comparing local electoral turnout in great britain and france

8/14/2019 Comparing Local Electoral Turnout in Great Britain and France

1/17

European Journal of Political Research 30: 241-257 (September 1996).996 Kluwer Academic Publishers. Printed in the Netherlands.

Comparing local electoral turnout in Great Britain and France:More similarities than differences?

VINCENT HOFFMANN-MARTINOT, COLIN RALLINGS &MICHAEL THRASHER2C ER VL -C N RS nstitut dEtudes Politiques de Bordeaux France; Local GovernmentChronicle Elections C entre University of Plymouth United Kingdom

Abstract. Electoral and census data from G reat B ritain and France are used in a comparativeanalysis of the levels an d determ inants of local election turnou t in the tw o countries. First, ninesimple propositions about variations in turnout are tested using as similar as possible variablesfor each country. These variables are then used as the bases for multi-variate regression analyses.The residual cases in each country are com pared in o rde r to explore the more qualitative factorswhich may explain why some localities have a level of participation so much above or belowthat exp ected statistically. Th e pape r en ds with an attempt to specify an exp lanatory model oflocal electoral turnout applicable to both countries.

IntroductionThe study of levels and trends in electoral turnout is a familiar theme forelections specialists in most Western states. Rates of political participationare seen, not least, as crucial indicators of the democratic health of thenation. Cross-national, comparative treatments of turnout are rarer, buthave been undertaken to emphasise the role that institutional and systemicvariables play in determining citizen participation Powell 1980; Morlan 1984;Blais & Carty 1990; Jackman & Miller 1995). Table 1demonstrates the starkvariations in turnout at the sub-national level, even among the membercountries of the European Union. It is obvious from these figures that factorssuch as the ease of voter registration; the proportionality of the electoralsystem used; and the normative pressures to cast a vote vary from state tostate and have an impact on levels of turnout.

What is less well understood, however, is the degree to which those factorswhich determine intra-national variations in electoral participation are similaracross countries. It is the purpose of this paper to test a number of hypothesesabout turnout with data drawn from municipal elections in England andFrance. Although the absolute levels of turnout at this type of election varyquite considerably between the two nations, there isprima facie evidence thatsimilar variables can explain variations by locality. An explicitly comparativemethod is adopted and the paper ends with an attempt to specify anexplanatory model of local electoral turnout applicable to both countries.


2/17

242Table 1 Average turnout in recent sub-national elections in EU countriesCountry Mean Y urnoutLuxembourg 93Sweden 90Italy 85Belgium 80Denmark 80France 68Ireland 62Greece 6Netherlands 54Great Britain 43

Germany 12Spain 64

Portugal 6

Source: Rallings, Temple & Thrasher (1994); various official sources and K eesings Contem por-ary Archive. Data unavailable for Austria and Finland. Figures are for the tier of governmentimmediately below the nation.

Data and methodThe French data comprise a record of voting turnout in all 382 metropolitani.e. non-overseas) municipalities with a 1983 population of more than 20,000inhabitants. Socio-economic data are taken from the 1982 French census

aggregated to the level of the municipality. The English electoral and censusdata are from 1991 and include a record of all 296 district councils, 36metropolitan districts and 32 London boroughs 1990 elections) with aminimum population of 25,000.

The same propositions are tested for each country with as similar a set ofvariables as possible.2 The results of the data analysis are discussed separatelyand comparatively for each proposition and in a final multi-variate regression.The residual cases in each country are then compared in order to explorethe more qualitative factors which may explain why some localities have alevel of participation so much above or below that expected statistically.

Testing some propositions about turnoutThe pattern and distribution of turnout in England and France seems quitesimilar, albeit that the mean level in France is some 17 percentage pointshigher. The histogram and frequency curve in Figure 1 reveals a meanturnout in France in 1983 of 64 percent with a median value of 64 percentand a standard deviation of 7 percent. Whereas about two-thirds of citizenswent to the polls in a city such as Verdun located at about the median, theturnout in the city with the highest participation rate Hazebrouk 85


3/17

243

45

41

M

32

27

n

i 1814

5

544 m 84 m 44

Figure 1 . Distribution of municipal voter turnout rates (France, 1983 .

