6

Political Voice across Issues

In the previous chapter we observed evidence consistent with what happens when political rhetoric is self-undermining. The people most sensitive to health care–related insecurity—those without health insurance or without the income to cushion a health-related economic shock—were the least likely to engage in political action that sought to reduce health care costs. A similar pattern emerged among people who were personally facing the burden of education expenses.

This pattern did not, however, signal a broader unwillingness to act politically. It did not suggest a broader lack of trust in the political system or an especially low level of political efficacy. The people who bristled at the idea of spending time and money on health care costs were (on average) perfectly motivated to act on other important issues that did not pertain to insecurity they were facing in their personal lives. In other words, the root of political inaction was uniquely linked to the self-undermining quality of rhetoric on economic insecurity issues.

Although the experimental results pointed in a clear and consistent direction, the cost and practicality of conducting each one necessitated certain limitations. Because of budget constraints, it was not possible to conduct an experiment for every single economic insecurity issue. Collectively, the experiments covered education costs and health care costs. But they did not touch on the other economic insecurity issues discussed in this book so far. Moreover, practical considerations meant that the time-based act (joining a listserv to spend time reading about the issue and opportunities to volunteer) involved only a subset of all possible time-based activities through which people might express political voice. It did not cover attending meetings, nor did it directly cover the act of volunteering. Practical considerations also dictated that the experiments could only be conducted over a restricted time period. It was obviously not possible to go back in time and see how people might have responded to similar experiments in the 1980s, 1990s, or earlier in the 2000s.

So, although the experiments permit clear causal conclusions, one might wish to see evidence across a broader range of activities and time. The purpose of this chapter is to address these concerns. My question is: If we look back over the past thirty years, have the people who consider insecurity issues to be most important (who also, based on the results in Chapter 2, are highly likely to be personally facing whichever insecurity issue they consider most important) also been less likely to spend resources on politics than those who consider other issues to be most important? Have they been less likely to donate money to political organizations? And, if they are in the labor force, have they been less likely to volunteer as well? Moreover, do these differences remain even after we take into account other differences between the types of people who prioritize economic insecurity issues (who tend to be facing them in their own lives) versus those who consider other issues to be most important?

INTRODUCTION TO THE DATA

As in Chapter 2, here I rely upon survey data from the American National Election Study (ANES). For over six decades the ANES surveyed a nationally representative sample immediately prior to and immediately after Election Day during presidential years. The data are particularly useful here because they include a measure of issue importance as well as several questions about respondents’ political behavior. Recall from Chapter 2 that my measure of issue importance is a question that asks respondents to identify the most important problem facing the country. I argued there why, despite the fact that the question asks about “problems” rather than “issues,” I believe it is reasonable to interpret responses to this question as an indicator of issue importance (with the implied understanding that “issue importance” in this case is shorthand for “an important issue that is perceived as the most important problem”). The suite of behavior questions includes precisely the types of money- and time-based activities that individuals most often consider taking, including monetary donations to campaigns, parties, and interest groups and time-intensive activities such as attending political meetings, rallies, or speeches and volunteering for a political organization during the campaign.1

My analysis uses ANES data from all presidential years from 1988 to 2008, excluding 2004 when the most important problem question was not asked. As I discussed in Chapter 2, this time period is advantageous because it includes comparable data on the key variables of interest and because it parallels the time period over which people consistently expressed concerns about insecurity issues.² With unemployment as a notable exception, very few respondents labeled insecurity issues as the most important problem facing the country prior to the 1980s. Looking at 1988–2008 means focusing on a time period in which families’ objective situation became riskier and in which the political footprint, at least in terms of issue importance, was ever more present.

The ANES’s most important problem question read as follows: “What do you think are the most important problems facing this country?” If respondents named more than one problem they were then asked to say which is the single most important problem that the country faces. The most important problem question has a number of advantages that I described in detail in Chapter 2. In brief, it asks respondents to identify the most important issue in their own words—what should be prioritized at that moment—and not what the survey designers thought was the most important. It also explicitly forces trade-offs, thereby preventing a situation in which people say that every issue is important.³ And because respondents are able to place only one issue at the top of the agenda, it is arguably a highly conservative measure of political importance—there are no doubt plenty of unmentioned issues that people find to be important, but only one can receive top billing.

I use data on Americans’ perceptions of the most important problem facing the country in five presidential elections—1988, 1992, 1996, 2000, and 2008—that vary in terms of both the broader economic circumstances as well as the partisan identification of the incumbent president. For the analysis I first separated people into theoretically meaningful groups based on their issue priority. Doing so was a two-step process. First, I classified their top issue as being related to the economy or not. This was relatively straightforward, as most issues that are not explicitly about the economy (such as foreign policy and social issues) are fairly easy to identify. Second, I further divided the economic issues into those related to household economic insecurity issues versus those that are not. Making this distinction was slightly trickier. The easy part was separating out issues related to the four main economic insecurity concerns at the heart of this book: unemployment, health care costs, retirement, and college costs. Everything else would be included in an “other economic” category. Most of the people in the latter category were chiefly concerned about government budgetary matters and the proper role of government intervention in the economy.4 The harder part was interpreting people’s responses when they reported that “the economy” was the most important problem facing the country. This was particularly a concern in 2008 as respondents were interviewed in the fall just as the news was filling with stories about financial firms collapsing, insurance giants on the brink of insolvency, and record one-day drops in the stock market. In their daily lives, Americans were confronting the increased possibility (and reality) of losing their jobs or having their hours cut. Not surprisingly, against this backdrop was an increasingly bearish outlook on both the nation’s economy and their personal finances.⁵ People were likely cutting back on expenditures in part because they had actually experienced worsened finances but even more because they worried about things getting worse in the near future. When the ANES asked them to report the most important problem facing the country, almost half (45%) simply encapsulated their number-one concern as “the economy.”6 This sentiment might reflect direct, personal experiences with having lost one’s job, having hours cut, losing health insurance, having premiums rise, losing a pension, and/or sending a child off to college. Or they might reflect concerns about the health of the broader economy, fellow citizens, and/or future generations. Or concerns about “the economy” might reflect all of the above. Because the answer was not obvious, I adopted a conservative strategy. All of the analyses in these chapters were conducted first by not including “the economy” as an economic insecurity issue and then second by including it as one. In almost all cases, I present results from both analyses just to demonstrate robustness. In a few cases, for the sake of brevity I display only the results assuming that “the economy” does not count as an economic insecurity issue (and note the results under the alternative assumption in a note).

