PA 552 Professor
Brian Stipak
Winter 2010 e-mail:
stipakb@pdx.edu
Tues 6:40-9:20 phone:
725-3043, 725-3920
http://web.pdx.edu/~stipakb
COURSE SYLLABUS: PA 552
Analytic Methods 2
Read this
syllabus carefully if you are taking this course. It defines some of your responsibilities and some of my
responsibilities.
This is a continuation of the PA 551 course. It is the second course in a two‑term
sequence covering analytic methods used widely in the public sector, focusing
specifically on the collection, analysis, and use of data. The aim of the course is not to train
analysts but rather to provide the understanding that administrators need in
their typical role as consumers of analysis generated by others. If you feel this is not relevant to an MPA
degree, then ask yourself what percent of administrators work in agencies that
record no quantitative data, include no statistics in any reports, and have
never had consultants or employees produce any statistics. Your answer will also answer the question
why you need an understanding of the course material as a professional public
administration.
Course
Materials
On-line materials available through my web site (see
below).
Paul D. Allison, Multiple Regression: A Primer, ISBN 0-7619-8533-6.
(continued use from PA551:)
Jane E. Miller, The Chicago Guide to Writing about Numbers, ISBN
0-226-52631-3
I recommend that you have a basic statistics book
available for reference, such as the Triola text (See "Additional
Reference Books" below.)
On-Line
Course Materials
Course materials and links to relevant web sites are on
my web site. Go to the page
"Resources / help for my students and for others". Carefully look over:
1) the PA552 help page
2) the links under "Data, Statistics, Statistical
Computing"
3) the
files in the "PA552" folder and the "Data" folder
The readings listed in the course outline that are
available on‑line through my web site have the designation "(web)"
in the course outline.
Since information available on the web is growing and
dynamic, please let me know when you find other on‑line materials that
are helpful for this course. You can
email me and/or you can post to the course listserv.
Course
Listserv
I have subscribed all students enrolled in the course
using the email address for each student in the PSU Information Sytem. If you
want to subscribe using a different email address then you will need to
subscribe yourself; do so within two days after the first class meeting.
I will use this listserv to send messages to the class members. You can also send messages to the class
members‑-for example, to solicit help and to help other students
requesting help.
To subscribe to the listserv go to the information page
for this listserv, which you can link to via my web site, or you can just go
directly to:
"https://www.lists.pdx.edu/lists/listinfo/pa552"
Additional
Reference Books
I recommend that you have a basic statistics text
available, and there are many textbooks that can help you as further
references, such as (I reference some of these under “Further Reference” in the
course outline):
Mario F. Triola, Elementary Statistics Using Excel,
ISBN 0-321-36513-5, or one of the other versions of this textbook.
Kenneth J. Meier and Jeffrey L. Brudney, Applied
Statistics for Public Administration.
Consult for simple presentation of basic statistics.
Anthony Walsh and Jane C. Ollenburger, Essential
Statistics for the Social and Behavioral Sciences. Clearly combines a conceptual treatment with instruction on
some of the important details of basic statistics.
Frederick Gravetter and Larry Wallnau, Essentials of
Statistics for the Behavioral Sciences.
A simple text for the basic statistics material.
Frederick Williams, Reasoning with Statistics. Good overview of statistical concepts.
Richard Jaeger, Statistics: A Spectator Sport.
Good general reference that covers a wide range of topics.
T. H. Wonnacott and R. J. Wonnacott, Introductory
Statistics for Business and Economics, 4th edition. I consider this one of the best texts in the
MBA stat market, but it is more difficult than the other references.
Course
Requirements
1) Learn the course material that is covered in class. There will be a final examination on that
course material.
2) Complete the computer assignments and all other
homework assignments.
3) Participate in any
classroom exercises.
4) Do a statistical
analysis project and turn in a paper presenting your analysis.
