ArticlePDF Available

Abstract and Figures

New techniques of ecological inference are utilized to estimate with confidence intervals francophone support in each federal electoral district in Quebec for the pro-sovereignty side in the 1993 and 1997 Canadian general elections and the 1992 and 1995 referenda. Analyzing the link between demographic and political contextual variables and support for the sovereignty of Quebec suggests that demographic factors, such as the proportion of farmers and government workers, influence francophone voting behaviour more often than political factors such as incumbency. Unlike in many other countries with ethnically based movements, francophone support for sovereignty actually rises as the francophone portion of the population rises. This finding indicates that the contact hypothesis probably applies to the Quebec sovereignty movement.
Content may be subject to copyright.
Context and Francophone Support
for the Sovereignty of Quebec:
An Ecological Analysis
David Lublin American University
D. Stephen Voss University of Kentucky
In the  nal decades of the twentieth century, and on into the twenty-
 rst, Quebec’s status has been at the centre of political debate in
Canada, yet mass opinion on the issue remains dif cult to gauge. The
same citizenry who only narrowly failed to pass a sovereignty referen-
dum in 1995, for example, routinely express overwhelming satisfac-
tion with life in Canada—two messages that seem dif cult to recon-
cile.1Nor is public inconsistency entirely to blame. Politicians and
activists often obfuscate the meaning of sovereignty, since opinion
shifts depending upon how this complex issue is framed.2The stream
1 Maurice Pinard, ‘‘Les déterminants psychosociaux,’in Maurice Pinard, Robert
Bernier and Vincent Lemieux, eds., Un combat inachevé (Sainte-Foy: Presses de
l’Université du Québec, 1997), 338-45.
2 Earl H. Fry, Canada’s Unity Crisis: Implications for U.S.-Canadian Economic
Relations (New York: Twentieth Century Fund, 1992), 28; Richard Johnston,
André Blais, Elisabeth Gidengil and Neil Nevitte, The Challenge of Direct
Democracy: The 1992 Canadian Referendum (Montreal: McGill-Queen’s Uni-
versity Press, 1996), 196-97; and Hudson Meadwell, ‘‘The Politics of National-
ism in Quebec,’’ World Politics 45 (January 1993), 225.
Acknowledgement: We are grateful for funding provided by the Government of
Canada under the Canadian Embassy’s Faculty Research Grant Program. We thank
Munroe Eagles, George Krause, Scott Piroth, Candace Redden, Pierre Serré and the
anonymous reviewers for this Journal for their help and comments on earlier ver-
sions of this manuscript. Of course, all conclusions and errors should be attributed
solely to the authors.
David Lublin, Department of Gove r m e n t , School of Public Affa i r s , American Unive r s i t y,
4400 Massachusetts Ave . , N.W., Washington, D.C., USA 20016;
D. Stephen Voss, Department of Political Science, 1615 Patterson Of ce Tow e r,
Unive r s i t y of Kentucky, Lex i n g t o n , Ke n t u c ky, USA 40506; dsvoss@uky. p o p . e d u
Canadian Journal of Political Science / Revue canadienne de science poli tique
35:1 (March/Mars 2002) 75-101
© 2002 Canadian Political Science Association (l’Association canadienne de science politique)
and / et la Société québécoise de science politique
of ambiguous public information has taken its toll. Many voters in
Quebec’s sovereignty referendum in 1995, for example, wrongly
thought that Quebec would remain part of Canada and continue to
send representatives to the federal parliament in Ottawa if it passed.3
In light of this sort of confusion and uncertainty, it is not entirely clear
what any particular expression of preferences really means.
Super cially, the sove r e i g n t y move m e n t might appear a prime
ex a m p l e of ‘‘symbolic politics, of the sort found to dominate ethnic
relations in the United States.4Supporters believe that continued surviva l
of the French fact in North America necessitates separation of Quebec
from the rest of Canada, so the desire to weaken political ties inevitably
invo kes a voter’s perception of threat, around which much work in the
social psychology of inter-group relations revo l ves.5If desire for sove r-
eignty is primarily a psychological phenomenon, demand for sove r e i g n -
ty would respond to the character and upbringing of individual franco-
phone voters, and self-interest would play little role.6
Interestingly, contextual effects heretofore have received little
attention despite the likely relationship between geographic context
and self-interest. If self-interest matters, then context would have a
disparate impact on support for sovereignty across Quebec, with the
costs and bene ts structured geographically. Action based upon self-
interest rather than on symbolic politics would not be particularly dif-
 cult for self-regarding voters, should such exist, and would reveal
itself in contextual voting patterns. Thus Quebec provides a valuable
opportunity to test theories of ethnic con ict outside the arenas in
which they were developed.7Determining which contexts enhance
francophone desire for sovereignty will also make a useful contribu-
3 Pinard, ‘‘Les déterminants psychosociaux,’’ 345-53.
4 It is worth noting, however, that some scholars question whether the symbolic
politics model even applies to the US case. See Lawrence Bobo, ‘‘Group Con-
 ict, Prejudice, and the Paradox of Contemporary Racial Attitudes,’in Phylis A.
Katz and Dalmas A. Taylor, eds., Eliminating Racism: Pro les in Controversy
(New York: Plenum, 1988), 85-114; and D. Stephen Voss, ‘‘The Rational Basis
of Symbolic Racism,’ paper presented at the annual meeting of the Midwest
Political Science Association, Chicago, 1999.
5 Fry, Canada’s Unity Crisis, 28-34; Thomas F. Pettigrew, ‘‘New Black-White Pat-
terns: How Best to Conceptualize Them?’’ Annual Review of Sociology 11
(1985), 329-46.
6 Donald R. Kinder and Lynn M. Sanders, Divided by Color: Racial Politics and
Democratic Ideals (Chicago: University of Chicago Press, 1996).
7 For the main exception, see Pierre Drouilly, Indépendance et démocratie:
Sondages, élections et référendums au Quebec 1992-1997 (Montréal: Harmattan,
1997). Paul Nesbitt-Larking has examined the impact of context on support for
protest parties in English Canada (‘‘The 1992 Referendum and the 1993 Federal
Election in Canada: Patterns of Protest,paper presented at the annual meeting
of the Canadian Political Science Association, Calgary, 1994).
76 David Lublin and D. Stephen Voss
Abstract. New techniques of ecological inference are utilized to estimate with con -
dence intervals francophone support in each federal electoral district in Quebec for the
pro-sovereignty side in the 1993 and 1997 Canadian general elections and the 1992 and
1995 referenda. Analyzing the link between demographic and political contextual vari-
ables and support for the sovereignty of Quebec suggests that demographic factors, such
as the proportion of farmers and government workers, in uence francophone voting
behaviour more often than political factors such as incumbency. Unlike in many other
countries with ethnically based movements, francophone support for sovereignty actually
rises as the francophone portion of the population rises. This  nding indicates that the
contact hypothesis probably applies to the Quebec sovereignty movement.
Résumé. Les nouvelles techniques d’inférence écologique sont utilisées ici pour esti-
mer avec  abilité les écarts de l’appui francophone au camp souverainiste, dans les cir-
conscriptions électorales fédérales du Québec, lors des élections générales canadiennes
de l993 et l997 et les référendums canadien de l992 et quebécois de l995. L’analyse de la
relation entre les variables contextuelles démographiques et politiques d’une part, et l’ap-
pui à la souveraineté du Québec d’autre part, suggère que les facteurs démographiques,
tels la proportion d’agriculteurs et de fonctionnaires, in uencent plus souvent le compor-
tement électoral des francophones que les facteurs politiques tels les mandats politiques.
Contrairement à ce qui se passe dans plusieurs pays où existent des mouvements eth-
niques, l’appui des francophones à la souveraineté augmente lorsque le poids démogra-
phique des francophones croît. Ce constat indique que l’hypothèse du contact s’applique
probablement au cas du mouvement souverainiste québécois.
tion to the more general question of which social conditions exacer-
bate cultural con ict.
The Quebec case is especially useful because the sove r e i g n t y issue
wa s implicated in seve r a l elections. Quebeckers had an unusually high
number of opportunities to express themselves on sove r e i g n t y at the
polls between 1992 and 1997: the referendum on the constitutional
Charlottetown Accord in 1992, against which sove r e i g n i s t s campaigned;
the federal elections of 1993 and 1997 contested by the sove r e i g n i s t
Bloc Québécois; and the 1995 sove r e i g n t y referendum promoted by the
Pa r t i Québécois.8The rapid recurrence of this issue in so many guises
aids an analysis of voter preferences, since the noise in any one vote
wa s h e s out, allowing a concentration on patterns that remain constant.
Supporters and opponents agree that sovereignty has certain clear
implications for Quebec. In particular, it means that the Quebec gov-
ernment will be strengthened, and the in uence of the federal govern-
ment in Ottawa reduced, if not eliminated. Quebec City, not Ottawa,
will have the  nal say over public policy. Attentive voters should be
able to weigh the stakes. Any systematic pattern found in all four elec-
tion results should re ect voter impulses on this federalism issue,
8 Jonathan Lemco, Turmoil in the Peaceable Kingdom: The Quebec Sovereignty
Movement and Its Implications for Canada and the United States (Toronto: Uni-
versity of Toronto Press, 1994), 49-53. Quebeckers could also support sovereign-
ty by voting for the PQ in the 1994 provincial elections for the Assemblée
nationale. The analysis excludes this election because of the absence of data.
rather than any intervening source of variation such as partisan loyal-
ties, strategic voting or ballot confusion. This article therefore takes
advantage of returns from the four campaigns to analyze links between
local context and support for sovereignty within Quebec.
