|
|
|
|
|
|
|
|
|
|
|
Syllabus
All material on this Web page is subject to change.
Course requirements will be final after 9/1/99
Time: Wednesday 4:15-6:15pm
Room: 2328N (2232N may also be used on occasion)
Course Description: The emphasis in this course will be on developing statistical reasoning: the ability to draw sensible conclusions from data. Research is never done in a vacuum, and you should expect to consult with colleagues for the rest of your life. At the same time, you should feel as comfortable being consulted as you do consulting with others. This course covers basic bivariate statistics and some models with multiple independent variables. Class time will be split between discussing the use and interpretation of various statistical models and actually analyzing data using the computers in the lab.
Objectives: By the end of the course, you should aim for the following.
Bachman, R. & Paternoster, R. (1997a). Statistical methods for criminology and criminal justice. New York: McGraw-Hill.
Bachman, R. & Paternoster, R. (1997b). Computer applications in criminology and criminal justice using SPSS. New York: McGraw-Hill.
Huff, D. (1993). How to lie with statistics. New York: Norton.
Required Software: You are required to complete homework assignments using SPSS for Windows. It does not matter which version or what platform (Windows, Mac, etc.) that you use. There is a student version available. This cannot be used for large data sets. I will order the SPSS Grad pack at the bookstore. This is fully functional, but available only to graduate students. Bachman & Paternoster (1997b) provides an introduction to using SPSS, as well as the data sets used in the course.
Course Flow: It is essential that you familiarize yourself with the reading material before the corresponding lecture. Lectures will summarize and clarify the reading. If you have questions about the reading, come prepared to ask them in class. General questions about the homework can be discussed at the beginning of class, specific problems are better discussed outside of class.
Examinations: The material from Bachman & Paternoster (1997) will be covered on three examinations. The examinations will not be cumulative but later material will always presuppose a familiarity with prior material. The examinations will follow a constructed response format. Content of the examinations will reflect the course objectives and will not be limited to computations. There will be no make-up examinations. You are allowed on 8.5 x 11 inch hand-written page of notes and a calculator to be used during each examination. Examinations will emphasize your ability to reason using statistical principles studied in the course.
Papers: You are required to read Huff (1993) and Abelson
(1995) on your own in preparation for the papers. Read Huff for the
first paper and Abelson for the second. For each paper, find a published
research report of interest to you. Use a different report for each
paper. Each paper must come from a peer-reviewed journal and carry
a 1999 copyright date. Include just the first page of the article
as an appendix to your paper. (If you are not sure if a given journal
is peer-reviewed, check the instructions to authors printed inside the
journal. These instructions will indicate if papers are subject to
peer review.)
Type at least 5 pages. After 8 pages I will
get grumpy. After 10 pages I will stop reading. If you find
yourself pressed for 5 pages, try broadening the concept or idea that you
have selected. If applicable, also consider using multiple examples
from the research article to apply the concept.
Late papers will be accepted with a 10% (= one letter
grade) penalty. Early papers are always accepted. If you have
a job that may prevent you from attending class when a paper is due, turn
it in early to avoid a misfortune.
Huff Paper: Under the heading Report Summary briefly review and summarize the research report. Under the heading Statistical Concept briefly describe a single idea or concept from the Huff (1993). Finally, under the heading Application describe in some detail how the statistical concept applies to the research report. Evaluate the report of the basis of the concept you have chosen. While writing the paper, bear in mind that your purpose is to demonstrate a clear understanding of the text in question and your ability to read research literature critically. (Critically does not mean negatively, it means making up your own mind based on sound reasons independent of the author's rhetorical allurements.) The third section of the paper will typically be longer than the each of the first two (but not both combined).
Abelson Paper: Complete the Report Summary section as with the Huff paper. In place of the Statistical Concept section, give a summary of Abelson's MAGIC principles using your own words. Finally, in the Application section, apply the MAGIC principles to the research report that you have chosen. Remember to use a different research report from the Huff paper.
Homework: You will be expected to run examples using SPSS and turn in printed output to demonstrate that you have done this. As such, you need to have a PC capable of running SPSS and a printer. It is understood that this is your first pass at the material, so homework assignments will be graded more for completeness than accuracy (not true of examinations and papers!). Homework is also an early warning system to help you evaluate your own understanding of the material. Late homework will not be accepted. Homework assignments will make use of the data sets from the disk included with Bachman & Paternoster (1997b).
Grading: Each of the three examinations is worth 20% of
your total grade. Each of the two papers is worth 15% of your total
grade. That leaves 10% for the homework assignments. Letter
grades will be assigned as indicated below.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Office Hours: Wednesdays & Thursdays 2:00-3:00 pm.