This classification resulted in three issue-importance groups: economic insecurity, other economic, and noneconomic. A full listing of each appears in the Table 6.1. The economic insecurity column includes responses from people who directly mentioned the four major economic insecurity issues at the heart of this book (jobs/unemployment, health care costs, Social Security/Medicare, and education costs), as well as a few other issues that likely go hand in hand with concerns about income/wages: high prices, debt, mortgage costs, housing, and child-care costs.⁷ The majority of people mentioned one of the first four. Of the latter set, only housing was mentioned by more than 1% of respondents.⁸ Similarly, the “other economic” group was dominated by a handful of concerns including the government budget, the deficit, and taxes.⁹ The noneconomic category is a catchall for everything else, including a diverse array of issues related to foreign policy, social policy, crime, governmental functioning, and terrorism. Again, the wide differences among the content of the noneconomic issues is less important here than the fact that they are all not about the economy and, in particular, not about financial security issues.

To get an overall sense of the prevalence of each of the three groups, I plot the proportion of people mentioning an issue in each of the three categories. Figure 6.1 presents results assuming that general mentions of the economy are not considered economic insecurity issues, whereas Figure 6.2 presents results assuming that they are. With the exception of 2008 (and to a lesser extent 1992), there are not exceptionally large differences depending upon how general mentions of the economy are coded. One of the most striking results is the lack of lopsided priorities. In each year, regardless of the health of the broader economy, a sizable proportion of respondents mentioned issues in each of the three categories. It is worth underscoring that point as it relates to economic insecurity: even in the “go-go” years of 1996 and 2000, approximately 30–40% of respondents still labeled some aspect of economic insecurity as the most important problem facing the country. This provides further evidence that insecurity is not simply a reflection of economic growth or decline.¹⁰ People lose their jobs, get sick, retire, and pay high tuition even when the broader economy is humming along.

TABLE 6.1
Issue importance categories

Economic Insecurity	Other Economic	Noneconomic
Jobs/Unemployment	The Economy	Honesty/Integrity/Morals
Health Care Costs	Recession	Environment
Social Security	Bailouts	War/Troops/Military
Medicare	Big/Small Businesses	Foreign Policy
High Prices (e.g., gas/oil/food)	Income Inequality	The President/Congress/Officials
Income/Wages	Banking	Terrorism
Household Debt	Federal Reserve	Political Awareness
Education Costs	Socialism	Civil Liberties
Mortgage Costs	Gov’t Budget	Cooperation/Gov’t Functioning
Middle-Class	Gov’t Deficit	Agriculture Policy
Econ Concerns		(Int’l Concerns)
Labor Union	Banking Crisis	Int’l Image
Concerns
Housing	Taxes	Partisanship
Child Care	Welfare/Poverty	Social Justice
Costs	Programs
	General Domestic	Church and State
	Economic Concerns
	Population	Immigration
	AIDS Epidemic	(Noneconomic Aspects)
		Energy
		Muslims
		Homosexuality/Gay Rights
		Energy Policy
		Crime/Drugs
		One’s Opponents
		Homelessness
		Abortion/Women’s Rights
		Discrimination/Equality
		Work Ethic
		Labor Union Tactics
		(Noneconomic)
		Self Interest/Greed/Moral Decay
		Law and Order
		Young People’s Attitudes
		Civil Rights/Racial Issues

Note: See p. 164 about classification of “the economy.”

Figure 6.1. Proportion of respondents who reported that these issues were the most important. (Source: ANES, 1988, 1992, 1996, 2000, 2008.) This assumes that general mentions of the economy are not coded as an economic insecurity issue.

Along these same lines, recall that throughout the book I’ve stressed how economic insecurity is not confined to people living at or near the poverty line. Concern about it (in terms of issue importance) is not either. Over three-quarters (78%) of the people who placed jobs, health care costs, retirement security, or education costs at the top of the political agenda were not in the very bottom (or very top) of the income distribution but instead situated between the 17th and 95th percentiles. This was perfectly comparable to the percentage of people in the other two issue groups—78% of those in the noneconomic group and 81% of those in the “other economic” group.

Figure 6.2. Proportion of respondents who reported that these issues were the most important. (Source: ANES, 1988, 1992, 1996, 2000, 2008.) This assumes that general mentions of the economy are coded as an economic insecurity issue.

MOTIVATIONAL CONSEQUENCES?

