5-6 pages reflective

11.Matthews.pdf

Home >Political Science homework help >5-6 pages reflective

11 When Partisans Are Attacked

Motivated Reasoning and the New Party System

J. SCOT T MAT THE WS

It is not always easy to be a partisan, that is, to sustain the affective commit- ment involved in party identification. In a competitive democracy, holding fast to one’s partisan convictions necessarily implies exposing oneself to the threat of disagreement – with opposition partisans and with those who reject partisan attachment altogether. Canadian political history, for ex- ample, abounds with moments of intense partisan disagreement. Indeed, in his early study of Canadian politics, André Siegfried remarked that “there can be few countries in the world in which elections arouse more fury and enthusiasm than in Canada” (1906, quoted in Carty 2006, 7). The ensuing history of party politics would supply countless examples to sup- port Siegfried’s early conjecture. A fractious debate over conscription dur- ing the First World War would ultimately lead the Liberals, in 1925, to assert that the election of a Conservative government would lead to the outbreak of war (Cairns 1968, 65). In 1988, Liberal leader John Turner as- serted that the Tory prime minister, Brian Mulroney, had “sold us out” in his pursuit of a Canada-US free trade agreement (Johnston et al. 1992, 27). Recent national elections have featured some of the most heated partisan rhetoric in memory: from the Liberals in 2006, a suggestion that the Con- servatives harboured (presumably nefarious) plans to increase the presence of the military in Canadian cities (Rose 2006); and, from the Conservatives in 2008, the assertion that Liberal leader Stéphane Dion was, in fact, “not a

Parties, Elections, and the Future of Canadian Politics, edited by Amanda Bittner, and Royce Koop, UBC Press, 2013. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/templeuniv-ebooks/detail.action?docID=3412833. Created from templeuniv-ebooks on 2018-01-21 17:58:10.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

232 J. Scott Matthews

leader” and “not worth the risk” (McLean 2008). The Liberals’ major narra- tive in the 2011 election – that the Conservatives had abused the country’s democratic institutions – was easily as aggressively partisan as any in memory (Norquay 2011).

This sort of intense partisan disagreement is not costless. Such disagree- ment may engender political withdrawal and a cynical orientation towards government and politicians on the part of citizens (cf. Ansolabehere and Iyengar 1997; Mutz and Reeves 2005). Yet the hidden and largely unrecog- nized cost of disagreement over partisan matters is the potential threat to the partisan’s self-concept: if you disagree with me, then what does this say about me as a person?

The focus of this chapter is an examination of partisan responses to “at- tacks” on the competence and integrity of the parties with which they identify. An attack, in this sense, is simply a message with obvious and intensely negative evaluative implications with respect to some object. For the purposes of this research, therefore, the focus is on “partisan attacks” – messages that imply obvious and strongly negative evaluations of polit ical parties.

The systematic study of partisans’ responses to partisan attacks offers two principal benefits to scholars of Canadian politics. First, a focus on such responses advances our understanding of partisanship as a feature of political cognition. Indeed, as this chapter will argue, such attacks provide a unique opportunity to adjudicate between contending views of partisan political cognition. In particular, the analysis supplies new evidence with regard to the role of political knowledge in sustaining partisan divisions. This has been an important theme in recent work on partisan effects on political attitudes and perception (Blais et al. 2010; Gaines et al. 2007). This chapter builds on this work by uncovering support for a specific interpreta- tion of the relationship between knowledge and partisan political cogni- tion: the theory of motivated political reasoning (Taber and Lodge 2006). In the process, the analysis raises questions about the dominant model of “polarization effects” in political science, John Zaller’s RAS model (1992).

Second, analysis of responses to partisan attacks offers a revealing view of the nature of Canadian partisanship. In an important sense, responses to partisan attacks represent the “high-water mark” of partisan cognition, compared with partisan responses to other political objects (such as issues) that have more equivocal partisan implications. Moreover, the data ana- lyzed here – the 2004, 2006, and 2008 Canadian Election Study (CES) data- sets – cover a critical time of change in the party system, which permits an

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

233When Partisans Are Attacked

investigation of the party system’s impact on partisanship. As others have noted (see Chapters 3 and 6 in this volume), Canadian electoral politics have, among other things, become more adversarial in recent years in the wake of the merger of two right-of-centre parties and a string of minority governments. Has the nature of Canadian partisan cognition registered the effects of these political developments? This chapter theorizes the pos- sibilities and evaluates the evidence.

At the most general level, the chapter speaks to continuities in the nature of Canadian partisanship, in spite of the paroxysms of the party system in recent years. The influence of partisan identities on political perceptions and attitudes is robust; that is, Canada remains “a party country” (Carty et al. 2000, 14), especially in the minds of its partisans.

Partisan Cognition: The Motivated Reasoning Account That partisan commitment shapes responses to political messages, such as partisan attacks, is a long-standing claim in political science (see Berelson et al. 1954; Lazarsfeld et al. 1948). The view of Campbell and colleagues on the centrality of partisanship in perceptual and evaluative processes is neatly summarized in their famous assertion that party identification “raises a perceptual screen through which the individual tends to see what is favorable to his [sic] partisan orientation” (1960, 133). Although the con- clusions of the “Michigan School” have been challenged from time to time (e.g., Bullock 2009; Gerber and Green 1999), theirs remains the standing view of partisanship’s role in political cognition (Bartels 2002; Gerber et al. 2010). In this regard, critical evidence has lately been supplied by scholar- ship informed by the late social psychologist Ziva Kunda’s theory of mo- tivated reasoning (1990). The theory stands out as a distillation of the distinctive implications of psychological models that assume biased, as op- posed to rational or “even-handed,” processing of information.

The core of the motivated reasoning theory is, naturally, the assumption that all perception is subject to motivation. In other words, the theory holds that, in forming representations of the external world, social perceivers seek to satisfy various goals, albeit to different degrees across individuals and situations. The first important goal is to perceive the world accurately, at least to an approximation, given finite cognitive resources (Kunda 1990, 481). In addition to accuracy goals, however, the motivated reasoner is also influenced by “directional goals,” or preferences for particular perceptual outcomes, regardless of the accuracy of such conclusions. Importantly, the theory is fairly open-ended as to the content of such directional goals.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

234 J. Scott Matthews

Indeed, Kunda (1990) notes diverse sources of directional goals, such as self-esteem needs, a desire to sustain positive expectations with regard to one’s performance in a given domain, and the desire to avoid cognitive dis- sonance. Whatever their source, the upshot is that preferences over the dir- ection of a given conclusion condition relevant reasoning processes and, ultimately, the conclusion at which one arrives.

It is the emphasis on “process” that distinguishes motivated reasoning theory from prior approaches to biased perception, such as Festinger’s theory of cognitive dissonance (1957). Kunda (1990) suggests that decisive evidence for the role of motivation in social perception is to be found not in the nature of beliefs and attitudes but in the nature of the cognitive oper ations the social perceiver undertakes in the course of forming beliefs and attitudes. The motivated reasoner, in this view, relies on a “biased set of cognitive processes: strategies for accessing, constructing and evaluating beliefs” (Kunda 1990, 480). At the same time, however, Kunda suggests im- portant limits on the magnitude of such processing biases: “People mo- tivated to arrive at a particular conclusion attempt to be rational and to construct a justification of their desired conclusion that would persuade a dispassionate observer. They draw the desired conclusion only if they can muster up the evidence necessary to support it” (Kunda 1990, 482-83, em- phasis added).

