PAF 502 Public Service Research II
Fall, 2004

• Course Outline
  • Dated 8/23/04
  • This is a planning tool--bring it to class and use it.

Introduction: To the Course and to Each Other

August 24--Week 1

Subjects : Introduction. The members of the class. The anatomy of the syllabus. Why a book on econometrics in a PA program? Why homework? Introduction to ethics of statistical research and presentation; begin statistics review

Readings: Sects. 6.4.1 Data Mining, 11.5 The Ethical Econometrician, start of Ch. 16

Part I: Fundamentals of Regression Analysis

August 31--Week 2

Readings : Material for last week (6.4.1 and 11.5), Using Econometrics , Ch. 16

Subject : Review of statistics underlying regression analysis

Especially Important Point : Don't confuse statistical significance with policy significance! (See 16.5.6, p. 560)

In-class Exercises p. 26 Q3(some?), p. 28, Q7b(only), p. 564 Q2, Q4, Q6, Q8, Q10, Q14

(did not get this done S04—only got to START of Z-scores) Don't always do the whole in-class process; sometimes have them look over, then you talk through

Study Tip: pp. 563-4, Q1 covers important concepts from the chapter

September 7--Week 3

Quiz 1—Covering the basic statistics reviewed in Ch. 16 and in class —at the end of class, after finishing review of underlying stat's

H, make one Q interp and then essay? Big stat sig, but little policy meaning

Readings : Using Econometrics , Ch. 16

Subject : Review of statistics underlying regression analysis

Especially Important Point : Don't confuse statistical significance with policy significance! (See 16.5.6, p. 560)

In-class Exercises p. 26 Q3(some?), p. 28, Q7b(only), p. 564 Q2, Q4, Q6, Q8, Q10, Q14 Don't always do the whole in-class process; sometimes have them look over, then you talk through

Readings : Using Econometrics , Ch. 1

Subject : Introduction to ‘trics; Quiz 1 returned and discussed

Especially Important Point : Check your data file! GIGO [Garbage In, Garbage Out]

In-class Exercises p. 26 Q2 (regression), p. 27 Q5, p. 32 Q12 use ALL these Don't always do the whole in-class process; sometimes have them look over, then you talk through

Homework 1 : p. 28 Q6, p. 29 Q8, pp. 31-32 Q11, pp. 32-33 Q13 (entire—enter data and run the regression—turn in data file and output), and p. 57 Q3

Handout: Campbell (1995) “Influencing Bureaucracy? A Research Note on Implications of Measuring Participation in Public Utility Rate Cases” JPART , 5(3): 319-329.

Discussion Questions: What is the research question? How does the author decide what variables to include? What is the method of statistical analysis? What are the policy implications?

Friday, Sep. 17: Unrestricted Withdrawal Deadline

Readings : Using Econometrics , Ch. 2

Subject : OLS (Ordinary Least Squares)—the basic regression model.

Especially Important Points : “quality of fit is only one of the measures of the overall quality of that regression” (p. 56). Conformation to theory is at least as important! Policy meaning is at least as important!

In-class Exercises p. 57 Q3, pp. 59-60 Q 7 (assume this is at ASU and just DISCUSS differences between them, rather than picking,) p. 62 Q13 REGRES for SURE

Homework 1 DUE

Homework 2: p. 57 Q4 but not (b) tell them why they're doing this ; pp. 58-59 Q6, for (e) explain how you decide ; p. 60 Q9

September 28--Week 6

Readings: Using Econometrics, Ch. 3, Campbell (1995) (prepare discussion questions)

Subject : How to do a regression study and how to read a regression article

Especially Important Point: Dummy variables aren't dumb (uses of dummy variables)

In-Class Exercises pp. 78-79 Q2 YES, pp. 79-80 Q3, p. 80 Q4, pp. 81-82 Q8 YES

Homework 2 DUE

Homework 3: pp. 109-110 Q7 (Due next week).