percent) was almost twice that in the city at the bottom of the list (Saint-Martin-d Heres - 46 percent) .In England, as Figure 2 shows, the mean turnout across all 364 localauthorities was 47 percent, with a median value of 47 percent an d a standarddeviation of 5 perc ent. Few er than half the electors voted in a local authorityat the median such as Worcester, and more than twice the proportion ofelectors took part in the election in the authority with the highest turnout(Derbyshire Dales -59 percent) as in that with the lowest (Kingston-upon-Hull - 28 percent) .The first two hypotheses in relation to such inter-authority variations inturnout focus on the spatial stability of the local population and the ex tentto which individual authorities a re located within larger urban area s. It seem sa reasonable apriori assum ption tha t a stron g and widely shared identificationwith a municipality is likely to be associated with relatively high levels ofturnout . T he degree to which a population is stable over t ime app ears to b ea good surro gate variable at th e aggregate level for such feelings of identity.Hence:

Proposition : The greater the spatial stability of the popu lation, the higherthe turnout.Variables to test this proposition are readily available for bo th England andFrance. In France there is a simple correlation of 0.35 between the level


4/17

244

Voter turnout (in )Figure 2. Distribution of municipal voter turnout rates (Great Britain, 1990191).

of turnout among municipali t ies and the proportion of the population whowere living in the same authority eight years before the census in 1975.Similarly in Eng land , although th e d ata only com pare s residence at the timeof the census with just one year before, there is a correlation of +0.22between turnout and residential stability (see Table 2). This propositiontherefore receives modest support for both countries.Several researchers have noticed that political participation is higher inisolated, traditional communities than in densely populated urban centres(Wolfinger & Rosenstone 1980; Verba & Nie 1972). Although social andeconomic, rather than spatial, factors may account for some of this relation-ship, it is likely that the inhabitants of such metropolitan areas may haveless of an identification with th e form al municipaMoca1 gove rnme nt structureas opp osed t o thinking of themselves as residents of a larger, more nebulouscity. H en ce :Proposition 2: The less a local authority is part of a large urban area, thehigher the level of local turnout.In France this proposition is tested by examining both the proportion ofthe total urban area population contained within the municipality and theproportion of the municipal population whose place of work is in the localauthority area. On both measures, the relationship is in the expecteddirection with the more autonomous and less mobile areas being associated


5/17

245Table 2. Correlations between turnout and selected variables in Great Britain and France*Proposition Gr ea t Britain Coeff. France Coeff.1.2.

3.4 .56 .7 .

8.

9.

resident in sameauthority in 1990population per hectare

YO manual workers in thepopulationself-employedhomeownersYO self-described as o therthan whitestrength of largest party inauthority

number of electors percouncillor

0.21-0.39

-0.28+0.510.42-0.31-0.2

-0.44

resident in samecommune in 1975municipal population inurban arearesiding and working incommunemanual workers in thepopulationself-employedhomeownersYO foreignersnumber of party listcandidaciesyears local majority partyin powermunicipalities in Nord-Pas-de-Calaismunicipalities in Provence-Alpes-Cotes dA z u r b n g u e d o c -Roussillon-

0.350.25

+0.240.15

+0.460.39-0.42-0.230.210.320.29

* All Pearsons correlation coefficients are significant at the 0.01 level.

with higher levels of turnout see Table 2). Cities whose electoral participa-tion is amongst the highest often have clear and well-established spatial anddemographic borders - being either isolated communes or core centres ofurban areas whose population shows low residential or travel to work mo-bility. Cambrai and Abbeville, with turnout levels of 80 and 79 percentrespectively, are good examples. On the other hand, municipalities whichare merely dormitories for the population of a large urban area have muchlower rates of local electoral participation as in Vaulx-en-Velin and Heou-ville-Saint-Clair with recorded turnouts below 50 percent.

For England it was necessary to examine this proposition by using dataon population density. The number of inhabitants per hectare is an acceptablemeasure of the urban-ness of a local authority, although it does not tap thedimension of functional relations with nearby authorities. Nonetheless, againas expected, there seems a clear negative correlation of -0.39 betweenpopulation density and electoral turnout. However, it should perhaps bepointed out that Kingston-upon-Hull, despite forming an isolated urban areawith a stable population and little workplace migration, had a level of turnoutin 1991 nearly 5 percentage points less than that of the next poorest perfor-


6/17


7/17

247those members of the community with a clear financial interest in the wayin which it was governed. Local business people, in particular, are seen tobe affected by the policies pursued by local authorities and would thus wantto take a full part in the process of the selection of that authority. Indeedthe mayor of Grenoble seems to have been voted out of office in 1983 asthe result of the mobilisation of shopkeepers in that city Brechon 1983).

