Wednesday, October 28, 2009

Deep partisanship

I spend a good bit of my academic life thinking about partisanship. What does it mean to be a partisan? How do partisans differ from other partisans and from non-partisans? What are the behavioural consequences of partisanship?

Our understanding of partisanship has changed over time. Beginning in the 1960s, academics understood partisanship to be a deep attachment to a party formed early in life which subsequently acted as a perceptual screen on the political world. Think of it as similar to religious affiliation. Individuals may stray from their familial religious affiliations, but for the most part these act as an anchor throughout the lifecycle and serve to influence how we experience and perceive the world.

This understanding of partisanship was later challenged by a 'running tally' perception in which partisanship was taken to be an active evaluation of the parties on offer, where an individual 'updated' their partisanship as new events unfolded and changed it as their affection for one party increased over another.

I think it's fair to say that the first view has better stood the test of time, both inside American and outside. Principal in the defense of this view is a great book by Green, Schickler, and Palmquist called Partisan Hearts and Minds.

My own work has examined the behavioural and material foundations of partisanship in Canada. For example, I've shown that revealed material concern for the well-being of other partisans explains much of the decision to vote in Canada. I've also shown that material concern for others varies with our own and others partisanship. Finally, I am working on a larger (though still very preliminary project) on whether behavioural differences characterize different partisans. My own contribution is very small.

The most important contribution in recent years, and this is the point of this post, has just been made by Alan Gerber and Gregory Huber in this paper. Here's the story: for a long time, we've had survey evidence that partisans have more positive economic expectations when their preferred party is in power in Washington. In other words, Republican partisans say they expect the economy to perform better when there is a Republican president than when there is a Democratic president. The same applies (in reverse) for Democrats. However, this could merely be an artefact of surveys. If partisanship really matters in a deep way, then what is needed is evidence that partisans behave differently when their preferred party is in office. Gerber and Huber provide evidence of this. They demonstrate that changes in the rate of expenditures at the county level following an election correlate with the partisanship of the county. So, more Democratic counties would increase their rate of spending more than Republican counties (or more accurately, decrease it less quickly, as spending is generally lower in the winter than in the fall) following the election of a Democratic President. The same applies for Republicans.

This is an extremely important finding as it shows that partisanship has deep behavioural consequences. It is not simply a tally of one's preferences, but instead an affiliation which influences how one approaches not only the political but also the commercial world.


Anonymous said...

So how does this fit with the 1993 Canadian election?

And more importantly, how do you square these findings with recent literature that finds that value bundling along partisan lines doesn't exist among individuals in Canada and western european countries?

Peter Loewen said...

Here's a question: how do you square findings from a survey with behaviourally generated findings?

Anonymous said...

True - no argument there.

I guess I'm just wondering about partisanship. I've always been a fan of it, especially the partisanship model of absent mandate. I guess your post put some things in perspective - specifically the idea that partisanship is kinda of a funny thing. It's almost tautological, isn't it? Liberals vote liberal. Of course!

But if you look at the value bundles, you don't get the same results, right?

Peter Loewen said...

See, I don't think it's a tautology in that it's true that Liberals generally vote Liberal, but not all who vote Liberal are liberal.

I think partisanship is a deep, affective attachment to a party which affects how one views the party, views other parties, views other partisans, and views the performance of things the government ostensibly controls, like the economy.

Now, the fact that people may not align their values with those of their party does not undermine partisanship, I think. First, I think it says that this attachment is prior to values, much like the way a person's religious identity can be very real while not actually squaring with what they believe (think of Catholics who support gay marriage). Second, political or ideological values are pretty second or third order things in my books. Most people just don't have all that many in a real, tangible sense, so it's not suprising that there would be a lot of noise in mapping back from partisanship to values (or even policy positions, a la Stephenson et al's latest and very interesting work).

Anonymous said...