Office: 2428N
Phone: 212-237-8795
Email: KMarkus@AOL.COM (preferred)
|
|
|
|
||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
In preparing for Exam Three, first review the chapters,
make sure everything in the summaries is crystal clear, that you know all
the terms in the list of key terms, and that you can do the problems in
the back of the chapter. Remember, however, that this is an undergraduate
text and the problems in the back are just a way to check your understanding.
They are not test samples.
Listed below you will find a chapter by chapter
breakdown of skills you should have developed in covering these chapters.
These lists to not replace the chapters. You are still responsible
for all the material in the chapters. Instead, they provide a checklist
for some of the skills that you should have going into the exam.
If you find a skill on the list that you do not feel comfortable with,
review the material and then pelt me with email questions or stop by my
office.
Chapter Twelve: Two-sample tests for means and proportions.
12.1 Recognize the difference between a one and two-group test.
12.2 Recognize the difference between independent and dependent (paired)
samples.
12.3 Recognize when a pooled variance estimate is appropriate.
12.4 Match the appropriate analysis to a research design.
12.5 Recognize the appropriate output in an SPSS output file.
12.6 Correctly interpret the results given in an SPSS output file.
12.7 When presented with a formula from the chapter, explain each component
of the formula. (Note: This applies more to definitional formulae
than computational formulae which tend to be conceptually opaque.)
12.8 Recognize and correct a mistaken interpretation or application
of a two-sample test procedure.
Chapter Thirteen: Analysis of Variance.
13.1 Recognize an appropriate problem for the application of ANOVA.
13.2 Recognize an inappropriate problem for the application of ANOVA.
13.3 Explain why ANOVA is preferable to doing a series of pairwise
t
tests.
13.4 Translate a symbolic representation of an ANOVA hypothesis into
words.
13.5 Translate a verbal representation of an ANOVA hypothesis into
symbols.
13.6 Given an alternative hypothesis, formulate the (nil) null hypothesis
in both symbols and words.
13.7 Translate between string of (in)equalities between means and ANOVA
notation representing deviations from a grand mean.
13.8 Interpret ANOVA notation as a model of the processes that generate
the data.
13.9 When presented with a formula from the chapter, explain each component
of the formula.
13.10 Interpret an F statistic as a ratio of explained to unexplained
variance.
13.11 Correctly associate WG and BG variance with the explained and
unexplained components of the variance in the DV.
13.12 Interpret Eta-squared (h2)
as a measure of association and effect size.
13.13 Interpret SPSS ANOVA output for a one-way ANOVA.
13.14 Recognize and correct an erroneous interpretation or application
of ANOVA.
Chapter Fourteen: Correlation and Regression for Two Variables.
14.1 Construct and interpret a scatterplot.
14.2 Explain and apply the assumptions required for the use of correlation
or regression.
14.3 Recognize the components of a regression equation even if rearranged
or expressed in notational variants of the textbook notation.
14.4 Interpret the components of a regression equation.
14.5 Translate a regression equation into words.
14.6 Translate words into a regression equation.
14.7 Recognize and articulate the relationships between b, beta
(b), and r in bivariate regression.
14.8 Recognize and articulate what is being minimized to determine
the values of regression parameters alpha (a)
and beta.
14.9 Explain the components of the formula for a correlation coefficient.
14.10 Interpret the coefficient of determination as a measure of association
and effect size.
14.11 Read and interpret SPSS output for bivariate correlation or regression,
including tests of statistical significance for regression parameters.
14.12 Recognize and articulate problems associated with nonlinearity
or restrictions in variability of the variables.
14.13 Recognize and correct misinterpretations or misapplications of
correlation or regression.
Chapter Fifteen: Multiple Regression & Partial Correlation.
15.1 Apply 14.3 through 14.6 to a multiple regression equation.
15.2 Decompose R2 into its components in terms of squared
partial correlations.
15.3 Read and interpret SPSS output for multivariate regression, including
changes in R2 for successive blocks in hierarchical regression.
15.4 Read and interpret hypothesis tests for both individual regression
parameters and an equation as a whole.
15.5 Explain the components of an F ratio in the context of
regression analysis.
15.6 Recognize the differences between bivariate correlations and regression
weights (did you realize that in multiple regression standardized regression
weights can be greater than one or less than negative one?).
15.7 Recognize and correct misinterpretations or misapplications of
multiple regression.
15.8 Not freak when presented with any of the above. (This comes
with exposure and successive desensitization. Playing with the sample
data sets is a good place to start.)
Exam Two
Here are the grades for Exam Two. Read the
column heads by parsing them into three pieces of information in accordance
with the following syntax: The letters and digits beginning with
'e' indicate the part of the exam. 'e2' indicates the in-class portion
and 'e2a' indicates the take-home portion. The next two characters
identify the question. Finally, the percent sign indicates that the
scores are given out of a possible 100 points.