The need to “muster up evidence” in support of desired conclusions high- lights the role of effortful cognitive processes in Kunda’s account of biased perception. One can imagine, for instance, an account of biased perception whereby information that challenges one’s directional goals is simply ig- nored (e.g., Zaller 1992; see discussion below). Obviously, this sort of “bull- headed” approach to realizing preferred conclusions would not leave the residues of biased memory retrieval and belief construction that are char- acteristically highlighted as evidence for the motivated reasoning theory. The focus on the need to “muster up evidence” for favoured conclusions also points to the importance of domain-relevant knowledge in motivated reasoning processes. The ability to “persuade a dispassionate observer” that one’s justification for her preferred conclusion is sound clearly may depend on one’s available store of relevant information and argumentation.

Evidence of motivated reasoning in political settings is robust (Gaines et al. 2007; Lau and Redlawsk 2006; Lodge and Taber 2000; Redlawsk 2002; Redlawsk et al. 2010; Rudolph 2006; Taber and Lodge 2006; Taber et al. 2009). Indeed, in their influential account of “motivated political reason- ing,” Taber and Lodge argue that politics is a highly likely domain in which

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

235When Partisans Are Attacked

to observe processes of motivated reasoning, reflecting the fact that, as a sphere of social perception, the political world is peculiarly “affectively- charged” (2006, 756). Following Kunda (1990), Taber and Lodge suggest that in addition to accuracy goals, individuals are motivated by “partisan goals” – that is, a desire to “apply their reasoning power in defense of a prior, specific conclusion” (2006, 756). With regard to mechanisms of biased processing, these scholars emphasize asymmetrical treatment of pro- and counter-attitudinal arguments and information. In particular, whereas pro-attitudinal messages are more or less accepted uncritically, counter- attitudinal information is likely to be subject to vigorous scrutiny and to lead to the generation of counter-arguments (757). Further, given the chance, Taber and Lodge (2006) suggest, individuals will “seek out” pro-attitudinal information.

The role of domain-relevant (i.e., political) knowledge and attitude strength in conditioning motivated political reasoning is also highlighted by Taber and Lodge (2006). The “politically knowledgeable,” they write, “because they possess greater ammunition with which to counterargue in- congruent facts, figures, and arguments, will be more susceptible to motiv- ated bias than will unsophisticates” (757). Likewise, strength of pre-existing judgments will index individuals’ motivation to make the “effort” to engage in selective counter-arguing, information seeking, and the other cognitive operations that underwrite perceptual bias (757).

Canadian research on motivated political reasoning is fledgling at best, a conclusion that applies generally to work on partisan effects in political judgment in Canada (Merolla et al. 2008, 674).1 Blais and colleagues’ recent study (2010) of Canadian reactions to the “sponsorship scandal,” which en- gulfed national politics and featured prominently in the federal elections of 2004 and 2006,2 constitutes the sole Canada-focused evaluation of the theory of motivated political reasoning. Notably, the findings of Blais and colleagues contradict motivated reasoning theory’s expectations, particu- larly regarding the role of domain-relevant knowledge, which they find to be insignificant. That said, while the study is trailblazing for Canadian stu- dents of partisanship, there are reasons to suspect that its conclusions are limited in their generalizability. As Blais and colleagues note, the proceed- ings of the commission of inquiry with regard to the scandal were “intense- ly covered by the media” in 2004-05 and “produced startling testimonies that left no doubt that there had been corruption” (2010, 4). Further, cover- age of the 2004 and 2006 election campaigns was mostly dominated by scandal-related issues of government accountability and corruption (Andrew

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

236 J. Scott Matthews

et al. 2006; Goodyear-Grant et al. 2004). Consequently, partisans’ capacity to “muster up evidence” (per Kunda 1990) to support their desired conclu- sions about the affair may have been constrained by an atypically intense informational environment: besides being plentiful, information was both relatively credible (i.e., dignified by an official public inquiry) and relatively engaging (i.e., featured “startling testimonies”).3 In keeping with this logic, the impact of the issue – at least in the 2004 election – was simply massive at the level of the individual voter, and views on the sponsorship scandal were “clearly a major factor in helping the Conservatives deny the Liberals another majority” (Gidengil et al. 2006, 15). The informational context of the sponsorship affair also clarifies the non-impact of political knowledge on the relationship between party identification and judgments relevant to the scandal; that is, given the intense media attention concerning the affair, its partisan implications were likely apparent to even the least knowledge- able identifiers.4

Partisan Cognition When Partisans Are “Under Attack”: Diverging Predictions For the most part, evidence in favour of motivated reasoning – in politics and in psychology generally – is experimental (e.g., Lau and Redlawsk 2001, 2006; Taber and Lodge 2006). This approach has obvious virtues with re- gard to measurement and internal validity. The trade-off, of course, is external validity. An alternative, survey-based approach relies on motiv- ated reasoning theory’s prediction that biased memory retrieval and belief construction processes are conditioned by domain-relevant knowledge (e.g., Blais et al. 2010; Gaines et al. 2007). The implication is that, given cer- tain informational and motivational assumptions (on this point, see dis- cussion below), partisan effects on political perceptions will increase with political knowledge, all other things being equal.

An obvious problem with this approach, however, is that motivated pol- itical reasoning theory is not alone in anticipating knowledge-partisanship interactions with respect to political perceptions. Indeed, much simpler processes, such as cue taking, are compatible with this expectation. In par- ticular, the most prominent alternative5 to the motivated reasoning account of partisan perception, Zaller’s “receive-accept-sample” (RAS) model (1992), also implies that partisan differences in political perception should gen- erally increase with political knowledge – subject to the constraint that the “contextual information” required to connect partisanship to relevant messages is sufficiently “obscure” (see note 4; see also Berinsky 2007). The

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

237When Partisans Are Attacked

logic is expressed in the second major premise of the RAS model, the “re- sistance axiom” (Zaller 1992, 44).

Unlike motivated reasoning, however, the RAS model rests on passive ac- ceptance and rejection of messages in light of partisan cues, rather than on biased evaluation of arguments or active generation of counter-arguments. As Zaller explains, “this postulate [i.e., the “resistance axiom”] makes no allowance for citizens to think, reason or deliberate about politics: If cit- izens are well informed, they react mechanically to political ideas on the basis of external cues about their partisan implications, and if they are too poorly informed to be aware of these cues, they tend to uncritically accept whatever ideas they encounter” (1992, 45). Importantly, this “mechanical” or “bull-headed” interpretation of political cognition implies a critical di- vergence across the predictions of the RAS and motivated reasoning theor- ies: in those settings where awareness of partisan cues is uniform, the theory of motivated reasoning predicts an interaction with political knowledge – as the knowledgeable are better able to generate proattitudinal arguments – while the RAS model does not.

It is in the evaluation of the above prediction that responses to partisan attacks – messages that imply obvious and strongly negative evaluations of political parties – are uniquely useful. Decoding the partisan implications of such messages does not require significant political knowledge. Indeed, if the message is received, then so should be its partisan implications. Consider, for example, the following partisan attacks:

The Conservative Party is a threat to Canada’s social programs. The Liberal Party’s Green Shift would really hurt the Canadian economy. The only reason the Liberals are cutting taxes is to buy votes. An NDP government would really hurt the Canadian economy. The NDP is out of touch with the times.

Arguably, for the typical Canadian, the partisan implications of these messages should be very clear. The party that is “a threat” to valued entitle- ments, for example, is clearly not the best party with which to identify one- self. Accordingly, the partisans of that party implicated in a given attack should find it easy to recognize the cues and reject such messages; that is, recognition should not be conditional on prior political knowledge. Like- wise, other partisans – those whose party is not targeted in a given attack – should be much more likely to passively accept the message, regardless of one’s pre-existing store of political knowledge.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

238 J. Scott Matthews

Therefore, if the RAS model fully describes the psychological under- pinnings of partisan bias in political perception, then political knowledge should not condition the relationship between party identification and re- sponses to partisan attacks. If, on the other hand, motivated reasoning pro- cesses are operating instead of, or even alongside, the more mechanical RAS dynamics, then political knowledge should condition partisans’ reactions to rhetorical attacks on parties. Assuming that the latter proposition holds yields the following hypothesis:

H1. The magnitude of partisan effects on responses to partisan attacks increases with political knowledge.