The goal of this chapter is to see if people who consider economic insecurity issues to be most important are less active in politics than those who label other issues as most important. This goal presumes that, in principle, people are motivated to act on the issue they consider to be most important. Is that a reasonable assumption? Recall again that the ANES asks a nationally oriented question—which issue is the most important problem facing the country—as opposed to a personally oriented one—which issue is most important to you personally.11 This is potentially problematic as the two are not measures of the same underlying attitude. When citizens report that an issue is personally important, they are thought to be revealing a latent willingness to act on that issue. Relative to others who consider an issue to be less personally important, they are more likely to donate money to and volunteer with organizations that are addressing that issue.¹² This is what led Penny Visser and her colleagues to conclude that “[issue] importance may be best characterized as a motivational variable—one that propels an individual to protect, use, and express an attitude by acting in accordance with it.”¹³

The question at stake here is whether people are just as behaviorally engaged when they find an issue nationally important. To see if that’s the case, Joanne Miller and her colleagues recently completed the first systematic comparison of national and personal importance. One of their major findings was that they are not two sides of the same coin. Instead, they appear to have distinct political footprints, with personal issue concerns motivating action far more frequently than national-level assessments. They concluded that “the roots of policy issue engagement for democratic citizens appear to be in personal issue assessments.”¹⁴

Based on their finding, some readers might question whether it makes sense to compare people’s level of activity based on what issue is most important nationally to them. This is a fair question and my decision to do so requires a bit of justification. One response is to empirically look for overlap between the two. Perhaps the people who consider an economic insecurity issue to be nationally important are also more likely to consider one to be personally important. Miller and her colleagues found some evidence of overlap on the issues they focused on, though it was by no means perfect. But given that I am looking at different time spans and surveys conducted at different points throughout the year, not to mention different issues, it is possible that the degree of overlap is greater.

To investigate this, I turned to the 2008 ANES. Unlike earlier surveys, this one asked people to report both the most important problem facing the country and the most important problem to them personally. And, as Table 6.2 shows, there was indeed an association between them. The people who mentioned some facet of economic insecurity as the most important national problem were by far the most likely to also mention some facet as the most important problem to them personally. The overlap is not perfect, but it does indicate a clear relationship.

TABLE 6.2
Relationship between most important problem facing country and most important issue to the respondent personally (2008)

Most important problem facing the country	% of respondents saying that some facet of economic insecurity is the most important issue to them personally
	Classifying “the economy” not as economic insecurity	Classifying “the economy” as economic insecurity
Economic Insecurity	28	51
Other Economic	8	32
Noneconomic	14	40

Moreover, there is actually good reason to believe that the results in Table 6.2 understate the potential overlap in people’s personal and national importance judgments. In the 2008 ANES the personal importance questions appeared immediately prior to the national importance questions. This aspect of the survey instrument potentially depressed the likelihood that people would answer with the same or even similar issues to both questions. The reason stems from the fact that people tend to apply the logic of conversations when answering surveys.¹⁵ Among other things this logic dictates that people do not repeat the same question multiple times and, to the extent that back-to-back questions might seem similar, it signals that the survey interviewer expects distinct responses (otherwise, the reasoning goes, why would the interviewer ask it twice?). Thus the fact that the personal and national importance questions were back-to-back might have indicated to respondents that they should answer the national one differently than the personal one. To be sure, this is speculative on my part—without an experiment that randomly assigns the order in which people received both questions there is no way to know for certain if people interpreted the situation this way. But past work on the psychology of survey responses suggests that it is a very real possibility.

Even if readers remain leery about the association between national level and personal judgments across my three categories, there is another response to potential concerns about the motivational influence of national-level judgments. Miller and her colleagues’ primary conclusion was that personal importance motivates political action more than national importance does. If my goal was only to trace the relationship between national concerns about economic insecurity and political action on those issues, then I would face the same problem that they uncovered. But that is not my aim. Instead, my goal is strictly comparative in nature—only to compare the likelihood of action across people with three different, theoretically relevant types of most important issues. For this comparative perspective, there is no reason to expect that this measurement-induced confound would not apply equally across all three categories. It still permits uncovering behavioral differences depending upon whether there is something unique about concern related to economic insecurity issues.

POLITICAL ACTION AND ECONOMIC INSECURITY CONCERNS

With those preliminaries in mind, the next step is to directly address this chapter’s key question: Over the past thirty years, have people most concerned about economic insecurity issues been less active in politics than those with other issue priorities? As in previous chapters, my analysis unfolds in a multistage process.

I begin by presenting simple bivariate comparisons focusing on people’s issue concerns, political behavior, and (where appropriate) labor force status.16 I present these first because they lie at the heart of my enterprise. At the same time, if I presented only bivariate results, several concerns would arise. We know that there are several other systematic factors that affect people’s willingness to get involved with politics, and there are several reasons why I might want to account for them. The bivariate comparisons might, for example, reveal large differences between the three issue-importance categories. But this relationship could simply be the result of a confound—something else that affects both people’s likelihood of getting involved and their issue priorities. Perhaps the people concerned about them simply have lower income (and thus less money for politics) or more children (and thus less free time for politics). Or they are disproportionately likely to identify as Democrats and perhaps have tuned out of national politics in recent decades. Democratic presidential candidates since Walter Mondale have stressed deficit reduction over major programs to restore economic security to American families, and so it could be that Democrats have been turned off by that and thus less willing to devote time and money to national-level politics. An altogether different problem with presenting just bivariate relationships arises if they do not provide strong evidence of a difference based on issue priorities. Here we might wish to control for other factors that help explain variation in people’s political behavior to achieve a greater ability to discern any effects.

Figure 6.3. Issue importance and donation behavior (Source: ANES). Figure shows likelihood of donating money based on which issue respondent considers most important (90% confidence intervals shown).

