Syllabus – Draft

 

Empirical Implications of Theoretical Models (EITM):

Complexity: Computational Models and Social Networks

 

Week 3 (July 9 – July 13).

 

Scott de Marchi (Duke)

James H. Fowler (UCSD)

 

Summary.

 

EITM, for many social scientists, involves developing a correspondence between game theory (i.e., the theoretical models half of the acronym) and parametric statistical work (i.e., the empirical implications half).  Implicit in this formulation is the idea that game theory, as an encoding, represents rational play, and can accommodate most phenomena of interest.  The challenge is that many game theoretic models lack clear empirical referents; thus, better tools are needed to test the results of these models.  This approach to EITM is certainly useful, but will not by and large be the focus of the second week.

 

Instead, we will be taking the road less traveled.  For us, theoretical models will most often be generated using computational experiments written in a programming language like R, C, or Perl.  For computational and complex systems models, deductive tractability is sacrificed for more verisimilitude, richer models with a greater number of testable implications, and the incorporation of dynamics (e.g., even if there is an equilibrium, how does a population of agents reach it – if ever?).  

 

Computational models are particularly attractive for those who study complex network phenomena.  Scientists are increasingly realizing that the interconnected nature of human social relationships presents special challenges that are frequently intractable in a sparse game-theoretic setting.  At the same time, the increasing availability of large-scale social network data provides an opportunity to study political phenomena at both the micro and macro level.

 

The main goal of the week will be to consider questions that are not normally asked within the confines of the game theoretic tradition, and consider what types of analytic and statistical methods are required to answer them.  Accordingly, much of our time will be spent learning new skills required for computational modeling and social network analysis.

 

First Assignment (due Tuesday, July 9).

 

1)      Before arriving at UCLA, read “A Tournament of Party Decision Rules” (available at http://jhfowler.ucsd.edu/).  This paper describes an Axelrod-style tournament in which players tried to program their own party with a strategy to win the most votes over time in competition with other party strategies, using only information about the location of other parties and their vote totals in previous elections

2)      You will participate in a similar tournament.  Each exercise group should compose and describe in words (250 words maximum) your own party strategy to be submitted against the strategies entered by your fellow participants at EITM.  This tournament will also include five “pre-entered” strategies: Sticker, Predator, Hunter, Aggregator, and KQ-Strat (the latter is the winner from the first tournament described in the paper).  Your goal will be to win as many votes as possible in a medium-length simulation.  The winning group will be announced on Friday and will receive a prize.

3)      Write your strategy in a programming language of your choice (R is the most preferred, but any language is acceptable).

4)      Use game theory to justify your strategy.  Is it possible to simplify this tournament in a way that makes it easy to understand what kinds of strategies tend to win? (write 1-2 pages)

5)      Think of ways to extend this simple tournament and explain how you would address/implement the extensions.  For example, imagine if you were to impose a social network on voters…

 

Assignment Parameters:  we’ll have a special pizza party on Sunday, July 7, before our week officially begins, to cover programming.  You will be asked to do this assignment in groups.  The assignment is due Tuesday morning.  If you crash and burn on any part of the assignment, don’t panic, but do write up an explanation of what went wrong.  We will take time to have each group present their findings / opinions Tuesday morning – so prepare a short statement.

 

Second Assignment (due Friday, July 13).

 

 

1)      Choose one of the computational models presented during the week and alter one or more assumptions in it.  Any proposed change should alter no more than about ten lines of code.  Less is more.

2)      Use game theory (and social network theory if applicable) to anticipate how the proposed change in assumptions will alter the outcome and implications of the model.

3)      Reprogram the model with the new assumptions.  Collect and analyze data from the new model to characterize its behavior using concepts discussed during the week.

4)      Each exercise group will present their model, prior expectations, and some rudimentary findings on Friday.
 

Readings + Themes

 

i.                     Before you come to EITM

Reading

·         Miller, John and Scott Page.  Complex Adaptive Systems.  Princeton, 2007.

·         Ballard, Dana.  Introduction to Natural Computation.  MIT, 1997.

·         Laver, Michael.  “Policy and the dynamics of political competition.” American Political Science Review 99:2 (May 2005): 263-281.

