Stat 250, Section 003, Extra Credit (Due 5/5/99 in class)
Each of the following questions is worth up to 20 points.
The policy for extra credit is that final grades will be determined
without looking at the extra credit points and cutoff point limits
will be set. Only thereafter, extra credits will be added. Therefore,
people that feel comfortable that they will be able to achieve the
grade they are looking for through regular course work do not
have to work on these extra credit questions. However, if you
feel uncertain about the outcome of your final grade, you may want to work
on one or multiple of these questions.
Please turn in each question starting on a different sheet.
- 1) We have seen several plots (e.g., the Weider empire)
so far that provide only very little of the information one
would like to have about the underlying topic. Can you
find similar plots in newspapers, magazines, on the Web, or even
your textbooks? If so, then state questions you would like to have answered
about the provided data but no such answers have been given in the
plot, the accompanying caption or the main text where the plot
is coming from. Indicate where the plot has been found
and turn in a photocopy (or printout) of this plot and the
accompanying text.
- 2) You have to use a Web browser to answer
this question. Use a search engine such as
http://www.excite.com/,
http://www.lycos.com/, or
http://www.yahoo.com/
and search for Webpages that relate to statistics and your
major subject.
On top of these pages you usually have a box
where you can type in the keywords you are looking for.
You can type in a single keyword, multiple keywords separated
by a blanc, e.g., kw1 kw2 kw3 (in this case, the search engine will look for
kw1 or kw2 or kw3), and multiple keywords with a leading +,
e.g., +kw1 +kw2 +kw3 (in this case, the search engine will look for
kw1 and kw2 and kw3). If you try "text" where text appears in double quotes
and may be a longer sentence, you should find pages that contain
these words in exactly the same order (but eventually with a few
other words inbetween).
As an example, I used
http://www.excite.com/
on 4/9/99 and tried
- +Statistics +Nursing (14647 hits)
- +Statistics +Nursing +Software (3116 hits)
- +Statistics +Poetry (6262 hits)
- "Government Statistics" (2135 hits)
As an answer to this question
- i) indicate which search engine you used and which keywords
you have searched,
- ii) indicate 3 to 5 interesting Web addresses that are
related to statistics and your major subject (please list this
subject again on your answer sheet), and
- iii) visit the pages from ii) above and provide a 2 or 3
sentence summary on what this page is about, e.g., providing
particular data, data about this subject (income distribution,
number of people with this profession), information on
(downloadable) software particularly helpful in this subject,
information on new statistical methodology for this subject, etc.
- 3) We have seen many small examples where JAVA applets have
been used, e.g., to demonstrate the effect of different bin widths
for histograms. The page
http://www.stat.sc.edu/~west/javahtml/Histogram.html
contains instructions how to modify this example and how to include
your own data. Can you create a Webpage that makes use of this
particular JAVA applet and gives an introduction to some
of our class data set(s) and allows to look at the effect
of a varying histogram for this data. It would be ideal if you
could create a page that contains two histograms (displaying payroll
and wins) as well as the regression applet from
http://www.stat.sc.edu/~west/javahtml/Regression.html.
As an answer to this question, just provide the URL where
your created Web page can be found.
- 4) We did not introduce any statistical package in this class
however many of you used statistical software for your homeworks.
The aim of this question is to explore an unknown package,
http://statlets.com,
and get it to work. You have to select 'Internet Access' first,
then the appropriate 'Menu Version' with respect to the capabilities of
your Web browser.
Also, read carefully through the on-line
documentation and use a small data set to get familiar with the
program first.
Move on one step and use the full "Cost of Victory: Payroll/Win" data set.
This data is available in electronic form at
http://www.galaxy.gmu.edu/~symanzik/teaching/1999_stat250/victories.dat.
Incorporate this data into
http://statlets.com
(using the 'Clipboard' option might be the easiest way to do so)
and do a full statistical analysis (graphical and numerical).
As an answer to this question, you can turn in printouts or
write down main results from the screen (make sure to maintain
rounding and classes used for drawings in exactly the way
http://statlets.com
does so I can see you really used this package).
Alternatively, you can indicate which steps (what is
the exact labeling of each button you selected at a given time)
you conducted to achieve a particular result. If printing does
not work for you, you should at least copy and paste numerical results
(what does the 'Statistical Interpreter' indicate)
from a http://statlets.com
window into your text processor window and print this text
later from a different computer. You can restrict yourself
to those statistical features known from class - you do not have to
explore every feature of this software package.
You cannot use any other package to answer this question!
Also note that
http://statlets.com
does not run on older machines and typically does work better
under Internet Explorer than under Netscape. Plan on enough
time to work on this question. If you cannot find an appropriate
computer, come to my office during my office hours - there
are several computers you can use for this purpose.
- 5) The following Web page from the Colorado Department of
Regulatory Agencies (DORA) describes, among others, the
number of registered nurses (RN) from 1/91 to 1/97:
http://www.dora.state.co.us/Nursing/Nur0017.htm
Do a complete statistical analysis (numerical and
graphical) of the RN data set and answer the
following questions:
- Based on the statistical model you have fitted, how many registered
nurses does your model predict
for 7/95 and for 1/99, respectively? Do these numbers
make any sense?
- Assess the quality of the statistical model you have fitted.
- Is there anything strange in the RN data?
- Is there any (strong) relationship between registered nurses (RN)
and nurse aides (NSE Aides)?
- Provide a written summary (about one half page) that includes
all your previous findings.
You can calculate your results by hand, use any statistical package
you like, or obviously use
http://statlets.com
that has been introduced in question 4) above. This question will
be graded with respect to correctness and completeness
(did you do all the steps that make up a full analysis).