More generally, many of the most controversial local issues of recent yearsin both England and France have directly concerned this sector. There hasbeen concern about the parallel expansion of the number of pedestrianprecincts in town centres and of shopping malls on green field sites. Localfiscal policy, involving a four-fold increase at constant prices in taxationbetween 1960 and 1980 in France and the introduction of the poll tax anduniform business rate in England, has also helped to mobilise politically theself-employed.

A larger category of citizens with a direct interest in local taxation issuesare homeowners. Such people have always been associated with support forright of centre parties, but they are also more likely to be involved inmunicipal political activities and to have an above average level of electoralparticipation. Studies have shown this to apply even having controlled foroccupational status Brechon & Cautres 1987; Denver & Hands 1985; Kings-ton, Thompson & Eichar 1984). Hence:Proposition 4: The larger the proportion of the population who are self-employed, the higher the level of local turnout.andProposition 5: The larger the proportion of the population who are propertyowners, the higher the level of local turnout.These propositions are clearly supported in England and France for bothtypes of citizen. In each case the simple correlations are among the strongestin the array of data that we have examined. In France local rightist leadersexploited concerns about taxation levels among these groups at the 1983municipal elections and seem to have been rewarded with both electoralsuccess and a high rate of participation Hoffmann-Martinot 1988). In En-gland, the introduction of the poll tax in the late 1980s- designed to spreadt h e burden of local taxation among all citizens rather than for it to falldisproportionately on property owners - caused a strong electoral reaction.Although the philosophy driving the policy could crudely be summarised asone of no representation without taxation, in the event most small busi-nesses and property owners discovered that their total local tax burden rosequite sharply and they became among the taxs most vociferous and motiv-ated opponents.

It has been noted in both Great Britain and France that immigrants and


8/17

248their children have a fairly low level of integration into the establishedpolitical system (Schna pper 1991; An wa r 199 4). Give n this, we wou ld expectthe members of these groups to participate less in local elections. Hence:Proposition 6: Th e larger th e proportion of the popu lation from ethnic mino r-ity comm unities, t he lower the level of local turno ut.This proposition was measured in France by th e proportion of the populationof each municipality born abroad and with non-French parents. It producesa negative an d significant correlation with levels of turnout . In Eng land, theproposition is tested by identifying those in the community who do notdescribe themselves as white. The simple correlation suggests that thegreater numbers of non whites are associated with lower levels of turnout .So far our profile of the way different variables are associated with levelsof turnout in England and France has concentrated on socio-economic anddemographic factors. Although there are clear similarities between the twocountries, these variables in themselves can no t fully explain eith er inter- orintra-country patterns of turno ut. W e now turn the refore to a considerationof the impact of the political context on participation. Here we may expectmore variation in the results given the different political traditions of thetwo countries. T he initial approach remains on e of the com parison of simpletwo-way correlations.It may be assumed that electors in local authorities pay at least someattention to the activities of political parties. Levels of turnout will beinfluenced by th e na ture and visibility of th e political d eb ate , bo th nationallyand locally. Indeed in both England and France gross differences in turnoutlevels at municipal elections can often readily be interpreted by reference toth e national political scene. In Fra nce , for example, th e reduction in munici-pal turnouts between 1983 and 1989 seems at least partly a function of achange in the intensity with which the left-right cleavage was fought out. InEngland, very high average levels of turnout throughout the country at theheight of the controversy ove r th e poll tax in 1990 becam e very low averagelevels of turnout just two years later when the local elections took placebarely a month after the 1992 general election.Much more local political contexts can also have an impact on participa-t ion. The nu mbe r of candidates and the degree of electoral competitivenessin individual contests and across local authorities would each seem likely tobe related to turnout . H ence:Proposition 7: The more intense the local electoral competition, the higherth e level of local turnout.T he different political systems in Englan d a nd Fran ce re quire this propositionto be tested in rather different ways. In France there is some evidence thatcompetition between the lists of just two parties represents a more