Interesting. I see where you are coming from. I guess I'm being influenced by this literature that suggests that value bundling matters, and that the way individuals bundle their values doesn't match up with their supposed partisan affiliations.

I think it's clear from the literature that for the most part, left-wingers tend to bundle their values coherently. But the same is not true of the right - the classic divide is social conservatives are not always fiscal conservatives.

If that's true, then it seems to me that this is a particular strong challenge to partisanship.

Does partisanship come before values? I'm not so sure. I don't know this literature as well as you and others do, but it seems to me that partisanship may exist alongside values, just as values may come before partisanship or vice versa.

Peter Loewen said...

If it's reasonable for social and fiscal conservatives to both support conservative parties, then why would we take the nonbundling of issues as evidence of the weakness of partisanship? It just says that the party does not perfectly represent either group, but it like represents them better than any other party. I don't see how this is a challenge to partisanship. What say you, masked inquisitor?

Anonymous said...

But isn't that the point? The social conservatives who are not fiscal conservatives, as well as fiscal conservatives who are no social conservatives, don't in general support the Conservative party at the federal or provincial level.

I also wonder about the interaction between partisanship and value bundling (which comes first, the chicken or the egg). Also, I would hazard an uneducated guess that although partisanship may form before coherent values, that this characterization changes over time, with partisanship weakening and value bundling getting stronger in terms of influencing choice.

Or maybe this is your point? That partisanship remains a powerful voice in the back of our head even as values come to the surface as we grow older?

Peter Loewen said...

Goren has the most convincing work on the chicken-egg, I do think:

As for your claim about voting rates, I am not sure the evidence supports it, but I am willing to be persuaded.

Anonymous said...

Thanks for the link. I'll be sure to have a read.

On the second point, I'll have to defer to you since it's not my area. But it seems to me that this would be a simple (and easy) test of this whole values vs. partisanship debate. Does value bundling matter for partisanship? If so, fiscal conservatives who are not social conservatives and social conservatives who are not fiscal conservatives would consistently vote PC and Conservative.

Anonymous said...

Actually, it looks like I should read the Goren piece first! Thanks.

LFC said...

What's the chain of causation or causal mechanism (or whatever you want to call it) for the Gerber/Huber results? Democrats feel generally happier about life when a Democrat wins, and 'happiness' connects to willingness to spend more freely? Or do they not get into the whole question?

Peter Loewen said...

No, I think it relates more to economic perceptions. If a Dem wins, then Democrats increase their evaluations of the future health of the economy and increase their consumption in that time period accordingly.

Anonymous said...

Isn't the main problem with this discussion, though, the false dichotomy? It does not defy values-based arguments to suggest that partisanship can have an effect on some of some people's opinions (as Campbell et al found) and behavior (Gerber and Huber), any more than it does to suggest that religion affects the opinions of some people about some issues.

Partisanship matters. No question.

But partisanship isn't the only influence on people's opinions, nor is it a wholly independent variable. There is nothing about being a Democrat that makes a person 11 percentage points more likely to be a woman. There is nothing about being a Republican that makes somebody twice as likely to be white.

Moreover, the American case is exceptional for its combination of a high number of partisans and, especially, for its stable two-party system. Why are we constantly citing the American case, when it is probably the most exceptional country in the Western world in terms of its stable two-party landscape?

Non-durable partisans, (i.e., the vast majority of the population) have beliefs about political issues. Where did those beliefs come from? If we acknowledge (as I suspect we all do) that there are non-partisan sources of opinion, then why would we assume that partisanship is "prior to values" for durable partisans on all issues? We probably wouldn't. This brings me back to the main point: I think the debate here a false dichotomy.

PS: "Second, political or ideological values are pretty second or third order things in my books. Most people just don't have all that many in a real, tangible sense,"

I agree with the observation, but not the inference. I think that this conflates the number of political/ideological values that people hold with the importance of the political/ideological values that they hold.

Peter Loewen said...

