voting behaviour and socio-economic characteristics: the middlesex east federal election, 1965

13
Voting Behaviour and Socio-Economic Characteristics: The Middlesex East Federal Election, 1965 Author(s): James W. Simmons Source: The Canadian Journal of Economics and Political Science / Revue canadienne d'Economique et de Science politique, Vol. 33, No. 3 (Aug., 1967), pp. 389-400 Published by: Wiley on behalf of Canadian Economics Association Stable URL: http://www.jstor.org/stable/139915 . Accessed: 16/06/2014 12:58 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. . Wiley and Canadian Economics Association are collaborating with JSTOR to digitize, preserve and extend access to The Canadian Journal of Economics and Political Science / Revue canadienne d'Economique et de Science politique. http://www.jstor.org This content downloaded from 185.44.78.156 on Mon, 16 Jun 2014 12:58:31 PM All use subject to JSTOR Terms and Conditions

Upload: james-w-simmons

Post on 15-Jan-2017

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Voting Behaviour and Socio-Economic Characteristics: The Middlesex East Federal Election, 1965

Voting Behaviour and Socio-Economic Characteristics: The Middlesex East Federal Election,1965Author(s): James W. SimmonsSource: The Canadian Journal of Economics and Political Science / Revue canadienned'Economique et de Science politique, Vol. 33, No. 3 (Aug., 1967), pp. 389-400Published by: Wiley on behalf of Canadian Economics AssociationStable URL: http://www.jstor.org/stable/139915 .

Accessed: 16/06/2014 12:58

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Wiley and Canadian Economics Association are collaborating with JSTOR to digitize, preserve and extendaccess to The Canadian Journal of Economics and Political Science / Revue canadienne d'Economique et deScience politique.

http://www.jstor.org

This content downloaded from 185.44.78.156 on Mon, 16 Jun 2014 12:58:31 PMAll use subject to JSTOR Terms and Conditions

Page 2: Voting Behaviour and Socio-Economic Characteristics: The Middlesex East Federal Election, 1965

VOTING BEHAVIOUR AND SOCIO-ECONOMIC CHARACTERISTICS: THE MIDDLESEX EAST FEDERAL ELECTION, 1965

JAMES W. SIMMONS University of Western Ontario

Although the interrelationship between socio-economic characteristics and electoral behaviour in Canadian elections is widely recognized, only recently have attempts been made to evaluate these interactions in a precise manner.1 The approaches used so far vary in methodology-using both interviews and voting results-and in the level of analysis-ranging from poll to constituency to province to nation. The results depend to some extent on the research approach. Particularly strong variations exist in the effects of social variables in different provinces2 and different constituencies.3

This study presents another research alternative, using multiple regression techniques to combine census small-area information with voting results at the poll level. Although the strengths and weaknesses of this approach are dis- cussed in detail by Meisel and Paquet,4 few Canadian studies have yet exploited its potential return. Problems of sampling and of variations in candidates and issues among constituencies are replaced by the difficulties of ecological correlation,5 but the use of regression procedures allows the evaluation of the predictive power of the explanatory variables.