Obviously we cannot study patterns of francophone voting unless
we have a means to determine how francophones voted. For this pur-
pose, we use an ecological inference technique that can estimate vot-
ing behaviour by primary language, not just for all of Quebec but also
for each of its ridings.9These estimates allow us to concentrate on
variation in francophone voting, as required for an understanding of
contextual group relations.10
Our results reve a l that the demographic character of a locality con-
sistently in uences sove r e i g n t y support leve l s . They imply that self-
interest drive s vo t i n g behaviour on sove r e i g n t y, at least when interest is
measured contextually; ethnicity does not trump all other personal con-
cerns. Howeve r, a community’s ethnic balance does shape francophone
attitudes. Voters show less sympathy with sove r e i g n t y when they reside
in ridings with a linguistic mix, consistent with the ‘‘contact hypothesis’
that others have applied to Canada.11 These results validate work by
other scholars, particularly Pierre Drouilly, using different methods
9 Gary King, A Solution to the Ecological Inference Problem: Reconstructing Indi-
vidual Behavior from Aggregate Data (Princeton: Princeton University Press,
1997); and D. Stephen Voss and David Lublin, ‘‘Ecological Inference and the
Comparative Method,’ APSA-CP: Newsletter of the APSA Organized Section in
Comparative Politics 9 (1998), 25-31.
10 Most geographic comparisons of support for sovereignty do not control for eth-
nicity, so it is dif cult to determine if ethnicity or other factors explain regional
differences in support for sovereignty. See Leo Driedger, Multi-Ethnic Canada,
Identities and Inequalities (Toronto: Oxford University Press, 1996). That is,
aggregation con ates ethnicity with other contextual in uences on the vote. We
focus on francophones, and strip out anglophones and allophones, since they
almost uniformly oppose sovereignty, and there is not sufcient geographical
variation for contextual study. See André Blais and Richard Nadeau, ‘‘To Be or
Not to Be Sovereignist: Quebeckers’ Perennial Dilemma,’Canadian Public Pol-
icy 18 (1992), 90-91; Drouilly, Indépendance et démocratie, 286-92; Fry,
Canadas Unity Crisis, 31-34, 39; and Maurice Pinard, ‘‘Le contexte politique et
les dimensions sociodémographiques,’’ in Maurice Pinard, Robert Bernier and
Vincent Lemieux, eds., Un combat inachevé (Sainte-Foy: Presses de l’Université
du Québec, 1997), 307-13. Drouilly and Pinard estimate that support for sover-
eignty among anglophones and allophones was not greater than 5-10 per cent in
either 1980 or 1995. Most First Nations Canadians opposed sovereignty, though
approximately one quarter of the Abénaquis and Hurons, who are more likely to
speak French as their second language, voted in favour.
11 See, for example, J. W. Berry, ‘‘Cultural Relations in Plural Societies: Alterna-
tives to Segregation and Their Sociopsychological Implications,’ in Norman
Miller and Marilynn B. Brewer, eds., Groups in Contact: The Psychology of
Desegregation (Orlando: University Press of America, 1984), 11-27.
78 David Lublin and D. Stephen Voss
and alternate levels of aggregation. The results contradict the popular
‘‘power threat hypothesis,’’ a common approach to inter-group rela-
tions that suggests proximity should fuel group con ict.12
Estimating Francophone Support for Sovereignty
Four campaigns between 1992 and 1997 called on Quebec’s voters to
express their preferences on sovereignty. Adoption of the Charlotte-
town Accord in 1992 would have signaled acceptance of the 1982
Constitution Act and foreclosed sovereignty as an option, although not
ev eryone opposing the Accord supported sovereignty.13 The 1993 and
1997 federal elections also had major implications for sovereignty,
ev en if they were not directly on the issue. Some non-sovereignists
undoubtedly supported the BQ, perhaps to place pressure on Ottawa,
but defeat for the BQ at the polls would have represented a major set-
back for the sovereignty movement.14 The 1995 referendum, mean-
while, dealt with sovereignty directly, even if voters did not always
understand precisely in what manner.15 Voting behaviour in each elec-
tion therefore might include sources of variation other than support for
political separation from Canada, but these differed in each election,
so any pattern that persists in all four votes presumably re ects the
process governing support for sovereignty itself.
Actual ballots are secret, so we cannot identify the demographic
characteristics of each voter. Analyzing the effect of context on fran-
cophone voting requires some method of estimation, and no approach
is perfect. Surveys might seem to be the natural choice, since gener-
ally one knows the personal characteristics of each respondent. How-
ev er, they also present clear disadvantages. Leaving aside the usual
complications with question wording, interviewer effects and response
bias, surveys also do not facilitate contextual studies because they con-
tain too few voters from each electoral district, creating excessive
sampling error within any one context.16 Contextual studies using sur-
12 Micheal W. Giles and Arthur S. Evans, ‘‘The Power Approach to Intergroup Hos-
tility,’Journal of Con ict Resolution 30 (1986), 469-86.
13 Johnston et al., The Challenge of Direct Democracy, 192-93.
14 Blais and Nadeau, ‘‘To Be or Not to Be Sovereignist,90; and Donald Charette,
‘‘Le choix est clair:’ Pour Duceppe, le résultat de l’élection vient d’anéantir les
espoirs des fédéralistes,’’ Le Soleil (Quebec City), June 3, 1997.
15 Pinard, ‘‘Les déterminants psychosociaux,’’ 343-53.
16 Researchers who analyze data on support for sovereignty have been laudably
forthcoming about the limitations of survey data. Blais and Nadeau provide a
valuable discussion of the complications (‘‘To Be or Not to Be Sovereignist,’
90-91). It is worth noting that, aside from the methodological dif culties dis-
cussed elsewhere, aggregate data studies suffer real substantive limitations. They
cannot include variables such as gender, lacking real geographical variation. Nor
Context and Francophone Support for the Sovereignty of Quebec 79
veys are forced to fall back upon multilevel models, which pose
known drawbacks, including steep data demands that usually rule out
complex multivariate models of the sort we need here.17 For this rea-
son, we chose to use aggregate data for our contextual analysis.
Aggregation Bias and Contextual Research
We do not know how francophones voted, either within individual rid-
ings or in Quebec as a whole. On the other hand, we do know overall
voter preferences in each of 75 federal ridings, as well as the linguistic
breakdown. The challenge of ecological inference is to estimate how
each linguistic group behaved according to how riding support for
sovereignty varied with a group’s share of the population.18 This is a
risky venture, since an unfortunate pattern of aggrega t i o n could produce
deceptive correlations between group density and voting behaviour.19
However, an ecological inference programme developed by Gary
King, called EI, surmounts most of the pitfalls common to ecological
inference. King’s approach allows voting estimates by linguistic cate-
gory not only for Quebec as a whole, but also for individual federal
ridings, and with con dence intervals around each estimate.20 EI
avoids imposing speci c assumptions about how any group behaved,
either within or across ridings, instead only requiring distributional
assumptions about the process governing voting. It is able to handle
can they explore the correlations among distinct political attitudes, since each
persons opinions are con ated into a single vote.
17 Anthony S. Bryk and Stephen W. Raudenbush, Hierarchical Linear Models:
Applications and Data Analysis Methods (Newbury Park: Sage, 1992); and
Marco R. Steenbergen and Bradford S. Jones, ‘‘Modeling Multilevel Data Struc-
tures,’’ paper presented at the annual meeting of the Political Methodology
Group, Columbus, Ohio, 1997.
18 Although some non-sovereignists vote for the BQ and some sovereignists voted
for other parties, ‘‘support for sovereignty’ is used throughout as a shorthand for
support for the pro-sovereignty position in each of the four elections.
19 Christopher H. Achen and W. Phillips Shively, Cross-Level Inference (Chicago:
University of Chicago Press, 1995); Brad Palmquist, ‘‘Ecological Inference,
Aggregate Data Analysis of U.S. Elections, and the Socialist Party of America
(unpublished Ph.D. dissertation, University of California at Berkeley, 1993); and
W. S. Robinson, ‘Ecological Correlations and the Behavior of Individuals,’
American Sociological Review 15 (1950), 351-57.
20 The choice of federal ridings as our lower-level unit of aggregation is not obvi-
ous, since the 1995 referendum was conducted using provincial ridings. How-
ev er, using electoral districts is only a data convenience, a proxy for the swirl of
local interests and in uences that would affect someone’s vote. Any approach
naturally introduces some noise, and there is no particular reason why the elec-
torally relevant riding would capture the demographic context better than an elec-
torally irrelevant one. More practically, data for all demographic variables in the
analysis were not available for the provincial ridings.
80 David Lublin and D. Stephen Voss
the greater across-riding variation among francophone preferences
compared to those of other Quebeckers (that is, heteroskedasticity). It
prevents absurd estimates of behaviour, such as rates of support for
sovereignty that fall below 0 per cent, that other forms of estimation
would  nd.21
The most important advance of King’s method, though, is that it
is partially robust to ‘‘aggrega t i o n bias.’If individual voting behaviour
is somehow related to a community’s linguistic composition, as most
theories of inter-group relations would predict, then other popular
forms of ecological inference (for example, Goodman’s ecological
regression, homogeneous unit analysis) would be facially invalid.
Given our speci c interest in contextual effects, those other methods
would also be useless unless signi cantly altered, and at any rate
should not perform as well as King’s solution.22
The key to King’s success derive s from the ‘‘method of bounds.’’
Fo r each riding, EI begins by identifying the complete set of values that
might describe voting behaviour. The obvious  rst limit is that turnout
rates for each linguistic group must fall between 0 per cent and 100 per
cent, but the range of possible values is actually much narrower. Using
Quebec itself as an example, 4.7 million people voted in 1995, and only
1.1 million people in Quebec are members of linguistic minorities, so at
least 3.6 million francophones must have voted. That is, the ‘‘lower
bound’’ on francophone turnout is about 61.5 per cent. The upper bound
is 81.1 per cent. Even if no linguistic minorities voted, at most 4.7 mil-
lion of the 5.8 million francophones could have turned out.
Such bounds, when applied to each riding, do a great job constrain-
ing voting estimates. Furthermore, if the range of possible values shifts
as the linguistic context shifts, there is strong evidence of aggrega t i o n
bias—that is, strong evidence that francophone voting responds to the
21 We hav e evaluated the bene ts of King’s solution much more thoroughly else-
where. See D. Stephen Voss, ‘‘Familiarity Doesn’t Breed Contempt: The New
Geography of Racial Politics’’ (unpublished Ph.D. dissertation, Harvard Univer-
sity, 2000), chap. 7; and D. Stephen Voss and David Lublin,‘Black Incumbents,
White Districts: An Appraisal of the 1996 Congressional Elections,’’ American
Politics Quarterly 29 (2001), 141-82. We also provide much more detail on how
his method would be applied to Quebec, showing how other methods of ecologi-
cal inference produce impossible estimates and how King’s approach clearly cap-
tures aggregation bias that those other methods do not (Voss and Lublin, ‘‘Eco-
logical Inference and the Comparative Method’’).
