Harvard Kennedy School

API-205

Deb Hughes Hallett Fall 2010

API-205

POLITICS AND POLICIES: WHAT CAN STATISTICS TELL US? Deborah Hughes Hallett Office: Littauer 102, phone: 617-495-9640 [email protected] Course Meetings: Tuesday and Thursday 10:10-11:30 pm in Littauer 140 Review Session: Friday 1:10 pm-2:30 pm in Starr Teaching Fellow Ellen Whitesides ([email protected]) Course Assistants Ronald Abraham ([email protected]) Tim Kotin ([email protected]) Jorge Mattar ([email protected]) Sileshi Woldeyohannes ([email protected]) Overview Intended for decision makers, the goal of this course is to enable you to read and interpret statistical information. We emphasize the use of statistical techniques to evaluate policies and programs. The course will discuss how to use data effectively to advocate and negotiate. Discussions will be based on case studies from international development, politics, education, health care, and the legal system. You will often be asked to explain a result in non-technical language, suitable for a news article or a parliamentary briefing. Course Meetings Lectures will focus on the development of concepts, using examples, case studies and discussion. Your active participation is warmly welcomed! Review sessions and office hours are designed to give you the opportunity to ask questions, get started on assignments, and check your understanding. We look forward to working with you. Textbook: Two Options for the Main Textbook The problems in both books are excellent. Thus the solutions manual may be a useful addition to the text. Statistical Reasoning for Everyday Life, 3rd edition, by Jeffrey Bennett, William Briggs, Mario Triola. (Addison Wesley, 2009, ISBN #0-321-28672-3.) Student Solutions Manual for Statistical Reasoning by Bennett, Briggs, Triola (Addison Wesley, 2009, ISBN #0-320128706-1.) The text and solutions manual are in the Coop and the KSG Library. You can also use the 2nd edition (Text: ISBN #0-201-77128-4, solutions: ISBN# 0-201-83846-X). Or Essential Statistics, by David. S. Moore (Freeman, 2010, ISBN# 1-4292-3369-9 or 1-4292-3446-6) Study Guide with Selected Solutions for Essential Statistics, by M. Fligner and R.S. Linder (Freeman, 2010, ISBN# 1-4292-3394-X) The text and solutions manual are at the Coop. 1

Harvard Kennedy School

API-205

Deb Hughes Hallett Fall 2010

Optional Additional Texts Statistics: Concepts and Controversies, 7th edn, by David Moore and William Notx (Freeman, 2010) Seeing Through Statistics, 3rd edn, by Jessica Utts (Thompson, 2005) How will we use the texts? Texts provide background and an additional approach to the topics. Reading assignments will be given from both books; students can choose which to read. The problems in each book provide additional practice; the solution manuals will allow you to check your solutions. The weekly assignments (for credit) will not be from the books. What is the difference between the two main texts? Are both necessary? It is not necessary to get both main texts. (Lecture notes will be available for each class.) The difference between the texts is that Statistical Reasoning is more basic. It has fewer formulas and is more talkative. The end part of API-205 is not in Statistical Reasoning—though by then you may not find a book to be necessary. Essential Statistics goes further—but may be harder going at the start. How to decide what is the right book to get? Look at both books at the Coop before choosing. Second hand versions of Statistical Reasoning are available on line, but probably not of Essential Statistics. However, Essential Statistics is less expensive than many texts. Case Studies Additional reading and data sets will be on the course website. These include: Election fraud: Allegations and evidence in Iran, US, Colombia, Venezuela Exit polls and polling regulation around the world African Governance Index Education of Afro-Caribbean boys in UK Set-asides of university places in India Legal cases in which statistics played a decisive role: Howland, Collins, Clark Civilian deaths in Iraq: The Lancet studies Human Development Index World Bank development indicators and World Bank governance indicators Labor market discrimination in the US New Haven firefighters; NCAA football coaches Roadside stops and racial profiling Madrassa enrollment in Pakistan Cross-country relationships between growth, wealth, and happiness Medical testing policy: Malaria and HIV; mammograms The death penalty Measles, mumps, rubella (MMR) vaccine controversy Swine flu: Spread and fatalities Do women judges vote differently? Sonia Sotomayor’s Supreme Court nomination

2

Harvard Kennedy School

API-205

Deb Hughes Hallett Fall 2010

Prerequisites The mathematical background required will be kept to a minimum, though students will be expected to be familiar with algebra and able to handle data with a calculator and Excel. (An inexpensive calculator is all that is needed.) The most important prerequisite is the willingness to ask questions and to grapple with ideas in a policy context. Assignments and Exams There will be weekly eight assignments to be handed in (three or four problems each); two midterms (Tuesday October 5 and Tuesday November 2) and a final exam on the afternoon of Wednesday December 8. The assignments can be worked in groups, but must be written up individually. They are due in class on Thursdays, except for the last one, and available the week before. There will be an optional extra assignment showing an application of the course material to a professional context of the student’s choice. You can decide much later in the semester whether you want to do it. Practice problems whose solutions are available will be assigned for most classes. These problems are not to be handed in, but are excellent practice for the midterm and final exams, so you can be confident of your preparation. Grading Weekly Assignments (eight) Midterm exams (two) Final exam Optional Extra Assignment

20% in total 20% each 40%. 10% Replaces half of one midterm or one quarter of the final if this gives a higher grade.

