Online Appendix: Coding of Variables anes

Download 184.16 Kb.
Date conversion29.04.2016
Size184.16 Kb.
1   2   3

Table A5: Party Identification and the Impact of the War in Iraq in Britain

Analysis of Respondents Who Were Undecided or Did Not Intend to Vote Labour, Conservative or Lib Dem in the Pre-Campaign Survey or Switched Votes from Labour, Conservative or Lib Dem after the Pre-Campaign Survey

Specific leadership qualities

Vote preference




Overall feelings



Liberal Democrat

War in Iraq

Labour identifier

.16 (.05)*

.24 (.05)*

.18 (.05)*

.06 (.04)

.04 (.09)

.05 (.04)

-.12 (.08)

Conservative identifier

.38 (.07)*

.31 (.08)*

.26 (.07)*

.05 (.06)

-.03 (.07)

.19 (.12)

-.33 (.11)*

Liberal Democrat identifier

.25 (.08)*

.47 (.10)*

.35 (.09)*

-.02 (.06)

-.12 (.15)

.31 (.09)*

-.36 (.16)*

War on terror

Labour identifier

.18 (.06)*

.27 (.06)*

.23 (.06)*

.03 (.04)

.06 (.10)

.10 (.05)#

-.05 (.10)

Conservative identifier

.33 (.07)*

.35 (.07)*

.22 (.06)*

.03 (.06)

.12 (.06)#

-.06 (.14)

.14 (.13)

Liberal Democrat identifier

.25 (.07)*

.28 (.09)*

.25 (.08)*

-.04 (.07)

.19 (.13)

-.10 (.09)

-.03 (.16)


Labour identifier

.44 (.06)*

.44 (.07)*

.18 (.06)*

.02 (.06)

-.10 (.12)

-.14 (.05)*

.20 (.11)#

Conservative identifier

.21 (.08)*

.03 (.09)

.05 (.08)*

.03 (.08)

-.02 (.06)

-.02 (.15)

-.11 (.15)

Liberal Democrat identifier

.50 (.10)*

.54 (.10)*

.16 (.09)#

-.01 (.06)

.09 (.11)

.00 (.08)

-.05 (.15)


Labour identifier

.32 (.06)*

Conservative identifier

.15 (.11)

Liberal Democrat identifier

.20 (.09)*


Labour identifier

.48 (.06)*

Conservative identifier

.72 (.08)*

Liberal Democrat identifier

.68 (.08)*


Labour identifier

.07 (.06)

Conservative identifier

-.06 (.08)

Liberal Democrat identifier

.10 (.07)

Overall feelings

Labour identifier

.27 (.07)*

-.16 (.04)*

-.01 (.07)*

Conservative identifier

.15 (.07)*

-.35 (.12)*

.31 (.10)*

Liberal Democrat identifier

.29 (.10)*

-.42 (.08)*

.13 (.12)


Labour identifier

-.23 (.05)*

-.34 (.05)*

-.27 (.05)*

.07 (.04)#

.41 (.09)*

.14 (.04)*

.27 (.08)*

Conservative identifier

-.56 (.05)*

-.62 (.05)*

-.52 (.05)*

-.13 (.06)*

.09 (.06)

.43 (.12)*

.46 (.11)*

Liberal Democrat identifier

-.38 (.08)*

-.52 (.08)*

-.35 (.07)*

.07 (.04)

.07 (.09)

.14 (.06)*

.71 (.12)*

n= 810, Chi2=.09

RMSEA = .033 (lower bound = .000, upper bound = .055)

Comparative Fit Index=.977

*p<.05 #p<.10

Notes: Coefficients are from structural equation model asymptotic distribution free estimates. Errors between specific leadership qualities are specified as correlated.

Additional Discussion and Analysis of the Australian Case
Footnote 32 of the paper refers to Australia as a potential third case in that an incumbent leader associated with the War in Iraq, John Howard, stood for re-election in October 2004 (and won a fourth term in office). Thus, Australia provides a possible additional case of a right-wing incumbent leader of a party with a hawkish reputation. However, I do not incorporate Australia into the main text of the paper for a variety of reasons.