Motivated Political Reasoning across the Party System While H1 should generally hold, two assumptions implicit in the account of motivated political reasoning developed above bear noting. Importantly, the assumptions imply that features of the party system may condition mo- tivated reasoning processes regarding partisan attacks.

The first assumption is that information conducive to biased evaluation of arguments or generation of counter-arguments in response to partisan attacks is relatively scarce. Clearly, this assumption may be violated under certain conditions – for instance, when the attack involves a particularly salient aspect of party conflict. Under these conditions, “evidence” sup- porting motivated reasoning processes may be relatively easy to “muster up,” making political knowledge a less discriminating variable.6

The second assumption is that, strength of partisanship aside, partisans are uniformly motivated to “bolster” (Taber and Lodge 2006) partisan at- tacks directed at other parties. In a two-party system (such as in the United States), this assumption should generally hold; in a multi-party system with plurality elections and highly variable electoral geography – that is, in Canada – the assumption seems problematic. The latter combination of circumstances implies, in general, important variation in the intensity of inter party conflict, as reflected in competitive conditions (Johnston 2008). Consequently, the degree to which positive evaluations of one party logically imply negative outcomes for another (and vice versa) also varies. Assuming that the perception of such contingencies underlies partisan processing goals, the motivation to bolster attacks directed at parties other than one’s own should, accordingly, partially reflect the pattern of competi- tion existing between the parties.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

239When Partisans Are Attacked

What does this logic imply with regard to partisanship in the Canadian party system between 2004 and 2008? For one thing, the electoral progress of the Conservatives should supply significant variation in the information- al environment. A straightforward possibility is simple quantitative change. If we assume that the level of information about a party roughly tracks the party’s relative dominance in the system,7 the availability of information relevant to the Conservative Party of Canada (CPC) should have increased steadily from 2004, when the party entered the system electorally, to 2008, when the party sought re-election as the government. Consequently, biased generation of arguments and counter-arguments regarding attacks on Con- servatives should have become easier over time. This implies (1) growing partisan differences within levels of knowledge (as relevant information is more plentiful), and (2) shrinking differences across levels of knowledge (as information is easier to find).8 The hypotheses, therefore, are:

H2. Controlling for political knowledge, the magnitude of partisan effects on responses to partisan attacks on Conservatives increases over time.

H3. The effect of political knowledge on the magnitude of partisan effects on responses to partisan attacks on Conservatives decreases over time.

For reasons developed in the following paragraphs, H2 and H3 can be evaluated for the major parties only (see, especially, the concluding para- graph of this section).

Apart from dynamics in informational context, the party system pro- vides variation in the competitive foundations of motivated reasoning. An obvious distinction is between major and minor parties. Apart from their unique place in the national-level competition for government, the pattern of riding-level competition between the Liberals and Conservatives sets them apart from both the NDP and the Bloc Québécois. An examination of the over-time dynamics of riding-level competition between the parties indicates that the major parties’ major competition has consistently been the other major party (whether Liberal or Conservative), while the major competition for the minor parties (NDP and BQ) was also the major par- ties. At the same time, the relative position of the Liberal and Conservative Parties as major competitors in relation to the NDP and the BQ changed significantly over time.9

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

240 J. Scott Matthews

This distinction – between the consistency of competitive relations be- tween the major parties and the variation in minor/major party competi- tive relations – should imply differences in the extent of motivated reasoning across major and minor parties. As the focus of the analysis to come is responses to attacks on the major parties, expectations are con- fined to this domain. In short, the motivation to bolster major-party attacks should be greater among major-party than minor-party partisans, inas- much as Liberals and Conservatives have consistently been each other’s major electoral foes, whereas competitive relations have been more variable between the major parties and either the BQ or NDP. This implies, among major-party partisans, (1) relatively larger partisan differences within levels of knowledge (as the motivation to bolster is greater), and (2) relatively lar- ger differences across levels of knowledge (as the informational returns to motivation are greater). Specifically:

H4. Controlling for political knowledge, the magnitude of partisan effects on responses to major-party partisan attacks is greater among major- party than among minor-party partisans.

H5. The effect of political knowledge on the magnitude of partisan effects on responses to partisan attacks is greater among major-party than among minor-party partisans.

Note, finally, that over-time variation in the competitive and informational foundations of motivated reasoning is confounded for the NDP and the BQ, which is the reason that H2 and H3 are assessed among major-party par- tisans only. Specifically, dramatic over-time shifts in the profiles of the leading contenders faced by minor-party candidates10 imply dynamics that should offset the influence of changes in the informational environment – that is, increased bolstering of Conservative attacks among BQ and NDP partisans after 2004 should coincide with improvements in the informa- tional context conducive to such bolstering.

Data and Methodology Data from the Canadian Election Studies (CES) allow an evaluation of these propositions concerning partisan responses to rhetorical attacks on parties. Indeed, it turns out that the CES investigators have been “at- tacking” Canadian partisans for years! More specifically, the CES has quer- ied reactions to each of the five attack statements quoted earlier during as many as three separate elections. These data provide a unique opportunity

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

241When Partisans Are Attacked

to consider the theory of motivated political reasoning in an observational setting. The analysis incorporates data from the 2004, 2006, and 2008 CES datasets. Only items from the telephone waves (the pre- and post-election surveys) are utilized. This yields 3,116, 3,107, and 2,426 cases for 2004, 2006, and 2008, respectively.11

It should be noted that the partisan attacks in these studies are, in one important regard, rather unlike the typical rhetorical attack encountered by partisans in real-world settings: the CES partisan attacks originate from a (presumably) credible source. Partisans in this context are (more or less) denied one of the easier means of counter-arguing or resisting rhetorical attacks: denigrating the source of the attack. Importantly, this dynamic should tend to inflate the impact of political knowledge on motivated rea- soning, as knowledgeable partisans will be relatively more advantaged in the generation of counter-arguments than in real-world settings.12

As noted above, five items over the past three iterations of the CES satis- fy this chapter’s conceptualization of partisan attack. In the present analy- sis, just the first two of the attacks listed above are examined – referred to henceforth as the CPC threat and the Liberals’ Green Shift attacks. The for- mer is the only one of the five items that can be examined for all years under consideration; it is selected because it permits analysis of over-time change. Selection of the Liberals’ Green Shift item adds a partisan attack concerning the other major party in the system. This item is substantively similar to the CPC threat attack, in that both items focus on policy commitments. Responses to the Liberals’ Green Shift attack can be examined for 2008 only, however, as the item reflects a pledge made in that year’s election cam- paign.13 In all years, attack statements were followed by the query “Do you strongly agree, somewhat agree, somewhat disagree, or strongly disagree?” Each variable was recoded to the (0, 1) interval. Canadians were, on aver- age, evenly divided on both attacks, whatever the year, and the measures have high variances.14

The principal individual-level independent variables in the analysis are party identification and political knowledge. The former is measured using the standard indicators of direction and strength of partisanship: “In fed- eral politics, do you usually think of yourself as a: Liberal, Conservative, N.D.P, Bloc Quebecois, or none of these?” and, for partisans, the follow-up query “How strongly [party] do you feel: very strongly, fairly strongly, or not very strongly?” In the statistical models, dummies for strong and weak iden- tifiers within each partisan group are included, where strong corresponds to those who feel “very strongly” and weak includes all other identifiers.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

242 J. Scott Matthews

Political knowledge also relies on standard measurement techniques: cor- rect responses to a battery of office recall measures are summed and then rescaled to the (0, 1) interval.15 Each model includes the eight dummies for combinations of party identification and attitude strength (4 parties × 2 at- titude strengths), along with interactions between these variables and pol- itical knowledge.