With those considerations in mind, Figure 6.3 displays the results for spending money and Figure 6.4 does so for time.¹⁷ The donation results present a clear pattern: people who considered an economic insecurity issue to be most important were less likely to donate money than anyone else. Between 1988 and 2008, 9% of them donated money to a campaign, party, or interest group, which was less than the 11.5% who considered a noneconomic issue to be most important (z = −4.35, p < .01). Those who considered another economic issue to be most important donated 13.5% of the time, which was statistically equivalent to those in the noneconomic group (z = 1.56, p = .12).

Figure 6.4. Issue importance and time expenditures. Figure shows likelihood of spending time on politics based on which issue respondent considers most important and also labor force status (90% confidence intervals shown).

The bivariate results for time are not as striking. Figure 6.4 displays the proportion of people who attended political meetings or volunteered with campaigns or other political groups, first for all respondents and then broken down by whether or not they are in the labor force. The proportions for all respondents are almost identical across all three groups: 9% for those prioritizing economic insecurity, 10.3% for those prioritizing another economic issue, and 9.5% for those prioritizing a noneconomic issue (p > .40 in all comparisons). But theoretically it is important to see if there are any differences depending upon one’s labor force status. I expect only those who care the most about economic insecurity and are working or looking for work to be less willing to spend time on politics. However, when I examine the results based on these factors, I continue to observe no statistically significant differences. Among those in the labor force, 8.9% of those who labeled an economic insecurity issue as the most important problem donated time to politics, as compared with 10.5% and 10% in the other two groups, respectively (again, p > .40 in all comparisons). A similar pattern occurs among those not in the labor force.¹⁸

ELABORATING UPON THE BIVARIATE RESULTS

The bivariate results are suggestive, particularly with respect to monetary donations, but we also want to take into account other factors that might be related to people’s issue priorities and likelihood of getting involved in politics. In the spirit of Achen’s argument for close analysis of theoretically relevant subgroups, I begin to do so by closely considering one of these alternatives.19 Afterward I will present results from a larger multivariate model that more directly reflects previous work in the study of political behavior.

This alternative explanation has to do with the wider political environment at the time these surveys were conducted. Perhaps the reason that people who considered economic insecurity issues most important donated less was simply because they were not presented with credible opportunities for political change. For example, if you were a Democrat in 2000, it is quite possible that you heard Al Gore choosing to spend a lot of time talking about how to improve the quality of the U.S. educational system.²⁰ If you were a Republican in 2000, you probably heard George W. Bush also talking about the quality of the education system and the educational reforms he enacted as governor of Texas. You probably also heard him presenting himself as a new kind of conservative—a compassionate conservative—who wished to help find common ground and be “a uniter, not a divider.” In either case, if you were part of the almost 40% of people who placed an economic insecurity at the top of the agenda during this time, it is quite possible that campaign rhetoric simply did not resonate with your primary issue concerns.

Contrast that with the situation in 1992 or 2008. Concern about economic insecurity issues in both of these years largely revolved around jobs and the cost of health care. Both were also key aspects of the campaign rhetoric, especially on the Democratic side. Consider Bill Clinton’s bid to win the White House in 1992. While the plurality of his advertising and stump speeches focused on the state of the economy and the job market, the next highest percentage of both mentioned domestic issues. Here, health care was front and center.²¹ Clinton was particularly interested in burnishing his credentials as a “New Democrat,” which entailed ambitious new plans to expand health care to the uninsured and make it more affordable without raising taxes.

Fast-forward to the 2008 campaign in which health care insecurity was once again a key campaign issue, especially on the Democratic side. During the nomination contest, a plurality (22%) of all Democratic ads that contained specific policy information focused on either health care or education. A much smaller percentage of ads were focused on leadership, experience, or even gas prices, and only 5% of Republican ads focused on either health care or education.22 In fact, by the time Obama accepted the Democratic nomination, he had decided to make health care a major issue in his general election campaign. And by the end of October, his campaign had spent $113 million on health care television advertising, with health care–related topics comprising over two-thirds of the total amount spent on television ads.²³ He had also signed on to the principles put forth by Health Care for American Now, which placed the affordability of health care front and center.

Thus it seems reasonable to suppose that, at least in 1992 and 2008, people who prioritized an economic insecurity issue would have also been relatively likely to perceive an opportunity for meaningful political change. If that’s the case, then we would like to see if people concerned about these issues were still less likely to donate in these two years. Let me underscore that although I am choosing to single out these two presidential election years, many Americans also expressed concerns about these issues during the other presidential election years that occurred closer to peaks in the business cycle (see Figures 6.1 and 6.2). My focus on the trough years is only meant to potentially indicate the presence of campaign rhetoric, which could indicate a heightened perception of political opportunity for change.

The top portions of Tables 6.3–6.5 contain these comparisons for people’s willingness to donate money, volunteer time if they are in the labor force, and volunteer time if they are not in the labor force. The cell entries are the differences between people based on which issues they consider most important, using the noneconomic group as my baseline.²⁴ The statistically significant differences are marked as indicated and largely reinforce the bivariate results presented earlier.

In 1992 and 2008, people who prioritized economic insecurity were 3 percentage points less likely to donate money than those who placed a noneconomic issue at the top of the agenda. In an electorate of over 230 million voting-age Americans in 2008, this would have translated to 6.9 million people.²⁵ There was no similar decrease in the other years examined or among people who placed another economic issue at the top of the political agenda.