The analysis indicates a considerable variation in the sensitivity of the

'The major studies of this type are: Robert R. Alford, Party and Society: The Anglo- American Democracies (Chicago, 1963), 250-86; the set of studies edited by John Meisel, Papers on the 1962 Election (Toronto, 1964), particularly the papers by George Perlin, "St. John's West," 3-18; Howard A. Scarrow, "Three Dimensions of a Local Political Party," 53-67; T. Peterson and I. Avakumovic, "A Return to the Status Quo: The Election in Winnipeg North Centre," 91-106; W. P. Irvine, "An Analysis of Voting Shifts in Quebec," 129-43; Robert R. Alford, "The Social Bases of Political Cleavage in 1962," 203-34; and John Meisel, "Conclusion: An Analysis of the National (?) Results," 272-88; also John Meisel, "Religious Affiliation and Electoral Behaviour: A Case Study," this JOURNAL, XXII (Nov., 1956), 481-96; Pauline Jewett, "Voting in the 1960 Federal By-Elections at Peter- borough and Niagara Falls," this JOURNAL, XXVIII (Feb., 1962), pp. 35-53; Peter Regenstreif, The Diefenbaker Interlude: Party and Voting in Canada (Toronto, 1965); Grace M. Anderson, "Voting Behaviour and the Ethnic and Religious Variables," this JOURNAL, XXXII (Feb., 1966), 27-37; John C. Courtney and David E. Smith, "Voting in a Provincial General Election and a Federal By-election: A Constituency Study of Saskatoon City," this JOURNAL, XXXII (Aug., 1966), 338-53. These works are cited hereafter by the author's name only, except where the author has written two or more cited studies. 2Alford, "Social Bases," 219-26. 3Note the marked differences between the characteristics of NDP supporters in Winnipeg North Centre, an old Socialist stronghold, and other areas: Peterson and Avakumovic, 101; Scarrow, 65; and Jewett, 38. 4John Meisel and Gilles Paquet, "Some Quantitative Analyses of Canadian Election Results: An Exercise in the Testing of Hypotheses," in J. Henripin and A. Asimakopoulos, Canadian Political Science Association Conferences on Statistics, 1962 and 1963: Papers (Toronto, 1964), 16-25; and the discussion by Muni Frumhartz, 32-38. The latter refers to several similar studies in the United States. 51Problems of analysing areal data are discussed in Otis D. Duncan, Ray P. Cuzzort, and Beverly Duncan, Statistical Geography (Glencoe, Ill., 1961).

This content downloaded from 185.44.78.156 on Mon, 16 Jun 2014 12:58:31 PMAll use subject to JSTOR Terms and Conditions

Page 3: Voting Behaviour and Socio-Economic Characteristics: The Middlesex East Federal Election, 1965

390 JAMES W. SIMMONS

l N 2 < t q p LONDON WEST

I ~~~~~~~~ ~ ~~~~~TOWNSHIP NISSOURI

<- TOWNSHIP

NORTH

WESTMINSTER DORCHESTER

TOWNSHIP TOWNSHIP

Q - MIDDLESEX EAST

'777

/ / C X i.

-..RESIDENTIAL AREAS /t

r NEW LONDON EAST RRIDING

FIGURE 1. Middlesex East

different parties to socio-economic characteristics 6 The largest amount of explained variation exists for the New Democrats, who are strongest in urban, New Canadian, youthful, and lower-class areas. The Progressive Conserva- tives are next, doing best in rural, Anglo-Saxon, and high-income polls. The Liberals showed very little variation among polls, except for high correlations with the college-educated. After eliminating irrelevant or redundant socio-

6Jewett, 38-9, made similar observations.

This content downloaded from 185.44.78.156 on Mon, 16 Jun 2014 12:58:31 PMAll use subject to JSTOR Terms and Conditions

Page 4: Voting Behaviour and Socio-Economic Characteristics: The Middlesex East Federal Election, 1965

RESULTATS ELECTORAUX ET CARACTERISTIQUES SOCIO- ECONOMIQUES DES ELECTEURS LORS DES ELECTIONS FEDERALES DE MIDDLESEX-EST DE 1965

JAMES SIMMONS

Pour cette etude l'auteur se sert de techniques de correlation et de regression en se basant sur des donnees statistiques officielles afin d'identifier les rapports qui existent entre les caracteristiques sociales des electeurs et les resultats electoraux. L'etude est fondee sur les resultats du scrutin de la circonscription electorale de Middlesex-est et prend comme point de depart les resultats des centres de vote de cette circonscription. Les donnees sociales sont tirees des resultats de recensements en fonction de zones d'enumeration.

De vingt a trente-cinq pour cent de la difference du suffrage obtenu par chaque parti s'explique a cause des variables de prediction. Les rapports enregistres confirment les resultats d'enquetes menees dans d'autres regions a l'aide de moyens differents. Les resultats de la regression sont ensuite appli- ques a la circonscription nouvellement delimitee en vue d'etablir des previsions pour des elections futures.