First, on the grounds that I argue the US and Britain are comparable, such as the impact of 9/11 and the extent to which Bush and Blair were personally associated with the build-up and initial conduct of the war, Australia falls short. Australia lost 11 people on 9/11, one-sixth of what was a historically large number of losses in a terrorist attack for Britain, and obviously considerably fewer than the US (Australia also ultimately suffered no military casualties in direct combat during the war in Iraq). John Howard was also much less personally associated with the war than Bush or Blair (Mueller 2005). To provide support for this claim I conducted a Lexis-Nexis search of media coverage in the six months before the outbreak of war in Iraq. Because Australian and British newspapers are likely to pay an unusual amount of attention to their leaders, I focused on coverage of the three leaders and the keyword “Iraq” in the same articles, in the six months leading up to the war in the third country, the US. I focused on five US newspapers: the New York Times, Philadelphia Inquirer, USA Today, Washington Post and Washington Times. There were 8 stories in which all three leaders and Iraq were mentioned, compared to 685 stories in which Bush and Blair featured in conjunction with Iraq but Howard did not, and 17 stories in which Bush and Howard appeared but Blair did not. I also looked at coverage of the efforts at a second UN resolution in particular, a key period in the build-up to the war: none of the coverage in these newspapers featured all three leaders, 27 mentioned Bush and Blair but not Howard, and 1 mentioned Bush and Howard but not Blair. Thus evidence from third country newspaper coverage supports the contention that the association of the war with Bush and Blair was much stronger than with Howard.

If that is the case, one would also expect that Howard’s approval ratings would show a different pattern because the public would not be responding to the same events—they would not be as salient to perceptions of the incumbent—during the onset of war and its aftermath. Figure A1, below, replicates Figure 1 of the paper and adds net satisfaction ratings for John Howard over the same period (plotted on the same y-axis as the ratings for Tony Blair). It clearly shows that, as expected, Howard’s satisfaction ratings were not as strongly correlated as Bush and Blair’s, neither after 9/11 up to the invasion of Iraq, nor following the invasion. In contrast to the correlation of .88 in the Bush and Blair series, Howard’s ratings over this period correlated at .10 with Bush and at .15 with Blair. There was a slightly stronger correlation after March 2003, .37 with Bush and .30 with Blair, but it is still far weaker than for the other two leaders and also further implies that 9/11 did not resonate in the same way with the Australian public.

Beyond the much weaker evidence for comparability with Australia there are also data differences in terms of timing and survey questions. First, the Australian Election Study (AES) in 2004 was a single post-election survey, which means that in contrast to the US and UK cases, where all the variables other than vote choice were measured before the election, perceptions of the qualities of the leaders, approval of the war in Iraq, party identification, vote choice etc. were all measured at the same time. This is undesirable for the model in terms of exogeneity and the reliability of causal relationships. Second, while the AES asked about approval of the war in Iraq, it did not ask about approval of the war on terror or of the incumbent government’s handling of the economy.

These reservations and the limitations with the data lead me to exclude Australia as a case in the paper but, nevertheless, I estimated parallel models for Australia to those in Tables 2-5 of the main paper for the US and Britain for this Appendix. I employed a measure of agreement that Australia should provide military assistance for the war on terror instead of approval of the war on terror and retrospective judgments of the state of the national economy compared to one year earlier in place of approval of the incumbent government’s handling of the economy. The AES also asked about perceptions of the major party leaders’ strength, knowledge, honesty, and trustworthiness, as well as overall feelings toward them and vote choice.

The results are supportive of the account in the paper in that: approval of the war in Iraq was positively related to perceptions of John Howard’s qualities as a leader, including his strength—it was a performance issue; perceptions of strength were a strong influence on overall feelings toward Howard, while perceptions of his knowledge were statistically significantly (at p<.06) less influential; total effects of approval of the war on vote choice were comparable in size to the US—increasing the probability of voting Liberal by about 32 points and decreasing the probability of voting Labor by about 20 points; the total effects for Labor (left-wing) party identifiers who approved of the war were to make them less likely to vote Labor and more likely to vote for the Liberal (right-wing) incumbent government and, also as in the US, the effects of approval of the war on the probability of voting for the left-wing party were much larger, and negative, for left-wing than for right-wing party identifiers; and Liberal identifiers were much more likely to approve of the war (81-19) than Labor identifers (22-78). The results also show a difference, however, in that, unlike the right-wing incumbent leader of the US, approval of the war in Iraq was as influential on perceptions of Howard’s relative honesty and trustworthiness as on perceptions of his strength (none of the differences were statistically significant in Wald tests).

None of these findings changed if I focused on defection from the vote in 2001 (following McAllister and Bean 2006, Table 1), or if I controlled for other issues such as perceptions of changes in health, education, taxation, prices and unemployment since 2001.
Figure A1: Approval/Satisfaction with George W. Bush, Tony Blair, and John Howard from October 2001-May 2005

US election

UK election

Saddam Hussein captured

War in Iraq begins

1   2   3

The database is protected by copyright © 2016
send message

    Main page