Background: Chapel Hill Expert Survey

Some context: parties, families and cleavages

Lipset and Rokkan: parties form around cleavages, and conflicts are “frozen” based on when democratization occurs

“Party Families” emerge around cleavages, and act as vehicles for their interests (i.e. socialist parties address class cleavage, agrarian parties represent rural interests etc.)
- note: this is a multi-dimensional view of party positions!
But what do they do with new issues that don’t have a basis in a deep cleavage? (such as EU integration)
Are parties really “frozen”?

Based on semi-annual surveys of area experts
Initially covered European Politics, now covers multiple regions (albeit in separate data sets)
Website

Primary goal: measure party positioning (initially on the EU, but subsequently across a broad range of issues)
Secondary goal: evaluating expert consensus as a measurement

Examples of other quantitative methods for measuring party positions include:

Scaling roll-call votes (NOMINATE)
Analysis of manifestos (Comparative Manifestos Project)
Voter surveys of party positioning (European Election Survey) and elite self-placement (EUandI)

Party-Year level data (one row per party per wave) with the average position, salience, amount of dissent and “blur”
Expert-Party-Year data (one row per expert-party-wave) with the same metrics

Experts contribute a significant but substantively small amount of variation to party placement
- In other words: experts disagree, but not so much that we would conclude that they lack a shared understanding of party positions.

Traditional party families held stable positions on the EU by the 90s
“New” families (Rad. Right and Greens) developed more cohesive positions over time

Immigration issues, rather than left/right positions, have become the strongest correlate of EU position taking

The “GAL-TAN” dimension has emerged as a new political cleavage that structures party positioning

(note: this isn’t really related to the CHES per se, but it’s relevant for studying political parties in general)

Are there really any true “two-party” systems? How should we count the number of parties in a given country?
The simplest method, just counting the number of parties on the ballot or in the legislature might not make much sense! (quick: how many parties were on the 2024 presidential ballot? How many mattered?)

Laakso and Taagapera (1979): weight parties by their vote share (electoral parties or ENEP) or seat share (parliamentary parties or ENEP)

\[ \text{ENP} = \frac{1}{\sum_{i=1}^{n}p_i^2} \]

(N is the Effective Number. P is the vote/seat share)

In R: if x is a vector of party seat shares:

50/50 split:

x <- c(.5,.5)

1 / sum(x^2)

[1] 2

Two major parties and one small one:

x <- c(.5,.4, .1)

1 / sum(x^2)

[1] 2.380952

The UK in 2024:

seats<-c(398, 120, 72, 15, 9, 6, 5, 4, 4, 4, 2,  7 , 1, 1, 1, 1)

seat_shares<- seats/sum(seats)

1/ sum(seat_shares^2)

[1] 2.367477

The CHES data excludes some small parties, ENPP will be slightly deflated from the “real” answer
For ease of comparison, most data sources (including CHES) only look at the lower house in bicameral systems

We’ll use the latest round of the Chapel Hill Expert Survey