Course
Project and Project Paper
For purposes of your statistical analysis project you can
use one of the datasets that I make available, or you can use another dataset
that you wish to analyze. You may want
to use that dataset throughout the term for some of the required
exercises. People can work together on
projects, but you have to turn in individual project papers that overlap only
to a minor extent. All writing and
final versions of tables or graphs contained in your project paper must be your
own individual work.
The project papers (5-20 pages) should explain the
purpose of the project, describe the data and variables used in the analysis,
and present and interpret the statistical results. Present your findings using appropriate tables and figures which
you explain in the text. If your
project uses your own data, attach computer printouts of the main statistical
results you report in your paper; if your project uses a dataset I provided,
you do not need to include computer output with the paper. Keep all computer results you use in your
project paper until you receive your course grade, in case I have questions
about your project. Write your email
address and phone number(s) on your paper so that I can contact you if
necessary.
For further information see:
1) DataAnalysisProjectInfo.doc,
"Course Project Information: Help on Getting Started", on my web site
2)
DataAnalysisProjectGuide.doc, "Course Project and Project Paper: Simple
Guidelines" (or "What to Do When You are Desperate"), on my web
site
3) Examples of past
course project papers, available for viewing in the PA Division Office
I encourage you to think about your project choice early
in the term, especially if you want to use data from work or other data that
requires additional preparation. I am available to consult with you about your
project throughout the term.
Computer
Exercises
The main software we will use this
term will be statistical software: SPSS and also the on-line (web-based) SDA
interface. The reason for using
statistical analysis software in PA552 is to provide an appreciation for the
power and utility of specialized analytic software, and also because SPSS and
SDA will provide useful tools for doing data analysis exercises and for
carrying out independent data analysis projects.
Computer software availability: At PSU SPSS and Excel are available for use in the various on‑campus
computer labs. Excel will also be on almost any computer to which you have
access. I recommend that you do SPSS work using the on-campus computers. If you
greatly desire to install SPSS on your own computer, you can 1) purchase the
full version of SPSS, called SPSS Graduate Pack, for a discounted educational
price, 2) rent SPSS very cheaply (get the Graduate Pack Base), 3) buy Student
SPSS, or 4) install a temporary free demo version of SPSS (See my web site for
more information on these options). Student SPSS differs from the full version
in that 1) it is limited to a maximum of 50 variables and 1500 cases, 2) lacks
some additional procedures, and 3) does not support SPSS syntax (command
language).
You may find working with other students helpful in doing
the computer assignments. However, the
computer assignments that you hand in require that you do your own computer
work for the assignments: each person
must sit at a computer and do every assignment. Similarly, a tutor may help you but must not do the assigned work
for you. Any student who turns in
computer work as his/her own, when in reality the computer work was done by
someone else, is committing academic dishonesty subject to disciplinary action
under university rules.
Course
Grades
I will assign course grades as
fairly as I can based on a computerized total points score and calculated
grade. I usually allow myself a maximum
leeway of one increment (for example, B to B+) from the calculated grade to
take into account class preparation, participation, and other special
considerations.[1] The total point score is computed from
scores on the following criteria (approximate weights in parentheses):
(40%) 1. Final
examination
(35%) 2. Course
project, including meeting paper due date
(25%) 3. Computer and other assignments, including
meeting due dates
I do not give all graduate students "A's". A "B" is also a passing grade at
the graduate level. A "C" or
lower means I feel your work in the course was below acceptable graduate
student standards. In practice, for
this type of course I typically assign about half of the grades in the A/A-
range, half in the B+/B/B- range, and occasionally assign a grade of C or
lower.
The university policy on assigning incomplete ("I")
grades restricts their use to special circumstances (see PSU Bulletin). Missing assignments will usually result in a
lower grade or an "X" grade, not an "I". If major assignments are missing a very low
grade may result. If you enroll in this
course but find you are not able to do all of the assignments, I recommend that
you drop the course. If you remain in
the course and do not complete all of the work, you need to write me (or email
me) to request an "I" grade.