The models also include a wide range of controls. The logic behind these terms is to control for other influences on responses to partisan attacks that may also be conditioned by political knowledge, particularly policy atti- tudes, political values, and various demographic terms.16 All models include controls for income taxation preferences (corporate, personal), political values relevant to the role of the government in the economy and society (regarding the role of the private sector in job creation and attributions of personal success/failure in the market), economic perceptions, age, income, and region (dummies for the West, Quebec, and Atlantic regions). The CPC threat models add controls for various social program spending attitudes (welfare, health, education, social housing17) and evaluation of the import- ance of social programs as an election issue. Similarly, the Liberals’ Green Shift model adds to the baseline set of controls measures of attitudes to- wards program spending on the environment and evaluation of the import- ance of the environment as an election issue.18 In addition – and critically – all models also include interactions between all control variables and pol- itical knowledge. (Coefficient estimates for the controls are not reported.)

Results The implications of the empirical results can be stated simply: there is sig nificant evidence of motivated political reasoning in Canadians’ responses to partisan attacks. There are large partisan effects on reactions to all of the partisan attacks examined. Moreover, political knowledge is an important moderator of the magnitude of these effects. This pattern of results is ro- bust to the inclusion of diverse controls. At the same time, there is notable variation in the nature and magnitude of partisanship-knowledge inter- actions. Some of the pattern makes sense in light of features of the party system, whereas other results are more difficult to interpret.

Partisan Responses to Partisan Attacks: “Bull-Headed” or Motivated Reasoning? Table 11.1 reports ordinary least-squares estimates for four regression models, one for each dependent variable. Predicted values based on these

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

TA B

LE 1

1. 1

Pa rt

y id

en ti

fi ca

ti o

n , p

o lit

ic al

k n

o w

le d

g e,

a n

d r

es p

o n

se s

to “

p ar

ti sa

n at

ta ck

s”

PC t

hr ea

t t o

PC t

hr ea

t t o

C PC

t hr

ea t t

o Li

be ra

l G re

en S

hi ft

so ci

al p

ro gr

am s

(2 00

so ci

al p

ro gr

am s

(2 00

so ci

al p

ro gr

am s

(2 00

w ill

h ur

t e co

no m

y (2

00 8)

St ro

ng L

ib er

0. 09

05 *

(0 .0

49 2)

03 50

.0 39

0. 06

(0 .0

51 7)

–

0. 04

(0 .0

51 4)

W ea

k Li

be ra

l 0.

03 31

.0 25

0. 01

(0 .0

27 0)

–

0. 02

(0 .0

32 0)

02 20

.0 31

8) St

ro ng

C PC

–

0. 23

4* **

.0 53

– 0.

24 9*

(0 .0

41 2)

–

0. 12

3* **

.0 46

0. 07

(0 .0

45 7)

W ea

k C

– 0.

10 6*

(0 .0

31 6)

–

0. 18

0* **

.0 29

– 0.

09 86

** *

(0 .0

31 9)

08 25

** *

(0 .0

31 8)

St ro

ng N

D P

0. 11

(0 .0

63 4)

15 0*

* (0

.0 64

0. 16

5* **

.0 52

0. 08

(0 .0

51 9)

W ea

k N

D P

0. 07

(0 .0

47 1)

12 6*

(0 .0

45 1)

07 96

* (0

.0 47

0. 11

4* *

(0 .0

46 8)

St ro

ng B

– 0.

09 67

.0 98

0. 11

(0 .0

60 5)

14 9

(0 .0

90 7)

00 21

4 (0

.0 90

3) W

ea k

B Q

05 77

.0 46

– 0.

00 36

9 (0

.0 39

– 0.

01 61

.0 55

0. 07

(0 .0

55 0)

St ro

ng L

ib er

al ×

P K

12 9

(0 .0

82 6)

20 2*

(0 .0

60 2)

18 7*

* (0

.0 77

– 0.

14 6*

.0 77

6) W

ea k

Li be

ra l ×

P K

07 11

.0 46

) 0.

08 05

(0 .0

40 9)

14 8*

(0 .0

53 5)

–

0. 12

6* *

(0 .0

53 1)

St ro

ng C

PC ×

P K

–

0. 07

(0 .0

86 6)

–

0. 08

(0 .0

61 3)

–

0. 21

9* **

.0 75

0. 18

7* *

(0 .0

74 9)

W ea

k C

PC ×

P K

–

0. 11

5* *

(0 .0

55 9)

–

0. 01

(0 .0

45 1)

–

0. 07

(0 .0

56 5)

07 60

.0 56

3) St

ro ng

N D

P ×

0. 21

(0 .1

11 )

0. 05

(0 .0

93 5)

05 42

.0 95

– 0.

11 1

(0 .0

94 4)

W ea

k N

D P

× PK

15 2*

.0 84

) –

0. 00

18 5

(0 .0

65 4)

04 47

.0 83

– 0.

26 4*

(0 .0

82 9)

St ro

ng B

Q ×

P K

34 8*

.1 86

) 0.

08 33

.0 99

0. 13

0 (0

.1 38

) –

0. 09

(0 .1

37 )

W ea

k B

Q ×

P K

02 39

.0 90

0. 10

2 (0

.0 63

0. 20

1* *

(0 .0

91 8)

–

0. 13

9 (0

.0 90

9) Po

lit ic

al k

no w

le dg

e 0.

00 29

2 (0

.1 22

) 0.

03 68

.1 00

) 0.

01 75

.1 35

) –

0. 10

4 (0

.1 19

)

C on

st an

t 0.

53 4*

(0 .0

69 )

0. 42

0* **

.0 65

0. 49

8* **

.0 80

0. 77

8* **

.0 65

3, 11

3, 10

2, 42

R 2

.2 4

N ot

es : C

o ef

fic ie

n t

(s ta

n d

ar d

e rr

o r)

. E st

im at

es f

o r

co n

tr o

ls n

o t

re p

o rt

ed . P

K =

p o

lit ic

al k

n o

w le

d g

e. *

p <

.1 , *

* p

< .0

5, *

** p

< .0

So ur

ce s:

C an

ad ia

n El

ec ti

o n

St u

d ie

s fo

r 20

04 , 2

00 6,

a n

d 2

00 8.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

0 .8

0 .6

0 .4

0 .2

00 0

.2 0

.4 0

.6 0

.8 1

0 .8

0 .6

0 .4

0 .2

00 0

.2 0

.4 0

.6 0

.8 1

0 .8

0 .6

0 .4

0 .2

00 0

.2 0

.4 0

.6 0

.8 1

0 .8

0 .6

0 .4

0 .2

00 0

.2 0

.4 0

.6 0

.8 1

0 .8

0 .6

0 .4

0 .2

00 0

.2 0

.4 0

.6 0

.8 1

0 .8

0 .6

0 .4

0 .2

00 0

.2 0

.4 0

.6 0

.8 1

S tr

o n g

p a rt

is a n s

W ea

p a rt

is a n s

(A ) 2

0 0

4 (B

) 2

0 0

6 (C

) 2

0 0

P o lit

ic a l k n o w

le d g e

P o lit

ic a l k n o w

le d g e

P o lit

ic a l k n o w

le d g e

Agreementwithattack Agreementwithattack

Li b er

a l

C P

C N

D P

B Q

FI G

U R

E 11

.1 Th

e in

fl u

en ce

o f

p o

lit ic

al k

n o

w le

d g

e an

d s

tr en

g th

o f

p ar

ty id

en ti

fi ca

ti o

n o

n p

ar ti

sa n

re sp

o n

se s

to t

h e

st at

em en

“T h

e C

o n

se rv

at iv

e Pa

rt y

is a

t h

re at

t o

C an

ad a’

s so

ci al

p ro

g ra

m s,

” in

t h

re e

d if

fe re

n t

n at

io n

al e

le ct

io n

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

245When Partisans Are Attacked

estimates are plotted in Figures 11.1 and 11.2, which indicate, for each dependent variable, the impact of partisanship by political knowledge for all partisan groups. For a given attack, the left panel indicates effects among strong partisans, while the right panel depicts effects among weak partisans.

