Rein Taagepera
UC Irvine
Winter 2001
PolSci 232B (65840)
TuTh 9:30-10:50
SSPB 5250
rtaagepe@uci.edu

Seminar and lectures, three hours. Prerequisite: graduate status or instructor's permission. Midterm, seminar presentations, and research paper. Comparative political institutions, especially analysis of electoral and party systems. Emphasis: How to produce, test and use logical quantitative models. The electoral systems are approached as an example of scientific study of politics: interaction between operationalization, measurement and models.

Office hours: SSPB 2225, Tu 11:15-12:15 and whenever I am in.

A. Lijphart, Patterns of Democracy, 1999.
R. Taagepera and M.S. Shugart, Seats and Votes,1989. [It's out of print. Duplicated copies will be supplied.]
Some papers by the instructor.

Lijphart, Electoral Systems and Party Systems, 1994.

Late midterm -- 30% -- open books and notes.
Research paper -- 30%.
Participation and seminar presentations -- 30%.
Whichever is highest -- 10%.

Research paper can be on any comparative institutions. This means it is not restricted to electoral and party systems. About 25 pp., including tables, graphs, etc. Some possible topics will be suggested, but do follow your own interests.

Tentative schedule (it depends on the number of students and their suggestions)

9 Jan. Patterns of Democracy: Westminster and consensus models (Ch. 1-4)

11 Jan. Ignorance-based quantitative models -- J. of Theoretical Politics 1999
15-minute presentations by students on Patterns of Democracy:

16 Jan. Ch. 5: Party systems: effective number of parties --
Ch. 6 and paper by RT: Cabinets: frequency of MW/OP --

18 Jan. Ch. 7: Executive-legislative relations: cabinet life --
Ch. 8: Electoral systems: disproportionality --

23 Jan Ch. 9: Interest groups: IP --
Ch. 10: Federal-unitary contrasts --
Ch. 11: Concentration of legislative power: bicameralism --

25 Jan Ch. 12: Constitutions: amendment and judicial review --
Ch. 13: Central bank independence --
Ch. 14: The two-dimensional map of democracy --

30 Jan Ch. 15: Macro-economic management and control of violence --
Ch. 16: Quality of democracy --
Ch. 17 and paper by RT: Conclusions
Seats and Votes -- discussion led by Rein Taagepera

1 Feb. Ch. 1-4: Description of electoral systems

6 Feb. Ch. 5-9: How to study electoral systems

8 Feb. Ch. 10-13: Measurement of electoral systems

13 Feb. Ch. 14-16: Interconnections between measured variables

15 Feb. Ch. 17-19: Conclusions

20 Feb. Party size baselines imposed by institutional constraints

22 Feb. Seats distribution in international assemblies

27 Feb Midterm test

1 Mar. Student research presentations

6 Mar. Student research presentations

8 Mar. Student research presentations

13 Mar. Research paper deadline.^* Student research presentations.

15 Mar Comments on research papers.
*The instructor will leave for Estonia in late March, returning in September. Hence it's advisable to adhere to the deadline.

• Looking at a map of presidential elections 2000, it would seem that 90% of US counties had a Bush majority. This goes also for states like NY with a marked Gore majority. Is big city vs. the rest the major cleavage in the US? Test by graphing the Bush(+Buchanan?)/Gore(+Nader?) ratio vs. urban % in the county.

• Switzerland: N=5 or N=2? Cf. L99:69-71 on counting closely allied parties.

• More generally, Lijphart's year-by-year data set underlying his 10 variables is available for various replays.

• In correlating with economic and other outputs, L most often deals with joint-power dimension only or with federal-unitary dimension alone. But what happens to the correlations when you average the indices on both dimensions? This needs economic etc. data I do not have.

• Testing seat-vote equation at district level (cf. T&S89:168) -- this removes the problems associated with varying district magnitudes within the same country. But one needs cases with M slightly over 1 (Finland, mean M=14, gives a curve too close to straight line in T&S89:168) -- Japan (I have data!), Ireland, Austria...

The following topics make sense in the framework of "Party Size Baselines..." (RT 2000) to be discussed very late in the quarter, of which the Figure below gives an overview. The broad goal is to test each link in a "complete theory of simple electoral systems".

• Testing p=(MS)^0.25 for all 30 country periods in Mackie & Rose (1991 and 1997) for which district magnitude M can be defined. This means graphing histograms like those in Fig.2 of RT 2000.

• Testing s₁=1/p^0.5 for all elections in Mackie & Rose -- being done by Kaili Roopalu (Tartu U), with median remarkably close to expectation.

• Testing s₁=(MS)^-0.125 for all 30 country periods in Mackie & Rose (1991 and 1997) for which district magnitude M can be defined. Graph histograms of all elections, analogous to Fig. 2 of RT 2000.

• Testing the assumption that the seat-winning party with the least seats has one seat, for all elections in Mackie & Rose (1991 and 1997) where no legal threshold applies. Graph histograms of all elections, analogous to Fig. 2 of RT 2000.

• Graphing s_R vs. R (on semilog), like Figure 3 of RT 2000, and compare to the straight lines

This could involve means for stable systems (as in RT 2000), or one could graph histograms of distributions at each rank, for a single country. Which and how many countries? It depends.