9/17

249competitive situation than the appearance of multiple par tie^.^ Certainly th esimple correlation between the number of candidacies and turnout isnegative. Equa lly, howev er, all but 22 of the 382 municipalities under studyin 1983 had fou r or fewer l ists of candidates and the mean turnout levels foreach of these categories vary only slightly. A second surrogate measure ofelectoral competition is the length of time a given party has been in powerlocally. The implication being that the dominance of one party will lead toa non-competitive electoral system a nd thus to lower levels of turnout. Thisrelationship does ap pear to hold in the French si tuation.In England electoral competition is measured in terms of the numericalmajority of the largest party o n th e council ov er all oth er parties. O n e partyhaving a large majority over all others may be seen as indicative of a lesscompetitive local political system; no party having an overall majority of aclosely fought set of elections. As expected there is a negative relationshipbetween the strength of a party an d th e level of tu rno ut, but it is no t a strongone. However it might be argued that English voters will have a betterperception of the closeness of the contest in their own particular ward rath erthan on th e council as a whole, a nd it will b e the likelihood of affecting theresult at that level which will determine their participation. Unfortunatelyward level data are beyond the scope of the present paper.Although both Great Britain and France constitute examples of fairlycentralised st ate structure s, variations in local political participation may alsobe explained by aspects of local political culture. In the United States, forexample, it has been shown that controlling for other relevant variables,tu rn ou t in mu nicipal elections is higher o ver time in th e old cities of the Eastan d Midwest than elsewhere in the country (Alford & Lee 1968). In Englandwe have noted before that local turnouts are higher in London than in themetropolitan districts areas even though at general elections that pattern isreversed (Rallings & Thrasher 1992). In France the long history of auton-om ous municipal governm ents in two geographical extremes he area northof Paris an d th e Mediterranean south - might lead o ne t o expect th e survivalof a more intense local political life. Th e M editerranean are a is also charac-terised by a continuing clientalist political culture which produces a highmobilisation of voters a t local elections (M edard 1981). Hence:Proposition 8: The longer the history of autonomous municipal governmentin a region and/or the more that region is characterised by a clientalistpolitical cu lture , the higher the level of local turnout.This proposition can only effectively be tested for France. It would beunrealistic to separate parts of England on either of these dimensions, al-though it is likely that areas with above average rates of participation willbecome apparent from our later analysis of residuals and will be examinedthe n. In France two dumm y variables were created to distinguish com mu nesin th e Nord-Pas-de-Calais from all oth ers an d those in Provence-Alpes-Cotes


10/17

250dAzur and Languedoc-Roussillon from all others. In both cases there arepositive simple correlations with turn ou t. M oreo ver, nine of the 10municipal-ities with the best voting records in 1983 came f rom o ne o r o ther of theseregions, with 26 out of 32 communes in Nord-Pas-de-Calais and 37 outof 41 in Provence-Alpes-Cotes dAzur and Languedoc-Roussillon having aturnout level above the national median.One contextual aspect of English local politics which could be worthexamining concerns the relationship between voters and their elected repre-sentatives. The number of electors in each council seat varies widelythroughout the country and it may be the case that smaller electorates areassociated with higher levels of tu rno ut. H enc e:Proposition 9: T he smaller the average num ber of electors in wards within alocal auth ority , the higher the level of local turnout.This proposition is quite strongly confirmed for England. It suggests eitherthat the relatively close contact between electors and council candidatesencourages greater participation or, of course, that smaller wards may bedisproportionately concentrated in those types of area predisposed t o aboveaverage tu rnouts for o ther reasons.