The lowest two weekly assignment scores will be dropped. If the final exam shows significant improvement over the midterms, it will be counted more heavily. If the second midterm is stronger than the final, it will be counted more heavily. Letter grades are determined in accordance with the HKS recommended grade distribution: http://www.hks.harvard.edu/degrees/registrar/faculty/exams-and-grading/grades

3

Harvard Kennedy School

API-205

Deb Hughes Hallett Fall 2010

Tentative Schedule: Date

Lecture Topic Reading: SR is Statistical Reasoning ES is Essential Statistics

Assignments

Week 1 Tues Aug 31

Shopping Election Fraud: Polls, Probabilities, and Statistical Evidence Case Discussions: Elections in Iran, US, Colombia, Afghanistan, and Venezuela

Thurs Sept 2

Lecture #1 Election Fraud: What Statistical Signals Are There? Histograms and Distributions. One Variable Numerical Descriptors: Mean, Median, Standard Deviation, Outliers Examples: Iranian Election; US income and EU incarceration rates; identifying risky investments; African Governance Indicators Reading: SR text: Sections 1.1, 3.1, 3.2 (Section 3.3 optional) 4.1, 4.2, 4.3 ES text: Chapter 1, p. 1-16, Chapter 2, p.29-32, p. 37-39. Background Information on Election Fraud: Iran “About Those Iran Polls” by Jon Cohen, Washington Post, June 15, 2009 “Signs of Fraud Abound, But Not Hard Evidence” Glenn Kessler and Jon Cohen , Washington Post, June 16, 2009 “New Analysis Points to Fraud in Iran” Jon Cohen, Washington Post, June 18, 2009 “Claims of Vote-Rigging in Iran Backed by British Academics’ Analysis”, Martin Fletcher, TimesOnLine, June 22, 2009 Venezuela “Black Swans, Conspiracy Theories, and the Quixotic Search for Fraud in Venezuela” www.venezuelanalysis.com Sept 20th 2004, by Mark Weisbrot, David Rosnick, and Todd Tucker - CEPR “Analysis of Voting Data from the Recent Venezuela Referendum”, Edward Felton, Aviel Rubin, Adam Stubblefield, Sept 1, 2004 4

Harvard Kennedy School

API-205

Deb Hughes Hallett Fall 2010

US Presidential Election 2004 “Those faulty exit polls were sabotage”, Dick Morris, The Hill, Nov 4, 2004 “A call to investigate the 2004 election” Steve Freeman and Joel Bleifuss, The Boston Globe, June 26, 2006 Week 2 Mon Sept 6 Tues Sept 7

Labor Day: Holiday Lecture #2 Normal Distributions Example: Heights and SAT scores; Body Mass Index (BMI); investments and loans; Baby weights: Indonesia and Paraguay Reading: SR Text: Sections 5.1, 5.2 ES Text: Chapter 3

Thurs Sept 9

Lecture #3 Assignments Sampling and Data Gathering: Random, Stratified, and Cluster Samples. #1 available Sampling frame; Internal and External Validity. Examples: Calling an Election (1948), Current Population Survey, Household surveys, MMR vaccine Case Study: Madrassa enrollment in Pakistan; Civilian deaths in Iraq Reading: SR Text: Section 1.1, 1.2. ES Text: Chapter 7 “The Human Cost of War” by Les Roberts. The Lancet Medical Journal “Religious School Enrollment in Pakistan: A Look at the Data” December 2005, by T.Andrabi, J. Das, A. Khwaja, T..Zajonc

Week 3 Tues Sept 14

Lecture #4 Observational Studies vs. Experiments. Experimental Design: Placebos, Controls, Hawthorne Effect, Confounding. Randomization and Natural Experiments. Ethical Issues in Experimentation. Cases and Examples: Policies to Encourage Marriage, Hormone Replacement Therapy, Physicians’ Health Study, Video-games and Disease, Identification of Cholera Source, Tuskegee Syphilis Study, Tearoom Trade, Tanzania HIV Study