TABLE 6.3
Differences in proportion of people donating money, based on issue most important to them

	1992 & 2008	All Other Years
Economic Insecurity vs. Noneconomic	−0.03*	−0.02
Other Economic vs. Noneconomic	0.01	0.04**

Note: † p < .10    ∗ p < .05    ∗∗ p < .01 (two-tailed tests)

TABLE 6.4
Differences in proportion of people volunteering time among those in the labor force, based on issue most important to them

	1992 & 2008	All Other Years
Economic Insecurity vs. Noneconomic	−0.01	−0.01
Other Economic vs. Noneconomic	−0.01	0.02

TABLE 6.5
Differences in proportion of people volunteering time among those not in the labor force, based on issue most important to them

	1992 & 2008	All Other Years
Economic Insecurity vs. Noneconomic	−0.01	0.03
Other Economic vs. Noneconomic	0.01	0.02

Lastly, focusing on the volunteering time results in Tables 6.4 and 6.5, we again fail to observe evidence of a statistically significant difference across issue groups in people’s willingness to volunteer, regardless of whether they were in the labor force. This lack of evidence parallels the earlier bivariate results, suggesting that perhaps there is nothing exceptional about these two campaigns in this regard.

To recap where we are: Over the past thirty years, many people have labelled job insecurity, health care costs, retirement insecurity, and college costs as very important issues. These people have also systematically been less active in politics, at least when it comes to donating money to campaigns, parties, and interest groups. At this point I do not have evidence of a parallel decrease in the realm of time expenditures. Part of that was expected. I did not expect any difference for those outside the labor force, even if they are personally facing insecurity. But I did expect a different pattern among people who are currently in the labor force (people working full-time, part-time, or looking for work). All of these people are likely to be reminded of a temporal constraint if they are also reminded of a financial constraint. Hence I expected to observe that, compared with people in the labor force who prioritize these other issues, they would have been less willing to spend time on politics. One possibility is that my expectation is simply not right, which would indicate a conflicting pattern of behavior relative to the experiments from the previous chapter. Yet another possibility is that the bivariate results presented so far are incomplete because they do not take into account other individual-level attributes that might be related to both people’s issue priorities and their political participation.

With those concerns in mind, I next turn to a series of broader multivariate models that take account of a much larger set of factors that also affect people’s political behavior. The set of factors is based on the model constructed by Steven Rosenstone and John Mark Hansen in their foundational book on political behavior, including: objective resources such as income and education; evaluations of parties and candidates; involvement in the community including whether one owns a home and how many years have been spent living there; whether one was contacted by a political organization during the campaign; one’s perceptions of political efficacy; and other demographic characteristics.26

To investigate whether there are behavioral differences based on people’s issue priorities, even after controlling for these other individual-level factors, I also include two dummy variables: one that indicates whether or not the person stated that an economic insecurity issue was the most important problem facing the country and another indicating whether or not the person stated that some other economic issue was most important. The missing category therefore includes noneconomic issues, and thus all of the main results are stated relative to people who stated a noneconomic issue was most important. Lastly, I add two other types of variables. One indicates whether respondents expressed a view that one party was better able to handle the most important issue. The other is a set of dummy variables for each year included in the analysis, which accounts for the fact that campaigns differ in the extent to which they are “about” economic insecurity relative to other issues. The raw probit results for both the money and time models appear in Appendix F. For the purpose of the main text, I restrict attention to the quantities of greatest theoretical interest (though I invite readers to examine the raw results).

TABLE 6.6
Donation results (1988–2008, based on full multivariate model)

	Model 1: Classifying “the economy” not as economic insecurity	Model 2: Classifying “the economy” as economic insecurity
Economic Insecurity Issue Most Important	−0.04∗	−0.03 ∗∗
Other Economic Issue Most Important	−0.01	−0.00

Note: ∗p < .05    ∗∗ p < .01 (two-tailed tests)

In Table 6.6, I present the main results for donation behavior. The cell entries indicate the change in the probability that individuals donated money to a political organization depending upon which type of issue they found to be most important.²⁷ Regardless of how I classified people’s concern with “the economy,” I find that people who prioritized economic insecurity issues were 3–4 percentage points less likely to have made a donation over the entire time series than those who prioritized a noneconomic issue (all else being equal). This was slightly larger than the 2.5 percentage-point estimate in the bivariate comparison. Put into perspective, Ansolabehere and his colleagues estimated that individuals donated $2.4 billion to candidates, parties, and PACs during the 1999–2000 election cycle.²⁸ They also estimated that the average donation was $115. If we extrapolate their numbers to the 2008 campaign, this would translate to 6.9–9.2 million fewer donors giving approximately $793–1,058 million less to political organizations during this election cycle. As expected, there is no similar demobilizing effect among people who find other economic issues, such as government spending or budgetary matters, to be most important.

Table 6.7 presents the results for spending time on politics. Here we see the first evidence in this chapter of time-related demobilization. The estimated magnitude of the effect is 1–2 percentage points, which, although seemingly not very large, could have large substantive consequences. In a voting-age electorate with over 230 million people in 2008, 1–2 percentage points could have translated into a large number of people that rivaled (and in many cases greatly exceeded) the size of many well-known citizen groups. I do not have consistent evidence of a similar effect among people outside the labor force or among those who place another economic issue at the top of the political agenda. These latter results likely reflect the relatively robust senior citizen activism on Social Security and Medicare, as well as activism by citizens (especially conservative citizens) who are concerned about taxes and government spending. In summary, once I account for several other factors that affect people’s propensity to volunteer time to politics, I am able to discern the strongest evidence of time-based demobilization precisely where I’d expect it: among the people for whom such issues are a reminder of the time-based constraints they face in their everyday lives.²⁹