Multi-variate analysisSo far we have demonstrated that most of the independent variables used toexamine our core propositions about variations in the level of turnout inFrance a nd G rea t Britain have been significantly correla ted with the dep en-de nt variable. Now it is necessary to evalua te the ir respective and cumulativeimpact on turnout independently of any inter-relationships that may existbetween the m . Th e statistical technique of mu ltiple regression analysis allowsus to identify both which of the range of independent variables contributeto o ur ability to predict turno ut an d what their relative importance is.T he regression analysis for F ranc e (see Table 3) shows only those variablesdeemed to have an individual and cumulative impact on the explanation ofturnout. All variables are as defined in Table 2 except for region whichtakes the value of 1 for municipalities within the Nord-Pas-de-Calais,Provence-Alpes-Cotes dAzur and Languedoc-Roussillon areas and formunicipalities elsewhere in France. The overall ability to predict variationsin turnout levels is quite high -R2 = 0.54. The region of residence; theproport ion of homeowners; the stability of the population; and the level ofparty competit ion each m akes an almost equa l and indepe nden t contributionto that proport ion of tur no ut which we can exp lain. C ities which score highlyon each of these dimensions do have the high predicted turnouts and thosewith the inverse characteristics d o have lower turn outs . Cases in point arethe example of Narbonne - predicted turnout 74.8 percent; actual turnout


11/17

25 1Table 3 Regression analysis of local e lectio n turnout: France 1983Variables B Betas T signif.

resident in same commune in 1975manual workers in the populationYOself-employedYOhomeownersYOforeignersnumber of party list candidaciesyears local majority party in powerregion

+0.19+0.06+0.35+0.17-0.24-1.86+0.07+5.7

+0.210.09+0.13+0.25-0.19-0.26+0.13+0.31

+4.41+1.93+2.98+6.29-4.43-7.05+3.44+7.81

O OO00.050.003O OO0O OO0O OO00.001O OO0

F = 55.98 S.E. .87 Adj. RZ= 0.54

Table 4 . Regression analysis of local election turnout: Great Britain 1991Variables B Betas T T signif.YOmanual workers in the population -0.07 -0.13 -2.8 0.005YOself-employed +0.3 +0.26 +5.14 O OO0YOhomeowners -0.0 +0.22 +4.75 o.Oo0number of electors per councillor 0.11 -0.27 -5.89 O OO0

F = 55.08 S .E. 4.07 Adi. R2 0.37

76.7 percent and Creil predicted turnout 52.1 percent; actual turnout54.1 percent.In the regression analysis for England see Table 4) only four variables

from the initial list prove to be independently significant. The ability topredict turnout is more modest than in France R2 0.37 - but the variablesof importance share similar characteristics. High levels of homeownershipand self-employment are each positively correlated with turnout, as separ-ately is the ratio of electors to councillors. Citizens really do seem to respondto the opportunity to vote in small electoral units. Turnout levels are close tothat specified by the regression equation in, for example, Caradon Cornwall)where there are above average numbers of homeowners and small businesspeople and a mean ward electorate size of less than 2,000 - predictedturnout 51 percent; actual turnout 51percent. Conversely, in Sandwell WestMidlands) which scores lower on such measures, the low level of turnout isalso predicted by the model - predicted turnout 39 percent; actual turnout39 percent.

Any regression model, however satisfactorily it acts as a predictor of thedependent variable, is only a best fit among all the cases available. Aninspection of plots of observed against expected values for turnout withinour regression equations for England and France shows most cases fallingrelatively close to the diagonal see Figures 3 and 4). The turnout in suchmunicipalities demands, in one sense, no further explanation. They behave


12/17

252

45 60 m 75 m 85 wAova l u lumrmt m in )

Figure 3 . Relationship between actual and predicted voter turnout (France, 1983).

10 0 70.0

60.0 m 0

XI 0 M 0

40 0 40 0

30.0 M.0

m.0 20 08

Actual voter turnou1 rate in 5 )Figure 4 . Relationship between actual and predicted voter turnout (Great B ritain, 1990/91).

according to the expectation of the m odel for their country. However th ereare plenty of local authorities whose levels of turnout are some way aboveor below th at which would have been predicted by th e models. T hes e outli-ers are worth examining to see if they provide clues to some of the morequalitative characteristics which may determine variations in turnout. Forboth countries we have restricted our analysis to a selection of those cases


13/17

253where the prediction is more than two standard deviations away from theactual turnout.

In France, three cities with a greater than predicted level of turnout areworthy of comment. In Draguignan residual = +13 percent) the electionswere characterised by an intense, even violent campaign by supporters ofthe long-serving mayor, Edouard Soldani, against the opposition candidate.In the event the Conseil dEtat nullified the election result because of seriousirregularities. In Sevres residual = +14 percent) parties of both the left andright were involved in the mobilization of the electorate. Both thought theycould win and both were using the elections as a trial of strength betweeninternal factions. In Abbeville residual = +11percent) a clientalistnetworkhad built up around the long-serving mayor, Max Lejeune. This ensuredactive support and participation from both the middle-class and municipalemployees.