Homework 4: Find and read 3 research articles on the topic you plan to do your final research paper on. Write a 2-4-page paper that could be similar to the literature review in your final paper. Explain what you have learned from this article that will help you develop your own regression analysis for the final paper. This explanation should be thoughtful and critical. What does the article do right? What does it do wrong? What do the results tell you about things you should do or not do? If a research article uses regression , explain the regression (see pp. 65-67, 3.1.2). Then, explain what changes you think you would make to the article's regression and why (or carefully explain why you wouldn't make any changes). If a research article does not use regression , nonetheless tell what you learn about variables you'll include (or won't) in the regression you will estimate in your paper. Turn in copies of the articles with your homework . DUE IN THREE WEEKS, OCTOBER 19.

October 5—Week 7

Readings : Using ‘Trics , Ch 4

Subject : The Classical Assumptions H, spend more time on sampling distribution concept

Especially Important Concepts : Thinking about the error term. Properties of estimat ors

In-Class Exercises p. 107 Q2 YES, p. 108 Q4, p. 109 Q6, p. 111 Q10 YES

Homework 3 DUE

October 12—Class Cancelled

The US Presidential Debate (between Bush and Kerry) will be held at ASU on October 13; we are told to expect tight security and horrendous parking. Work on Homework 4.

Readings : Using ‘Trics, Ch. 5

Subject : Hypothesis testing in regression analysis

Especially Important Concept: The standard test of statistical significance—whether the 95% confidence interval around the estimated beta-coefficient overlaps zero or not—is NOT a test of policy importance or theoretical correctness (see Sects. 5.4.1 and 5.4.2). Also, note the importance of sign prediction—often the 2-sided t-test is not the most relevant.

In-class Exercises pp. 146-147 Q1?, p. 147 Q3, Q4 tell what-sided, Q5a(only), p. 148 Q9 (use Marianne's article to show error in use of hypothesis tests? espec re 5.4.2) Q13? Q15?

Homework 4 DUE

Study for Quiz 2 (next week)

Homework 5: p. 148 Q8, pp. 148-149 Q10a, d, e; AND Run a regression on a dataset the professor will give you. Carefully explain why you chose the specific variables you did. Predict signs BEFORE you run the regression and explain the predicted signs and why. Interpret estimated coefficients, t-tests, F-test, and R-squared. ONLY RUN ONE REGRESSION (unless you simply make a mistake). Then, write a 1-2-page policy memo explaining to an appropriate decision-maker what the implications of the analysis are. Assume the actual analysis would be an appendix; make simple and clear meaning of your findings. DUE IN TWO WEEKS.

Handout: Campbell, Gerdes & Steiner (2004) “What's Looks Got to Do With It?”

Discussion Questions: What is the research question? How do the authors decide what variables to include? What is the method of statistical analysis? What statistical issues are addressed? What are the policy implications?

QUIZ 2 --Basic OLS through Hypothesis testing (Test on Part I of Using ‘Trics ); Study Hint: Question 1 of each chapter covers key concepts of each chapter

Professor will be at the Association for Public Policy Analysis (APPAM) conference. No class after Quiz.

Fri., Oct. 29: Restricted Course Withdrawal Deadline

Part II: Functional Form

November 2--Week 10

Readings : Using ‘Trics, Ch. 6; Campbell, Gerdes, & Steiner (2004) (prepare discussion questions)

Subject : Specification issues

Especially Important Point : Don't data-mine! (usually)

In-Class Exercises pp. 182-183 Q3, p. 186 Q7 (but PA instead of Business) YES, p. 189 Q12d YES

Homework 5 DUE

Homework 6: p. 191 Q15

Handout: Brooks (2003) “Taxes, Subsidies, and Listeners Like You: Public Policy and Contributions to Public Radio” PAR , 63(3): 554-559.

Discussion Questions: What is the research question? How does the author decide what variables to include? What is the statistical method of analysis? What are the policy implications? What statistical issues are addressed?

November 9--Week 11

H, you must do a much better job on the functional form stuff; they didn't get it S04; work this out better!!

