PSTAT 120C: Data analytic report 1 Due: Nov 2, 2021 before class
Please submit your report as a pdf, word or image file. Please submit your R
code in separate file(s). Please attach figures from R to illustrate your
answers.
1. (10 points) For the presidential poll in 2016, explore the poll in Michigan,
Georgia and North Carolina from August 1, 2016 to November 2 in 2016.
Use the data to answer the following questions.
a. Who is ahead in each of these three states? What is the percentage
difference for each state?
b. Run a paired t test of the counts in polls for each of the state. Who is
in favor of winning based on the test? Is the test significant? Is there
potential problem?
c. Run a Wilcoxon signed-rank test of the counts in polls for each of the
state. Who is in favor of winning based on the test? Is the test signifi-
cant? Is there potential problem of the test?
d. Fit a linear model of the percentage difference with respect to date of the
polls separately for each of these states. Show a plot of the observations
of the polls, fitted values and confidence interval of the fitted line for
each of these state. From the linear model and observations, which state
may have the closest election (in terms of percentage difference)?
e. From the real results of 2016 election, which state has the smallest mar-
gin (in terms of percentage difference)? Discuss at least two reasons that
are different than what polls indicate. (You may check Wikipedia for
2016 US presidential election to find out the real voting results for each
state.)
f. Do polls correctly predict the candidate who wins these states? Discuss
the bias of polls in these states. Name a few possible reasons.
2. (10 points) Redo Question 1 (a)-(f) for the same three states for the pres-
idential polls in from August 1 to November 2 in 2020. (You may check
Wikipedia for 2020 US presidential election to find out the real voting re-
sults for each state.)
3. (10 points) Explore the poll data from September 1, 2016 to November 2,
2016 and September 1, 2020 to November 2, 2020 to answer the following
questions.
a. Graph the percentage difference of polls in each state of US for 2016 and
2020. Compare the difference.
b. Name 10 battleground states (states with closest percentage difference
between two candidates) in 2020 based on the plots for (a). Explain
your reasoning.
PSTAT 120C: Data analytic report 1 Due: Nov 2, 2021 before class
c. Compare the difference of the polls in 2016 and in 2020 for states in US.
d. Do polls underestimate the percentage of the real votes (in terms of
percentage) received from one candidate in 2016? How about 2020?
Discuss some reasons that may explain the bias in polls.
4. (10 points) Use data to explore states may change their electoral votes to
another candidate from a different party and answer the following questions.
a. Use figures or tables to compare the state level polls in 2016 and 2020.
b. Draw your conclusion and name 5 states that may change their electoral
votes in 2020.
c. Are these 5 states Arizona, Georgia, Michigan, Pennsylvania and Wis-
consin (which elected another candidate from a different party)? If not,
please give your reasons. If so, based on the polls, name one or two other
states that may elect another candidate from a different party in 2020
as well but did not happen in reality. Explain the reason.
5. (10 points) Compare the polls in Florida and Iowa in 2016 and 2020.
a. Are most of the polls in these two states accurate to predict the elected
candidates? If not, please give some reasons.
b. For Iowa, is there a poll that approximately correct for the final outcome
of the election in Iowa? What is the name of this poll? You may search
the internet to know more some information about this pollster.
c. Name a few possible reasons that account for the bias in polls for these
two states.
d. Discuss some possible ways to improve polls for political election.

学霸联盟































































学霸联盟