Well, I think much of the point of the new partisanship research is to show that it matters in more ways than we ever thought, and does so in a lot more countries than the US. I think this is the case, and I don't think the US is _that_ exceptional, though is probably a debate over degrees rather than kind.

As for the dichotomy, I am not sure which one you're talking about. But if the question is whether partisanship matters or not in a meaningful way, independent of a lot of other factors, then I think the evidence suggests that it does.

Anonymous said...

Hi Peter: I agree. The false dichotomy being that "either" partisanship matters "or" values matter. Partisanship matters, but it is not causally prior to values, even if it is causally prior to some values for some people.

The behavioral genetics literature, that we both find quite interesting, is suggestive. Hatemi et al. (2006) find that genetics, rather than environment, shapes the intensity of PID. They also find that environment, rather than genetics, shapes the party to which that PID is directed. Yet, Alford, Funk and Hibbing (2005) find stronger genetic than environmental correlates of opinion. Something doesn't add up here with the "hard" partisanship model: shared genetics is not correlated with the party to which people belong, but it is correlated with the political opinions that people hold. At the very least, this suggests, as AFH say, that we need to think "more broadly" about the origins of opinions, by which I think they almost certainly mean rethinking the partisanship account.

Anonymous said...

"Here's a question: how do you square findings from a survey with behaviourally generated findings?"

Here's another question: how do you square claims about individual-level behaviour with aggregate-level?

I think the most important thing about Gerber and Huber's piece is that it clearly demonstrates the ecological fallacy isn't a concern for reviewers at APSR anymore. Good to see we're making progress.

Anonymous said...

3:26 and 4:48 here: I thought of the same thing, but then asked: "how else could they have done it?"

If there are individual-level data on spending patterns throughout the US, then fine. They should have used those. But if not, as I suspect, then what?

It's an honest question.

The partisanship theory has aggregate-level implications, those implications are observable, and so they tested them.

Personally, I had no problem with the level of analysis.

Anonymous said...

Nov. 5, 11:58 here.

"See, I don't think it's a tautology in that it's true that Liberals generally vote Liberal, but not all who vote Liberal are liberal."

I'm still not quite convinced that partisanship isn't a tautology, especially the second part of your sentence which says that "not all who vote Liberal are liberal."

What does this sentence mean? By small "l" liberal, do you mean liberal in terms of values? Or are you saying non-partisans or partisans from other political parties vote for the Liberal party?

I had a discussion with a colleague of mine on some of these issues and he pointed out a very important issue - If these dynamics we are speaking about is a function of the parties (especially brokerage parties!) doing a poor job of aggregating/expressing/matching individual partisan interests, then how can partisanship come before values?

If the existing political parties that we have don't fit/match with individual liberal partisanship, then how can individual liberal partisanship exist?

Instead, I think values have to be first (and most important). From these values, citizens may develop partisanship as an information shortcut to make decisions during elections. But when it comes down to it, citizens have a number of core values that trump partisanship. It's why non-partisans or partisans from other parties vote liberal. It's also why new political parties emerge and why partisans push for new political parties (like the Reform; Socreds; Progressive; even the the new Wild Rose Alliance in Alberta). The old parties don't fit comfortably with the core values of some partisans, and so they develop new parties.

Hence, what really matters is values.

Anonymous said...

8:58am Here: Again, though, I would reiterate that I don't think there is anything in the observation that values shape partisanship which precludes the possibility that partisanship shapes the opinions of some people on some issues.

Evangelicals probably aren't going to change their opinions about abortion if the Republican party does, but they might follow their party on opinions about, say, immigration. Hardcore racists probably aren't going to change their opinions about immigration if the Republican party does, but they might follow their party on opinions about, say, abortion.

As Brody, Sniderman and Tetlock warn us: "the mistake [is] to suppose that the analytic problem [is] how the characterize the political reasoning of the public as a whole". I like the Goren piece, but I certainly see it from this perspective.