22 Achen and Shively, Cross-Level Inference; James E. Alt, ‘Persistence and
Change in Southern Voter Registration Patterns, 1972-1990,’’ paper presented to
the Workshop on Race, Ethnicity, Representation and Governance, given by the
Center for American Political Studies of Harvard University Government Depart-
ment, Cambridge, Massachusetts, 1993; Voss, Familiarity Doesn’t Breed Con-
tempt, chap. 7; and Voss and Lublin, ‘‘Ecological Inference and the Comparative
Context and Francophone Support for the Sovereignty of Quebec 81
linguistic mix of the riding.23 King’s EI is able to pick up the pattern
and apply it when computing voting estimates for each riding, preserv-
ing at least some of the meaningful contextual variation. Having to
take into account this additional complication naturally adds uncer-
tainty to the riding estimates, but we can evaluate how much once the
estimates are in hand. Furthermore, to whatever extent EI consistently
underestimates aggregation bias,24 it would only weaken the apparent
contextual effects, not create signi cant contextual  ndings.
Ecological Estimates for Four Campaign Outcomes
We performed an ecological analysis for each of the four contests,
using data on the proportion of francophones in the population from
the 1991 Census.25 We prompted the software to model aggregation
bias, where present, so that the estimates would be sensitive to contex-
tual voting among francophones.26
Table 1 shows the estimated proportion of francophones voting
against the Charlottetown Accord in the 1992 referendum and for the
Bloc Québécois in the 1993 federal election, as well as con dence
intervals around these estimates for the riding boundaries used in
1993. Table 2 similarly presents estimates and con dence intervals
around these estimates of proportion of francophones voting for sover-
eignty in the 1995 referendum and for the BQ in the 1997 general
election in the riding boundaries in place in 1997.27 The size of the
con dence intervals around the estimates is generally very small. The
median size of 95 per cent con dence intervals for the federal elec-
tions is .013 for 1993 and .018 for 1997. The median size of the same
con dence intervals for the referenda is .025 in 1992 and .018 in 1995.
Although the con dence intervals for some ridings are considerably
larger, the majority of ridings in all elections have very small con -
dence intervals that permit a high degree of certainty about estimated
francophone support for the pro-sovereignty position.
23 It does not provide evidence, however, that context causes the shifts.
24 Wendy Tam Cho, ‘‘Iff the Assumption Fits . . . : A Comment on the King Eco-
logical Inference Solution,’’ Political Analysis 7 (1998), 143-63.
25 Munroe Eagles, James P. Bickerton, Alain-G. Gagnon and Patrick Smith, The
Almanac of Canadian Politics (Toronto: Oxford University Press, 1995); Pro le
of Federal Electoral Districts—Parts A and B (Ottawa: Statistics Canada, 1992).
26 Exact globals were, for 1993’s rst stage, _Eeta=1, _EalphaB=(0˜.15). For 1992
we added _Estval=(0|0|-2|-2|-2|0). The second stage for 1995 also added _Eis-
fac=2. For 1997s second stage we used _Eeta=3, _EalphaB=(0˜.15), _Eal-
phaW=(0˜.15), _Estval=(-0.2|0|-1|.1|-2|-2|-.2). The remaining analyses used basic
EI. Replication buffers are available from the authors.
27 No data are available on the results of the 1992 referendum in the 1997 riding
boundaries. Similarly, data are unavailable on the results of the 1995 referendum
or the 1997 federal election within the 1993 boundaries.
82 David Lublin and D. Stephen Voss
Table 1
Proportion of Francophones Supporting the BQ in 1993 and Opposing the 1992 Charlottetown Accord in the
Riding Boundaries of 1993
For the BQ in 1993 Opposed to the Charlottetown Accord 1992
French Home
95% CI Lower
Bound Estimate
95% CI Upper
95% CI Lower
Bound Estimate
95% CI Upper
Abitibi .840 .541 .548 .550 .740 .762 .773
Ahuntsic .720 .605 .621 .626 .618 .639 .654
Anjou-Rivière-des-Prairies .728 .576 .588 .592 .591 .624 .640
Argenteuil-Papineau .879 .534 .537 .538 .649 .661 .668
Beauce .995 .364 .364 .364 .551 .552 .552
Beauharnois-Salaberry .869 .583 .589 .591 .667 .679 .685
Beauport-Montmorency-Orléans .991 .582 .582 .582 .676 .677 .677
Bellechasse .997 .408 .408 .408 .578 .580 .579
Berthier-Montcalm .962 .631 .633 .633 .676 .680 .682
Blainville-Deux-Montagnes .903 .652 .656 .658 .710 .718 .723
Bonaventure-Îles-de-la-Madeleine .871 .380 .383 .385 .550 .561 .567
Bourassa .786 .525 .532 .534 .598 .618 .629
Brome-Missisquoi .784 .507 .517 .520 .587 .600 .609
Châteauguay .825 .690 .700 .703 .728 .751 .764
Chambly .904 .654 .659 .660 .677 .685 .690
Champlain .970 .502 .503 .503 .656 .659 .660
Charlesbourg .984 .603 .603 .604 .661 .662 .663
Charlevoix .972 .636 .637 .638 .687 .690 .691
Chicoutimi .990 .645 .645 .645 .739 .740 .740
Drummond .986 .556 .557 .557 .642 .643 .644
Context and Francophone Support for the Sovereignty of Quebec 83
Table 1 (continued)
For the BQ in 1993 Opposed to the Charlottetown Accord 1992
French Home
95% CI Lower
Bound Estimate
95% CI Upper
95% CI Lower
Bound Estimate
95% CI Upper
Frontenac .986 .592 .592 .592 .622 .623 .624
Gaspé .933 .482 .484 .484 .629 .633 .637
Gatineau-La Lièvre .905 .387 .389 .390 .457 .465 .468
Hochelaga-Maisoneuve .900 .677 .681 .682 .719 .731 .737
Hull-Aylmer .766 .340 .352 .355 .484 .496 .504
Joliette .988 .670 .671 .671 .726 .727 .728
Jonquière .990 .682 .683 .683 .783 .784 .785
Kamouraska-Rivière-du-Loup .998 .530 .530 .530 .627 .627 .627
Lévis .991 .620 .621 .621 .701 .702 .702
La Prairie .706 .586 .603 .610 .613 .641 .658
Lac-Saint-Jean .997 .758 .758 .758 .773 .773 .773
Lachine-Lac-Saint-Louis .366 .498 .539 .556 .589 .660 .725
LaSalle-Émard .550 .562 .594 .605 .624 .689 .729
Laurentides .961 .630 .631 .632 .670 .674 .675
Laurier-Sainte-Marie .812 .749 .757 .761 .774 .798 .815
Laval Centre .852 .642 .646 .648 .664 .675 .683
Laval Est .891 .578 .581 .582 .631 .639 .645
Laval Ouest .689 .607 .619 .624 .607 .644 .662
Longueuil .925 .710 .713 .714 .724 .734 .738
Lotbinière .993 .542 .542 .542 .632 .632 .632
Louis-Hébert .961 .578 .549 .580 .640 .643 .645
Mégantic-Compton-Stanstead .856 .516 .520 .522 .562 .572 .578
Manicouagan .788 .686 .694 .698 .789 .814 .831
84 David Lublin and D. Stephen Voss
Matapédia-Matane .996 .575 .575 .575 .670 .671 .671
Mercier .915 .638 .642 .644 .711 .721 .726
Mont-Royal .226 .249 .291 .308 .449 .497 .520
Notre-Dame-de-Grâce .254 .442 .529 .558 .604 .701 .812
Outremont .552 .633 .667 .677 .665 .732 .783
Papineau-Saint-Michel .652 .580 .596 .601 .619 .656 .678
Pierrefonds-Dollard .349 .427 .480 .498 .552 .632 .677
Pontiac-Gatineau-Labelle .755 .433 .442 .446 .599 .620 .631
Portneuf .980 .546 .547 .547 .666 .667 .668
Québec .965 .556 .557 .558 .644 .647 .649
Québec-Est .986 .604 .604 .605 .670 .672 .672
Richelieu .989 .672 .672 .672 .644 .646 .646
Richmond-Wolfe .956 .545 .546 .547 .625 .628 .630
Rimouski-Témiscouata .997 .601 .601 .601 .672 .672 .672
Roberval .987 .607 .608 .608 .681 .682 .683
Rosemont .868 .721 .724 .726 .733 .746 .755
Saint-Denis .491 .678 .726 .745 .691 .776 .855
Saint-Henri-Westmount .402 .429 .457 .470 .596 .667 .724
Saint-Hubert .807 .692 .701 .705 .710 .731 .745
Saint-Hyacinthe-Bagot .992 .578 .579 .579 .625 .626 .626
Saint-Jean .958 .581 .583 .584 .645 .648 .650
Saint-Léonard .544 .480 .496 .503 .558 .592 .613
Saint-Laurent-Cartierville .436 .386 .412 .421 .478 .570 .549
Saint-Maurice .992 .408 .408 .408 .653 .654 .654
Shefford .965 .575 .577 .577 .624 .627 .629
Sherbrooke .928 .405 .408 .408 .617 .624 .627
Témiscamingue .952 .583 .584 .585 .673 .677 .679
Terrebonne .972 .707 .709 .709 .747 .749 .751
Trois-Rivières .988 .540 .540 .541 .634 .635 .636