Two municipalities with below expected turnouts were Saint-Martin-dHeres residual = -14 percent) and Dijon residual = -12 percent). Saint-Martin-dHeres is located in a Department with a history of high electoralabstention and had been controlled by one party, with no threat to its tenure,for almost 40 years. Dijon, similarly, presented a non-competitive electoralenvironment together with a divided opposition.

In England turnouts much above or below the expected levels tend to fallinto two categories. On the one hand, there are those municipalities whoseturnout is consistently deviant which suggests an almost cultural explanationfor their be ha ~ io ur .~n the other hand, each election produces instances oflocal authorities where particular circumstances produce one off unex-pectedly high or low turnouts. It is possible to specify examples of eachtype from our 1990/91 data. Kingston-upon-Hull residual = -11 percent) isconsistently the English authority with the lowest absolute level of turnout.In 1991 it also recorded the largest residual. Hull is a geographically isolatedcity with a stable population and a long-standing, one party dominant Labourcouncil. However, as the residual suggests, even these circumstances are notsufficient to explain its low levels of electoral participation. The authoritywith the highest turnout, Derbyshire Dales residual = +9 percent), has beendominated by the Conservatives in local elections but also has a history ofstrong electoral participation at both local and general elections. Indeed,four of the fifteen highest turnouts in Great Britain at the 1992 generalelection were recorded by constituencies in Derbyshire.

The local authority with the second largest positive residual +9 percent)was the London borough of Wandsworth. Its actual turnout level of 53percent easily exceeded that for all previous elections and placed it third topamongst all 36 boroughs. What happened in Wandsworth in 1990 is easilyexplained, bu t it involves nothing which could contribute to our regressionmodel. Simply, Wandsworth was in the cockpit of a dispute between centralgovernment and the local authorities over the new local tax, the poll tax.Because of favourable treatment by government, allied with a policy of


14/17

254

privatising the delivery of many local services, Wandsworth was able to levythe lowest poll tax in the whole country. This was favourably received byConservatives and furiously condemned by opposition parties. The subse-quent frenetic election campaign had a clear and untypical impact on turnout.By contrast, Hart Hampshire), an affluent middle-class local authority inthe south of England with party political competition in all its wards, had ahigh negative residual -9 percent) and an absolute level of turnout consider-ably less than at the equivalent elections four years previously. N o readyexplanation is available for this behaviour.

The question must now be asked whether these outliers have anything incommon and, if so, whether such characteristics could be included in animproved model of turnout? Our discussion of cases for both Britain andFrance suggests that it is qualitative rather than quantitative considerationsthat are influential. Clearly, the salience and visibility of the local campaignsand aspects of local political culture are important, but obtaining sufficientinformation of this kind for all authorities is likely to prove impracticable.For the foreseeable future it will be possible to operationalise such factorsonly as a post hoc way of explaining turnout variations and differences.

ConclusionWhat has been fascinating about our analysis is the way in which a goodproportion of the variance in turnout between local authorities in Englandand France can be explained by the examination of the same propositions.At the level of both simple correlation and regression equations similarvariables show themselves to be significant and to work in the same direction.Spatial factors, the vigour of party competition and the presence of propertyowners and business people all have a clear statistical influence on turnoutin both countries. The one area of divergence has to do with the role ofmanual workers. In England local authorities with heavy concentrations ofmanual workers tend to be associated with low levels of turnout; in Francethey demonstrate higher rates of electoral participation. This variation isprobably explained by the cultural differences in how parties and organis-ations of the left mobilise their supporters and in how they form and encour-age networks which encompass many aspects of a workers life.6 In otherrespects, however, those municipalities which tend towards high turnouts inEngland will show similar characteristics to those that do so in France, andvice versa.

