X
WX
X
W
X
WX
WW
X
W
XX
W
48551
48553
W
X
W
X
WX
XX
W
W
X
W
48519
W
X
W
XX
W
WX
W
W X
X
48529
48507
48507
48473
·
Legend
54 Results greater than 15 parts per billion of Lead
W
X
State of Michigan
510 Results less than or equal to 15 parts per billion of Lead
Flint w/Franchise Boundary
Flint Area Zip Codes
1
0.5
0
1
2
Date: 03/01/2016
Sources: Esri, HERE, DeLorme, USGS, Intermap, increment P Corp., NRCAN, Esri Japan, METI, Esri China (Hong Kong), Esri (Thailand), MapmyIndia, © OpenStreetMap contributors, and the GIS User Community
Figure 1: The map of the locations of sentinel sites.
2
3 Miles
2. Provide a description of the steps taken to identify your best model (or models). Do
not give all the details of your search. Do not submit any computer output in this
section. Simply outline the issues you considered, your decisions, and the sequence
of steps you took to develop a model.
3. Provide a description of your best model (or models), including estimates of parameters and their standard errors, and the statistical inference you performed. You
may copy small parts of computer output into this part of the report. Discuss and
interpret any important features of your model and state you conclusions in the
context of the problem.
4. Provide convincing evidence that your analysis is based on a good model. Discussion
of residual plots and other diagnostic checks would be appropriate. You may attach
graphs, but lists of raw computer output should not be submitted and will be ignored.
5. You may submit one to two more paragraphs outlining additional analyses that you
would have done if you had more time. You will earn points for good suggestions
and lose points for suggestions with little potential value. (This is optional.)
3