As one would expect, the estimates imply stark differences across par- tisans in levels of agreement with the partisan attacks. Disagreement be- tween Liberals and Conservatives is especially obvious: the two groups of partisans are far apart in their responses to the partisan attacks. And while differences among strong, major-party partisans are consistently larger than those among weak partisans, partisan effects are substantively large regardless of attitude strength. Differences between Conservative and minor-party partisans are also consistently large, in some cases even larger than those between Liberals and Conservatives. Between Liberal and minor-party partisans, however, differences are fairly modest. For the CPC threat models, this is intuitive: all non-Conservative partisans should re- spond affirmatively to the attack on the Conservatives. For the Liberals’ Green Shift model, on the other hand, the pattern is more puzzling; we will come back to this point in the discussion below.

Overall partisan differences aside, the moderation of these differences by political knowledge is key to assessing the motivated reasoning account of partisan perception (H1). In this regard, thirty-two of the coefficients

FIGURE 11.2 The influence of political knowledge and strength of party identification on partisan responses to the statement “The Liberal Party’s Green Shift/carbon tax would really hurt the Canadian economy,” in the 2008 national election

0.8

0.7

0.6

0.5

0.4

0 0.2 0.4 0.6 0.8 1.0

Liberal CPC NDP BQ

0.7

0.6

0.5

0.4

0 0.2 0.4 0.6 0.8 1.0

Political knowledge Political knowledge

Strong partisans Weak partisans

A g re

em en

t w

it h

a tt

a ck

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

246 J. Scott Matthews

are of interest: the estimates for the eight partisanship-knowledge inter- actions in each model. Nearly two-thirds of the thirty-two partisanship- knowledge interactions are in the direction predicted by H1 and of sufficient magnitude to be reliably distinguished from zero.19 Despite the obvious partisan implications of the CPC threat and the Liberals’ Green Shift at- tacks, political knowledge has important moderating effects.

To convey the substantive thrust of the results, consider panel (C) in Figure 11.1, which plots the estimated partisan effects for the 2008 CPC threat model. At all levels of knowledge and attitude strength, Conservatives are much less inclined to agree that their party is a threat to Canada’s social programs. At the same time, the gap between Conservatives and other par- tisans clearly grows as political knowledge increases (between strong Liberals and Conservatives, for example, the gap grows from less than 0.2 to 0.6 units). Between Liberals and Conservatives, furthermore, the polar- ization is highly symmetrical: as knowledge increases, strong Liberals grow in their approval of the CPC threat attack at roughly the same rate as strong Conservatives grow in their disapproval of the attack. In the language of Taber and Lodge (2006), strong Liberals bolster to the same degree that strong Conservatives counterargue the sentiment. New Democrats and Bloquistes also appear to increasingly bolster the attack on the Conservatives as knowledge increases, but the effects are not statistically significant.

Among weak partisans, a congruent pattern of effects is observed. In- deed, if anything, the effect of knowledge is somewhat more pronounced. Polarization between Liberals and Conservatives increases by about five times across the range of knowledge: the difference grows from about 0.05 to 0.25. Note also that, again, the polarization is roughly symmetrical among Liberals and Conservatives. Knowledge also has a powerful effect on differ- ences between weak Conservatives and Bloquistes: the distance between these groups increases about sixfold across the range of knowledge.

The relatively larger impact of knowledge among weak partisans com- pared with strong partisans may reflect unanticipated floor and ceiling effects. For instance, the Liberal-Conservative gap in response to the CPC threat attack in 2008 is tripled among strong partisans but quintupled among weak partisans. At the same time, among strong partisans, initial partisan differences between Liberals and Conservatives are four times as large as those among weak partisans. This implies, among other things, less room for detectable growth in partisan polarization.

The overall pattern fits the motivated reasoning theory. These results, furthermore, are fairly typical of the findings across the models, as depicted

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

247When Partisans Are Attacked

in Figures 11.1 and 11.2. Importantly, the pattern is generic across both the CPC threat and Liberals’ Green Shift attacks – at least among Liberal and Conservative partisans. Indeed, the left panel of Figure 11.2 indicates that, among strong partisans, it is with regard to this attack that the greatest in- crease in polarization between Liberals and Conservatives is observed: the difference between the groups grows from 0.10 to over 0.40 units across the knowledge scale – a 400 percent increase! The relationship between Liberal- Conservative polarization and political knowledge is, again, even sharper among weak partisans: the difference between these groups increases roughly sixfold (from approximately 0.04 to approximately 0.25 units) across the range of knowledge.

Among partisans of the minor parties – the NDP and BQ – the pattern of responses to the Liberals’ Green Shift attack does not fit expectations nearly so neatly. For both strong and weak minor-party partisans, know- ledge effects are trivial and, worse, incorrectly signed: statistical signifi- cance aside, these partisans appear to resist rather than bolster the attack on the Liberals. And, as noted, differences between Liberals and minor- party partisans are modest overall. The pattern is sharpened among weak New Democrats: rather than polarizing against the Liberals (as these groups polarized against the Conservatives in the CPC threat models), these partisans polarize against the Conservatives. Whereas the least knowledgeable of the weak NDPers accept the attack on the Liberals quite readily – more readily, in fact, than the least knowledgeable Conservatives – the most knowledgeable weak New Democrats reject the attack even more categorically (0.40) than knowledgeable, weak Liberals (approximate- ly 0.45)!20

The inconsistency of effects across major and minor parties in the Liberals’ Green Shift model highlights a more general feature of the results: while roughly two in three of the estimates of the partisanship-knowledge interactions provide reliable support for H1, one in three estimates do not – either because the coefficients are too imprecisely estimated to warrant serious consideration or because they are incorrectly signed. This suggests that motivated reasoning processes may be moderated in significant ways by variables other than knowledge and strength of partisan affiliation. In keeping with earlier discussion, the obvious candidate is the party system.

The Impact of the Party System H2 and H3 reflect the theorized effects of an informational environment increasingly propitious to motivated reasoning about the Conservatives.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

248 J. Scott Matthews

H2 implies that, controlling for political knowledge, partisan differences in responses to the CPC threat attack should be larger in 2006 and 2008 than in 2004. H3 implies that the effect of knowledge on partisan differences should shrink over the same period.

Put simply, neither hypothesis is supported by the data. Contrary to H2, overall differences between major-party partisans appear to shrink, rather than grow, between 2004 and 2008. Indeed, at lower levels of political know- ledge, partisan differences (between Liberals and Tories) are roughly one- third smaller in 2008 than in either 2004 or 2006. The general pattern holds, moreover, for both strong and weak major-party partisans, at least to an approximation. That said, in the top rank of political knowledge, partisan differences are at least as large at the end of the period as at the beginning.

