Introduction and Overview of the Course
Part 1: Preliminaries
The Meaning and Measurement of Public Opinion
What is public opinion?
How should it be defined? How should it be measured? Are these separable questions?
What do we mean by Public? Do we mean the mass or the attentive public? Important question, because it gets at the meaning of democracy.
Why did the sociological tradition fade? Have we lost something essential about the meaning of public opinion?
What is the effect of increasing rationalization (Herbst)?
Do you like Key's distinction between latent and active opinion?
Do you think that Converse fairly characterizes Blummer's criticism of the aggregation method?
Should politician care about public opinion? Should they care about opinion polls?
Part 2: Questions of Competence
Political Ignorance: Causes and Effects
Are there limits to the cue-taking strategies discussed by Lupia? What is the role of trust and expertise - we need information to identify relevant cues. The key to Lupia's argument is the mechanism of contextual information. You need to identify the insurance company's positions. Is it realistic to expect that this will happen in most circumstances, or was this a special case? Also, this was a non-random sample. Does that create problems for his inferences?
What do you make of the On-line vs. Memory-Based debate?
Some questions/problems with Zaller. Zaller says that both "cuing" and "persuasive" messages are carried in elite discourse. But he doesn't really draw a clear distinction between them - what is what. Also Zaller talks a lot about messages that are congruent or discordant with people's underlying predispositions. But where do these predispositions come from?
What are the problems created by the unequal dispersion of political information throughout society? Are there solutions?
Some authors argue that it is a mistake to give people "tests" to measure their information levels. Do you agree? If so, what is a more sensible method?
Think back to last week and the definition of the public. Do this week's readings affect how you think we should define the public?
If we have a poorly informed citizenry, what is the starting point for democracy?
The Search for Overarching Principles: Ideology and Political Reasoning
What is "ideology"? What is a "belief system"? What are we looking for? Converse doesn't presume a L/R dimension (he gets a bad rap for that). He's looking for any structure.
How different is Lane's "Morselizing" from Converse's "fragmentation."
Is ambivalence a bad thing? Gets at Lane's contention that ideology is not necessarily a good thing. Do we want people with fixed attitudes? Elite discourse on many issues is slippery and difficult to pin down - draws on different faces of the American political tradition. Is this reflected in people's attitudes?
Pluralist view of politics. How does Converse fit into this? Is he right about issue publics (think back to Delli Carpini and Keeter).
If Achen is right, is this fatal for Converse? What about other forms of evidence - open-ended, cross-issue constraint, mass/elite differences… Converse's argument rests on many legs. Finds that:
People don't use ideological terms in open-ended proves.
People fail recognition tests of ideological terms - 50% recognize, and that's generous.
Constraint is weak - inter-issue correlation is relatively low; and higher among elites.
Over time instability; this is the piece that Achen deals with - issue answers seem to be random; therefore can't be driven by fragmented (but coherent) belief systems.
Let's look at Achen closely. What's his argument? Work through the math. Key test is at end when he tries to model response variance. Says that doesn't follow patterns of political sophistication as we would expect if Converse is right. Everyone is variant, therefore, problem is with the question. But concerns: measures (Converse's criticism of stacking the deck against findings), missing data problem (pp. 1225; knocks out 1/3 of the sample who don't express an opinion; seems like he's stacking the deck in his favor).
What do you make of Lane's technique? The citizens seem sensible and somewhat coherent. How much of that is real, how much is reaction to method.
Do we want people to be "ideological" Lane sees it as a bad thing - rigidity of thought. He would rather see people as flexible thinkers.
Part 3: Building From the Ground Up: Opinion Ingredients and The Micro-Foundations of Public Preferences
Material Interests (Or Symbolic Politics?)
How should we define self-interest? (Sears - Narrow, short-term; Downs - Broad, long-term: also, what counts as an interest? What is self? You alone, family, group?) How should we measure self-interest? Are these linked? An important question because the studies of self-interest find that it (generally doesn't matter). Is that because it really doesn't matter, or because we have poorly conceptualized and measured the concept? (Central questions: what is self? What is an interest? Subjective vs. objective).
When does self-interest matter?
- Sears and Citrin study of prop 13 - matters when costs are clear and made salient by a campaign.