Les techniques multivariables utilises pour des etudes fondees sur les resultats de centres de vote semblent etre un moyen valable pour l'etude des caracteristiques sociales et politiques, tout en ayant le grand avantage de pouvoir etre appliques 'a I'analyse d'elections passees.

economic variables, a set of regression equations were obtained and used to simulate future outcomes in a new riding defined by redistribution.

The constituency

Middlesex East is a large heterogeneous riding including the eastern half of Middlesex county and the eastern part of the city of London (Figure 1). About 15 per cent of the polls are outside the city; another 25 per cent are pre-Second- World-War urban, and the rest are suburban areas developed since 1945. Although Middlesex East includes all economic levels, lower income groups are more heavily represented than in the city as a whole. The riding has a long history of Conservative allegiance, but both the Liberals and New Democrats have been gaining strength, culminating in the tight 1965 race which was won by the Liberal candidate.

The campaign was a three-way fight between the incumbent, Cameron Millar (Conservative), who, despite victories in 1962 and 1963, was not well known; James Lind (Liberal), a local lumber dealer, active in municipal politics and a candidate in 1963; and Kenneth Bolton (NDP), an Anglican theologian, running for the first time. The NDP organization supported Bolton with a massive door-to-door campaign. James Watson (Social Credit) obtained only 400 votes and will be ignored in the analysis.

This content downloaded from 185.44.78.156 on Mon, 16 Jun 2014 12:58:31 PMAll use subject to JSTOR Terms and Conditions

Page 5: Voting Behaviour and Socio-Economic Characteristics: The Middlesex East Federal Election, 1965

TABLE I MIDDLESEX EAST ELECTION RESULTS, 1953-65*

Total votest CCF-NDP Liberal Progressive Conservative Num ber of _ _ _ _ _ _ _ __ _ _ _ _ _ _ _ _ _ _

voters % of %O f % of % of Year on list number eligible number total number total number total

1953 36,291 24,116 66.3 3,137 13.0 8,952 37.2 12,027 49.9 1957 43,064 32,852 76.1 3,242 9.6 9,323 28.4 20,287 62.0 1958 45,085 35,870 79.5 3,125 8.7 7,849 21.8 24,896 69.4 1962 54,603 39,570 72.4 7,246 18.3 13,231 33.5 19,003 48.1 1963 57,158 42,985 75.2 5,092 11.9 18,043 42.0 19,850 46.2 1965 63,281 46,607 73.8 13,073 28.0 17,675 37.9 15,859 34.0

*Dominion of Canada, Report of the Chief Electoral Officer: Twenty-second General Election, 1953, etc. (Ottawa, 1955). tTotal votes for the three parties.

This content downloaded from 185.44.78.156 on Mon, 16 Jun 2014 12:58:31 PMAll use subject to JSTOR Terms and Conditions

Page 6: Voting Behaviour and Socio-Economic Characteristics: The Middlesex East Federal Election, 1965

Voting Behaviour 393

The data

Two different sources of data were used: the unofficial returns for each candi- date for each poll were combined to produce the total vote; the percentage vote for each candidate was mapped using the description of the polls pro- vided in the voters' list. These maps indicated that there was considerable spatial contiguity in the results, reflecting the contiguity of social characteristics, and suggested socio-economic variables which might explain the voting pat- terns. The turn-out was examined by comparing votes cast with the number on the voting list.

The socio-economic data were gathered for a different set of spatial units, the enumeration areas defined by Dominion Bureau of Statistics for the 1961 Census. They are roughly the same size (500-1,000 persons) as the polls (about 300 voters), but the boundaries do not generally coincide. In this respect the analysis depends on the spatial contiguity of the two sets of variables. Both the voting statistics and the socio-economic variables were converted to percentages and the two sets of areas were matched as well as possible. In some cases two or more enumeration areas grouped together in order to cover a poll, which was the unit of observation. In addition to the spatial discrepancies there is a time lapse between the 1961 socio-economic data and the 1965 election results. However, the major kind of change, growth, does not necessarily affect social characteristics which are usually stable within each sector of a city. The areas which will change most radically are those rural areas which have been invaded by suburban development, but these are relatively few.