Turning to H3, the results, again, do not accord with expectations. If anything, knowledge effects are, on the whole, larger in 2008 than in either of the previous years, regardless of attitude strength. The most dramatic, and curious, disconfirmation of the hypothesis is among strong Conserv- atives: knowledge effects in this group grow threefold between 2004 and 2008. The implication, from a motivated reasoning perspective, is that the Conservative Party’s electoral success made it more, not less, difficult for the party’s partisans to counter-argue partisan attacks. An interpretation of this unexpected result is offered in the conclusion.

Whereas the first two party-system-level hypotheses fail, the remaining two hypotheses fare much better. Pursuant to H4, differences between major-party partisans in response to attacks on the major parties are, on average, larger than differences between minor-party and the relevant group of major-party partisans.21 There are significant differences between Liberals and Conservatives in responses to all of the partisan attacks, but differences between minor-party partisans and the relevant group of major- party partisans are unstable. The indicative case has already been noted: Figure 11.2 depicts the trivial differences between Liberals, on the one hand, and New Democrats and Bloquistes, on the other, in response to the Liberals’ Green Shift attack. Presumably, the shifting competitive land- scape, particularly the relative decline of the Liberals as leading contenders in viable BQ and NDP ridings (see note 9), had diminished the intensity of associated partisan conflicts by 2008.

Regarding the moderation of partisan effects by political knowledge, as predicted in H5, the effect of knowledge is greater among major-party than minor-party partisans, as evidenced by the strength and statistical

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

249When Partisans Are Attacked

significance of coefficients in Table 11.1 and the steeper slopes of major- party lines in Figures 11.1 and 11.2.22

Thus, the findings on the impact of the party system are perfectly mixed. Notably, those predictions concerning the motivational consequences of the party system (H4 and H5) fare better than those relating to the infor- mational consequences of party system change (H2 and H3). Of course, firm conclusions about the influence of the party system must await analy- sis involving more party-system-level cases. More generally, it is signifi- cant that attention to features of the party system appears to resolve some of the inconsistency in the results as they pertain to H1. If we restrict the analysis of H1 to partisans of the major parties, then the supportive evi- dence for the motivated reasoning theory is impressive indeed.

Conclusion Rhetorical attacks on parties – and, by extension, on their partisans – are a common feature of democratic politics that offers an uncommon oppor- tunity to explore the cognitive dimension of partisanship. It is precisely the familiarity and simplicity of such verbal assaults in politics that makes them a useful instrument for uncovering evidence of deeper, more involv- ing features of political reasoning in Canada.

Within the largest partisan groups, the Liberals and Conservatives, an- alysis of responses to partisan attacks suggests an image of the Canadian partisan as a motivated political reasoner – as an “intuitive lawyer” acting in cognitive defence of her pre-existing partisan commitments. For these major-party partisans, responses to messages that imply obvious and strong- ly negative evaluations of their political parties are conditioned by political knowledge, such that partisans implicated in a given attack are more likely to reject – and other partisans more likely to accept – the attack with in- creased political knowledge. Although such partisanship-knowledge inter- actions are consistent with Zaller’s RAS model in certain contexts, the present set of results – assuming uniform awareness of the partisan impli- cations of partisan attacks – is uniquely predicted by the motivated reason- ing theory. This makes a notable contrast with the one previous analysis of motivated reasoning among Canadian partisans (Blais et al. 2010).

An important caveat, however, concerns the apparent moderating in- fluence of the party system. As indicated, the best evidence of motivated reasoning is found among major-party partisans. Among minor-party partisans, on the other hand, the evidence is mixed. The consistent influ- ence of partisanship and its interaction with political knowledge among

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

250 J. Scott Matthews

Liberals and Tories contrasts with uneven effects among Bloquistes and New Democrats. This pattern, of course, is sensible in light of the differing qualities of competitive relations involving the major parties compared with those involving the minor parties.

Other features of the party system do not seem to have such sensible ef- fects. Most striking is the apparent over-time growth in knowledge effects regarding attacks on the Conservatives – a finding that extends even to Conservative partisans. It seems unlikely that the key informational prem- ise underlying the expected decline in knowledge effects is faulty: we have good evidence that media coverage of the Conservatives did indeed grow after 2004 (see Daku et al. 2009). It is conceivable, however, that the theor- etical logic is too crude. A more complex dynamic combines quantitative and qualitative change in information about the Conservatives. In this regard, the critical fact would be the Tories’ 2006 election win: at this point, the party switched from opposition to government status, in the process becoming a likelier target of negative political rhetoric. This should have made it more difficult for Conservative partisans to resist rhetorical attacks on their party.23

A second unexpected result concerns weak New Democrat partisans, who, in response to the attack on the Liberals’ Green Shift, polarized against – rather than with – the Conservatives, at least beyond the lowest levels of political knowledge. A similar effect among weak BQ partisans approach- es the 0.10 threshold ( p = 0.126). In broad terms, the inconsistency fits theoretical expectations: given their more variable competitive relations with the major parties, minor parties’ partisans should have less motivation than their major-party counterparts to bolster major-party attacks. But this leaves unexplained the intriguing tendency, especially among weak New Democrats, to seemingly counter-argue the attack on the Liberals. One possible interpretation is that, at very low levels of knowledge, weak New Democrats (and perhaps Bloquistes) react much as Conservatives do, in that they accept (and perhaps bolster) the attack on the Liberals, as a straight- forward interpretation of interparty relationships would imply. At higher levels of knowledge, however, these partisans recognize various deeper implications of the attack on the Liberals’ environmental policy commit- ments, including implications for the environmental platforms of their own parties. Consequently, such sophisticated New Democrats and Bloquistes may conclude that the attack on the Liberals is simultaneously an attack on their parties, and react accordingly. This is highly speculative, but it does

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

251When Partisans Are Attacked

indicate the complexities of motivated reasoning regarding partisan at- tacks in a multi-party system.

Finally, an unexpected but sensible pattern is seen in the existence of apparent floor and ceiling effects. That is, counter-arguing and bolstering processes appear to exert greater effects when initial differences between partisan groups are smallest (and therefore when there is the greatest scope for increasing attitude extremity). This dynamic appears in comparisons of strong and weak partisans in three of the four statistical models. How to explain this finding? A mundane possibility is that they are a measurement artifact: with a limited number of response categories, the dependent vari- ables may place constraints on the degree of attitude extremity that can be observed (Taber and Lodge 2006, 757). A more substantive interpretation is that strong partisans, even at low levels of knowledge, engage in significant motivated reasoning that, given cognitive bounds, creates limited room for further cognitive reinforcement.24

What are this chapter’s broader implications? First, it adds to our picture of the Canadian partisan. Party identification in Canada is sometimes thought to be a rather inert affair – unusually flexible and lightly held, quite unlike the fiery partisanship of our Republican and Democratic neigh- bours to the south (for a recent statement of this view, see Bélanger and Stephenson 2010). Presumably, the turmoil in the party system in recent years – particularly the transformations among programmatic conserva- tives – strengthens the case for this view. A generally unstated but rather natural implication of this “flexible” view of party identification is that Canadian partisans are unlikely to engage in the defensive cognitive gym- nastics that characterize the motivated reasoner. Indeed, why would a “flex- ible partisan” be so “inflexible” in her views? Inasmuch as this chapter lends support to a motivated reasoning view of Canadian partisanship, it also troubles (even though it does not dispose of) the image of the fickle Can- adian partisan.