Context and Francophone Support for the Sovereignty of Quebec 85
Table 1 (continued)
For the BQ in 1993 Opposed to the Charlottetown Accord 1992
French Home
95% CI Lower
Bound Estimate
95% CI Upper
95% CI Lower
Bound Estimate
95% CI Upper
Vaudreuil .642 .560 .593 .601 .621 .652 .676
Verchères .976 .689 .689 .690 .702 .704 .705
Verdun-Saint-Paul .685 .594 .613 .620 .664 .697 .718
86 David Lublin and D. Stephen Voss
Table 2
Proportion of Francophones Supporting the BQ in 1993 and 1997 and Supporting Sovereignty in 1995
For the BQ in 1993 For the BQ in 1997 For Sovereignty in 1995
95% CI
Bound Estimate
95% CI
95% CI
Bound Estimate
95% CI
95% CI
Bound Estimate
95% CI
Abitibi .814 .494 .534 .559 .392 .434 .455 .650 .656 .659
Ahuntsic .689 .501 .571 .619 .354 .418 .455 .539 .550 .555
Anjou-Rivière-des-Prairies .734 .500 .562 .596 .347 .407 .440 .542 .552 .556
Argenteuil-Papineau .880 .497 .519 .531 .418 .447 .464 .565 .568 .570
Beauce .995 .363 .364 .365 .266 .267 .268 .439 .439 .439
Beauharnois-Salaberry .874 .530 .566 .582 .403 .436 .452 .417 .420 .421
Beauport-Montmorency-Orléans .991 .581 .583 .584 .431 .434 .435 .566 .566 .566
Bellechasse-Etchemins .997 .409 .410 .410 .333 .333 .334 .463 .463 .463
Berthier-Montcalm .960 .638 .647 .652 .529 .542 .548 .655 .656 .657
Bourassa .773 .459 .505 .532 .351 .396 .421 .519 .529 .532
Brome-Missisquoi .784 .469 .512 .536 .289 .331 .352 .517 .523 .527
Brossard-La Prairie .721 .512 .594 .632 .328 .419 .451 .544 .554 .558
Chambly .940 .638 .660 .669 .493 .514 .524 .628 .630 .631
Champlain .971 .492 .499 .502 .440 .449 .453 .611 .612 .612
Charlesbourg .984 .603 .606 .608 .385 .389 .390 .552 .553 .553
Charlevoix .972 .623 .633 .638 .545 .553 .557 .684 .685 .685
Châteauguay .819 .613 .672 .704 .469 .522 .553 .666 .673 .676
Chicoutimi .989 .641 .644 .645 .428 .431 .432 .687 .687 .687
Compton-Stanstead .852 .486 .525 .542 .332 .367 .384 .538 .542 .544
Drummond .986 .552 .556 .557 .423 .426 .428 .581 .581 .582
Frontenac-Mégantic .986 .571 .574 .576 .371 .375 .376 .517 .517 .518
Bonaventure-Gaspé .898 .420 .446 .456 .401 .442 .458 .618 .622 .624
Gatineau .912 .356 .376 .384 .202 .221 .228 .305 .307 .308
Hochelaga-Maisonneuve .868 .580 .621 .639 .462 .506 .524 .627 .630 .632
Context and Francophone Support for the Sovereignty of Quebec 87
Table 2 (continued)
For the BQ in 1993 For the BQ in 1997 For Sovereignty in 1995
95% CI
Bound Estimate
95% CI
95% CI
Bound Estimate
95% CI
95% CI
Bound Estimate
95% CI
Hull-Aylmer .765 .306 .339 .354 .218 .250 .267 .322 .328 .331
Joliette .991 .660 .662 .664 .466 .468 .469 .664 .664 .664
Jonquière .990 .679 .682 .683 .486 .489 .490 .717 .717 .717
Kamouraska-Rivière-du-Loup .998 .528 .529 .529 .383 .383 .384 .568 .568 .568
Notre-Dame-de-Grâce-Lachine .379 .401 .528 .606 .247 .368 .422 .544 .583 .600
Lac-Saint-Jean .996 .756 .758 .758 .635 .636 .636 .744 .744 .744
Lac-Saint-Louis .257 .327 .409 .461 .343 .544 .827 .349 .389 .408
LaSalle-Émard .564 .446 .549 .601 .294 .379 .431 .567 .588 .597
Laurentides .960 .616 .627 .633 .458 .471 .477 .609 .610 .610
Laurier-Sainte-Marie .843 .648 .730 .764 .533 .610 .647 .744 .749 .752
Laval Centre .852 .582 .625 .647 .384 .445 .464 .592 .596 .598
Laval Est .891 .540 .567 .582 .394 .419 .432 .556 .559 .561
Laval Ouest .689 .516 .579 .621 .292 .401 .443 .559 .572 .577
Lévis .991 .620 .622 .623 .452 .455 .456 .592 .592 .592
Longueuil .929 .674 .696 .707 .512 .531 .540 .647 .649 .650
Lotbinière .991 .553 .555 .556 .370 .373 .374 .531 .532 .532
Louis-Hébert .961 .566 .575 .580 .401 .411 .415 .540 .541 .541
Manicouagan .826 .577 .651 .682 .485 .539 .568 .756 .763 .765
Matapédia-Matane .966 .491 .498 .501 .451 .460 .463 .611 .612 .613
Mercier .916 .606 .634 .644 .507 .536 .551 .650 .653 .654
Mont-Royal .223 .185 .269 .302 .087 .155 .182 .258 .284 .297
Outremont .570 .475 .604 .683 .350 .432 .490 .623 .638 .647
Papineau-Saint-Denis .540 .468 .587 .655 .346 .456 .526 .643 .664 .674
Pierrefonds-Dollard .341 .452 .686 .896 .184 .269 .312 .396 .427 .442
Pontiac-Gatineau-Labelle .770 .379 .422 .441 .337 .388 .414 .477 .487 .490
Portneuf .979 .542 .547 .549 .436 .442 .444 .555 .555 .555
88 David Lublin and D. Stephen Voss
Québec .967 .547 .554 .558 .445 .456 .460 .570 .571 .572
Québec-Est .986 .498 .601 .603 .395 .398 .399 .542 .542 .542
Repentigny .971 .700 .709 .714 .565 .575 .579 .687 .688 .688
Richelieu .988 .661 .665 .667 .548 .552 .553 .609 .610 .610
Richmond-Arthabaska .964 .525 .532 .536 .370 .378 .382 .573 .573 .574
Rimouski-Mitis .995 .611 .612 .613 .470 .472 .472 .643 .643 .643
Roberval .987 .604 .606 .607 .518 .521 .523 .672 .673 .673
Rosemont .826 .602 .680 .713 .455 .539 .573 .673 .679 .681
Témiscamingue .952 .567 .578 .584 .469 .483 .490 .587 .588 .589
Saint-Eustache-Saint-Thérèse .904 .617 .641 .653 .478 .501 .512 .646 .649 .650
Saint-Bruno-Saint-Hubert .846 .614 .656 .681 .464 .505 .530 .635 .640 .642
Saint-Hyacinthe-Bagot .992 .573 .575 .576 .428 .430 .431 .556 .557 .557
Saint-Jean .959 .571 .583 .589 .462 .475 .480 .578 .580 .580
Saint-Lambert .769 .555 .620 .656 .383 .464 .503 .610 .617 .620
Saint-Laurent-Cartierville .436 .239 .309 .338 .206 .258 .288 .353 .375 .386
Saint-Léonard-Saint-Michel .539 .356 .417 .447 .197 .271 .303 .435 .452 .460
Saint-Maurice .992 .413 .415 .415 .441 .442 .443 .568 .568 .568
Shefford .963 .555 .563 .566 .358 .369 .372 .546 .547 .547
Sherbrooke .928 .391 .404 .410 .302 .316 .323 .561 .563 .564
Terrebonne-Blainville .943 .661 .678 .686 .504 .525 .533 .663 .664 .665
Trois-Rivières .988 .532 .535 .536 .426 .430 .431 .571 .571 .571
Vaudreuil-Soulanges .752 .522 .586 .621 .349 .413 .442 .560 .566 .570
Verchères .976 .677 .685 .689 .540 .547 .550 .640 .641 .641
Verdun-Saint-Henri .700 .491 .563 .600 .372 .434 .472 .596 .605 .610
Westmount-Ville Marie .328 .278 .394 .441 .195 .296 .343 .440 .475 .493
Context and Francophone Support for the Sovereignty of Quebec 89
Some of the largest con dence intervals appear in ridings with a
mixed population. This is quite natural: we know the least about fran-
cophone voting in such places, since the presence of so many linguis-
tic minorities conceals in the totals what francophones are doing. Nev-
ertheless, EI has disclosed what information the bounds did provide
for these ridings, and otherwise borrowed strength from the other rid-
ings (and the trend in their bounds) to choose the most likely voting
estimates. Figures produced this way are never as strong as one might
like, but they are precise compared to what would be possible with a
typical opinion poll.28 Without a specialized (and costly) survey
intended to maximize contextual variation, EI is the best option avail-
Support for sovereignty varied considerably across Quebec rid-
ings in all campaigns. Estimated support among francophones for the
BQ ranged from 30 per cent in Mont Royal to 77 per cent in Laurier-
Sainte Marie in 1993, and from 15 per cent in Mont Royal to 64 per
cent in Lac Saint Jean in 1997. Francophones in the median riding
gave 59 per cent of their votes to the BQ in 1993 and 44 per cent in
1997. The standard deviation around these estimates was 10 per cent
in 1993 and 9 per cent in 1997. Francophone opposition to the Char-
lottetown Accord in 1992 ranged from 46 per cent in Gatineau-La
Lièvre to 78 per cent in Jonquière. The median level of opposition to
the Accord was 65 per cent with a standard deviation of 7 per cent.
Estimates indicate that 74 per cent of francophones in Lac Saint Jean
but only 24 per cent of francophones in Mont Royal voted for sover-
eignty in the referendum of 1995. Fifty-seven per cent of franco-
phones in the median riding supported sovereignty in 1995, with a
standard deviation of 11 per cent.
