1. The phrase comes from Crosby 1998: 3. On the same page, Crosby’s argument for electoral reform implicitly addresses both exit and voice. Crosby argues that contemporary electoral forms "do little to empower the average citizen to understand … which candidates are most likely to steer it in the proper direction." He then adds that a weak public voice makes it impossible to bridge public and special interests in crafting policy. [BACK]

2. In this view, the civic skills and responsibility developed through social movements might make electoral reform less necessary. See Evans and Boyte 1992. [BACK]

3. For this view, see, e.g., Barber 1984; Crimmins 1995; Kemmis 1990; Mathews 1994. Only Barber makes a direct connection between "democratic talk" and election reform. [BACK]

5. In a sense, a model of representation should assume citizens are more analogous to "auditors" than active observers (Arnold 1990, 1993). "The monitorial citizen engages in environmental surveillance more than information gathering" (Schudson 1998: 311). [BACK]

6. Putnam 1995b. Note that Robert Putnam has acknowledged some overestimation of the decline of citizens' organizational memberships due to miscalculation (see Helliwell and Putnam 1996). [BACK]

7. Bennett 1998 [BACK]

8. The term citizen panel has been used elsewhere, such as in Crosby 1998. By using only a small number of random sample forums, I am answering one of the challenges put forward by Michael Bailey (1996: 14–16). As described by Bailey, liberal democratic theorists doubt that "widespread and intense participation… can be achieved with little or no coercion or growth in bureaucracy." To achieve full deliberation, "hundreds of thousands of public meetings must be created across the nation" and "surely abuse will occur. Rather than reducing the state’s influence, … its arbitrary power might increase." The panels I propose require only specific and very limited "incentives to deliberate" for small subsamples of the public, so there is no need for further public intrusion into the private sphere to inspire full participation in deliberation. [BACK]

9. Building on his experience with citizen juries, Ned Crosby (1998) has suggested a citizen election forum (CEF) that resembles the panels proposed later

10. I do not mean to imply that public financing is a bad idea per se. In fact, it might complement the citizen panels. The point is simply that reformers are proposing and enacting far more expensive reform packages for state and federal elections. [BACK]

11. Robert Dahl (1989) suggests a year-long "minipopulous" of 1,000 citizens, and Jim Fishkin (1995) recommends that deliberative polls require 400 or more participants. Peter DeLeon (1997: 112–13) also appears to lean toward policy panels with over 100 participants. There is a trade-off between panel size and error margin, as well as between the duration and depth of deliberation. As panels become larger and their deliberations become more lengthy, there is also a loss of the potential for face-to-face deliberation among all participants, and there is an increased risk of attrition as participants serving on relatively uncohesive panels become tired of the process or must tend to other obligations. Especially for single-issue or candidate judgments, I believe that I have chosen a balance that makes the citizen panels manageable and affordable without making them too small or hasty in their judgments. [BACK]

12. Some random sample forums use "attitudinal quotas," in which previously collected survey data provide baseline quotas for attitudes relevant to panel deliberations (e.g., "50 of the 400 panelists must oppose handgun registration"). The rationale for attitudinal quotas is that they protect against skewed attitudinal samples that can occur within demographically representative samples. These quotas would be more subject to fraud and error than demographic measures, and because of public skepticism about surveys, joining them to the citizen sampling might actually undermine the validity of the panels. Partisan quotas serve the same purpose by matching the random sample to data on party registration (e.g., "the panel should consist of 150 Democrats, 140 Republicans, 100 nonpartisans, and 10 Libertarians"). In states where voters register by party, this might be feasible, but a national sample could not do this, because many states do not register by party. When partisanship is reduced to unofficial self-reporting, it is not very different from other survey attitude measures, and potential panelists could lie about their party affiliation, as some have done in previous random sample forums. In an October interview with Dennis Foley, Ombudsman at the Orange County Register, I learned that the Register forums included a participant who lied about his party affiliation intentionally to throw off the quotas. [BACK]

13. Experimentation with different sizes of honoraria could determine the appropriate level. The point is to get a very high participation rate among the individuals initially contacted, which gives one more confidence that the sample is not skewed. It may turn out that $600 is too low. Over ten years ago, $50 or even $100 payments were not uncommon for focus-group participation (see Morgan 1988: 44), and a citizen panel is far more demanding. [BACK]