The choice of variables was restricted to the limited information provided by the census for enumeration areas. They include age-sex structure, religion, ethnic background, and education characteristics. Within the seventy-five possible variables the selection was guided by the recognition of three major dimensions of urban socio-economic variation.

Social class is a closely linked set of occupation, education, and income phenomena, represented in this study by two education variables: the pro- portion of the population, not attending school, with less than eight years education; and the proportion of the population, not attending school, with at least one year at university. These variables are slightly distorted because they ignore the effect of the varying proportions of children still attending school in different areas. In London, as in most cities, social class varies in a sectoral pattern.

Urbanization describes a set of demographic and life-style variables which differentiate between city centre and suburban areas. For instance, suburban areas have more young families, fewer women working, and fewer apartment dwellers. Two variables, the percentages of the population over sixty-five and under fourteen, were used in this study.

The third dimension, that of segregation, refers to the whole mix of racial, ethnic, and religious variables. The variables selected were the proportion born outside the country, the proportion of British descent, and the per- centages of Catholics and Anglicans, the last variable included because of

This content downloaded from 185.44.78.156 on Mon, 16 Jun 2014 12:58:31 PMAll use subject to JSTOR Terms and Conditions

Page 7: Voting Behaviour and Socio-Economic Characteristics: The Middlesex East Federal Election, 1965

394 JAMES W. SIMMONS

Bolton's profession. In addition to the dimensions represented by these vari- ables, the proportion of farm residents was included as an additional source of variation within Middlesex East.

The regression results

The variables defined above were analysed as a multiple regression problem in two stages, first to sort out unnecessary independent variables, and then to obtain regression equations which could be used to simulate future electoral behaviour.7 The dependent variables in each run were, in turn, TOTAL, NDP,

LIB, and PC with the socio-economic measures as independent variables. The matrix of simple correlation coefficients is presented in Table III,

indicating the association of the input variables for the units of observation. None of the correlations is very high. Consider the subset of correlations between the political variables. Since each is expressed as a percentage of the total vote each party is negatively related to the other two, with the com- plementarity greatest between the NDP and the Conservatives, the NDP and the Liberals, and the Liberals and the Conservatives in that order. In other words, as the percentage of the NDP vote increases the percentage of the Conservative and Liberal vote decreases. Of special concern to the New Democrats is the negative relationship between their percentage of the vote and the total per cent voting. Areas where they are strongest have the weakest turn-out.

The socio-economic variables are closely interrelated, particularly the intra- dimensional ones (underlined in the matrix). The FARM variable is also closely linked to the other characteristics. It is difficult to identify the effect of any one variable because of the complex interactions, many of which will be working against each other on the political variables.

TABLE II VARIABLES USED IN STUDY

Standard Symbol Mean deviation

NDP Per cent of vote for New Democratic party 28.6 9.7 Lib. Per cent of vote for Liberal party 37.2 8.0 PC Per cent of vote for Progressive Conservative party 34.2 8.7 Total Voters as a Per Cent of those eligible 72.0 7.0 FARM Per cent of population living on farm 7.2 17.5 OUTSIDE Per cent of population born outside Canada 18.6 4.5 c of E Per cent of population Church of England 26.2 6.8 RC Per cent of population Roman Catholic 17.0 5.9 AGED Per cent of population over 65 years of age 5.8 7.0 BRIT Per cent of population of British origin 75.9 7.0 PUB SCH Per cent of population with eight years or less of

schooling and not currently attending school 19.1 8.3 COLLEGE Per cent of population with at least one year of

college and not currently attending school 3.7 3.7 CHILDREN Per cent of population under 14 years 35.4 7.1

7A standard multiple regression program provided by the University of Western Ontario Computer Centre was used. For a discussion of the mathematics of correlation and regression analysis consult Hayward R. Alker, Jr., Mathematics and Politics (New York, 1965) or any introductory statistics text.