28 Consider that, in a random sample of 1,500 Quebeckers, only 42 would be fran-
cophones residing in the six west-Montreal ridings where they are a linguistic
minority. Whereas the EI 95 per cent con dence intervals in these ridings aver-
aged less than +/- 8 per cent in 1993 and 1995, the equivalent margin of error for
a proportion (p) computed from a survey of only 42 people is 2Ö` ```````
p(1 -p)/42 or
as much as +/- 16 percentage points for two standard errors; see Thomas H. Won-
nacott and Ronald J. Wonnacott,Introductory Statistics (5th ed.; New York: John
Wiley, 1990), 207. Controlling for other contextual variables would drive the
hypothetical sample size to minimal levels. Any generalization about franco-
phones in mixed ridings would thus be much less reliable with surveys, unless
they consciously oversampled that group, and when considered this way EI’s per-
formance is impressive.
90 David Lublin and D. Stephen Voss
Context and Support for Sovereignty
The contextual variables theorized to in uence support for sovereignty
are primarily demographic, although we also capture whether candi-
dates held of ce prior to the 1993 and 1997 elections. The demo-
graphic variables are aggregated characteristics of individuals in the
riding, such as their income or education. These demographics are
more than crude measures that parallel the individual characteristics
utilized in survey data analysis. Rather, we hypothesize that voters are
in uenced by the context in which they liv e,29 especially when sover-
eignty might carry indirect economic consequences for most residents
of a locale. Government workers may vote against sovereignty if they
believe that it will cost them their jobs, but reduced government
employment would also carry serious implications for the entire econ-
omy in ridings with a high percentage of government workers. Self-
interest might fuel widespread opposition among all voters in such a
context. A similar logic applies to other demographic variables: highly
educated francophones may in uence those with less schooling when
the former are numerous. Some of the in uences of a youthful popula-
tion may rub off on the older voters who reside near them.
For most of the contextual variables, we cannot separate the
direct in uence of individuals from the indirect in uence of context.
For example, if ridings with a younger population tend to support sov-
ereignty, we cannot say to what extent that pattern stems from younger
voters differing from older voters, rather than places with a youthful
population differing from those with an older population. This does
not pose a major barrier to the analysis, though, because it does not
apply to our variable of prime theoretical interest: linguistic context.
For that variable we used EI to estimate individual behaviour, so any
remaining systematic relationship between a riding’s linguistic mix
and its vote should be purely a contextual phenomenon.
Proportion Where French is the Home Language
Francophones in mixed ridings might be the strongest supporters of
sovereignty because they are more likely to compete with non-franco-
phones for jobs or for control of their localities. They are presumably
most likely to hear English idioms creeping into their children’s
speech, or to endure conversations spoken in English in public places.
29 Robert Huckfeldt, ‘‘Political Participation and the Neighborhood Social Con-
text,’American Journal of Political Science 23 (1979), 579-92; and Christopher
B. Kenny,‘Political Participation and Effects from the Social Environment,’
American Journal of Political Science 36 (1992), 259-67.
Context and Francophone Support for the Sovereignty of Quebec 91
Proximity fuelling separatist sentiment would accord with the ‘‘power
threat’’ approach to inter-group relations.30
On the other hand, francophones in mixed ridings may have
adjusted to the stresses of living between two worlds already, and may
have economic or social ties to the Canadian majority. They might fear
that their ridings will secede from Quebec if Quebec secedes from
Canada, a possibility already raised by federalist politicians and sup-
ported in some opinion polls.31 Those living in heavily francophone
areas, meanwhile, may feel besieged without the mitigating effect of
cross-cultural contact. They could have less to lose from severing
provincial ties with Ottawa. They naturally might view sovereignty as
less disruptive, and perceive less kinship with non-francophones,
either in Quebec or in other Canadian provinces. Proximity lessening
the sense of threat francophones might feel about cohabiting with the
rest of Canada would correspond to the ‘contact hypothesis’’ popular-
ized by Gordon Allport and applied to Canada by J. W. Berry.32 It
would validate the  ndings reported by Drouilly’s study, which used a
different methodology.33 We therefore include an explanatory variable
30 Giles and Evans, ‘‘The Power Approach to Intergroup Hostility. The threat
hypothesis was once dominant in the study of US race relations, especially in the
South, but it has been on the defensive even in the region from which it origi-
nated. See Michael Curt Corbello, ‘‘Searching for Evidence of ‘Racial Threat’ in
Louisiana,’American Review of Politics 19 (1998), 163-74; David Lublin and
D. Stephen Voss, ‘‘Racial Redistricting and Realignment in Southern State Legis-
latures,’American Journal of Political Science 44 (2000), 792-810; Jeffrey D.
Sadow, ‘‘David Duke and Black Threat: Laying to Rest an Old Hypothesis,
Revisited,’’ American Review of Politics 17 (1996), 59-68; D. Stephen Voss,
‘‘Beyond Racial Threat: Failure of an Old Hypothesis in the New South,’ Jour-
nal of Politics 58 (1996), 1156-70; D. Stephen Voss and Penny Miller, ‘‘Follow-
ing a False Trail: The Hunt for White Backlash in Kentucky’s 1996 Desegre-
gation Vote,’State Politics and Policy Quarterly 1 (2001), 63-81; and Voss and
Lublin, ‘‘Black Incumbents, White Districts.’’
31 Edward Greenspon, ‘‘Liberal unity agenda part of the election debris,’’ Globe
and Mail (Toronto), June 3, 1997; Lemco, Turmoil in the Peaceable Kingdom,
69-70; and ‘‘ ‘Plan B’ for Quebec developed by Trudeaus cabinet,’ Politics
Canada, June 16, 1997,
32 Gordon W. Allport, The Nature of Prejudice (Reading: Addison-Wesley, 1954).
Some francophones have compared their status in Canada to that of blacks in the
US; see Hudson Meadwell, ‘‘The Politics of Nationalism in Quebec,’World Pol-
itics 45 (1993), 207 n8, 218; and Berry, ‘‘Cultural Relations in Plural Societies.’
See also, Pierre Vallières, Nègres blancs d’Amérique (Montreal: Editions Parti
Pros, 1968). It may therefore seem strange to apply a hypothesis developed to
describe majority attitudes. However, francophones are clearly a majority in Que-
bec, forming roughly 85 per cent of the population. The threat mechanisms
described by these proximity theories would therefore still apply.
33 Drouilly’s study compared ridings within the Montreal region and throughout
Quebec separately (Drouilly, Indépendance et démocratie, 287-96). Support for
the contact hypothesis is stronger in the Montreal region, although that may be a
92 David Lublin and D. Stephen Voss
to capture the linguistic mix: percentage of homes with French as a
primary language.34
Income and Education
Harold D. Clarke and Allan Kornberg found no statistically signi cant
relationship between education or income and support for the BQ in their
analysis of voting behaviour in the 1993 election.35 Nevertheless, sup-
port for sovereignty and the BQ may decline as median family income
rises because af uent residents (as well as people who depend on con-
sumption by the af uent for their livelihoods) may fear the cost of sov-
ereignty to their incomes.36 Since leading sovereignists have stated
their intention to maintain, and perhaps even to expand, existing social
welfare programmes and government employment, and Quebec pays
less in taxes to Ottawa than it receives in federal transfers, wealthy
Quebeckers could fear that a ‘‘sovereignty tax’’ would be levied to
make up the difference. They also might fear a temporary reduction of
economic growth during the  rst decade of sovereignty, especially if
international bond rating agencies view Quebec as a risky debtor.37 On
the other hand, if ethnicity trumps economic concerns, there may be
little relationship between the median family income of a riding and
its support for sovereignty after controlling for other variables.38
Income and education are often highly related at the individual
level, but less so at the aggregate level. The correlation between
statistical result of the city’s greater linguistic diversity rather than a behavioural
34 The inclusion of the proportion French home language variable necessitates
excluding several other closely related variables that are effectively proxies for
the ethnic composition of the riding. The correlation between proportion French
home language and proportion English home language was -.94 in 1997. The
correlations with proportion Catholic and proportion immigrant were .94 and
-.89, respectively.
35 Harold D. Clarke and Allan Kornberg, ‘‘Partisan Dealignment, Electoral Choice,
and Party-System Change in Canada,’Party Politics 2 (1996), 467.
36 Drouilly,Indépendance et démocratie, 297-98.
37 Fry, Canada’s Unity Crisis, 28; Lemco, Turmoil in the Peaceable Kingdom,
68-69, 81-84, 93, 97, 114, 133-34. Although most scholars agree that Quebec is
economically viable, they disagree over the economic implications of sovereign-
ty. Lemco is essentially a pessimist. See Pierre Fournier for a more optimistic
conclusion, though he also acknowledges business concerns (A Meech Lake Post-
Mortem: Is Quebec Sovereignty Inevitable [Montreal: McGill-Queens University
Press, 1991], 104-19, 133-37).
38 Blais and Nadeau, ‘‘To Be or Not to Be Sovereignist,’ 96-97; Fry, Canada’s
Unity Crisis, 39; Johnston et al., The Challenge of Direct Democracy, 198-200;
Pierre Martin, ‘‘Générations politiques: rationalité économique et appui à la sou-
verainteté au Québec,’’ this Journal 27 (1994), 347-55; and Meadwell, ‘‘The
Politics of Nationalism in Quebec,’’ 212-28.
Context and Francophone Support for the Sovereignty of Quebec 93
median family income and the proportion of adults with university
degrees was .41 for the 1993 ridings and .48 for the 1997 ridings. Uni-
versity students and professors have been in the vanguard of the sover-
eignty movement.39 Particularly after controlling for income, ridings
with relatively educated populations may exhibit greater support for
sovereignty and the BQ. However, analyses of survey data suggest that
educational attainment is unrelated to support for sovereignty after
controlling for other factors. André Blais and Richard Nadeau found
that socio-economic differences between federalists and sovereignists
virtually disappeared after 1990.40 Maurice Pinard argues that educa-
tional differences in voting greatly eroded during the 1995 referendum
battle, particularly after the BQ leader, Lucien Bouchard, entered the
Proportion Aged 65 or Older
Ridings with relatively large populations of voters aged 65 or older
should demonstrate relatively high levels of support for federalist par-
ties. Quebec’s ‘‘Quiet Revolution of the 1960s and the subsequent
economic advancement of French Quebeckers, particularly the rise of
a francophone business elite, increased con dence that Quebec could
succeed as a sovereign nation. The decline of religious conservatism
among francophones has meant that younger francophones tend to
look to secular elites, including the Quebec government, rather than
the Roman Catholic Church, to protect French language and culture.42
Younger francophones grew up in an era during which sovereignty
was widely discussed as a viable and legitimate option, and survey
research has consistently documented that they support sovereignty
and the BQ at a higher rate than their elders.43 However, these results
could re ect that young anglophones have been leaving Quebec;44 it is
not clear that the age pattern should appear within linguistic cate-
39 Drouilly,Indépendance et démocratie; and Fry, Canada’s Unity Crisis, 39.
40 Blais and Nadeau, ‘‘To Be or Not to Be Sovereignist,’’ 94-99; Johnston et al., The
Challenge of Direct Democracy, 198-200; and Richard Nadeau,‘Le virage sou-
verainiste des Québécois,’Recherches sociographiques23 (1992), 13, 24.