Although the models can account for many individual-level differences, they cannot account for every single one. For example, there is still the possibility that those most concerned about insecurity might not believe that there is a feasible and worthwhile public policy response to it. A concern like this could manifest itself in several ways. One is the possibility that they are incredulous that the political system will actually produce new policies to address the problems. Gridlock, partisan polarization, war, entrenched interests—all of these might suggest that even if the perfect policy solution exists, and even if a few lawmakers support it, it might not pass the legislative gauntlet. Despite this, we saw in the previous chapter (as well as in the 1992 and 2008 results presented earlier and some of the examples in Chapter 1) that mobilization around economic insecurity remained difficult even during periods of heightened political opportunity. So even when President Obama stated that health care was a priority of his administration and health care reform–minded Democrats held a filibuster-proof majority in the Senate, people most concerned about insecurity issues like this one were still demobilized by rhetoric focused on the issue.

TABLE 6.7
Spending time results (1988–2008, based on full multivariate model)

	Model 1: Classifying “the economy” not as economic insecurity	Model 2: Classifying “the economy” as economic insecurity
	Among People in the Labor Force
Economic Insecurity Issue Most Important	−0.02*	−0.01**
Other Economic Issue Most Important	−0.02	−0.02
	Among People Not in the Labor Force
Economic Insecurity Issue Most Important	−0.01	−0.02†
Other Economic Issue Most Important	−0.04	−0.03

†p < .10    ∗p < .05    ∗∗ p < .01 (two-tailed tests)

Another aspect not accounted for by the models is that perhaps people do not have a firm sense of what kinds of policies would help shore up insecurity. Some issues are simply easier than others. For example, issues that involve more symbolic content, as opposed to technical content, are more easily understood by people regardless of their level of political information and/or overall interest in politics.30 Issues like abortion and gay marriage elicit a gut response from most people. Even taxes would often fit the bill for some Americans, particularly if they favor lower taxes above all else. On the other hand, short of the government simply handing out money, policies that would reduce insecurity tend to be more technical. And some people tend to bristle at that—witness how those opposed to the Affordable Care Act frequently lampooned its twenty thousand pages of regulations. All of this might suggest that people are less willing to become active because they do not have a good idea of what should be done. Yet, as we saw in Chapter 1, there does appear to be a fairly wide consensus on what should be done, at least on a general level. Majorities favor the government increasing spending on protecting and improving the nation’s health, Social Security, and education. There is also broad support for having the government reduce job-related insecurity by funding retraining programs for those who have lost their jobs, improve health care by having a national health insurance plan, ensure the financial health of Social Security, and help make college affordable for everyone who wants to attend.

As a final note on the multivariate models, it is worth briefly addressing two potential methodological concerns. One is about the direction of causality. Throughout I have posited that people who consider jobs, health care costs, retirement security, and education costs to be most important are less likely to be involved in politics than those who prioritize other issues. This interpretation presumes that the decision to act follows from the attitude. But, in models like these that rely upon observational data, it is always possible for the direction of causality to be reversed. Perhaps instead of the attitude affecting people’s decisions about whether to become politically active, the reverse was the case—perhaps the act of donating or volunteering changed people’s perceptions of what the political agenda should look like. In theory, this interpretation of the data is entirely possible. But, in order for it to be substantiated in this instance, we would have to believe that donating and volunteering have wholly divergent effects on people’s issue priorities if they are not in the labor force, yet have the same effect if they are in the labor force. I have a hard time coming up with a reason for why both of these patterns would simultaneously emerge.

Another methodological issue worth mentioning, a perennial concern with models like these, is the possibility of omitted variables that might bias the estimates. While I have sought to make theoretical and methodological choices that are strongly consistent with the literature, some of the assumptions result from unavoidable limitations of the data and design. In particular, one might worry that the specification omits a variable that affects both the likelihood of finding certain issues to be most important and people’s propensity to donate time or money to politics (again with the caveat that it would depend upon the specific activity as well as one’s labor force status). In theory, a set of possible omitted variables might include more specific personal experience measures than what are available in the ANES. This might include more details on one’s job insecurity (e.g., history of job loss, health of one’s company, set of transferable skills), health care insecurity (e.g., health risk factors, size of premiums and deductibles, whether they have health insurance), retirement insecurity (e.g., amount of savings, availability of employer-provided retirement health insurance), or education insecurity (e.g., amount of tuition bills or expected financial aid). I noted in Chapter 2 that the ANES includes some markers of relevant personal experience, but they are certainly not as detailed as this list would suggest. And it is reasonable to suspect that each of these variables could affect one’s issue concerns as well as one’s likelihood of donating money or time to politics.

While I cannot “test” for the presence of an omitted variable that would bias my estimates, I can assess the magnitude of the threat to the results it might pose using a sensitivity analysis. My goal in doing so is to provide an idea of how large the correlation between an omitted variable (e.g. amount of retirement savings) and the included variables (i.e., issue concern and political behavior) would have to be in order to invalidate the results.31 The results would be invalidated if the omitted variable produced bias that was sufficiently large to wipe out the substantive effect (i.e., make it zero). For the purpose of the main text, I’ll simply state that with the exception of situations with very high correlation, I find that a potential omitted variable would not wipe out the substantive effects reported earlier.³²