14. Bessette 1994: 149. [BACK]

15. I recommend holding the priority panel before the primary election to ensure that voting records become an issue during even the primary election. This means that a panel would consider all bills since the previous panel, which would mean that some key votes at the end of a term would not be considered until the subsequent election. [BACK]

16. Barone and Ujifusa 1998: 16 [BACK]

17. Indirectly, this system gives some encouragement to third parties, which I believe enrich the political process by articulating alternative points of view. Were there national voter registration by party, a voter registration threshold could be used to determine which parties could introduce bills. When the legislative panel is applied to states with such registration, I would recommend using registration rather than the one-seat-in-the-legislature requirement because winner-take-all districting makes it difficult for third parties to win seats, even if they have as much as 15 percent of a state’s registered voters. [BACK]

18. Some reviewers of the priority panel proposal have suggested that citizen panelists themselves might each have the opportunity to introduce one bill for consideration. If the initial citizen panel includes 400 members, this would permit another 400 bills to appear before the panel. My main concern is that this would unduly complicate the citizens' deliberations by overcrowding their agenda. It would also require citizens to take action prior to meeting one another, and I prefer their first official task occurring only after they convene. If citizens were allowed to introduce bills, it might be prudent to require that any bill suggested by a citizen panelist not be authored by their own representative. The panel’s purpose is to move beyond the concerns of individual districts and traditional political allegiances, and this restriction might prevent exceedingly local interests from entering the process at this stage. [BACK]

19. A reviewer of an earlier draft of this book suggested a fourth criterion— controversy. In this view, it is important that panelists pick bills about which candidates will disagree. I chose to leave out that criterion for two reasons. First, panel ratings of candidates should be high when there is a broad public (and candidate) consensus on how to approach the major problems of the day. Second, because they are deliberative, panel judgments will not always reflect conventional party views and public opinion. Thus, discarding "noncontroversial" legislation may cause panels to skip important issues about which there should be debate, and bills chosen for their controversial value may, nonetheless, result in little conflict during deliberation.

Another suggested criterion would ask panelists to look for bills that might lead to consensus among legislative panelists (described later in this chapter). I do not include this criterion because it asks too much of the priority panels. They could not complete their main task if they were also asked to discern the deliberative judgments likely to be reached on numerous bills. [BACK]

20. On these influences in relation to spatial arrangement, see Davies 1994. [BACK]

21. A vote could break any ties resulting from identical average ratings. [BACK]

22. In either system, I prefer rankings to ratings because they give each citizen equal weight in the final evaluation: that is, each citizen’s average ranking is 5.5. If, by contrast, citizens rated each bill on a ten-point scale, some citizens

23. David Morgan (1988: 44) discusses the difficulty of generating truly representative samples for focus groups, which are even smaller than legislative panels. The citizen jury process uses quotas to create well-balanced small groups (Crosby 1995). [BACK]

24. On the other hand, simultaneous meetings would permit all deliberation to take place closer to the election, which would reduce the likelihood of legislation "slipping past" the panel process late in the final session. Such slippage is inevitable, however, because the panels would have to come early enough to permit ballot printing, and so on. Because citizens can consider bills that have not even had committee votes, the panels might nonetheless deliberate upon bills that ultimately wound up on the floor. In any case, the next round of panels could consider legislation that came after the deadline for the previous panels. [BACK]

25. Crosby 1998 discusses the citizen panel in detail. Crosby described the witness procedure to me in a November 4, 1998, telephone interview. [BACK]

26. Lupia and McCubbins 1998: 225. The authors argue that this and other safeguards can provide a valuable "threat of verification." [BACK]

27. In a November 3, 1998, interview, Ned Crosby made the following suggestion regarding the panelists' control over the moderator role. If a panelist wanted the moderator to become more or less active, the panelist could have one minute to explain why the change was necessary, another panelist could give a counterargument for a minute, then a quick majority vote could determine whether to adopt the change. On the problems that commonly emerge in small democratic groups, see Gastil 1993. [BACK]

28. Straw polls could interfere with full deliberation by committing citizens too early to particular views. If one changed this process to include a straw poll, it should be by secret ballot, for the same reasons given earlier. If the vote were public, it should still be taken by simultaneous ballot to avoid the conformity problem posed by sequential straw polls taken in groups where the initial voice votes happen, by chance, to be the same. On the problems with straw polls, see Davis, Hubert, and Au 1996. [BACK]

