Supplementary Appendices (To Be Posted Online)

Supplementary Appendices (to be posted online).

Supplementary Appendix A. The organization of mayoral elections

Mayoral elections are held every six years in each of the 36,000 French towns and villages. However the voters, who must be at least 18 years old and French[1], do not directly elect the mayor. Instead they vote for the members of the municipal council who then elect the mayor and his/her deputies. Such a system is possible because candidates in mayoral elections compete on lists, with the head of the list being the would-be mayor of the town.

It must be noted that there are some differences in the organization of the mayoral elections, depending on whether the town has more or less than 3,500 inhabitants. The election procedure is also slightly different in the three largest French towns, i.e., Paris, Lyons and Marseilles.

In towns with less than 3,500 inhabitants, a two-round majority voting procedure is used to elect the members of the municipal council. In the first round, the absolute majority of the votes is needed to be elected, provided that at least a quarter of the registered voters voted. In the second round, only the relative majority is needed to win. Moreover, the members of the municipal council are elected on lists where the number of candidates is exactly to the number of positions in the municipal council. An exception is however made for towns with less than 2,500 inhabitants where individual candidacies and incomplete lists are allowed. Furthermore, in towns with less than 3500 inhabitants, voters can choose to vote for some, but not all the candidates, on a list.

However, in towns with more than 3500 inhabitants, voters cannot modify lists. In addition, a two-round majority voting procedure is also used in those towns but only complete lists of candidates are allowed to run. If a list obtains more than 50% of the votes in the first round, it receives half of the seats in the municipal council. The remaining positions are awarded between all the lists (including the winning list) following the proportional representation with the highest mean. If not, a second round is organized where it is usually the case that only two lists remain, for at least two reasons. First, only the lists which received more than 10% of the votes (out of the total number of registered voters) can run in the second round. Second, modifications of lists between the first and the second round are allowed. This usually allows two lists which are closely-aligned to merge between the first and the second round of the elections. Following the vote count of the second round, the list with the highest share of the votes obtains half of the seats in the municipal council. The remaining seats are awarded between all the lists (including the winning list) following the proportional representation with the highest mean.

Finally, in Paris, Lyons and Marseilles, the rules are similar to those in towns of more than 3500 inhabitants, except that elections are carried out by districts within these three towns. This means that the seats to the city council are attributed to the different lists, given the electoral outcomes in each district. Furthermore, the elections serve to designate members of district councils who elect a district mayor.

Supplementary Appendix B. Ruling parties in France, 1958-2007

Table SA1 provides the political affiliation of the Presidents of the French Republic, the Prime Ministers as well as the Presidents of the lower and upper houses of the French Parliament between 1958 and 2007.

[Insert Table SA1]

Supplementary Appendix C. The CE judges’ rulings before and after 1981

In this appendix, we demonstrate that the CE judges behaved differently when handing out decisions on the validity of controverted elections before and after 1981. For this purpose, we use the methodology of Bai and Perron (1998, 2003, 2006) which identifies changes between periods as breaks in the means of time series data.[2]

Our analysis focuses on two variables: the percentage of elections where the CE judges identified irregularities out of the total number of valid requests to the CE (the Cancelled Elections out of Valid Requests variable) as well as the percentage of cancelled elections out of the total number of valid requests to the CE (the Elections with Irregularities out of Valid Requests variable). In Table SA2, we provide the numbers of invalid and valid requests between 1958 and 2007, the numbers of elections with irregularities and of cancelled elections, as well as the percentages of irregular and cancelled elections out of the total number of valid requests. In addition, we give descriptive statistics for these two variables in Table SA3.

[Insert Table SA3]

The Bai and Perron (1998, 2003, 2006) method is a purely data-driven procedure which allows for a multiple number of breaks, and does not require any a priori assumption on the exact number or the location of breaks. In other words, we do not need to select historical events a priori but can let the model determine the events which mattered.

