Homework Assignments

You are encouraged to use software such as SPSS, R, Excel, etc. The raw data for virtually all of the homework problems are both on the CD in the back of your book, and on Howell's website: (http://www.uvm.edu/~dhowell/StatPages/StatHomePage.html).

On the CD, you will find them in several formats, including tab-delimited text, Excel, and SPSS.

General Notes on the homework assignments:
- The response to each item should be in the form of a short results section or essay, as appropriate.
- The specific questions and requested graphs are items to be encorporated into your answers, not items to be addressed separately.
- Grades for each week's homework will be on a 10 point scale, with 5 points alloted to presentation and clarity, and 5 points to completness and content.
- Brevity and clarity make for good scientific writing.
- Embedding your figures in your answers (as opposed to having them all at the end) is strongly encouraged.

Here is an example homework. The only formatting improvement I might suggest for this would be embedding the figures in the text, rather than placing them at the end.

Assignments will be updated as we go...

I. (Ch. 1):  21 but, before addressing (a) through (e) from the book, use these data and do 1) either stem-and-leaf plots or histograms of the two variables, 2) do a scatterplot of disease rate (hdrate - in incedence per 1000) vs. cigarette consumption (smk - in cigerettes per day), and 3) (as always) write a brief summary of the results.

II. (Ch. 2):  1, 2, 4, 6, 7, 41, 42, 43, 44, 45, 50, 51 - note: taken as a whole, this set of questions just boils down to "analyse and summarize the data given in 2.1 and 2.4."
The data are on the CD in the back of your textbook and on Howell's website .

Week 2

I. (Ch. 3):  Make a histogram of the reaction time data from Table 2.1 / Fig. 2.1 (ignore the other two variables) and overlay a normal distribution (or, if you prefer, you can do a cumulitive frequency distribution and overlay a cumulative normal). Then, compare the two and discuss the implications of using a normal approximation along the lines of problem 3.20.
Extraspecialsupersecret hint: you may want to look into something called a Quantile-Quantile plot (a.k.a. QQ plot).

II. (Ch. 4):  Address the problem outlined in 4.7, 4.8, 4.9, and 4.13, and include a plot or sketch of the appropriate normal distribution with the location of the questionable score clearly marked.

III. (Ch. 4):  Answer question 4.21. Make sure you explain the logic behind your approach ("I would do a correlation and see if it was significant" would not, e.g., be an acceptable answer).

Week 3

I. (Ch. 7): Address questions 1 through 5 (the data are on the web & cd as usual). Your answer should take the form of an instructional essay on CLT. The Rice Virtual Statistics Lab (see Links page) would be a good place to build your intuition and check your work. Follow your nose to the "sampling distribution simulation" applet.

II. (Ch. 7): Address the questions raised in 25, 27, 28, 31, and 32 in the form of a results section.

III. (Ch. 8): Do 1, 2, and 3. Include a sketch for the 0.90 case as well. Essentially, you will be defending your choice of a sample size, as you will have to do in your dissertation proposal and many grant proposals...

Week 4

Exam I - no homework assigned

Week 5

I. (Ch. 5): 14, 15

II. (Ch. 5): 21, 22, 23
Some people have trouble with the wording of this one, so let me try to clarify with a more concrete example. An experimenter is trying to determine how many trials it will take subjects to discover which of 5 symbols is the "correct" one (a " * ", say). One each trial, the subjects are shown something like
   *    %    #    @    $
and they have to pick the initially-unknown (to them) correct symbol. Obviously, they have no choice but to make a random guess on the first trial. They are then told whether they are right or wrong, and the next trial is run...
So basically, you need to determine what you would expect to see if all the subjects are guessing, so that you can then determine how many subjects getting the correct answer on a given trial would constitute a significant departure from the "everybody is still guessing" hypothesis.

III. (Ch. 5): 33

IV. (Ch. 6): 5, 6, 7

V. (Ch. 6): 34, 35

Week 6

I. (Ch. 9): Analyze (i.e. write a results section around) the data of problem 1. You can use 2, 3, and 4 as hints about the kinds of things you should look at. Also, after you sketch the suggested "chi-by-eye" line, compute and draw the actual OLS (ordinary least squares) best-fit line on the same graph.

As a secondary analysis, look at the data as suggested in problem 37, but do so (separately) for the three conditions. In a final paragraph, address the general issue of whether changes over time (like learning or fatigue) would be expected to be linear.

(If you are using SPSS, you may have to change some variables to "scale" and, to get the change to stick, you may have to save the file, then quit and restart SPSS.)

Week 7

I. (Ch. 11): Complete the summary table as per 22. Write a little results section based on this summary table, and include a graph of speed vs. number of people listening including error bars approximating 95% confidence intervals (which you should be able to estimate - and you can draw them in on your graph by hand if need be). Does this plot suggest a method of analysis besides an ANOVA?

II. (Ch. 11): Using the data from "epineq.sav" (or "epineq.dat") write a brief results section along the lines suggested in 29 and 30 (28 and 29 in 5th Ed.). The focus of the results section, however, should be a graph of response ("errors") vs dose for the three delays. The requested ANOVA's should play a supporting role. As above, do the nature of these data suggest an analysis other than an ANOVA?

III. (Ch. 12): Write a brief results section on the data of question 1. Include the analyis requested in 1a, but put the answers to 1b and 1c in an appendix.

IV. (Ch. 12): Write a brief results section on the data of 25 incorporating 25a through 25e.

Week 8

Exam II - no homework assigned

Week 9

I. (Ch. 13): Write a results section for the data of 13.5, including the elements suggested in 13.6 through 13.10. As always, however, the focus of a results section should be a sensible plot or plots of the data.

Week 10

I. (Ch. 15): Write a short results section for the data of 15.4

 Note: The description of the Mireault.dat file in the appendix of the textbook doesn't match the actual data set I have. In my data file, there is simply one variable per column.

Week 15 or so

Exam III - no homework assigned - here's the study guide!