Second, this chapter generates a clear forecast regarding the quality of partisan reasoning over the life of the new Parliament and, most import- antly, during the next election. The signal political development resulting from the 2011 election was the profound change in the competitive situa- tion among the parties, particularly the rise of the NDP to the status of Official Opposition and the collapse of the Liberal Party. On the theoretical logic developed in this chapter, these developments imply, among Con- servatives and New Democrats, an intensification of motivated reasoning

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

252 J. Scott Matthews

processes in relation to each other’s parties, candidates, and partisans. To be sure, Conservative/NDP competition had begun to intensify at the riding-level prior to 2011 (see note 9), yet the novel national-level competi- tive dynamics appear likely to provide these partisan groups with new mo- tivation for biased cognition. Conversely, Conservatives and Liberals may exhibit somewhat less evidence of motivated reasoning processes in rela- tion to each other’s parties and partisans – although, of course, riding-level competition between these parties is likely to be fierce in certain parts of the country, such as southern Ontario.

Finally, the findings described here have implications for party strategy. As the analysis reveals, immunity to partisan attacks – at least those com- ing from a reasonably credible source (e.g., the academic survey inter- viewer) – depends greatly on political knowledge, even for the strongest partisans. This result provides important counsel to strategists of party pol- itical communication: it may be important to respond to an attack directed at your party in order to supply less knowledgeable identifiers with the cognitive ammunition they need to resist the attack’s evaluative implica- tions. Of course, for any given partisan attack, the source may be insuffi- ciently credible or the volume of the attack insufficiently loud to have much effect on the least knowledgeable partisan. But when both of these condi- tions are absent, it is critical not to leave emotionally committed but cogni- tively disengaged partisans defenceless.

Acknowledgments The author thanks Amanda Bittner and Royce Koop for organizing the workshop that gave rise to this book and for their detailed feedback on an earlier draft of this chapter. Thanks also to Marc-Andre Bodet for supplying riding-level electoral data.

Notes 1 Merolla and colleagues’ review (2008) of work on partisanship is focused on re-

search that explicitly conceptualizes partisanship as an “information shortcut” or “heuristic.” Even so, their review encompasses most of the important work on par- tisan effects in political attitudes and perceptions in Canada.

2 Blais et al. (2010, 3-4) provides a concise review of the details of the scandal. 3 The possibility that perceptual bias is constrained under certain informational

conditions fits with a range of studies that conclude that motivated perception hinges on informational ambiguity (e.g., Berelson et al. 1954; Festinger 1957; Lord et al. 1979; Redlawsk et al. 2010).

4 This follows Zaller’s claim that the impact of political awareness (which Zaller meas- ures using indicators of factual knowledge) on partisan selectivity in information

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

253When Partisans Are Attacked

processing is limited to those cases “in which the contextual information necessary to evaluate an issue in light of one’s predispositions is, for one reason or another, obscure” (1992, 47-48). See also discussion below under “Motivated Pol itical Reasoning across the Party System.”

5 On this point, see discussion of the theory’s citations and accolades in Dobrzynska and Blais (2008, 261n1).

6 Consider, in this regard, Blais and colleagues’ findings (2010) regarding partisan effects on judgments concerning the sponsorship scandal, discussed above.

7 Evidence in support of this proposition is supplied by analysis of the nature of media coverage of the parties in elections since 1993 reported in Daku et al. (2009, see especially Table 3). In a similar vein, see Nevitte and colleagues’ analysis (2000) of awareness of the NDP in the 1997 federal general election.

8 Conversely, biased generation of arguments and counter-arguments regarding the Liberals should have become more difficult. The hypotheses, therefore, would be reversed in relation to responses to attacks on the Liberal Party. Data limitations prevent the evaluation of these hypotheses in relation to the Liberal Party, and I therefore focus on Conservative attacks only.

9 These conclusions derive from analysis (not shown) of the partisan distribution of each party’s leading competitors at the riding level within ridings where a given party was viable – that is, placed either first or second on election day. Both Liberals and Conservatives faced leading competitors from the other major party in at least 50 percent of their viable ridings in each of the three elections. On the other hand, both New Democrats and Bloquistes were far more (less) likely to face Conservative (Liberal) leading competitors in 2008 than in 2004. Data for this analysis was pro- vided by Marc-Andre Bodet.

10 See note 9 above. 11 The data were obtained from the Canadian Opinion Research Archive (CORA) at

Queen’s University at Kingston (http://www.queensu.ca/cora). Further methodo- logical details can be obtained through the CORA website.

12 An alternative interpretation has parallel implications. It could be that knowledge- able partisans may be more aware than others of the “real” sources of the CES team’s partisan attacks – that is, the parties themselves – insofar as the CES items draw on common features of the discourse of the party system. If this is so, then the knowledgeable will be in a relatively better position to engage in source denigration processes (perhaps even prior to the survey interview).

13 It should be noted that the “Green Shift” attack was actually presented in two dif- ferent ways. Random halves of the CES sample were asked their reactions to (1) the statement presented above or (2) this statement: “The Liberal Party’s Carbon Tax would really hurt the Canadian economy.” For the purposes of this chapter, item (2) is treated as conceptually equivalent to item (1), and so discussion throughout the chapter is in terms of the “Green Shift” phrasing only.

14 Data not shown. 15 Details of these questions are available upon request. 16 Evidence of cognitive heterogeneity of this sort is legion. The classic citation is

Sniderman et al. (1991). For Canadian results, see Bittner (2007) and Roy (2009).

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

254 J. Scott Matthews

17 The 2008 model does not include the social housing spending attitude, as it is not measured in that year of the survey.

18 Except for age, which is measured in years, all variables are scaled to the (0, 1) inter- val. Details of all question wordings and coding decisions are available upon request.

19 Of these thirty-two estimates, thirteen are statistically significant at the .10 level or better, four at the .15 level, and three at the .20 level. In all, fully twenty-seven coefficients are correctly signed. Just 2 coefficients are incorrectly signed and sig- nificant at the .20 level.

20 It bears emphasizing that this result is robust to controls for various indicators of policy preferences relevant to the environment. It is difficult, therefore, to attrib- ute these effects to unmeasured policy differences between the Conservatives and the other non-Liberal parties.

21 The relevant groups are Conservative and Liberal partisans for the CPC threat and Liberals’ Green Shift attacks, respectively.

22 All sixteen of the relevant coefficient estimates applying to major-party partisans (the partisanship/knowledge interactions) are correctly signed, with nine of these statistically significant at the .10 level and five at the .20 level. Among minor-party partisans, just nine of the relevant interactions are correctly signed, with five of these significant at the .10 level and two significant at the .20 level. Importantly, two of the minor-party partisanship/knowledge interactions at or approaching conven- tional significance thresholds are incorrectly signed: those applying to strong NDP and weak BQ partisans in the Liberals’ Green Shift model.

23 Of course, this would have simultaneously made it easier for opposition partisans to bolster anti-Conservative claims, and there is little indication of this in the results.

24 Matthews’s analysis (2010) of the impact of election campaigns on partisan bias in economic perception has parallel implications; see especially p. 234.

References Andrew, Blake C., Antonia Maioni, and Stuart Soroka. 2006. “Just When You

Thought It Was Out, Policy Is Pulled Back In.” Policy Options 27 (3): 74-79. Ansolabehere, Stephen, and Shanto Iyengar. 1997. Going Negative: How Political

Advertisements Shrink and Polarize the Electorate. New York: Free Press. Bartels, Larry M. 2002. “Beyond the Running Tally: Partisan Bias in Political Per-

ceptions.” Political Behavior 24 (2): 117-50. Bélanger, Éric, and Laura B. Stephenson. 2010. “Parties and Partisans: The Influence

of Ideology and Brokerage on the Durability of Partisanship in Canada.” In Voting Behaviour in Canada, edited by Cameron D. Anderson and Laura B. Stephenson, 107-38. Vancouver: UBC Press.

Berelson, Bernard, Paul Lazarsfeld, and William McPhee. 1954. Voting. Cambridge, MA: Harvard University Press.