41 Pinard, ‘‘Le contexte politique,’’ 305-07.
42 Stéphane Dion, ‘‘Explaining Quebec Nationalism,’’ in R. Kent Weaver, ed., The
Collapse of Canada? (Washington: The Brookings Institution, 1992); Fry,
Canadas Unity Crisis, 39; and Meadwell, ‘‘The Politics of Nationalism in Que-
bec,’ 207.
43 Blais and Nadeau, ‘‘To Be or Not to Be Sovereignist,’’ 93; Clarke and Kornberg,
‘‘Partisan Dealignment, Electoral Choice,’’ 467; Fry, Canada’s Unity Crisis, 39;
Martin, ‘‘Générations politiques,’ 6; Nadeau, ‘‘Le virage souverainiste des
Québécois,’’ 13; and Pinard, ‘‘Le contexte politique,’’ 301-05.
44 Lemco, Turmoil in the Peaceable Kingdom, 86-89, 99.
94 David Lublin and D. Stephen Voss
Proportion Employed in Government
Quebec contains two areas with a high proportion of government
workers: Quebec City and Hull. Quebec may not have the  nancial
resources to hire all former Canadian federal employees, especially
since supporters typically argue that sovereignty would eliminate
needless duplication of government services by Quebec and the fed-
eral government.45 Regardless, sovereignty would undoubtedly be dis-
ruptive to economic activity in ridings with large numbers of govern-
ment employees. Evidence on the relationship between attitudes and
government employment has varied over time, but its success is con-
sistent enough to warrant inclusion in our model.46
Proportion Employed in Agriculture
Farmers may fear that a sovereign Quebec will have to cut farm subsi-
dies. Farmers also might lose protected access to Canadian markets, as
well as bene ts from federal regulation of the Canadian market.47
Non-agricultural workers in ridings with signi cant agricultural
employment undoubtedly would also suffer if farm incomes declined.
Finally, agricultural regions may be more traditionalist than other rid-
ings, and therefore less in uenced by, or imbued with, the con dence
generated in the business sector by the Quiet Revolution. Conse-
quently, ridings with high proportions of agricultural workers may
support the BQ and pro-sovereignty positions in referenda at relatively
low rates.48
Studies of elections for the House of Commons in the United King-
dom and the House of Representatives in the United States have
revealed that incumbents often fare better than other candidates at the
polls.49 Party identi cation is far more  uid in Canada than in those
45 Ibid., 68-69, 97-100.
46 André Blais and Stéphane Dion, ‘‘Les employés du secteur public sont-ils dif-
férents?’Revue française de science politique 37 (1987), 279-317; André Blais
and Richard Nadeau, ‘L’appui du Parti québécois: évolution de la clientèle de
1970 à 1981’’ and ‘‘La clientèle du ‘oui,’ in Jean Crête, ed., Comportement
electoral au Québec (Chicoutimi: Gaëtan Morin, 1984), 323-34; Edouard
Cloutier, Jean H. Guay and Daniel Latouche, Le virage l’évolution de l’opinion
publique au Québec depuis 1960 (Montreal: Québec-Amérique, 1992); Fry,
Canadas Unity Crisis, 35; and Meadwell, ‘‘The Politics of Nationalism in Que-
bec,’ 208-09.
47 Fry, Canada’s Unity Crisis, 35.
48 Drouilly,Indépendance et démocratie, 294.
49 Bruce Cain, John Ferejohn and Morris Fiorina, The Personal Vote: Constituency
Service and Electoral Independence (Cambridge: Harvard University Press,
Context and Francophone Support for the Sovereignty of Quebec 95
two countries.50 The federal party system does not extend into the
provinces, and elites switch parties in Canada more often than their
American or British counterparts. So it is possible that incumbency
would carry great weight in Quebec. To measure the effect of incum-
bency on francophone voting behaviour, we use three dummy vari-
ables, coded one for incumbents and zero for non-incumbents—one
each for the Bloc Québécois, Liberal or Progressive Conservative can-
didates. The coef cient on the BQ variable should be positive , indicating
stronger pro-sove r e i g n t y candidates, while the Liberal and Conservative
va r i a b l e s should return nega t ive coef cients because they indicate
stronger anti-sove r e i g n t y candidates.51
Table 3 presents the results of regressing the estimated pro-sovereignty
position in each of the four campaigns on the demographic and politi-
cal contextual factors. The impact of the independent variables on sup-
port for the pro-sovereignty side, as measured by the direction of their
coef cients, is largely consistent across elections. Con dence in the
coef cients does vary from model to model.
As found by Drouilly, francophone support for sovereignty con-
sistently rose with the proportion of francophones.52 Raising the per-
1987); James Campbell, Cheap Seats: The Democratic Party’s Advantage in U.S.
House Elections (Columbus: Ohio State University Press, 1996); J. Curtice and
M. Steed, ‘‘The Voting Analyzed,’’ in Dav i d E. Butler and Dennis Kava n a g h , eds.,
The British General Election of 1983 (London: Macmillian, 1983); Andrew Gel-
man and Gary King, ‘‘Estimating the IncumbencyAdvantage Without Bias,Amer-
ican Journal of Political Science 34 (1990), 1142-64; Philip Norton and David
Wo o d , C o n s t i t u e n cy Service By Members of Parliament: Does it Contribute to a
Personal Vote?’ Parliamentary Affairs 43 (1990), 196-208; and David Wood and
Philip Norton, ‘‘Do Candidates Matter? Constituency-Specic Vote Changes for
Incumbent MPs, 1983-1987,’’ Political Studies40 (1992): 227-38.
50 Harold D. Clarke and Allan Kornberg, ‘‘Risky Business: Partisan Volatility and
Electoral Choice in Canada, 1988,’ Electoral Studies 11 (1992), 138-56; Harold
D. Clarke and Allan Kornberg, ‘Support for the Canadian Federal Progressive
Conservative Party Since 1988: The Impact of Economic Evaluations and Eco-
nomic Issues,’’ this Journal 25 (1992), 29-54; Harold D. Clarke and Allan
Kornberg, ‘‘Evaluations and Evolution: Public Attitudes toward Canadas Federal
Political Parties, 1965-1991,’’ this Journal 26 (1993), 287-312.
51 Thanks to the party’s loss of all but two seats nationally in 1993, Jean Charest,
the Conservative leader, was the only Conservative incumbent in Quebec in 1997
so the analysis of the 1997 elections excludes the Conservative incumbency vari-
52 On the surface, the data appear to support Drouillys conclusion that the relation-
ship between proportion francophone and proportion of francophones voting
‘‘yes’ is strongest in Montreal (Drouilly, Indépendance et démocratie, 287-90).
Correlations from the 1995 referendum suggest a strong relationship in Montreal
96 David Lublin and D. Stephen Voss
Table 3
Contextual Factors and Proportion Francophone
Support for Quebec Nationalism
For the BQ
in 1993
For the BQ
in 1997
For No in
For Yes in
Intercept .75a.45a.91a.80a
(.14) (.16) (.09) (.12)
Proportion French home language .18b.23a.15a.38a
(.08) (.10) (.05) (.06)
Median family income ($1000) -.003 -.002 -.005a-.007a
(.002) (.002) (.001) (.002)
Proportion with university degrees .33 .25 .46a.53a
(.23) (.23) (.16) (.19)
Proportion aged 65 or older -1.13a-.69b-1.20a-1.33a
(.43) (.42) (.29) (.34)
Proportion employed in government -.95a-.78a-.79a-1.24a
(.25) (.25) (.16) (.19)
Proportion employed in agriculture -1.42a-1.22a-1.17a-1.88a
(.40) (.36) (.26) (.29)
BQ incumbent .10a.03
(.04) (.02)
Liberal incumbent -.08b-.03
(.04) (.03)
Conservative incumbent -.02
R-Squared .51 .44 .45 .65
Adjusted R-Squared .44 .37 .41 .62
Standard Error of the Regression .08 .07 .05 .06
Number of Cases 75 75 75 75
Note: Diagnostic tests conducted on the OLS results include the Jarque-Bara test, the
White test, and the Ramsey-RESET test. The Jarque-Bara test indicates that the resid-
uals are not normally distributed for the 1993 models. The White test suggests that
heteroskedasticity is a problem for the 1997 models. These models were reestimated
using a nonparametric bootstrap regression technique that resamples the OLS residu-
als 1000 times to produce more accurate estimates based on the empirical distribution
of the data. Since the bootstrap results are virtually identical to the OLS results, the
table presents only the OLS results.
ap <.01 for one-tailed tests.
bp <.05 for one-tailed tests.
Context and Francophone Support for the Sovereignty of Quebec 97
centage of francophones by one standard deviation, 19 per cent,
increased francophone support for the BQ by approximately 3 per cent
in 1993 and 5 per cent in 1997. The impact of a similar change on
opposition to the Charlottetown Accord was slightly less than 5 per
cent. The ethnic composition of the population had the greatest impact
on support for sovereignty in 1995, with a one standard deviation
increase in the proportion of francophones causing the proportion of
francophone ‘‘oui’’ voters to rise by nearly 10 per cent. Francophones
who reside in a linguistically mixed community do not exhibit hostile
backlash against their neighbours or greater perceived threat from
anglophones; some combination of contact, assimilation and self-
interest may even make them more supportive of retaining political
ties to Canada. Furthermore, these  ndings indicate that the sovereign-
ty movement cannot be viewed purely as the continuance of ethnic
con ict through the federalism debate. Though  rm conclusions
require attitudinal data, support for sovereignty seems more likely to
re ect the belief that a sovereign Quebec can manage and protect fran-
cophone interests better than a federal Canada can.