This content downloaded from 185.44.78.156 on Mon, 16 Jun 2014 12:58:31 PMAll use subject to JSTOR Terms and Conditions

Page 8: Voting Behaviour and Socio-Economic Characteristics: The Middlesex East Federal Election, 1965

TABLE III SIMPLE CORRELATION MATRIX (N = 233)

NDP Lib. PC Total FARM OUTSIDE c of E RC AGED BRIT PUB SCH COLLEGE CHILDREN

NDP 1.00 Lib. -.53 1.00 PC -.63 -.32 1.00 Total -.47 .30 .24 1.00

FARM - .38 .00 .43 .15 1.00 OUTSIDE .39 -.18 -.26 -.35 -.39 1.00 cofE .14 .03 -.18 -.03 -.56 .12 1.00 RC .28 .02 -.33 -.23 -.33 .30 .02 1.00 AGED -.25 -.01 .28 - .05 .07 .24 -.03 -.24 1.00 BRIT -.26 -.01 .29 .14 .15 -.43 -.02 -.38 .07 1.00 PUB SCH .06 -.32 .24 - .36 .37 .37 -.31 -.04 .51 -.10 1.00 COLLEGE -.45 .40 .12 .41 -.21 -.16 .28 -.21 .15 .16 - .48 1.00 CHILDREN .28 -.06 -.25 -.06 .04 -.34 -.02 .21 -.72 -.08 -.34 -.41 1.00

This content downloaded from 185.44.78.156 on Mon, 16 Jun 2014 12:58:31 PMAll use subject to JSTOR Terms and Conditions

Page 9: Voting Behaviour and Socio-Economic Characteristics: The Middlesex East Federal Election, 1965

396 JAMES W. SIMMONS

The rest of the correlation matrix relates electoral behaviour to socio- economic data. The New Democratic party is most sensitive to these variations as is indicated by the higher correlations. Table V shows that 49 per cent of the variation in NDP percentage is explained by four variables. The Con- servatives and Liberals are less closely associated with socio-economic vari- ables which explain 30 and 18 per cent of their variation respectively. The New Democrats are particularly weak in rural areas (FARMM),8 in older areas of the city (AGED ) ,9 and areas with the highest social class (COLLEGE ) ,10 but they show strength where there are foreign-born (OUTSIDE)," and Roman Catholics (RC).12

Only a small proportion of the variation in support for the Liberal party can be explained by socio-economic factors. They draw a consistent pro- portion of the vote in all polls and are particularly effective in areas with a high proportion of university educated (COLLEGE).13 The Conservatives, how- ever, vary with almost every socio-economic indicator. Strong in rural polls (FARM)14 and in polls with a large proportion of elderly (AGED),15 they are weak where there are foreign-born (OUTSIDE ) ,16 CATHOLICS (RC) 17 or SUBURBA-

NITES (CHILDREN).

Table IV lists the partial correlation coefficients, the effect of each socio- economic variable after the effects of the others have been eliminated. Several variables in the initial run are redundant and may be deleted. Examination of Tables III and IV led to the withdrawal of several independent variables so that the next round of regression analysis used only four: FARM, OUTSIDE,

AGED, and COLLEGE. Table V contains the results in the form of regression equations, in which each dependent variable: TOT, NDP, LIB, and Pc, in turn is expressed as a linear function of the independent variables. For instance, the NDP percentage: NDP = 27.7 - 0.21 FARM + 0.46 OUTSIDE - 0.28 AGED - 1.23 COLLEGE.

