APPLIED STATISTICS

PROJECT ONE

This project is worth 15% of your final grade. It is due on Friday April 11 by 5pm. No late
projects will be accepted. Please do not discuss the project with others. The answers to the
project questions (i.e., what you hand in) should take a similar form to the posted tutorial
solutions. Each question is worth 20 marks. Within each question the parts are worth equal
marks. The data files can be found on the oracle.

Question One (ex 7.30 from text)

Studies over the last two decades have shown that activity can effect the reorganisation of the
human central nervous system. For example, it is known that the part of the brain associated
with activity of a finger or limb is taken over for other purposes in individuals whose limb or
finger has been lost. In one study, psychologists used magnetic source imaging (MSI) to
measure neuronal activity in the brains of nine string players (six violinists, two cellists, and
one guitarist) and six controls who had never played a musical instrument, when the thumb
and fifth finger of the left hand were exposed to mild stimulation. The researchers felt that
stringed instrument players, who use the fingers of their left hand extensively, might show
different behaviour in the brain as a result of this extensive physical activity than
individuals who did not play stringed instruments. The file violin.csv contains data on the
neuron activity index from the MSI and number of years that the individual had been playing
a stringed instrument (zero for the controls).

a). Is the neuron activity different in the stringed musicians and the controls?

b). Is the amount of neuron activity associated with the number of years the individual
had been playing the instrument?

Question Two (ex 8.25 from text)

The U.S. presidential election of November 7, 2000 was one of the closest in history. As
returns were counted on election night it became clear that the outcome in the state of Florida
would determine the next president. At one point in the evening, television networks
projected that the state was carried by the democrat nominee, Al Gore, but a retraction of the
projection followed a few hours later. Then, early in the morning of November 8, the
networks projected that the Republican nominee George W. Bush, had carried Florida and
won the presidency. Gore called Bush to concede. While on route to his concession speech,
though, the Florida count changed rapidly in his favour. The networks once again reversed
their projection, and Gore called Bush to retract his concession. When the roughly 6 million
Florida votes had been counted, Bush was shown to be leading by only 1738 and the narrow
margin triggered an automatic recount. The recount, completed in the evening of November
9, showed Bushs lead to be less than 400.

Meanwhile, angry Democrat voters in Palm Beach County complained that a confusing
butterfly lay-out ballet caused them to accidentally vote for the Reform Party candidate Pat
Buchannan instead of Gore. The ballot, shown below, listed presidential candidates on both a
left-hand and right-hand page. Voters were to register their vote by punching the circle
corresponding to their choice, from the column of circles between the pages. It was suspected
that since Bushs name was listed first on the left-hand page, Bush voters likely selected the
first circle. Since Gores name was listed second on the left-hand page, many voters who
already new who they wished to vote for did not bother examining the right-hand side of
the ballet and consequently selected the second circle in the column; the one actually
corresponding to Buchannan. Two pieces of evidence supported this claim: Buchannan had
an unusually high percentage of the vote in that county, and an unusually large number of
ballots (19000) were discarded because voters had marked two circles (possibly inadvertently
voting for Buchannan and then trying to correct the mistake by then voting for Gore.)


The file vote.csv contains data on the number of votes for Buchannan and Bush in all 67
counties in Florida.

a). Produce plots of the number of Buchannan votes versus the number of Bush votes and
another plot for the log of these two variables. Does the log-log transformation appear better
for performing a simple linear regression?

b). Analyse the data without the Palm Beach County results to obtain an equation for
predicting Buchannan votes from Bush votes.

c). Use the residual and normal probability plots from the regression in (b) to check the
adequacy of the model.

d). Obtain a 95% prediction interval for the number of Buchannan votes in Palm Beach
assuming the relationship is the same in this county as the others.

e). Comment on the result in (d) given that Buchannan actually received 3407 votes in
Palm Beach County.


Question Three (ex 10.29 from text)

The file wage.csv contains a dataset from the 1988 March U.S. Current population Survey.
The set contains weekly wages in 1987 (in 1992 dollars) for a sample of 25632 males
between the age of 18 and 70 who worked full-time along with their years of education, years
of experience, an indicator variable for whether they were black, an indicator variable for
whether they worked in a standard metropolitan area (i.e., in or near a city), and a code for
the region in the U.S. where they worked (northeast, midwest, south, and west).

a). Does this data provide evidence that black males were paid less than nonblack males
in the same region and with the same levels of education and experience? Provide a 95%
confidence interval for the average amount less that black males were paid. (Include all the
explanatory variables in your model. Do not worry about transformations and interactions at
this point)

b). Does the extent to which blacks were paid differently than nonblacks depend on
region? Provide a p-value in your answer.


Question Four (ex 11.10 from text)

As part of a study to investigate reproductive strategies in plants, biologists recorded the time
spent at sources of pollen and the proportion of pollen removed by bumblebee queens and
honeybee workers pollinating a species of lily. This data is contained in the file pollen.csv.

a). Draw a scatterplot of the log of the proportion of pollen removed relative to the
proportion unremoved [if p is the proportion removed, take Y=log(p/(1-p)] versus the log of
the duration of visit, use a different plotting symbol for workers.

b). Fit the regression of Y (as described in (a)) on the two explanatory variables and their
interaction, and obtain a residual plot. Does the residual plot indicate any problems?

c). Obtain a set of case influence statistics leverage, studentized residuals, and Cooks
distance. Are there any problems? What is the most advisable course of action?

d). Does a significant interaction appear to exist, or can the simpler parallel regression
lines model be used?



Question Five

A study of the relationship of body fat to several physical measurements was conducted on a
sample of 20 healthy women aged 25-34 years. The explanatory variables were the triceps
skinfold, the thigh circumference, and the midarm circumference. This data is contained in
the file btf.txt. To read this data in use the use the command read.table(btf.txt,header=T).

a). Fit the regression of body fat on the three explanatory variables. Comment on the sign
of the coefficient for thigh circumference.

b). Calculate the variance inflation factors for each of the explanatory variables. Why are
these values so large?

c). Fit the regression of body fat on triceps and midarm. Recalculate the variance
inflation factors. Comment on the difference between the variance inflation factor values in
(b) and (c).


The remaining questions are for students enrolled in STAT8001 only!

Question Six

In the case of simple linear regression it is often of interest to fit a model that passes through
the origin. Recall, however, that this assumption implies more than the fact that the origin is a
reasonable intercept for the data, but also that the linear relationship between the two
variables under study holds not only in the range of the collected data, but also all the way
back to the origin. If such an extrapolation is justified then, we might want to model our data
as:

(Y|X)=X

In this situation, find the least squares estimator of . Further, assuming constant variance
and independence, find the variance of this estimator. (show your working).

Question Seven

Show that the least squares estimator of found in Question 6 is the same as the maximum
likelihood estimator of . (Hint: under the SLR assumptions
2
i i i
Y|X ~N(X , ) )