Berinsky, Adam. 2007. “Assuming the Costs of War: Events, Elites, and American Public Support for Military Conflict.” Journal of Politics 69 (4): 975-97.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

255When Partisans Are Attacked

Bittner, Amanda. 2007. “The Effects of Information and Social Cleavages: Ex- plaining Issue Attitudes and Vote Choice in Canada.” Canadian Journal of Political Science 40: 935-68.

Blais, André, Elisabeth Gidengil, Patrick Fournier, Neil Nevitte, Joanna Everitt, and Jiyoon Kim. 2010. “Political Judgments, Perceptions of Facts, and Partisan Effects.” Electoral Studies 29 (1): 1-12.

Bullock, John G. 2009. “Partisan Bias and the Bayesian Ideal in the Study of Public Opinion.” Journal of Politics 71: 1109-24.

Cairns, Alan. 1968. “The Electoral System and the Party System in Canada, 1921- 1965.” Canadian Journal of Political Science 1 (1): 55-80.

Campbell, Angus, Philip Converse, Warren E. Miller, and Donald E. Stokes. 1960. The American Voter. New York: John Wiley and Sons.

Carty, R.K. 2006. “The Shifting Place of Political Parties in Canadian Public Life.” IRPP Choices 12 (4): 5-13.

Carty, R.K., William Cross, and Lisa Young. 2000. Rebuilding Canadian Party Pol itics. Vancouver: UBC Press.

Daku, Mark, Adam Mahon, Stuart Soroka, and Lori Young. 2009. “Media Content and Election Campaigns: 2008 in Comparative Context.” Paper prepared for presentation at the annual meeting of the Canadian Political Science Asso- ciation, Ottawa, May.

Dobrzynska, Agnieszka, and André Blais. 2008. “Testing Zaller’s Reception and Acceptance Model in an Intense Election Campaign.” Political Behavior 30: 259-75.

Festinger, Leon. 1957. A Theory of Cognitive Dissonance. Evanston, IL: Row, Peterson.

Gaines, Brian J., James H. Kuklinski, Paul J. Quirk, Buddy Peyton, and Jay Verkuilen. 2007. “Same Facts, Different Interpretations: Partisan Motivation and Opin- ion on Iraq.” Journal of Politics 69 (4): 957–74.

Gerber, Alan S., and Donald P. Green. 1999. “Misperceptions about Perceptual Bias.” Annual Review of Political Science 2: 189-210.

Gerber, Alan S., Gregory A. Huber, and Ebonya Washington. 2010. “Party Affiliation, Partisanship, and Political Beliefs: A Field Experiment.” American Political Science Review 104 (4 November): 720-44.

Gidengil, Elisabeth, André Blais, Joanna Everitt, Patrick Fournier, and Neil Nevitte. 2006. “Back to the Future? Making Sense of the 2004 Canadian Election out- side Quebec.” Canadian Journal of Political Science 39: 1-25.

Goodyear-Grant, Elizabeth, Stuart Soroka, and Antonia Maioni. 2004. “The Role of the Media: A Campaign Saved by a Horserace.” Policy Options 25 (8): 86-91.

Johnston, Richard. 2008. “Polarized Pluralism in the Canadian Party System: Presi- dential Address to the Canadian Political Science Association, June 5, 2008.” Canadian Journal of Political Science 41 (4): 815-34.

Johnston, Richard, André Blais, Henry Brady, and Jean Crête. 1992. Letting the People Decide: Dynamics of a Canadian Election. Montreal and Kingston: McGill-Queen’s University Press.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

256 J. Scott Matthews

Kunda, Ziva. 1990. “The Case for Motivated Reasoning.” Psychological Bulletin 108 (3): 480-98.

Lau, Richard R., and David P. Redlawsk. 2001. “Advantages and Disadvantages of Cognitive Heuristics in Political Decision Making.” American Journal of Political Science 45 (4): 951-71.

–. 2006. How Voters Decide: Information Processing during Election Campaigns. New York: Cambridge University Press.

Lazarsfeld, Paul, Bernard Berelson, and Hazel Gaudet. 1948. The People’s Choice. New York: Columbia University Press.

Lodge, Milton, and Charles Taber. 2000. “Three Steps toward a Theory of Motivated Political Reasoning.” In Elements of Reason: Cognition, Choice, and the Bounds of Rationality, edited by Arthur Lupia, Matthew McCubbins, and Samuel Popkin, 183-213. London: Cambridge University Press.

Lord, Charles, Marc Ross, and Mark Lepper. 1979. “Biased Assimilation and Atti- tude Polarization: The Effects of Prior Theories on Subsequently Considered Evidence.” Journal of Personality and Social Psychology 37 (11): 2098-2109.

Matthews, J. Scott. 2010. “Enlightenment, Equalization or What? Campaign Learn- ing and the Economy in Canadian Elections.” In Voting Behaviour in Canada, edited by Cameron D. Anderson and Laura B. Stephenson, 211-41. Vancouver: UBC Press.

McLean, James. 2008. “The Messenger Is the Message.” Policy Options 29 (10): 12-17. Merolla, Jennifer, Laura Stephenson, and Elizabeth Zechmeister. 2008. “Can

Canadians Take a Hint? The (In)Effectiveness of the (In)Effectiveness of Party Labels as Information Shortcuts in Canada.” Canadian Journal of Political Science 41 (3): 673-96.

Mutz, Diana, and Byron Reeves. 2005. “The New Videomalaise: Effects of Televised Incivility on Political Trust.” American Political Science Review 99 (1): 1-15.

Nevitte, Neil, André Blais, Elisabeth Gidengil, and Richard Nadeau. 2000. Unsteady State: The 1997 Canadian Federal Election. Toronto: Oxford University Press.

Norquay, Geoff. 2011. “The ‘Ballot Question’ in the 2011 Election: Two Wins, Two Losses.” Policy Options 32 (6): 47-51.

Redlawsk, David. 2002. “Hot Cognition or Cool Consideration: Testing the Effects of Motivated Reasoning on Political Decision Making.” Journal of Politics 64 (4): 1021-44.

Redlawsk, David, Andrew Civettini, and Karen Emmerson. 2010. “The Affective Tip ping Point: Do Motivated Reasoners ever ‘Get It’?” Political Psychology 31 (4): 563-93.

Rose, Jonathan. 2006. “The Liberals Reap What They Sow: Why Their Negative Ads Failed.” Policy Options (June-July): 50-54.

Roy, Jason. 2009. “Voter Heterogeneity: Information Differences and Voting.” Can adian Journal of Political Science 42: 117-37.

Rudolph, T.J. 2006. “Triangulating Political Responsibility: The Motivated Formation of Responsibility Judgments.” Political Psychology 27 (1): 99-122.

Siegfried, André. [1906] 1966. The Race Question in Canada. Toronto: McClelland and Stewart.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .

257When Partisans Are Attacked

Sniderman, Paul M., Richard A. Brody, and Philip E. Tetlock. 1991. Reasoning and Choice: Explorations in Political Psychology. Cambridge: Cambridge Univer- sity Press.

Taber, C.S., and M. Lodge. 2006. “Motivated Skepticism in the Evaluation of Political Beliefs.” American Journal of Political Science 50 (3): 755-69.

Taber, Charles S., Damon Cann, and Simona Kucsova. 2009. “The Motivated Processing of Political Arguments.” Political Behavior 31: 137-55.

Zaller, John. 1992. The Nature and Origins of Mass Opinion. New York: Cambridge University Press.

C o p yr

ig h t ©

2 0 1 3 . U

B C

P re

ss . A

ll ri g h ts

r e se

rv e d .