Note that the independent variables are of greatly different magnitudes (see the means in Table II). It is apparent that the effectiveness of socio- economic variables as predictors of voting behaviour is related to the total

8As shown by Alford, "Social Bases." 9Confirming Jewett, Regenstreif (each reference to this book refers to the series of tables ofn pp. 14, 33, 37 and 38), Anderson, and Courtney and Smith. ?0The lack of appeal of the CCF-NDP to the upper levels in education, occupation, and

income is universal: Alford Party and Society, Jewett, Regenstreif, Scarrow, Alford "Social Bases," and Courtney and Smith. 11Alford, "Social Bases," gives support to this finding, but Anderson found little support

for the NDP on the part of first-generation immigrants. 12This relationship is inconsistent with the other studies: Regenstreif, Alford "Social Bases," Anderson, and Courtney and Smith. -3Education is gradually emerging as a significant trait of Liberal supporters: see Perlin, Irvine, and Courtney and Smith. The weak correlations with Rc and OUTSIDE run counter to strong evidence from Meisel "Religious Affiliation," Regenstreif, Alford "Social Bases," Anderson, and Courtney and Smith. One might also have expected the suburban variable (CHILDREN) and its correlate (AGED) to have more effect: see Perlin, and Alford "Social Bases." 14Confirming Jewett, Alford "Social Bases," and Regenstreif. 15See Jewett, Perlin, Alford "Social Bases," Regenstreif, Anderson, and Courtney and Smith. 16As in Scarrow, Alford "Social Bases," and Anderson. 17As in Meisel "Religious Affiliation," Regenstreif, Perlin, Alford "Social Bases," Anderson, and Courtney and Smith.

This content downloaded from 185.44.78.156 on Mon, 16 Jun 2014 12:58:31 PMAll use subject to JSTOR Terms and Conditions

Page 10: Voting Behaviour and Socio-Economic Characteristics: The Middlesex East Federal Election, 1965

Voting Behaviour 397

amount of variation in the results for each party. The unexplained variation for the three parties is approximately the same and may indicate a hard-core of uncertainty which cannot be further probed using aggregate data. Error is introduced by use of enumeration areas which do not coincide with polls and out-of-date census material, in addition to the variation of socio-economic characteristics within polls and the psychological aspects of electoral be- haviours. The standard error of the regression estimate for each poll outcome is about 7 per cent for each party. However, the estimate of party success for a larger area, obtained by averaging results from a number of polls, has a much smaller standard error (OjN'2) equal to the standard error for one poll (o-) divided by the square root of the number of polls aggregated (N).

Despite the size of error these equations may be useful for predicting out- comes in other areas and other elections where information on social variables is available. Their application where different issues and different personalities modify the relationships between social characteristics and political parties depends on the stability of the regression parameters (Table V). If similar analyses in different areas and elections produce similar results or consistent variations in the parameters over time and space it would be possible to simulate the variations in electoral behaviour over time, or as a result of social change18 or redistribution.19

The regression equations also measure differences between ridings and elections, which also have utility for voting simulation. For example, consider the effect of the recent redistribution of constituencies in the London area. Middlesex East will undergo major changes (Figure 1) with the removal of the rural polls and the addition of some of the older working class areas of the City of London to form the new riding of London East, while the rural polls plus the old riding of Middlesex West form the new constituency of Middlesex. The voting outcome in the new London East riding could be obtained by aggregating the poll-by-poll results for the relevant parts of the old ridings of London and Middlesex East. However, the candidates and intensity of campaigning in the two old ridings were quite different so that the combination of the two sets of actual results provide an inaccurate estimate of the joint outcome.20 Alternatively we can apply the Middlesex East regres- sion parameters, reflecting the issues and personalities of that riding, to the polls from the London riding. The results of this simulation, shown in Table VI, provide an estimate of the outcome in London East under the conditions of the 1965 election.

Conclusions

The study is in general agreement with the results of other studies of voting

1:-Unfortunately, our knowledge of social change, as provided by the decennial census, generally lags behind our knowledge of voting changes. 19An interesting attempt at electoral simulation is Ithiel de Sola Pool, Robert P. Abelson, and Samuel L. Popkin, Candidates, Issues and Strategies: A Computer Simulation of the 1960 Presidential Election (Cambridge, Mass., 1964). 20The NDP, waging only a nominal campaign, obtained about 15 per cent of the vote in the adjacent riding of London, with the Liberals obtaining 37 per cent and the Con- servatives 46 per cent.