The Bai and Perron (1998, 2003, 2006) model consists of a multiple linear regression where the number of breaks is limited to m

with ,(A1)

with (A2)

…………………………………..

with (A3)

where is the observed dependent variable at time t, and are vector of covariates respectively of size and , and , with are the corresponding vector of coefficients, T the number of observations. The m break points are treated as unknown, and are estimated together with the unknown coefficients.

Two types of regression are possible: a partial change model suggests that the changes between the periods are limited while a pure change model implies that major changes take place. The above model is a partial change model since not all the coefficients are subjected to change: does not vary and is estimated over the entire sample. However, if , the model is a pure change model: all coefficients are allowed to change.

For simplicity, we estimate the partial change model as follows

with (A4)

while the pure break model is estimated as

with (A5)

In both equations (A4) and (A5), there are m+1 intercept terms denoted by so that the m breaks manifest themselves by shifts in the intercept terms.

In addition, to ensure that we correctly assess the number and the dates of the breaks in each series, we rely on the statistical tests suggested by Bai and Perron (1998, 2003, 2006). Thus, we use the “double maximum” UD and WD tests which allow us not to pre-specify a number of breaks. Both are tests of the null hypothesis of no structural break against an unknown number of breaks: UD and WD are respectively the weighted and the asymptotic versions of the “double maximum” tests.

Another way to determine the number of breaks is to employ the supFT(0|l) statistic, which is a Wald test assessing the presence of no break versus the alternative that there are l break(s), with l=1,…,m. If there is evidence that there is at least one break, the exact number of breaks can be computed by the supFT(l+1|l) test, whose null hypothesis is the existence of l break(s) against the alternative that there are l+1 breaks.

Finally, the number of breaks can also be determined by using information criteria such as the traditional Bayesian Information Criterion (BIC), the LWZ criterion of Jian et al. (1997), which is a modified version of the Schwarz's criterion, and the Bai-Perron Sequential Procedure, which relies on a sequential application of supFT(l+1|l) test as discussed by Bai and Perron (1998). For simplicity, we choose to report the number of breaks following the Bai-Perron Sequential Procedure. As a robustness check, we compute 95% confidence intervals for the lower and upper bounds of the occurrence of each break.

The results of our tests are shown in Table SA4. For each variable, i.e., Cancelled Elections out of Valid Requests and Elections with Irregularities out of Valid Requests, we test for the presence of structural breaks under a partial change and a pure change model. However, at this stage, a couple of caveats are in order because there are some econometric limitations to our study, as in any applied work. First, we limit the number of breaks to five, but this does not appear to limit the relevance of our analysis since we find less than five breaks in each series. Second, given the size of our dataset, which comprises 49 yearly observations, the econometric requirements of Bai and Perron (1998, 2003, 2006) approach impose that the minimum number of observations between two breaks equals 7 years.

[Insert Table SA4]

In three out of our four tests, the Bai-Perron Sequential Procedure, the BIC and the SupFT(.) tests suggest the presence of one structural break in our series, which occurred around 1980.[3] In other words, the Bai-Perron methodology for detecting structural breaks suggest that the judges’ decision patterns in ruling over elections with irregularities and cancelled elections were different between the 1958-1980 and 1981-2007 periods. It thus validates our approach to run separate regressions over these two periods.

Supplementary Appendix D. Robustness checks on cancelled elections

Table SA5 reports regressions which rely on the sample of the rulings where the CE did not use the vote differential to draw conclusions on the validity of the elections. They confirm the general partisan of the CE after 1981.

With respect to Table 6, Table SA6 includes interaction variables between the vote differential and the affiliation of the election winners in our baseline specification. The regressions in Table SA6 confirm that the elections of mainstream-right wing, mainstream left-wing and communist politicians were likely to be upheld except when the vote differential was small. However, none of the interaction variables is significant, thus suggesting that the CE judges did not treat differently elections won by mainstream left-wing or mainstream right-wing candidates.