- Matters when look at behavior/action rather than opinion (Green and Cowden).
Why doesn't self interest matter?
What do you think of the alternative - symbolic politics? What are the problems with that? Is symbolic politics just a culmination of long-term interests? (Problems with Panel data used by Sears et al).
What role does elite discourse play here. Maybe we need elites to point us to our interests - Prop 13 case.
Make sure not to confuse self-interest axiom and rational axiom. In the first, we're talking about what goes into a utility function; in the second, we're talking about the process of utility maximization. These concepts are separable.
How unusual are the cases we saw in this week's readings where self-interest plays a role in shaping participation patterns? Busing and Social Security might be different - what general lessons can we draw?
Groups in American Politics: Us and Them
What does the "racial divide" mean? Is a dummy on race enough to capture differences?
How does group identification fit with previous work? The authors of the American Voter seem to take a somewhat dim view of group identification (as does Converse '64). Is group identification an adequate basis for political reasoning?
What is interesting about the literature on groups to you? What should the next steps be? Too focused on race? Do we want to continue with an atomistic approach to studying groups?
What does group membership mean? How should we measure it? Ingroup vs. outgroup.
Principles and Values
Is there a "national creed?" If so, how can we have variance in opinion? If everyone holds the same values, how can we arrive at different ends?
Let's talk about methods. Here we have a number of different methodological cuts into the question of values: Quant Analysis (Feldman), Quant Analysis of Codes (Feldman and Zaller), Open-ended questions (Chong). What are the strengths and weaknesses of these different approaches? Think about Feldman - I ask you - vs. Luker and Feldman and Zaller - you tell me. What can they singularly and together tell us about public opinion?
More about methods: problems with defining what values are. What's equality? (728 types). This becomes even more difficult as we move from conceptualizing to operationalizing. It's hard to define values, but it's harder to measure them. Bottom line: from the get go, there are a lot of problems with the concept of values.
What is the line between ideology and values? Why does one seem to work and the other not? What is the distinction? How much of the results are methods-driven? (May be ideology comes in when values clash - tells you how to reconcile competing arguments).
Talk about Feldman's methods. What do you make of his measurement strategy? But, he finds big coefficients. What about that?
Question of causal direction on values. Do you adopt the values of the people you like (such as Reagan) or does the arrow go the other way?
在1988 （Feldman）至1992 （Feldman and Zaller）年間，「平等」有什麼改變？1988年，平等是大事，但到了1992年，人們不怎樣談到平等，反而更多談到個人主義（這在1988年是寂寂無聞的）。
What happened to equality between 1988 (Feldman) and 1992 (Feldman and Zaller)? In 1988, equality is the big thing. But in 1992, people don't say much about equality. Instead, they talk about individualism (which had no pop in 1988).
How are links made between values and policies? The authors don't spend much time on this. Gets to issues of framing and elite discourse that we'll take up in the next section of the course.
What happened to value ordering? Value conflict - see some of this in the Alverez and Brehm piece when they talk about value conflict. But most work on values has proceeded in a somewhat piecemeal fashion - look at one value (or maybe two) at a time; don't look at values in relation to each other.
Part 4: The World Out There: Society and Politics
How do the different authors conceive of norms - what is a norm? How should we define norms? Are they useful constructs?
What do you think of Mutz's tests of the different mechanisms? She chooses issue areas and political contexts to vary the level of engagement with politics (same thing that I do), but she doesn't look at inter-individual variation as much.
What does it mean for someone to express different attitudes in surveys depending on the context? Is there a particular context that allows us to assess "true attitudes"? Or are different expressions just different? Gets back to the Sanders reading we did that first week.
What do you make of the Mendelberg and Oliver findings? Are you convinced? Is this the best way to do contextual studies? Compare their strategies to those of Huckfeld and Sprague?
Does the work this week address the Blumer concerns about context? Or is there still something missing?
How should we measure the effects of context? What are the appropriate methods? Measuring aggregate indicators of context (Mendelberg and Oliver), measuring the opinions of discussion partners (Huckfeld and Sprague), or inferential methods (Berinsky)? What are the strengths and weaknesses of these different approaches?
How does Mutz fit with the bell-curve of political information. Nothing sinister happens in the middle (there it's just a check). But how big is that middle - like Lupia; the middle is okay, but is almost everyone in the middle? Normative implications?