Readings : Using ‘Trics , Ch. 7, pp. 198-216 (thru Sect. 7.3) and Brooks (2003) handout (discussion questions)

Subject : Choosing a functional form: transforming variables

In-class Exercises p. 225 Q1(but not f) S04 this took a lot of time, p. 226 Q3?, pp. 226-227 Q4?, pp. 230 Q11 but not (d)

Handout: Berry & Feldman (1985), pp. 57 – 72 from Multiple Regression in Practice , Sage: Newbury Park, CA. (4/27/04 this wasn't as helpful as I expected)

Homework 6 DUE

No homework; LOTS of reading

November 16--Week 12

Readings : Using ‘Trics , Ch. 7, entire; Hardy (1993) Regression With Dummy Variables , pp. 1-12 and 18-48 (S04 I think I should cut this off earlier—got hard to understand)

Subject : Choosing a functional form: using dummy variables

In-Class Exercises p. 225 Q1(f), pp. 227-228 Q6 YES, p. 229 Q8?, pp. 232-233 Q14?make this a HW prob?, p. 234 Q15 YES

Homework 7: Write a paper that is approximately 2-5 pages long. Explain the topic on which you plan to do your final research paper, the conceptual model to be estimated, the data you believe you can get to control for the elements of the conceptual model, the predicted signs of the effects, and any functional form issues that you think may be relevant. If you plan to do your Final Project in a two-person group, you should do this paper in the same two-person group.

Part III: Violating the Classical Assumptions—Improving Estimation by Using More Information

November 23--Week 13

Readings : Using ‘Trics , Ch. 8

Subject : Multicollinearity

Especially Important Point: Multicollinearity does not cause bias! It makes it harder to reject the null hypothesis

In-Class Exercises

p. 273, Q3, Q4a-d, pp. 273-274 Q6?, p. 278 Q14, pp. 274-275 Q8(regression), pp. 281-309 Appendix 8.7 FOLLOW DIRECTIONS (will take longer than most) YES

Handout: To be determined S04 used the water paper; it addresses almost every topic in the class—hetero, simultaneity, multicollinearity, etc. It's a bit long, though But, it worked pretty well. H do a PowerPoint presentation making complex simple with strict limit to illustrate?

Homework 7 DUE

Homework 8: pp. 403-407 Hedonic Housing Price Interactive Model. Decide based on theory. Reveal how many regressions you run. Explain your reasoning for each. Don't consider serial correlation, but do consider heteroskedasticity.

November 30--Week 14

Readings : Using ‘Trics , Ch. 10

Subject : Heteroskedasticity

Especially Important Point : Heteroskedasticity does not cause bias in the estimated coefficients; it increases the variance of the estimator (OLS is no longer “best”) and causes bias in the estimated standard errors

In-Class Exercises

(remember some of these may be “impure”) p. 377 Q2a,b?, Q3, pp. 377-378 Q4(regression)?, OR Q5(regression), p. 383 Q11?, pp. 383-384 Q12?

Homework 8 DUE

Homework 9 (last homework): Write approximately one page discussing if heteroskedasiticy and serial correlation are issues for your research paper. Explain why or why not.

Quiz 3 next week (and Final Project Papers due in 9 days)

December 7--Week 15 —H, 3/22 add some hyp testing (mnt to be here?)

QUIZ 3 —Hypothesis Testing, Functional Form, Muticollinearity, & Heteroskedasticity (emphasis on Weeks 10 through 14)

Readings : Using ‘Trics , Ch. 9 and Handout from Week 13 (Nov. 23)

Subject : Serial Correlation (Autocorrelation)

Especially Important Point : Serial Correlation does not cause bias in the estimated coefficients; it increases the variance of the estimator (OLS is no longer “best”) and causes bias in the estimated variances and standard errors

In-Class Exercises

p. 336 Q1 OR pp. 337-8 Q7 NOT BOTH, p. 336 Q2a,b,cYES, p. 337 Q5YES, pp. 379-380 Q9a(?)

Homework 9 DUE H, grade while they take the Quiz and give right back

Final Project Papers due in 2 days

Wed., Dec. 8: Reading Day (according to university regulations, no classes or exams may be held on this day except with agreement by all students)

Final Project Due: Use outline in syllabus. Turn in a physical copy to Wilson Hall. Because Wilson Hall is an office building, it is locked at 5:00 pm (sometimes a bit earlier if staff are eager). Papers will lose 10% per day late and will not be accepted after Dec. 14.

Research Paper Presentations (probably 10 minutes each). Use presentation grading criteria in syllabus.

Please note that the students of the School of Public Affairs have started a journal, Perspectives in Public Affairs . Papers from this class may be suitable for submission to the journal. To learn more about the journal, visit http://asu.edu/mpa/journal.htm .