I consider academic dishonesty a serious offense and will
penalize offenders to the full extent possible under university policies. Academic dishonesty includes cheating on
examinations, copying or stealing other people's work, turning in work done by
someone else as one's own, plagiarism, and other kinds of
misrepresentation. Turning in computer
assignments done by another person would be an example. If you know of any of these problems
occurring in this class, please let me know so that I can investigate. The vast majority of you would never do
this, and my commitment to you is if I should learn about such cheating I will
deal with it severely, regardless of who the student is and what excuses the
student offers.
Return of
Assignments, Examinations
Any assignments I have not returned to you by the last
class meeting you can pick up from me after finals week. You will not get back copies of
examinations, but a key exam will be available for you to see immediately after
completing exams, and your scored exams will be available for you to examine in
my office.
My
Availability Outside of Class
I am available to consult with you about course matters
by phone and in person outside of class hours.
To see me in person, schedule a time with me. Also, feel free to stop by my office if you are in the CUPA
Building; if I am in, I will see you then if I can. This combined by-appointment/flexible-office-hours approach
provides a more practical way to arrange consultation than limited specific
office hours, given the varied work schedules of students in our program.
Course
Outline
Note: Readings with the designation "(web)" are available
through my web site. Authors
names/abbreviations refer to references listed earlier.
1. Introduction and
Overview
Overview
of Course
2. Brief
Re-Consideration of the Role and Meaning of Methods of Basic Statistical
Inference
Hypothesis
Testing vs. P-values vs. Confidence Intervals
The
"Significance" of Statistical Significance
•Statistical Significance vs. Importance
•Statistical Significance and Sample Size
•Sample Size and Statistical Power
•Non-Plausibility of most Null Hypotheses
•Hypothesis Testing vs. Estimation
Triola,
Ch. 7-8
Miller,
pp. 40-52
3. Overview of
Research Design
Choice
of Research Methods
Concepts
of Research Design
Causal
vs. Spurious Statistical Relationships
Experimental
versus Observational Studies
Statistical
Controls vs. Experimental Controls
Triola,
Ch. 1
Miller,
pp. 33-40
Further
Reference: G & W, Ch. 1; W
& W, Ch. 1; Jaeger, Ch. 6; Research
Methods Knowledge Base (web), other links on PA555 help page and under
"Program Evaluation" (web)
4. Statistical
Relationships
Statistical
Independence vs. Association
Comparison
of Means, Percents, Rates Tables
Crosstabulation
(Contingency) Tables
Scattergrams
Types of
Relationships, Terminology
Chi-Square
Test of Statistical Significance
Analysis
of Variance
Causal
Analysis
Miller,
pp. 26-31, 33-40, 190-199, 235
Triola,
Ch. 10 (sec. 10-3)
T-Test for Difference in Means: How to Do in SPSS and How
to Interpret the Output (web)
Crosstabulation
Table Analysis / rules for crosstab analysis (web)
Statistical Interaction explanation and Statistical
Interaction exercise (web)
Further
Reference: Statsoft (web),
basic-statistics/crosstabulation; G & W, Ch. 16; W & O, Ch. 10; Jaeger,
Ch. 12; M & B, Ch. 14-16; Bruce Bowen and Herbert Weisberg, An
Introduction to Data Analysis, Ch. 6.
5. Correlation
Rank-Order
Correlation
Pearson
Correlation
Correlation help, various (web)
Triola,
Ch. 10-1, 10-2
Miller,
pp. 191-192
Statsoft
(web), basic-statistics/correlations
Further
Reference: Statsoft (web), Basic
Statistics / Correlations; G & W, Ch. 15; W & O, Ch. 11; Williams,
Ch. 11; Jaeger, Ch. 4
6. Simple
Regression Analysis
Least
Squares Estimation
Basic
Regression Model
Interpretation/Tests
of Regression Coefficients
Triola,
10-3, 10-4
Allison,
sec. 1.1-1.5, 1.8-1.10, 1.12, 5.1-5.4
Equation for a straight line, on-line tutorial (web)
Basic Regression Model (web)
Further
Reference: Statsoft (web),
linear-regression; Gravetter & Wallnau text, Ch. 15; Williams, Ch. 12;
Jaeger, pp. 333-339; M & B, Ch. 17; W & W, Ch. 11-12, 15 pp.