This content downloaded from 185.44.78.156 on Mon, 16 Jun 2014 12:58:31 PMAll use subject to JSTOR Terms and Conditions

Page 11: Voting Behaviour and Socio-Economic Characteristics: The Middlesex East Federal Election, 1965

TABLE IV PARTIAL CORRELATION COEFFICIENTS (N = 233)

FARM OUTSIDE C of E RC AGED BRIT PUB SCH COLLEGE CHILDREN

NDP -.33 .21 .02 -.10 -.10 -.04 .02 -.34 .08 Lib. .12 -.05 -.05 -.05 .14 -.09 -.16 .23 .01 PC .22 -.15 .04 .04 -.04 .13 .16 .12 -.08 Tot. .17 -.10 -.07 -.07 -.07 -.06 .24 .15 -.08

TABLE V REGRESSION EQUATIONS (N = 233)

Dependent- Original Explained Unexplained variable Intercept FARM OUTSIDE AGED COLLEGE variation variation variation R2

NDP 27.7 -.21 .46 -.28 -1.23 21707 10643 11065 0.49 (.03) (.12) (.07) (.14)

Lib. 36.7 +03 -(.14 - .07 +90 14861 2427 12119 0.18 (.031 (.13) (.07) (.14)

PC 35.0 +.19 -.29 +.34 +.33 17506 5219 12287 0.30 ( 03) (.13) (.08) (.14)

Tot. 74.9 +.06 -.32 -.07 +.80 11441 3058 8383 0.31 (.02) (.11) (.06) (.12)

This content downloaded from 185.44.78.156 on Mon, 16 Jun 2014 12:58:31 PMAll use subject to JSTOR Terms and Conditions

Page 12: Voting Behaviour and Socio-Economic Characteristics: The Middlesex East Federal Election, 1965

TABLE VI PREDICTED EFFECT OF REDISTRIBUTION

Total Voters' vote NDP Lib. PC

list % % % %

1965 vote in polls remaining in London East* 30,600 20,950 68.4 7,150 34.2 7,300 34.9 6,500 31.0 No. of voters in area to be added to London Eastt 11,150

Total vote predicted (regression equation) 7,350 65.9

Party support (regression equation) ?, 2,550 34.7 2,450 33.4 2,300 31.3

Totalvotes 41,750 28,300 67.7 9,700 34.3 9,750 34.5 8,800 31.2

*Unofficial results, 1965. tVoters list. tPredicted total vote was obtained by substituting socio-economic data for each new poll into the regression equation for per cent total vote Table V) and then multiplying the number on the voters list to obtain an estimate of the number of votes. ?Predicted vote for each party was obtained by applying the regression formulae in Table V to the socio-economic characteristics of each poll to obtain per cent of vote. The per cent vote was then multiplied by the number of voters obtained in the preceding step. jlActual votes in these polls: Total, 8,038; NDP, 1,450; Lib., 2,357; PC, 3,231.

This content downloaded from 185.44.78.156 on Mon, 16 Jun 2014 12:58:31 PMAll use subject to JSTOR Terms and Conditions

Page 13: Voting Behaviour and Socio-Economic Characteristics: The Middlesex East Federal Election, 1965

400 JAMES W. SIMMONS

behaviour in Canada, indicating the validity of the approach. There is an interesting variation in the relationships over time, with the earliest studies showing the Conservatives and Liberals virtually identical except for religious adherence,21 and later work indicating the development of distinctive occu- pational and demographic bases for their support,22 as is evident in Middlesex East.

The Middlesex East results differed most from the other studies in the relative weakness of religious adherence as an explanatory variable. This may reflect a unique situation in this riding or in the form of analysis, or-as a hypothesis for further work-a continuing decline over time of the religion- voting relationship. Fortunately, correlations based on ecological units are readily obtainable for other points in time.

2IMeisel "Religious Affiliation," (using 1953 and 1955 data), Regenstreif, and Alford Party and Society, chap. IX. 22AIford, "Social Bases," Perlin, Irvine, Regenstreif, and Courtney and Smith.

This content downloaded from 185.44.78.156 on Mon, 16 Jun 2014 12:58:31 PMAll use subject to JSTOR Terms and Conditions