THE REQUESTS PEOPLE RECEIVED

At various points in the book I have stressed the organizational side of politics in addition to the individual level. While it is true that individual citizens are ultimately the ones who act (or don’t), organizations are critical because people’s decisions to donate or volunteer are often occasioned by requests they send. Such requests might come from a personal friend involved with the organization, as was often the case for voluntary membership federations that proliferated starting around the time of the Civil War and up until the mid-1960s.³³ Or they might come from direct mail, which thanks to mainframe computing databases became a popular recruitment tool for Washington based advocacy organizations after the 1960s.³⁴ Or they could come via the Internet, which many organizations (those begun during the Internet age or before) routinely use to reach out to potential new members and mobilize existing ones.³⁵

These requests are often tied to issue concerns. Indeed, advancements in information technology mean that campaigns can now send far more tailored messages to their supporters than ever before. Unlike television ads and stump speeches that reach a broad audience at once, other forms of communication such as direct mail, email, telephone calls, web-based ads, and text messages can be far more personalized. Whereas for decades campaigns have collected information like past voting history, age, and income, since the 1990s they have started complementing that with a host of “lifestyle” data. For example, as key strategists of the 2004 Bush campaign acknowledged, “[The Bush campaign’s voter database in 2004] allowed them to track millions of voters based on their confidential consumer histories. If you’re a voter living in one of the 16 states that determined the 2004 election, the Bush team had your name on a spreadsheet along with your hobbies and habits, vices and virtues, favorite foods, sports, and vacation venues, and many other facts about your life.”36 But “life targeting” was hardly new in 2004. For instance, in 1996 the Clinton campaign started keeping track of things like what types of magazines people read and what kinds of cars they drove.³⁷ The presumption is that each of these lifestyle choices indicates the issue priorities voters hold, preferences they espouse, and as a result the political messages to which they might be responsive. In addition, we also know that campaigns routinely canvass supporters regarding their perceptions of the political agenda and which issues are important.³⁸ Such information enables campaigns to appear responsive to the concerns of their current and potential supporters.

How does all of this affect how we interpret the results from this chapter? Although it is reasonable to assume that the people most concerned about economic insecurity would have been targeted based on those concerns (especially in 1992 and 2008, using the arguments I presented earlier), ideally we would like to have actual data on this. It could still be the case that economic insecurity issues were never featured in any solicitations during the campaigns covered by my data set, despite the fact that the issues were commonly mentioned in both Clinton’s and Obama’s television ads.³⁹ Unfortunately the ANES does not include measures of that kind of information, and organizations do not typically make communications strategies like these publicly available.

We do, however, have the benefit of a unique data set that contains information about donation and volunteer requests at the individual level. In 2004 Brigham Young University conducted the Campaign Communication Study (CCS), a survey that asked a nationally representative sample of registered voters to collect all pieces of fund-raising and volunteer request mail that they received during the last three weeks of the campaign.⁴⁰ This includes mail from all national and congressional candidates, political parties, and interest groups. Overall, 1,606 people returned 213 unique pieces of mail that specifically asked the recipient to donate money and 76 unique pieces that specifically asked them to volunteer time. Each appeal was then coded in a variety of ways, including which issues it mentioned. The appeals that were returned have a high degree of face validity. As we would expect, the volume was higher in battleground states than in nonbattleground states, and the five most commonly mentioned issues—taxes (36% of solicitations), jobs/unemployment (25% of solicitations), health care (20% of solicitations), abortion (20% of solicitations), and terrorism (20% of solicitations)—were all top public concerns in 2004.⁴¹

TABLE 6.8
Percentage of appeals that mentioned select economic insecurity issues during 2004 campaign

Issue	Percentage of donation appeals featuring issue	Percentage of volunteer appeals featuring issue
Jobs/Unemployment	24	36
Health Care	20	33
Social Security	10	22
Medicare	11	22
Prescription Drugs	12	24

Source: Campaign Communication Study.

The first column of Table 6.8 lists the percentage of fundraising solicitations that mentioned unemployment, health care costs, and retirement costs.⁴² Given the wide range of concerns that people have and the wide range of possible issues that could be mentioned in each request, we would not expect any one issue to be mentioned all that often. It is thus quite noteworthy that each one was featured in at least 10% of the solicitations during this time frame.

To give a flavor for what they looked like, consider the following two examples from opposite sides of the campaign. On October 1, 2004, the Democratic National Committee sent a letter urging people “to rush as generous a donation as you can possibly afford to support your Democratic candidates up and down the ballot.” The letter went on to discuss several reasons why people needed to urgently support John Kerry and the Democrats running for Congress with a monetary donation, including several issues that no doubt reminded some recipients of financial constraints in their own lives:

George W. Bush has presided over one miserable failure after another—and he has misled us on the most important issues of our day. Here’s the record Bush and the Republicans don’t want to discuss:

Almost two million jobs lost
More than 5 million more Americans without health care
Gas prices out of control and America dependent on Middle
East oil
Medicare premiums rising a record 17 percent

In addition to requesting a donation—after it reminded recipients about job insecurity, health care insecurity, and pain at the pump—this letter also encouraged them to “get involved in your own community” via the extensive get-out-the-vote operation that was being set up at the time.

Here is a second example. The Republican National Campaign Committee sent out a letter asking for donations to “help protect and strengthen our Republican Majority in the U.S. House of Representatives.” It provided information about the most pressing issues of the day before then asking people to fill out a survey expressing their policy preferences. The issues ran the gamut, with several related to bread-and-butter economic security issues facing American families, including health care costs, prescription drug costs, and Social Security shortfalls. On health care, for example, the letter struck an upbeat yet sobering tone:

America has the best health care system in the world. However, 43 million Americans still lack the health insurance coverage they need to properly access the system and millions have problems paying for prescription drugs.

Given the prevalence of high prescription drug costs, it is likely that many of the people who read this were reminded of a financial constraint before then being asked to donate money.