482-496
7. Multiple
Regression Analysis
Multiple
Regression Model
Model
Specification
Multiple
Correlation, Variance Explained
Multicollinearity
Allison,
Ch. 1-7
Triola,
10-5, 10-6
Further
Reference: Statsoft (web), linear-regression; Multiple Regression with Ren & Stimpy (web); Williams, Ch. 13;
M & B, Ch. 20; W & W, Ch. 13, 15 pp. 496-506
8. Regression
Extensions
Dummy Variables
Nonlinear Effects
Interaction Effects
Dichotomous
Dependent Variables
How
to use and interpret Dummy Variables (web)
Allison,
Ch. 8
Further
Reference: Williams, p. 165; W
& W, Ch. 14, pp. 435-445; W & O, Ch. 12
9. Applications of
Regression Analysis
Causal
Modelling, Path Analysis, Econometric Models
Contextual
Analysis
Allison, 2.1, 2.2, 2.3, 2.4, 2.5, 3.6
(Note: Following three items may or may not
be covered.)
Brian Stipak, "Citizen Satisfaction with Urban
Services: Potential Misuse as a Performance Indicator", Public
Administration Review (Jan/Feb, 1979).
John Pucher et al., "Impacts of
Subsidies on the Costs of Urban Public Transport", Journal of Transport
Economics and Policy (May, 1983).
"Aggregate Data Regression Analysis of the 1968 Los
Angeles Rapid Transit Vote", see file
"RegressionSCRTDvoteExample.doc" (web)
Further
Reference: W & W, Ch. 13, pp.
417-433
10. Time-Series
Analysis/Forecasting
Decomposition of a Time
Series
Regression Models
Moving Average Smoothing
(Note: Following three items may or may not
be covered.)
Cliff Ragsdale, Spreadsheet Modeling and Decision
Analysis, pp. 525-530, 536-541
handout, "Exponential Smoothing Forecasting
Techniques"
"Forecasting Error Analysis in Spreadsheets",
see file "ForecastErrorAnalysis.doc" (web)
Further
Reference: M & B, Ch. 19; W & W, Ch. 24
11. Brief Overview
of Other Statistical Techniques
Allison,
Ch. 9
Further
Reference: Williams, Part 6; Jaeger, Ch. 15
PA 552 Winter
2010
Approximate Class Schedule*
|
|
Date |
Assignments Due** |
Topics / Activities |
|
1 |
1/5 |
|
Overview /
Miscellaneous Reconsider Statistical Inference Research Design Statistical
Relationships |
|
2 |
1/12 |
|
Statistical
Relationships |
|
3 |
1/19 |
CA1 |
Statistical Relationships |
|
4 |
1/26 |
CA2 |
Statistical
Relationships |
|
5 |
2/2 |
CA3 |
Correlation Regression |
|
6 |
2/9 |
|
Regression |
|
7 |
2/16 |
CA4 |
Multiple Regression Regression Extensions |
|
8 |
2/23 |
|
Regression Extensions Regression
Applications |
|
9 |
3/2 |
|
Forecasting |
|
10 |
3/9 |
3/12 Project Papers Due |
Forecasting Overview of other techniques Review |
|
11 |
3/16 |
Course Evaluations, Final
Examination |
|
|
|
|
||
*This class schedule is
approximate, and adjustments will occur during the term. Some topics above appear for more than one
date because a topic may take more than one class period, or to allow leeway
when the topic is covered.
**Due dates for computer
assignments may change, and will be announced in class and/or on the course
listserv.