WikiSummary, the Social Science Summary Database

Kirchgaessner and Schulz: Expected closeness or mobilisation

From WikiSummary, the Free Social Science Summary Database

 

Kirchgaessner and Schulz. 2004. Expected closeness or mobilisation: Why do voters go to the polls? Empirical results for Switzerlan. CESifo Working Paper No. 1387.

In Brief

We have heard it argued that turnout goes up when elections are expected to be close. But K&S argue that this overlooks a key intervening variable: Mobilization. Close contests generate intensive efforts at mobilization (advertising, etc). It is this mobilization that boosts turnout, not the contest's closeness per se.

K&S study 142 Swiss referenda and initiatives from 1981-1999. They show that including mobilization variables renders closeness variables insignificant.

Variables

Several of K&S's operationalizations seem questionable:

  • Without explanation, the authors use a bizarre measure of "expected closeness" of an initiative: -a*ln(a) - (1-a)*ln(1-a), where a = YES/(YES+NO). Huh?
  • "Mobilization" is measured as the number of ads on either side of an initiative taken out in the six largest Swiss papers. But is advertising the same as mobilization?
  • K&S wish to account for how dramatically an initiative/referendum will move policy. To do so, they measure the expected financial consequences of each proposal. But is this remotely valid? The initiatives that had the highest turnout--one to abolish the Swiss Army, another to join the European Economic Area--had primary consequences that were far from financial. (K&S's financial measure would probably make tax proposals appear more dramatic than these two major proposals.)

Case Selection

The authors examine 142 Swiss referenda and initiatives spanning 18 years. These proposals are clustered--several proposals were voted on at any particular election. Clearly high turnout for one proposal will increase turnout for other proposals on the same ballot. K&S use good statistical techniques to control for this clustering, but is their effective N (the number of election days) high enough to produce reliable results?

Findings

The authors find what they expected: Closeness does not affect turnout once mobilization efforts are controlled for.

Comments and Criticism

They never present evidence that closeness actually boosts mobilization, only that closeness is insignificant when mobilization is in the equation. So based on their findings, is it really appropriate to dismiss evidence that closeness boosts turnout?

How do we know that mobilization really increases turnout? Perhaps interesting bills prompt both spending and turnout, creating a spurious correlation between spending and turnout. [[]]