The direction of the remaining demographic variables re ects the
theories outlined in the previous section, although the results cannot
distinguish between individual- and aggregate-level effects. Franco-
phones in af uent ridings seem more risk averse, since they are less
likely to support sovereignty. After controlling for income, those in
ridings with a disproportionate share of university graduates support
sovereignty at a higher rate. A change of one standard deviation in
median family income, or $7000, corresponds to 4 per cent fewer bal-
lots for sovereignty in the 1995 referendum and 3 per cent fewer for
the Charlottetown Accord in 1992. Increasing income by a similar
amount reduced support for the BQ by 2 per cent in 1993 and over 3
per cent in 1997. Raising the proportion of university graduates by one
standard deviation, 7 per cent, reduced francophone support for Char-
lottetown in 1992 by nearly 3 per cent and increased francophone sup-
port for sovereignty in 1995 by the same amount. Similar increases in
the share of university graduates resulted in 2-3 per cent more support
for the BQ in both 1993 and 1997.
As expected, francophones in ridings with relatively elderly pop-
ulations support sovereignty at a lower rate. Reducing the percentage
of elderly in the riding’s population by one standard deviation, 3.5 per
cent, causes a rise in francophone support for the pro-sovereignty
(r = .77) and a weak relationship elsewhere (r =.17). After controlling for other
variables, however, the relationship between proportion francophone and propor-
tion francophone voting ‘‘oui’’ is strong even for non-Montreal ridings. When the
regression is performed for just non-Montreal ridings, ß on Proportion French
Home Language equals .36 (standard error of .10).
98 David Lublin and D. Stephen Voss
position of approximately 3-4 per cent in all four votes. Survey
research suggests that generational replacement, rather than aging,
explains this  nding.53 Generational replacement may aid the sover-
eignty movement in future federal elections and referenda.
Francophones in ridings with a high proportion of government
workers voted against the sovereignist position at a greater rate than in
other ridings. Those in ridings with relatively high numbers of agricul-
tural employees demonstrated similar sensitivity to their economic
base. The negative relationship was strongest for the 1995 referendum,
likely because sovereignty posed a greater threat to the livelihoods of
government employees and farmers than a vote for the BQ or against
the Charlottetown Accord. By casting ballots against Charlottetown or
for the BQ, francophone voters could stand up for Quebec, rebuke the
federal government, and maintain sovereignty as an option for the
future without actually risking the dissolution of Canada. Increasing
the percentage of government workers by only 4.5 per cent, a standard
deviation, resulted in a decline in francophone support for the BQ by 4
per cent in 1993 and 1997. Similar increases in the share of govern-
ment employees reduced the francophone pro-sovereignty vote by 3
per cent in 1992 and 6 per cent in 1995. Raising the percentage
employed in agriculture by a standard deviation of 3 per cent reduced
the pro-sovereignty vote among francophones by 4 per cent in 1992
and 1993, and 5 per cent in 1995 and 1997.
It is worth noting that, while these results con ate both individual
and contextual patterns, they provide fairly clear evidence that context
in uences francophone support for sovereignty. Coef cients greater
than 1.0 or less than -1.0 should not appear if the correlations being
studied are entirely individual-level effects. Increasing the percentage
working in agriculture by one point should at most decrease the per-
centage voting in a particular way by one point as well, unless some of
the non-agricultural workers also behave differently on average in that
changed context. Coef cients exceed this range at least once on the
age, government employment and agricultural employment variables.
This is additional evidence that francophone voters are somehow
being in uenced by the demographic makeup of their local communi-
ties when formulating an opinion on sovereignty.
Incumbency appears to have strongly in uenced the vote for the
BQ in 1993, but not in 1997. BQ incumbents receive d 10 per cent more
of the francophone vote in 1993 than other BQ candidates after control-
ling for other factors. BQ candidates running in ridings with Liberal
incumbents receive d 8 per cent less of the francophone vote than other
53 Nadeau, ‘‘Le virage souverainiste des Québécois,’’ 14-16; and Pinard, ‘‘Le con-
texte politique,’’ 304.
Context and Francophone Support for the Sovereignty of Quebec 99
BQ candidates. Re ecting the meltdown of Conservative party fortunes,
the presence of a Conservative incumbent did not in uence francophone
vo t i n g behaviour in 1993. Neither the presence of a BQ incumbent nor
a Liberal incumbent in uenced the share of the vote receive d by the BQ
in 1997. The 1997 results are consistent with the party-centred nature of
Canadian elections. The strength of the incumbency advantage in 1993
may re ect that there were only nine BQ and eleve n Liberal incum-
bents. Representing bastions of support for their parties, these incum-
bents may have bene ted disproportionately from the electoral swing
against the Conservative s . BQ incumbents were founding members of
their party and most likely attracted disproportionate media attention
and public support through their by-election victories or when they dra-
matically abandoned federalist parties.
This article examined the effect of local context on francophone sup-
port for sovereignty in four Quebec campaigns: the 1992 vote on the
Charlottetown Accord, the 1995 sovereignty referendum, and the 1993
and 1997 federal elections. We isolated francophone voting behaviour
using King’s solution to the ecological inference problem, and then
ran multivariate OLS models predicting the estimated francophone
vote one contest at a time.
Our main interest was to test competing theories of ethnic con-
 ict. The power threat (or con ict) approach predicts that franco-
phones in mixed-language ridings should feel most threatened. The
contact hypothesis, by contrast, predicts that those in mixed ridings
would have their fears assuaged by exposure to, assimilation with and
interdependence with English speakers. The latter approach wins con-
sistent support across all four elections, con rming Drouilly in a man-
ner that should instill great con dence.
These results  t the con ict over Quebec’s sovereignty into a
much larger (and growing) literature on ethnic relations. Research
almost consistently targets homogeneous ethnic communities as a cat-
alyst for inter-group friction. In the United States, for example, the
sources of cultural con ict between blacks and whites are enhanced in
all-white neighbourhoods or counties.54 The relationship between the
ethnic composition of a riding and support for sovereignty in Quebec
54 Donald P. Green, Dara Z. Strolovitch, and Janelle Wong, ‘‘Defended Neighbor-
hoods, Integration, and Racially Motivated Crime,’American Journal of Sociol-
ogy 104 (1998), 372-403; and Voss, ‘‘Beyond Racial Threat,’’ and ‘‘Familiarity
Doesn’t Breed Contempt.’
100 David Lublin and D. Stephen Voss
also resembles patterns of support for the Plaid Cymru in Wales. Sup-
port for the Plaid correlates positively with the percentage of the popu-
lation in a constituency that speaks Welsh.55
More generally, Quebec’s francophones respond to context, as do
voters in other nations. Uniquely Canadian institutions and history
condition their response to that context, as do local demographics and
the incentives that those conditions bring to bear in the sovereignty
debate. Francophone voters appear quite sensitive to the potential eco-
nomic impact of sovereignty. Ridings with large numbers of govern-
ment employees apparently fear that sovereignty might eliminate jobs
and undermine the economic base of their region. This  nding
explains the high level of opposition to sovereignty in the Ottawa/Hull
region. Agricultural regions similarly appear to fear the economic
impact of sovereignty. That the francophone vote re ects such sensi-
tivity to local context, in predictable ways corresponding to hypothe-
sized self-interest, gives a more optimistic picture of voter rationality
than implied by opinion poll results.56
55 Robert Waller and Byron Criddle, The Almanac of British Politics (6th ed.; Lon-
don: Routledge, 1999).
56 Pinard, ‘‘Les déterminants psychosociaux,’’ 345-53.
Context and Francophone Support for the Sovereignty of Quebec 101
... 4 It would, of course, have been possible to extend this study further, covering all presidential elections in American history. We chose not to do so, however, because of the relative frequency with which parties split, new parties arrived on the scene and parties changed names during the period prior to 1860. 5 The specific variables chosen for each pair of elections, and their summary statistics, can be found in Appendix I. 6 Only for one of the studies was the total of black population not available, see Appendix 1. Lublin and Voss, 2002). Its extensions in King et al. (1999) and Rosen et al. (2001) have subsequently solved problems found in the original model (Rosen et al., 2001). ...
... Given that ecological inferences are made from population groups, particularly useful covariates are group characteristics from the moment the aggregate data was collected. For example, census data has been used to estimate voting behavior (King, 1997;Calvo and Escolar, 2003;Rosen et al., 2001;Lublin and Voss, 2002). ...
Full-text available
Political scientists have long debated theories of electoral party realignments. In this paper, we apply ecological inference methods to statistically analyze the transfer of votes within counties in US presidential elections since 1860. Through this analysis we are able to identify the major periods of party realignment in US history and the counties where these shifts took place. As a result, we are able to provide new insights into American electoral history, and provide strong evidence that the 2008 presidential election did not represent a realigning election as the phrase is generally understood.
... Polarized voting outcomes certainly are not unique to the U.S. (Lublin and Voss, 2002). Still, most literature and research on the topic is on American elections and observations or analyses thereof (Key, 1949;Blalock, 1967;McCrary, 1990;Giles and Buckner, 1993;Forbes, 1997;Voss, 1996Voss, , 2000Orey, 1998Orey, , 2001Voss and Miller, 2001;Tolbert and Grummel, 2003;Rocha and Espino, 2009;Donovan, 2010;Orey et al., 2011). ...
Full-text available
This paper advances a model of racially polarized voting that captures the intervening effects of urbanization and residential segregation on white voters’ political behavior. The model is tested for a 2011 referendum election in the U.S. state of Mississippi. Using King’s method of ecological inference and weighted least squares regression, we find that regional minority population size impacts white opposition to minority-preferred political alternatives both directly and indirectly through an effect on residential racial segregation. Importantly, these influences hinge on intra-regional patterns of urbanization. The findings have important implications for understanding spatial variation in regional political behavior and intergroup relations.