Of course there are examples in both countries of authorities whose levelsof turnout fail to match that predicted by our quantitative models. Theexplanation for their behaviour has a more qualitative base, but again similarfactors seem to come to the fore. On the one hand, electorates in Englandand France do respond to especially high profile campaigns and issues toproduce exceptional turnouts on a one-off basis. On the other hand, some


15/17

255communes appear over time to have inculcated in their citizens a culturethat predisposes them to recurrent patterns of unexpectedly high or lowturnouts. Such instances are hard to predict statistically, but d o merit furtherstudy if we are to improve our model of turnout and, more important ly,understand the processes whereby electoral participation might be increased.However, there remains one crucial difference between England andFrance and that we have left until last. As i s apparent in Table 1 and inour subsequent more detailed discussion, local turnout levels in France areabsolutely higher than in En gland by some 15 to 25 percent. This appliesboth at the mean and in a comparison between high and low participatingmunicipalities in the two coun tries. Mo reover, such a contrast is not re pe atedin g eneral elections. Th e turn ou t in th e 1981 French Presidential election (72percent) was just 8 percent above the mean figure for the 1983 municipalelections. T h e turn ou t at th e British general election in 1992 (78 percent)was fully 31perce nt high er th an in the 1990-91 local elections. This similarityin national election turnouts suggests that the cross-national variations inlocal tur no ut a re unlikely to be accounted for simply by the different distribu-tion in England and France of those factors which we have shown to hav e avalue in explaining intra-nation variations. The explanation for them mustbe soug ht in a sepa rat e cross-national analysis of institutional and systemicvariables beyond the scope of this paper. However, some final speculativeremarks are in order.In F ran ce local governm ent has a more secure constitutional base. Munici-palities have wide discretion in the policies they pursue and mayors arefigures of significance who can attract both loyalty and disfavour. Citizensidentify with their communes and believe that their vote is worth casting inits own right (Thiebault 1976). In England local government is seen moreas a relatively powerless arm of central government whose decisions arecon strained . In such circum stances the incentive to vote is less (M iller 1988).Even the structural features of the system in France encourage participationwhen compared with those in England. Voting in France takes place on aSunday; in England on a normal working weekday - Thursday. In France,am end me nts to the tradit ional double ballot electoral system gua rantee re-presentation to all parties which obtain the support of ten percent of regis-tered electors in the first round and of five percent of voters in the secondround. In England the first past the post system of election heavily favoursthe largest party in an y locality a nd can lead to opposition groups, an d theirsup po rters , feeling permanen tly excluded from the d emo cratic process. T ha tcontrast alone is likely to account for a significant part of the difference inturnout .

Concern ove r the level of local turnout and that it may be declining hasbeen articulated in both countries. In a comparative sense, however, Francehas little t o worry ab ou t. Cho osing local governm ents is still something tha ta majority of citizens value sufficiently to participate in. Turnout levels inEngland are not unambiguously dropping, but the continuing low absolute


16/17

256level gives pause for thought. It may be that France can offer some pointersas to how this situation may be improved.

Notes1. See, for exam ple, Rallings & Thrasher (1990); Lancelot (1968); Hoffmann-Martinot (1992).2. Turnout in France is defined as the number of votes cast as a percentage of the adultpopulation. n England it is the number of votes cast as a percentage of the registeredelectorate.3. Crewe 1981 shows a clear correlation (+0 .55) to exist between the mean turnout rate in 19democracies since 1945 and the mean proportion of votes cast for the traditional working-

class party(ies).4. Indeed Capron & Krusem an (1988) suggest that in any election there is an optimum numb erof parties - oo few OR too many and electors will be disinclined t o participate.5. This claim for England is based on historical data contained in Rallings & Thrasher (1993).Similar information is not available for France.6. Even in Fran ce, thou gh, this variable is the least significantof those loading in to the regressionequation. It could be that trends towards working-class fragmentation in France might havelead to a result similar to that for England if the French data had been collected a decadelater.7. Experimental evidence for this has been produced in Great Britain by Bochel & Denver(1971) and Pimlott (1973).

ReferencesAlford, R. & Lee, E. (1968). Voting turnout in American cities, American Political ScienceAnwar, M. (1994). Race and elections. Wanvick: Centre for Research in Ethnic Relations.Blais, A & Carty, R. (1990). Does proportional representation foster turnout?, EuropeanBochel, J. & Denver, D. (1971). Canvassing, turnout and party support, British Journal ofBrechon, P. (1983). Comment expliquer Itchec de la municipalitt Dubedout h Grenoble,Brechon, P. & Cautres, B. (1987). Linscription sur les listes electorales: indicateur de socialis-Capron, H . & Kruseman, J.-L. (1988). Is political rivalry an incentive to vote?, Public ChoiceCrewe, I . (1981). Electoral participation, p p. 216-263 in D. Bu tler, H. Penniman & A. RanneyDenver , D. & Hands, G. (1985). Marginality and turnout in general elections in the 1970s.