While both of these examples focused heavily on persuading people to donate money, volunteer requests also regularly featured economic insecurity issues. The second column of Table 6.8 lists the percentage that mentioned them. We again see that each one regularly appears, attesting to the fact that they were routinely part of the campaign to get people to spend time (most often with get-out-the-vote operations). Each of them also appears more frequently than they did in the donation requests, which reflects the fact that volunteer requests were on average more packed with issue content than fund-raising ones (on average they mentioned 5.6 issues versus 4.4 for fund-raising letters).

Although these solicitation data are limited to one year, it is important to keep in mind that 2004 was not an exceptionally low watermark for the economy. Thus what we see is that even when the broader economy is growing, political organizations have attempted to convince people to become more active by appealing to issues that remind many of them of financial concerns they face every day. I would argue that this observation helps us better interpret the ANES results given earlier in the chapter. Although we do not have details on every request that people received over the multidecade time span of the ANES data set, we now have some evidence that in general solicitations during campaigns do not categorically shy away from economic insecurity. We can thus tentatively rule out the possibility that the only reason that people who care the most about economic insecurity issues participated less was because they never received solicitations mentioning such issues.

COMPARISON OF POLICY PREFERENCES

The conclusion to emerge so far is that there is something unique about economic insecurity issues relative to others. The findings suggest that we should expect the number of people advocating on these issues to be smaller than it should be relative to the degree of concern about them. Part of that will likely be reflected in the contours of the political agenda, as fewer voices make it less likely that the issues end up on the broader political agenda.

Nevertheless, sometimes insecurity-related issues do make it on the agenda as a result of exogenous shocks. One recent example involved the spike in unemployment following the 2008 financial crisis. Between September 2008, when Lehman Brothers went bankrupt and insurance giant AIG received an unprecedented government bailout, and October of the following year, the unemployment rate in the United States rose from 6.1% to 10%. This was America’s first brush with double-digit unemployment since the early 1980s. In response, the federal government extended the time period over which people could receive unemployment benefits, up to ninety-nine weeks. For a moment at least, there was some extra relief for ordinary Americans facing heightened labor market insecurity.43

Situations like this immediately raise the following question: When such issues do happen to make it onto the political agenda, does it matter if the people personally affected by them are less vocal? Put differently, do the people who are most active in politics hold similar preferences on the issues anyway? Is it reasonable to claim that those most concerned about economic insecurity issues (who are highly likely to be personally affected by them based on Chapter 2’s findings) have effective surrogates speaking out on their behalf? One way to examine that question is to compare the policy preferences of people who prioritize economic insecurity issues with those of everyone else.

To do that I investigate responses to three questions that concern the government’s role in the economy. One is what should be done about rising health care costs, and in particular what role the government versus the private sector should play in covering all medical and dental expenses. The responses, which appear in Figure 6.5 were coded 1–7, with 1 most favoring a government insurance plan and 7 most favoring private insurance. Regardless of how I code general concerns about “the economy,” the basic pattern is the same: those most worried about economic insecurity issues issues adopt a more liberal view. They are significantly more likely to favor a government-based response rather than a private one to escalating health and dental costs (t=6.08, p < .01). Similar patterns emerge when we look at preferences regarding job creation in Figure 6.6 and government spending in Figure 6.7. In both cases we observe substantial differences. With respect to the government’s role in job creation, those who consider economic insecurity issues to be most important are more likely to say that they favor the government guaranteeing jobs and a good standard of living as opposed to just letting each person get ahead on his/her own (t=6.24, p < .01). With respect to general attitudes toward government spending, those who consider economic insecurity issues to be most important are more likely to say that the government should provide many more services and increase spending a lot as opposed to reducing spending and services (t=7.03, p < .01).⁴⁴

Figure 6.5. Comparison of preferences regarding health insurance between those who find economic insecurity issues to be most important versus some other issue.

Figure 6.6. Comparison of preferences regarding guaranteed jobs/standard of living between those who find economic insecurity issues to be most important versus some other issue.

Figure 6.7. Comparison of preferences regarding government spending between those who find economic insecurity issues to be most important versus some other issue.

Looking at these policy preferences adds texture to the earlier conclusions. Differences in participation patterns arise when we focus on action. They take on especially poignant political importance once we dig even deeper. The people who are less active in the public square also have systematically different views on what direction policy should take. I noted in Chapter 1 that approximately two-thirds to three-quarters of the electorate supports the government reducing job-related insecurity by funding retraining programs, improving health care by spearheading a national health insurance plan, ensuring the financial health of Social Security, and helping make college affordable for everyone who wants to attend. That provided evidence in support of the oft-cited idea that although Americans might be philosophical conservatives (thereby espousing conservative principles associated with freedom and limited government in the abstract), they are operationally liberal when faced with specific policy proposals.45 Yet the broad pattern is by no means a complete consensus. Here we see suggestive evidence that the people more opposed to those proposals are probably those who are more active in politics.

CONCLUSION

Overall what we have observed in this chapter helps to bolster the experimental patterns we witnessed with U.S. PIRG, ACSCAN, and IHA from Chapter 5. Looking across a wider period of time and a broader set of actions and issues, we see how over the past thirty years the people most concerned about unemployment, health care costs, retirement insecurity, and college costs have systematically been less involved in politics than people most concerned about other issues. The evidence for this point was especially strong for monetary donations. There was also evidence in the realm of time expenditures, though that only came into focus with the multivariate models. Lastly, as expected, the major exception to the latter pattern concerns people outside the labor force.