5

Harvard Kennedy School

API-205

Deb Hughes Hallett Fall 2010

Reading: SR Text: Sections 1.3, 1.4 ES Text: Chapter 8 Sources: “America's Toe-Tapping Menace” by Laura MacDonald, New York Times, Sept 2, 2007 “Statistical Models and Shoe Leather”, by David A. Freedman “No Link Between MMR Vaccine and Autism” Chris Guillot, N.C. Journal of Law and Technology, Feb 18, 2009 “Estimating the Impact of the Hajj: Religion and Tolerance in Islam’s Global Gathering” D. Clingingsmith, A. Khwaja,. M. Kremer, 2008 Thurs Sept 16

Lecture #5 Probability, Properties and Measurements. p-values and Significance Levels Examples: Vitamin A Study. Coaching for SAT Scores Reading: SR Text: Sections 6.1, 6.2, (to end p.244), 6.3 (p.251, p.254-256) ES Text: Chapter 9

Week 4 Tues Sept 21

Lecture #6 Independence and Conditional Probability Policy Questions: Should Afro-Caribbean boys in UK be educated separately? Should Set-Asides in Indian Colleges be Expanded? Reading: SR Text: 3rd edn: Section 6.5 (to mid p.270), Section 4.4 (mid p. 179182) 2nd edn: Section 10.2 p. 419 (mammograms) to p. 421. ES Text: Chapter 11, p. 199-207 Sources: “Bad attitudes; Race and education” Economist, Mar 10, 2005 “Tackling low educational achievement” by Robert Cassen and Geeta Kingdon, Joseph Rowntree Foundation, 2007 “Under-Representation of Disadvantaged Classes in Colleges: What Do the Data Tell Us?” by Rana Hasan and Aashish Mehta. Economic and Political Weekly, Sept 2, 2006, 3791–3796

6

Assignments #1 due #2 available

Harvard Kennedy School

Thurs Sept 23

API-205

Deb Hughes Hallett Fall 2010

Lecture #7 with Arash Nekoi, Economics Dept, FAS Iranian Election: Probabilistic Evidence

Assignments #2 due #3 available

Case Discussion: 2009 Election in Iran: Analyzing the Claims of Fraud http://online.wsj.com/article/SB124523854750623001.html (Interactive Graphics tab) “The Devil Is in the Digits” By Bernd Beber and Alexandra Scacco, Washington Post, June 20, 2009 Week 5 Tues Sept 28

Lecture #8 Sampling Distribution of Proportions, Central Limit Theorem. Data types. Case Study: Venezuela’s 2004 referendum on the recall of Hugo Chavez. New Haven firefighters; head coaches for US college football “Who You Know & Who Knows You: The Hiring Process & Practice of NCAA FBS & FCS Head Coaching Positions” Keith Harrison, Black Coaches and Administrators, 2008 Reading: SR Text: Section 8.1 ES Text: Chapter 18: p. 327-329

Thurs Sept 30

Lecture #9 Sampling Distribution of Means, Central Limit Theorem. Data types.

Assignments # 3 due

Example: Repeated sampling; 17.5 lb Russian baby and siblings Reading: SR Text: Section 5.3, 8.1 ES Text: Chapter 10 Week 6 Tues Oct 5 Thurs Oct 7

Midterm Exam Lecture #10 Confidence Intervals: Means. Margin of Error and Sample Size. Example: Radar detectors and speeding Case Discussion: Civilian Deaths in Iraq: Iraq Body Count and two Lancet studies Reading: SR Text: Section 8.2 ES Text: Chapter 13, p. 231-241, Chapter 14, p. 257-264

7

Assignments #4 available

Harvard Kennedy School

API-205

Deb Hughes Hallett Fall 2010

“The Human Cost of War” by Les Roberts. The Lancet Medical Journal Week 7 Tues Oct 12

Lecture #11 Confidence Intervals: Proportions. Margin of Error and Sample Size. Examples: Presidential approval ratings; Mexican election data; Kissing study Reading: SR Text: Section 8.3 ES Text: Chapter 18, p. 330-334

Thurs Oct 14

Lecture #12 Hypothesis Testing: Means: One Sample. Practical and Statistical Significance

Assignments #4 due #5 available

Example: Malaria and drug testing; slide in SAT scores Reading: SR Text: Sections 9.1–9.2 ES Text: Chapter 13, p. 238-249 Chapter 16, p. 285-295 Week 8 Tues Oct 19

Lecture #13 Hypothesis Testing: Means: Two samples Examples: Do poets die young? Do women talk more than men? Case Discussion: Did the Washington Consensus work? Comparison of growth rates of developing countries; Rule of Law and development Reading: ES Text: Chapter 17 “Goodbye Washington Consensus, Hello Washington Confusion?”, by Dani Rodrik (2006) pp. 1-2

Thurs Oct 21

Lecture #14 Hypothesis Testing: Proportions: One Sample Examples: Hispanics and Jews on Juries; Army suicide rates; testing a HIV vaccine Case Study: Madrassa Enrollment in Pakistan Reading: Text: Section 9.3 “Pakistan’s Madrassas, Extremism and the Military,” Crisis Group, July 29, 2002/July 15, 2005 “Religious School Enrollment in Pakistan: A Look at the Data” 8