29. Some advocates of deliberation stop short of voting, and the public forums and community deliberation initiatives described in chapter 6 work toward consensus more than they risk dividing into majority and minority viewpoints. In informal settings, it may be appropriate to avoid voting, but the citizen panels would not be able to make strong (and straightforward) recommendations without voting. As Lisa Disch argues in presenting her model of "power-sensitive deliberation," it is necessary to create "formal mechanisms to delineate the limits of consensus" (Disch 1996: 14). [BACK]

30. In a sense, the two-thirds requirement is a way of compensating for the small size of the citizen panel. With a sample of 50 people, there is an enormous margin of error (approximately 14 percent) surrounding the average view of the panel in relation to the views a full public. Requiring two-thirds agreement dramatically increases the odds that the small panel’s decision will match that of the entire population (had it gone through the same deliberations). By calling

31. Sanders 1997: 373. [BACK]

32. Past random sample forums I have observed have had some unenthusiastic participants, but most take their task very seriously. Even though participating in only a trial-run citizen conference closer to a private focus group, one participant was so conscientious that she mailed back (at her expense) the notepad she had not completely filled out during her deliberations on radioactive waste. Stories like that are common and reflect the fact that even "lazy" and "selfish" citizens can become very motivated and thoughtful for short periods of time, such as when called upon to serve on a jury or as a participant in a random sample forum. This is similar to the argument I made in Democracy in Small Groups (Gastil 1993: 123). A small group might manage to approach the democratic ideal, but it can probably only do so for a short period of time. [BACK]

33. Moon 1990: 5. [BACK]

34. Conover and Feldman 1989. [BACK]

35. See Rahn 1993; Moon 1990. [BACK]

36. Plutzer and Zipp 1996. [BACK]

37. On the use of ballot-based cues in low-intensity elections, see Gastil 1998. I believe that cue reliance is greatest for low-awareness voters in lowintensity elections (see chapter 3). Ellen Riggle argues that voters respond to simple cues in the full spectrum of elections. In moderate- and high-intensity races, voters might use cues because of the opposite problem: a media-saturated campaign can create "an information overload such that citizens resort to the use of party or image heuristics in order to cope with the overabundance of campaign trivia" (Riggle 1992: 243). [BACK]

38. Lupia 1994: 72 [BACK]

39. Though he denies the accusation, I believe that Perry Deess shares responsibility for the development of this idea. I have elsewhere called it a "legislative batting average," but I have not done so here to avoid confusing those who might not understand the baseball metaphor. The use of a rating has another advantage: it puts even more pressure on candidates to take both official and unofficial votes because it would treat abstentions as a non-match with the legislative panel verdict. Thus, a candidate who ignores this unofficial voting process altogether would be guaranteed a rating of zero. In cases where legislative panels deadlock on more than five bills, it might be more appropriate to print no summary ratings on the ballot. Otherwise candidates' ratings might differ by 25 percent or more based on only one disagreement. [BACK]

40. Miller and Krosnick 1998. Random ordering is better than the practice common in some areas, such as Salt Lake County, Utah. There, it is customary for a clerk to list first the candidates from the clerk’s own party (see Associated Press Political Service, "Demos Object to Republicans Being Listed First," Cam paign News, November 1, 1990). [BACK]

41. There are pros and cons associated with using the jury pool. An ideal jury pool would be drawn from all residents, but courts have no such list. Registered voter lists would produce an acceptable pool, but it would be too exclusive to use a list of registered drivers, as is done in some states. The next problem is "leakage" out of the original jury pool. In New Jersey, for example, Kadane 1993 found that only 30 percent of those contacted initially became available to serve. In locales where jury pools are unrepresentative, it might be necessary to use paid random samples instead. [BACK]

42. See Fred Bales, "Venerable Town Meeting Is Slowly Losing Its Voice," USA Today, April 14, 1998, 3A. Bales also reports that in towns with populations under 1,000, town meeting attendance is usually near 25 percent, but in towns with over 2,500 people, attendance goes as low as 5 percent. [BACK]

43. David Broder has argued that cutting taxes has become a "driving passion—a good in itself" for the electorate, because voters "are more affiuent than their nonvoting fellow citizens" ("Catatonic Politics," Washington Post, November 11, 1997). Antigovernment sentiment spans the population, however, as demonstrated by the widespread cynicism toward public officials documented in chapter 2. [BACK]

44. Threlkeld 1998: 5–6. [BACK]