To refresh your memory, e2q1 and e2aq1
involved computing probabilities. e2q2 involved determining
which statistic was most likely to give a value below zero. e2q3
mistakenly involved a two-sample t test and was scored as a bonus
question. The remaining two questions involved interpreting categorical
data analyses.
e2totpct = sum( max(e2q1,e2aq1), e2q2,
e2q3, max(e2q4, e2aq2) )/3 where sum() gives the summation
of the items listed in the parenthesis and max() gives the maximum of the
items listed in the parenthesis. The list of four scores is divided
by three (not four) because of the bonus question. e2totpct gives
your final grade on Exam Two.
secret code name e2q1% e2aq1% e2q2% e2q3% e2q4% e2aq2% e2totpct
Tango
100% 88% 75% 0% 100%
100% 92%
Annie
33% 50% 88% 0%
50% 100% 79%
CHELSEA
33% 13% 100% 0% 54%
100% 78%
Mookie
33% 100% 75% 0% 92%
100% 92%
BOO
0% 88% 50% 0%
0% 100% 79%
Cookie Monster 33% 75%
75% 33% 88% 100% 94%
guapo
33% 88% 100% 0% 100%
100% 96%
"ralph"
33% 100% 100% 33% 81% 100%
111%
LOKI
33% 100% 50% 33% 27%
100% 94%
dolphin
33% 88% 75% 0%
88% 100% 88%
STATE
0% 88% 50% 0%
75% 100% 79%
EmmaLou
33% 75% 75% 0%
81% 100% 83%
Scully
33% 100% 100% 0% 100% 100%
100%
Paper Two
The following gives a rough
indication of the scoring rubric for the second paper:
0% = Failed to turn in the paper, or did not follow the assignment
in any recognizable way.
70% = Paper contained serious flaws, but at lest partially fulfilled
the assignment.
80% = Paper included at least two substantially weak subsections or
failed to demonstrate clear understanding, but overall the paper fulfilled
the assignment.
90% = Paper adequately fulfilled the assignment and demonstrated adequate
understanding of the material with no more than one weak subsection.
100% = Paper demonstrated exceptional understanding of material and/or
outstanding insight in applications.
secret code name paper 2
Tango
100%
Annie
90%
CHELSEA
100%
Mookie
95%
BOO
85%
Cookie Monster
95%
guapo
95%
"ralph"
90%
LOKI
90%
dolphin
100%
STATE
100%
EmmaLou
90%
Scully
90%
Exam Three
I recognize that there is nothing I can say here
that is going to stop you from scrolling down to find your letter grade
first. So, go ahead, follow your impulse. I'll insert a paragraph
break after this sentence so that when you scroll back up you will know
where to resume reading.
Okay, now that you know the end of the story, let
us rewind and start at the beginning. The column heads follow the
same syntax as for Exam Two. As you have already fathomed, course%
gives your overall percent grade in the course and lettergrd gives
your letter grade from the above table. This is the only part of
the process that I cannot automate, so please take a moment to check that
your letter grade matches the tabled value.
The four questions are equally weighted and the
bonus question counts as half a question. As such, e3tot%
equals the sum of the four questions plus one half the bonus question all
divided by four. The course percent is computed using the weights
given in the syllabus.
secret code name e3q1% e3q2% e3q3% e3q4% e3qb% e3tot% course%
lettergrd
Tango
100% 100% 83% 100% 100% 106%
97% A
Annie
83% 67% 58% 75% 50%
78% 88% B+
CHELSEA
67% 50% 58% 17% 0%
55% 85% B+
Mookie
67% 100% 58% 92% 100% 86%
93% A-
BOO
100% 67% 33% 42% 50%
65% 82% B
Cookie Monster 58% 67%
0% 17% 0% 35%
79% B-
guapo
100% 67% 67% 92% 100%
90% 95% A
"ralph"
58% 67% 50% 75% 100%
69% 88% B+
LOKI
58% 67% 75% 75% 50%
78% 87% B+
dolphin
67% 67% 42% 50% 100%
61% 89% B+
STATE
67% 75% 67% 100% 100% 85%
88% B+
EmmaLou
67% 67% 50% 92% 100%
75% 86% B+
Scully
67% 100% 75% 67% 100% 86%
96% A
If you did less well on the second two questions than you would have liked, I would suggest reviewing the linear regression chapters before diving into logistic regression next semester. Many of the same basic concepts apply. I have no way of knowing how many of you made use of the exam preparation materials that I posted to this Web page, but grading the bluebooks gave me a sense that some of you made more use of them than others. Obviously I wish that I had thought of that earlier in the semester. Difficulties with Exam Two notwithstanding, however, you did well both collectively and individually. Enjoy the rest of the break and keep up the good work.