Do Huckfeld and Sprague answer the questions you think are interesting? What else could they have done?
Aggregate vs. individual-level analysis. Potential problems (talk about Goodman's problem/EI).
What matters most: personal influence or impersonal influence?
Framing, Priming and Media Effects
How should we study media effects? Cross-sectional secondary analysis (I+K), experiments (Iyengar and Kinder, Miller and Krosnick Nelson, Berinsky+Kinder), interviews (Gillens)? What are the advantages and disadvantages of the different approaches? Also, this is our first brush with experimental research - what do you make of it as a research tool?
Should we be concerned about the existence of framing effects? What does it say about the meaning of public opinion and question of democratic worthiness - in many ways we're back to Converse, but with an Achen-like twist. Changing the presentation of political alternatives can move opinion.
What do you make of the Miller and Krosnick and the Nelson et al. findings on accessibility. Do you find the research convincing? What does this say about our Zaller-centric view of public opinion? If these authors are right, it seems that people have more agency over their opinions than we give them credit for.
What about the role of "understanding." Most authors deal with the end-state of the opinion formation process - where do you stand on a given issue? Berinsky and Kinder are trying to deal also with the process of political understanding. What do you think of that?
The Miracle of Aggregation and Enlightened Preferences (Is Anyone Right?)
Problems in applying Condorcet's theorem. Page and Shapiro argue that "the simple process of adding together or averaging many individuals' survey responses . . . tends to cancel out the distorting effects of random errors in the measurement of individuals' opinions." But is this the case. As Larry notes. As Bartels notes, "if sources of error affect the entire electorate (or a significant fraction of the entire electorate) in similar ways, the resulting errors will simply not cancel out, no matter how large the electorate may be."
Other problems in applying Condorcet's result. Condorcet assumes that individuals seek a common goal. In the case of juries, this goal is the search for the truth about a case. However in a heterogeneous society, there may not be a comparable goal. Even if individuals had perfect information about the impact of particular policies, for example, fundamental disagreements regarding their wants, needs, and desires might lead them to opposing answers. Also question of independence - Bryan gets at this. Condorcet's result requires that individuals make their choices independently. I the real world, however, individuals share common information and may communicate with each other. Thus, the assumption of independent action may be unrealistic. This flaw may not, in practice, prove fatal. Ladha (1992) has extended Condorcet's result to the case with correlated votes.
Converse argues that aggregation works its magic by riding on the backs of the well-informed. So he would say that it's the people with "real opinions" (as opposed to non-attitudes) who drive Page and Shapiro's analysis of the question - in effect, Page and Shapiro find the same thing as Converse. Is this a reassuring result? What do we make of Page and Shapiro in light of this? Aggregate stability comes from a stable central tendency. Is this measurement error canceling out, or is it the stability of the most informed driving result?
What is the proper level to examine public opinion. Individual level? Aggregate? Something in-between? Would look at different background characteristics in both cases - over-time versus cross-sectional. Think about Converse. Rationality comes in net shifts. What happens if groups of people move in radically different directions? The change will get washed out in the aggregate. So we're missing important trends there. Sub-aggregate trends can get washed out as noise, even though they are not. Maybe we should be looking at aggregates of groups.
What is the right way to look at the aggregate? Should we just make the leap up? Seems like it's better to start with the individual, then aggregate up. Need a better connecting story. That's the problem with a lot of this work - doesn't start with a theory of the individual - better to build up.
Democratic deliberation. Page and Shapiro talk about the role played by discussion and division of labor in society. What evidence do they present (none)? Is this a good idea?
What do you make of the strategy proposed by Bartels and Athaus? Are their problems (non-random assignment of information; potential problems in measurement brought on by the Mondak work,…). What about the Gilens' strategy? Does that solve the problem. What do you think about the enterprise of looking for "enlightened preferences"? Is this something that we should be doing?
Public Opinion and Public Policy
Public Opinion and Foreign Policy
Part 5: Reflections
Final Thoughts: The Place of Public Opinion in Democratic Politics
What is the best of the research we read this semester?
Where should we go next as a field?
Are the unexplored/under-explored avenues? Are there dead horses we should stop beating?