Eagles, M. & ErfIe, S. (1989). Com mun ity cohesion a nd voter turnou t in English parliamentaryGiblin-Delvallet, B. (1986). Le Nord-Pas-de-Calais, in Y . Lacoste (ed.), Geopolitiques desHastings, M. (1991). Halluin la rouge 1919 1939: aspects dun comunisme identitaire. Lille:

Review 62: 796-813.

Journal of Political Research 18: 167-182.Political Science 1: 257-269.Cahiers dHistoire 28: 59-78.ation ou de politisation, Revue Francaise de Science Politique 37: 502-525.56: 31-43.(eds.), Democracy at the Polls. Washington: AEI.British Journal of Political Science 15: 381-388.constituencies, British Journal of Political Science 19: 115-125.regions francaises. Paris: Presses de le FNSP.Presses Universitaires d e Lille.


17/17

257Ho ffm ann -M artino t, V. (1988). Gestion modem iste a Nimes: construction dune image de ville,Hoffmann-Martinot, V. 1992). La Participation aux elections municipales dans les villes fran -Jackman, R.W. & Miller, R. (1995). Voter turnout in the industrial democracies during theKingston, P., Thompson, J. & Eichar, D . (1984). Th e politics of home ownership, AmericanLancelot, A . (1968). LAbsten tionnisrne electoral en France. Paris: A Colin.Medard, J.-F.1981). Political clientalism in France: the centre-periphery nexus re-examined,pp. 125-171 in S. Eisenstadt & R. Lemarchand (eds.), Political clientalism, patronage anddevelopment. Beverley Hills: Sage.

Les Annales de la Recherche Urbaine 38: 95-103.

caises, Rev ue Francaise de Science Politique 42: 3-35.1980s, Compara ive Political Studies 27: 467-492.Politics Qu arterly 12: 131-150.

Miller, W. (1988). Irrelevant elections? Oxford: Clarendon Press.Morlan, R. (1984). Municipal versus national election voter turnout: Europe and the UnitedStates, Political Science Quarterly 99Pimlott, B . (1973). Local party organisat t and marginality, British Journ al of PoliticalScience 3: 252-255.Powell, G.B. (1980). Voting turnout in thirty democracies: Partisan, legal and socio-economicinfluences, in R. Rose (ed.) , Electoral participa tion: A comparative analysis. Beverley Hills:Sage.Rallings, C. & Thrasher, M. (1990). Turnout in English local elections - an ag gregate analysiswith electoral and contextual data, Electoral Studies 9: 79-90.Rallings, C. & Thrasher, M. (1992). The impact of local electoral systems, Local GovernmentStudies 18: 1-8.Rallings, C. & Thrasher, M. (1993). Local elections in Britain: A statistical digest. Plymouth:Local Gov ernm ent Chronicle Elections Centre.

Rallings, C ., Temple, M . & Thrasher, M. (1994). Co mm un ity identity and participation in localdemocracy. Lond on: Comm ission for Local Democracy.Schn apper, D. (1991). La France de lintegration: Sociologie de la nation en 1990 Paris: Galli-rnard/NRF.Thieba ult, J.-L. (1976). Le s incertitudes du pouvoir local dans une ville mo yen ne. Lille: Univers-ite de Lille I1 thesis.Verba, S. & Nie, N. (1972). Participation in America: Political democracy and social equality.Chicago: Chicago University Press.Verba, S . Nie, N. & Kim, J. (1978). Participation andp olitica l equality:A seven-nation com pari-son Chicago: C hicago University Press.Wolfinger, R. & Rosenstone, S. (1980). W h o votes? New Haven: Yale University Press.

Address fo r correspondence: D r Colin Rallings, Dep artmen t of Politics, University of Plymouth,Plymouth PL4 8AA UKPhon e: +44 1752 233200; Fax: 44 1752 233194

comparing local electoral turnout in great britain and france

Documents