Assignments #5 due #6 available

Harvard Kennedy School

API-205

Deb Hughes Hallett Fall 2010

December 2005, by T.Andrabi, J. Das, A. Khwaja, T..Zajonc Week 9 Tues Oct 26

Lecture #15 Hypothesis Testing: Proportions: Two samples Case Studies: LAPD and roadside searches; Labor market discrimination; Miller-El vs. Cockrell Example: Swine Flu fatalities Reading: Handout “Driving While Black in the City of Angels” by Lily Khadjavi, in Chance, 2006 “Are Emily and Greg More Employable than Lakisha and Jamal? A Field Experiment on Labor Market Discrimination” by Marianne Bertrand and Sendil Mullainaithan, American Economic Review, 2004

Thurs Oct 28

Lecture #16 The Search for Causation

Assignments #6 due

Policy Question: Should We Have the Death Penalty? Reading: SR Text: Section 7.4 “Does Death Penalty Save Lives?” NY Times, Nov 18, 2007 (plus links) “The State of the Death Penalty: Evaluating Fairness and Accuracy in State Death Penalty Systems” American Bar Association, Oct 2007 Week 10 Tues Nov 2 Thurs Nov 4

Midterm Exam Lecture #17 Two variables: Scatter Plots, Association, Correlation and Causation: Confounding Examples: Human Development Index: Life expectancy and GDP; Pornography and violent crime, Voice pitch and reproductive success Examples: Analysis of news headlines Case Study: Social cohesion between Jews and Arabs in Haifa, Israel Reading: SR Text: Section 7.1, 7.2 ES Text: Chapter 4 “Positive prejudice: Really loving your neighbour” Economist, 2007 9

Assignments #7 available

Harvard Kennedy School

API-205

Deb Hughes Hallett Fall 2010

“Allophilia” by Todd Pittinsky, in Compass, 2005 Week 11 Tues Nov 9

Lecture #18 Regression: Least squares. Making Predictions Examples: Olympic records, Age of first marriage; World Bank Governance Indicators, Inequality (Gini Index) and Economic Growth; Relationship Between Health, Wealth and Happiness (Cross-country) Reading: SR Text: Section 7.3 ES Text: Chapter 5 “Are Health, Wealth and Happiness Linked Worldwide?” by Melissa Lafsky in New York Times, Aug 31, 2007 “Income, Aging, Health and Wellbeing Around the World: Evidence from the Gallup World Poll”, by Angus Deaton, NBER working paper

Thurs Nov 11

Veterans’ Day: No Class

Week 12 Tues Nov 16

Lecture #19 Bivariate Regression and Hypothesis Testing Examples: Cigarettes and lung cancer death Case Study: The 2002 Alabama gubernatorial election Reading: ES Text: Chapter 22 “Strange Case of an Imprisoned Alabama Governor” New York Times, Sept 10, 2007 “A Statistical Analysis of Possible Electronic Ballot Box Stuffing; The Case of Baldwin Country Alabama Governor’s Race in 2002”, by James Gundlach

Thurs Nov 18

Lecture #20 Multiple Regression and Hypothesis Testing Examples: Cross country comparison of Central Bank independence and inflation; Voice pitch and reproductive success

Week 13 Tues Nov 23 May be rescheduled

Lecture #21 Reading a Regression Table: Coefficients and Significance Examples: Death Penalty; Climate change 10

Assignments #7 due #8 available

Harvard Kennedy School

API-205

Deb Hughes Hallett Fall 2010

Case Discussions: Impacting Teachers’ Performance “Teacher Performance Pay: Experimental Evidence from India” K. Murualidharan, V. Sundararaman, NBER Working Paper, Sept 2009 http://www.econ.ucsd.edu/~kamurali/teacher%20performance%20pay.pdf

Thurs Nov 25

Thanksgiving Vacation

Week 14 Tues Nov 30

Lecture #22 Reading a Regression Table: Interpretation of Coefficients; Dummy Variables Case Discussions: Are women less corrupt than men? What is the impact of women judges? “The Waves Minority Judges Always Make” New York Times, May 30, 2009 Sources: “Female Judges Matter: Gender and Collegial DecisionMaking in the Federal Appellate Courts” by Jennifer Peresie, http://yalelawjournal.org/114/7/1759_jennifer_l_peresie.html “Gender and Corruption” by Anand Swamy,, Stephen Knack, Young Lee, and Omar Azfar, Journal of Development Economics, Volume 64, Issue 1, February 2001, pp. 25-55

Thurs Dec 2

Lecture #23 Review and Summary: Connections Between the Big Ideas

Wed Dec 8 2- 5 pm

Final Exam

11

Assignments # 8 due