Hug and Sciarini: Referendums on European integration
Disclaimer. Don't rely on these old notes in lieu of reading the literature, but they can jog your memory. As a grad student long ago, my peers and I collaborated to write and exchange summaries of political science research. I posted them to a wiki-style website. "Wikisum" is now dead but archived here. I cannot vouch for these notes' accuracy, nor can I say who wrote them.
Hug and Sciarini. 2000. Referendums on European integration: Do institutions matter in the voter's decision?. Comparative Political Studies 33:3-36.
Question: What is the effect of referendum institutions on voter behavior?
Hypothesis 1: Government supporters vote more strongly in favor of a government proposal if the referendum is nonrequired than if it is required.
Hypothesis 2: Government supporters vote more strongly in favor of a government proposal if the referendum is binding than if it is nonbinding.
Hypothesis 3: The effect of the binding/nonbinding distinction on the role of partisanship is stronger than the effect of the require/nonrequired distinction.
Assumptions: The gov't is supporting the referendum proposal, and is not suffering from serious internal decisions. Also, only looking at government supporters. (Mostly because dealing with the opposition is too difficult to code/gather data).
The 14 Votes: 4 in Denmark, 3 in Ireland, 2 in Norway, 1 in Great Britain, 1 in France, 1 in Switzerland, 1 in Austria, 1 in Finland, 1 in Sweden. The votes were taken in the 1970s, 1980s, and 1990s. The were voting on the EC membership, voting on the Single European Act, first and second votes on the Maastricht treaty, voting on EU membership, voting on the EEA treaty.
Data on Voting: from surveys, to find out if voted for or against their (gov't) party; and for or against referendum.
Methods: Percentages of voters for or against (for broad claims). Regression analysis.
Results that support actually studying this:
- Partisanship seems to matter (people do vote with their preferred party).
- But, the loyalty of government voters can't be taken for granted (vary from 48 to 96% when clear party cues given).
- Other studies used the outliers when making claims about government and opposition capacities to affect vote/outcome/unity.
- (Table 3) the degree to which parties are supported varies as much for government parties as opposition parties, when position is controlled for.
- Broadly, gov't voters are more likely to support treaties than others.
Results for their hypotheses (formal definitions):
- Gov't support is higher in required referenda (p=.74 vs p=.63). (does not support h1).
- Gov't support is higher in binding referenda (p=.75 vs. p=.59). (supports h2 nicely).).
- Gov't support is higher in nonrequired/binding votes (p=.85), then binding/required (p=.73), then nonbinding/nonrequired (p=.59). (support h3, h2 and h1).
Results for their hypotheses (political definitionsÃ¯Â¿Â½sometimes unavoidable when not required):
- Results are weaker, but differ only marginally from other tests.
- Binding plays the most role (p=.71 vs. p=.50). (supports h2)
- BUT, gov't support is higher if binding/required (p=.72), but not statistically distinguishable from binding/nonrequired (p=.69); nonrequired/nonbinding (p=.50). (supports h2, h3, not h1)
- "voters react quite differently in varying institutional contexts to partisan cues provided by their government." p. 26.
- When using formal definitions, all 3 hypotheses corroborated. When using political definitions, h1 is not corroborated. "This suggests that gov't voters take the formal characteristics more strongly into account than political ones when deciding how to vote. This suggests that voters see through the strategies of political parties."
- Eliminating the 4 votes where the gov't did not support the theory (against the assumption they have in the model), the results hold up.
- The 14 votes are not necessarily independent. For example, two of Denmark's votes were on the same thing: the first failed, the second passed.
- There is no control for historical period or type of vote.
- Voter data is underexplained. Different surveys for different countries (can we compare across time and space?). Are people answering surveys accurately? Is it ok to be "anti-Europe" in a poll? How about extrapolating people's responses to unasked questions from several questions that were asked?
- They use 4 votes where the gov't doesn't support (or at least didn't negotiate the treaty; but they assume that they do).
- I believe that they have found a hole in the literature. Their logic is great. Their hypotheses make intuitive sense. I want to believe in their conclusions. Alas, I'm not sure their data is good enough to support their conclusions. Alas, I'm not sure their models are good enough to support their conclusions. Their broad interpretive conclusions are a bit overreaching. Poor Hug and Sciarini.