... In other words, an increased presence of the "other" in a locality does not increase the likelihood of voting for the BQ, whereas it does increase the likelihood that nationalism will be an important predictor of support for the BQ. This corroborates findings in the literature (Piroth et al., 2006;Lublin and Voss, 2002) that an increase in the share of non-francophones at the local level either has no effect (outside of Montreal) or a negative effect (in Montreal) on francophone's support for Quebec sovereignty. In terms of our second contextual variable our preliminary analysis shows that the percent of unemployment has a significant negative effect on support for the BQ, suggesting that the higher the percentage of unemployed people in a locality the smaller the likelihood that an individual will vote for the Bloc Qu eb ecois. ...
This article explores the individual-level correlates of nationalist party vote choice and the extent to which these correlates are conditioned by an individual’s local context. We argue that the influence of individuals’ policy positions on nationalism should vary in importance for predicting voting for nationalist parties in localities where voters feel threatened culturally or economically. To test this argument we use the case of support for the Bloc Quebecois in the Canadian province of Quebec and data from the 2011 Canadian Vote Compass. We show that voters’ policy positions on nationalism become more important in predicting a vote for the Bloc Quebecois when the percentage of English speakers (our proxy for ethno-cultural threat) increases in their locality. By contrast, we find that the relationship between nationalism and support for the Bloc Quebecois is not conditioned by economic hardship in the place where an individual lives. To test the robustness of our findings, we reestimate our models using a different dataset from multiple elections – the Canadian Election Study as well as an additional modelling approach. Our findings contribute to the broader vote choice literature by examining the role that local context plays in individuals’ choice of parties. Furthermore, our findings lend support to arguments made in the literature on the importance of an ethno-cultural calculus among voters voting for nationalist parties.
Ethnic settlement patterns and other forms of everyday interethnic social contact have the potential to influence voter preferences for ethnic tribune parties who position themselves as the most strident protectors of, and flagbearers for, their respective ethnic groups. Previous studies on this topic have come to rival conclusions, with some finding that increased intergroup contact and residential mixing produce a corresponding increase in support for tribune parties and others finding the opposite. This study uses a combination of data from elections to the Northern Ireland Assembly from 1998 to 2017 and survey data from 2016 to evaluate these rival arguments and assess the extent to which the broader institutional and demographic context in which political competition takes place condition responses to intergroup contact. Our findings indicate that voters in both declining and ascending demographic groups respond similarly to intergroup contact, expressing less support for tribune parties in contexts where residential patterns and social networks provide more opportunities for intergroup contact. These results highlight the conditional nature of the effectiveness of consensus-based institutions in divided societies: they can create incentives for moderation, but those incentives are most likely to be realized in contexts where rival groups experience a high level of integration.
Scotland’s 2014 Independence Referendum affords a rare opportunity to examine public support for the break-up of a long-established, stable democracy. Analyses of support for Scottish independence reveal that while issues of national identity loomed large in the vote, they were not the only factors involved. Questions around the economic and political direction of the state, and around uneven development, ideology and trust in established politicians also influenced voters’ decisions. Partisanship also mattered, as voters were more likely than not to follow the lead of their party in what had become a highly partisan contest. But some parties – especially Labour – saw large minorities of their supporters vote against the party’s line to support independence.
This study examines whether East-to-West variation in attitudes towards linguistic duality (English/French) in Canada can be explained by regional variation in the amount of contact with the other linguistic group. Results indicate that all of the contact variables tested, especially the ability to understand the other official language at a good-to-excellent level, contribute to a reduction in regional variation.
How do varying levels of inter-group contact affect voter preferences in connection with ethnically radical political candidates and parties? Two competing hypotheses have emerged in the last 60 years: the first, known as the group threat hypothesis, argues that voters from an ethnic or religious group in more ethnically or racially heterogeneous districts will exhibit stronger preferences for ethnically radical political candidates. The contact hypothesis argues that groups living in mixed localities are actually less likely to support ethnic radicals. Both perspectives have found empirical support, but no previous study has offered a theoretical explanation for two seemingly contradictory conclusions. We specify just such a theory, arguing that the effect of district level integration is conditioned by the direction of a group’s share of the national population. We test this theory quantitatively using electoral data from Northern Ireland between 1983 and 2010.
Instant-Runoff Voting (IRV/RCV) is a relatively new reform to the rules of voting and elections that has been adopted by several cities. The impact of IRV/RCV balloting on racial group voter turnout in urban elections has not been subject to rigorous empirical analysis. In order to evaluate the relationship between IRV/RCV and voter turnout, I analyze precinct-level racial group voter turnout rates in five San Francisco mayoral elections from 1995 to 2011. I find significant declines in turnout under IRV/RCV rules, especially among Black voters. Variation in racial group turnout is best explained by factors related to candidate race and racial group competition. In addition, I find that IRV/RCV exacerbates turnout disparities between older and younger groups in the population and between those with high education compared to those with low education.
Gary King's ecological inference method represents a major breakthrough for analysts working with aggregate data because it is sensitive to contextual behavior patterns that previous methods had to assume away. King's critics underestimate the value of his method because they apply it to patently uninformative data, do not look for the contextual patterns that would improve estimation, and do not take advantage of the contextual knowledge that substantive experts would bring to an analysis. This chapter offers new diagnostic trials for the estimates from King's EI software, applied to data that have emerged from a genuine research agenda but for which the true values are known. Not only does the method do a superb job with typical precinct-level voting data, it even manages to produce solid estimates with inedequate county-level data once the analyst takes into account insights provided by the relevant scholarly literature. King's approach can and should be improved, but the imperfections provide no justification for using older methods of aggregate-data analysis or for relying solely on survey data such as exit polls. INTRODUCTION Debate over Gary King's proposed solution to the ecological inference problem (King, 1997) has begun to emerge in journals oriented toward statistics and political methodology.
The practice of using point estimates produced by the King ecological inference technique as dependent variables in second-stage linear regressions leads to second-stage results that, in general, are inconsistent. This conclusion holds even when all assumptions behind King's ecological technique are satisfied. Second-stage inconsistency is a consequence of the fact that King-based point estimates of disaggregated quantities contain errors correlated with the true quantities the estimates measure. Our findings on second-stage inconsistency, as well as a fix that we propose, follow from econometric theory in conjunction with an analysis of simulated and real ecological data sets.
Is political choice under conditions of uncertainty influenced more by expected losses than by expected gains? Does the relative importance of economic rationality in political choice vary between generations? The analysis of individual opinion on sovereignty in Quebec sheds new light on these questions. Among determinants of individual support for sovereignty, apprehensions of possible losses play a more important role than expectations of possible gains. The relative importance of the evaluation of economic consequences in the decision to support sovereignty, however, differs across three generations. These differences can be explained with reference to the economic context that prevailed when members of each generation reached adulthood.RésuméLe choix politique dans l'incertitude est-il plus influencé par la crainte de perdre des acquis que par la perspective de réaliser des gains? La part de la rationalité économique dans la décision politique varie-t-elle selon les générations? L'analyse de l'opinion relative à la souveraineté au Québec apporte un éclairage nouveau sur ces questions. Dans l'explication de l'opinion sur la souveraineté, la crainte de pertes économiques est plus déterminants que la perspective de gains. On observe cependant des différences entre générations pour ce qui est de l'importance relative de l'éevaluation des conséquences économiques dans la décision d'appuyer la souveraineté. Ces différences s'expliquent par le contexte économique qui prévalait au moment du passage de ces générations à l'âge adulte.
This article employs national survey data gathered over the past quarter century to analyze the evolution and present state of public attitudes toward Canada's federal political parties. A 1991 survey employing new questions on evaluations of party performance reveals that these evaluations are structured in terms of two dimensions, and that negative judgments on both dimensions are pervasive. The significance of the current negativism is assessed using 1965–1991 data on Canadians' feelings about and identifications with the federal parties. Although for a long time party affect has been lukewarm at best, and partisanship has been weak and unstable, negative trends have magnified the disaffection and dealignment. The discontent has accelerated in recent years, as the percentage of Liberal and Progressive Conservative identifiers has plummeted, and the non-identifier group has swelled to record levels, particularly in Quebec. The article concludes by considering the implications of these findings for the future of the federal party system.
Former Ku Klux Klansman David Duke ran three statewide campaigns in Louisiana. This note presents a GLS analysis of his voting support. It suggests that whites living in racially heterogeneous parishes were no more likely to support Duke than those in less-diverse locales, contrary to the “racial threat hypothesis.” Suburban whites were particularly likely to back Duke. Giles and Buckner (1993) present contrary results because their analysis contained measurement error and omitted variable bias, aggregated data for urban areas inappropriately, and used OLS on grouped data.
Liberal unity agenda part of the election debris Lemco, Turmoil in the Peaceable Kingdom, 69-70; and '' 'Plan B' for Quebec developed by Trudeau's cabinet
  • Edward Greenspon
Edward Greenspon, ''Liberal unity agenda part of the election debris, '' Globe and Mail (Toronto), June 3, 1997; Lemco, Turmoil in the Peaceable Kingdom, 69-70; and '' 'Plan B' for Quebec developed by Trudeau's cabinet,'' Politics Canada, June 16, 1997,
To Be or Not to Be Sovereignist 96-97; Fry, Canada's Unity Crisis, 39 The Challenge of Direct Democracy
  • Nadeau Blais
  • Johnston
38 Blais and Nadeau, ''To Be or Not to Be Sovereignist, '' 96-97; Fry, Canada's Unity Crisis, 39; Johnston et al., The Challenge of Direct Democracy, 198-200;
Recherches sociographiques23
  • Québécois Des
verainiste des Québécois,'' Recherches sociographiques23 (1992), 13, 24. 41 Pinard, ''Le contexte politique,'' 305-07.
Turmoil in the Peaceable Kingdom
  • Lemco
Lemco, Turmoil in the Peaceable Kingdom, 86-89, 99.