·         Fowler, James H. and Michael Laver.  “A Tournament of Party Decision Rules.”

Programming

·         start fiddling with the programming language R.   If you have an interest in or experience with python, c/c++ or ruby, fiddling around with these will also be helpful

·         be sure to install R on your machine.  For information go to http://cran.r-project.org/.

 

 

ii.                   July 8 (Sunday).  Theme: Basic Programming Skills


6 – 9  Pizza and Programming – bring your laptop! (de Marchi, Fowler)

 

iii.                  July 9 (Monday).  Theme: An Overview of Computational Models.

Readings:

 

·         de Marchi, Scott.  Lifting the Curse of Dimensionality: Computational Modeling in the Social Sciences.  Chapters 1-3.

·         Ken Kollman, John H. Miller, and Scott Page, ``Adaptive Parties in Spatial Elections,'' American Political Science Review 86 (December, 1992): 929-37.

 

9 – 10, Introduction (Fowler)

            10 – 12, Overview of Computational Models (and their place in methods generally, de Marchi)

            1 – 3, All you need to know about non-parametric statistics, equivalence classes, optimization, and other sundry topics (de Marchi)

            3 – 4, Group Work on first assignment

             

 

iv.                 July 10 (Tuesday).  Theme: Computational Models Applied to Elections.

Readings:

 

·         Laver, Michael.  “Policy and the dynamics of political competition.” American Political Science Review 99:2 (May 2005): 263-281.

·         de Marchi, Scott.  Adaptive Models and Electoral Instability. 1999. Journal of Theoretical Politics. 11(3): 393-419.

 

Recommended:

de Marchi, Scott, Michael Ensley, and Michael Tofias.  "District Complexity and Congressional Incumbency Advantage," Working Paper.

 

            9 – 11, Group presentations of first homework assignment

12 – 3, Elections, computational style (de Marchi)

            3 – 4, Group Meetings

 

 

v.                   July 11 (Wednesday).  Theme: Turnout and Social Network Theory.

Readings:

 

·         Fowler, James H. “Habitual Voting and Behavioral Turnout.” Journal of Politics 68 (2): 335-344 (May 2006)

·         Fowler, James H. “Turnout in a Small World.” in Alan Zuckerman, ed., Social Logic of Politics, Temple University Press, 269-287 (2005)

·         Kanthak, Kristin.  “Gender Relations in the U.S. Congress.”  manuscript

 

Recommended:

Bendor, Jonathan, Daniel Diermeier, and Michael Ting. “A Behavioral Model of Turnout.” American Political Science Review 97(2): 261-280 (2003)

S. Boccalettia, V. Latorab, Y. Morenod, M. Chavez, D.-U. Hwang, “Complex networks: Structure and dynamics,” Physics Reports 424 (2006) 175 – 308

Hoff PD, Raftery AE, Handcock MS, “Latent space approaches to social network analysis.”  Journal of the American Statistical Association 97 (460): 1090-1098 (Dec 2002)

 

            9 – 12, Turnout models and an introduction to social networks (Fowler)

            1 – 3, Gender relations in the U.S. Congress (Kanthak)

            3 – 4, Group Meetings

 

 

vi.                 July 12 (Thursday).  Theme: Applied Social Network Theory.

Readings:

·         Fowler, James H. “Connecting the Congress: A Study of Cosponsorship Networks.” Political Analysis 14 (4): 456-487 (Fall 2006)

·         Saiegh, Sebastian M.  “Political Prowess or Lady Luck? Evaluating Chief Executives’ Statutory Policy Making Abilities.”  manuscript

 

                  9 – 12, Applications of networks to social science (Fowler)

            1 – 3, Measuring chief executive abilities (Saiegh)

            3 – 4, Group work on second assignment

 

 

vii.                July 13 (Friday).  Theme: Computational Models of Social Network Phenomena in Political Science


No Reading – student presentations

 

            9 – 11, Group presentations of second homework assignment

            11 – 12, Concluding remarks and discusion (Fowler)

            12 – 1, Concluding remarks and discussion (de Marchi)

            2 – 4, Meetings and personal project time