[Insert Table SA5] [Insert Table SA6]

Table SA1. Presidents of the French Republic, Prime Ministers, and Presidents of the lower and upper houses of the French Parliament, 1958-2007.

Name / Year / Political affiliation
Presidents of the FrenchRepublic
Charles de Gaulle / 1958-1969 / mainstream right-wing (gaullist)
Georges Pompidou / 1969-1974 / mainstream right-wing (gaullist)
Valery Giscard d'Estaing / 1974-1981 / mainstream right-wing (center right Christian-Democrat)
Francois Mitterrand / 1981-1995 / mainstream left-wing
Jacques Chirac / 1995-2007 / mainstream right-wing (gaullist)
Prime Ministers
Michel Debré / 1958-1962 / mainstream right-wing (gaullist)
Georges Pompidou / 1962-1968 / mainstream right-wing (gaullist)
M. Couve de Murville / 1968-1969 / mainstream right-wing (gaullist)
Jacques Chaban-Delmas / 1969-1972 / mainstream right-wing (gaullist)
Pierre Messmer / 1972-1974 / mainstream right-wing (gaullist)
Jacques Chirac / 1974-1976 / mainstream right-wing (gaullist)
Raymond Barre / 1976-1981 / mainstream right-wing (center right Christian-Democrat)
Pierre Mauroy / 1981-1984 / mainstream left-wing
Laurent Fabius / 1984-1986 / mainstream left-wing
Michel Rocard / 1988-1991 / mainstream left-wing
Edith Cresson / 1991-1992 / mainstream left-wing
Pierre Beregovoy / 1992-1993 / mainstream left-wing
Edouard Balladur / 1993-1995 / mainstream right-wing (gaullist)
Alain Juppé / 1995-1997 / mainstream right-wing (gaullist)
Lionel Jospin / 1997-2002 / mainstream left-wing (socialist)
Jean-Pierre Raffarin / 2002-2005 / mainstream right-wing (gaullist)
Dominique de Villepin / 2005-2007 / mainstream right-wing (gaullist)
Presidents of the lower house of the French Parliament
Jacques Chaban-Delmas / 1958-1969 / mainstream right-wing (gaullist)
Achille Peretti / 1969-1973 / mainstream right-wing (center right-wing)
Edgar Faure / 1973-1978 / mainstream right-wing (center right-wing)
Jacques Chaban-Delmas / 1978-1981 / mainstream right-wing (gaullist)
Louis Mermaz / 1981-1986 / mainstream left-wing
Jacques Chaban-Delmas / 1986-1988 / mainstream right-wing (gaullist)
Laurent Fabius / 1988-1992 / mainstream left-wing (socialist)
Henri Emmanuelli / 1992-1993 / mainstream left-wing (socialist)
Philippe Seguin / 1993-1997 / mainstream right-wing (gaullist)
Raymond Forni / 1997-2002 / mainstream left-wing
Jean-Louis Debré / 2002-2007 / mainstream right-wing (gaullist)
Presidents of the upper house of the French Parliament
Gaston Monnerville / 1959-1968 / mainstream left-wing
Alain Poher / 1968-1992 / mainstream right-wing (center right Christian-Democrat)
Rene Monory / 1992-1998 / mainstream right-wing (center right Christian-Democrat)
Christian Poncelet / 1998-2008 / mainstream right-wing (gaullist)

1

Table SA2. Number of valid requests judged by the Conseil d’Etat, of elections with acknowledged irregularities and of cancelled elections.

Requests to the CE / Valid Requests to the CE / Elections with Irregularities / Cancelled Elections / Percentage of elections with Irregularities out of the total number of Valid Requests / Percentage of Cancelled Elections out of the total number of Valid Requests
1958 / 1 / 1 / 1 / 1 / 100 / 100
1959 / 7 / 7 / 6 / 5 / 85.71 / 71.43
1960 / 2 / 2 / 1 / 0 / 50 / 0
1961 / 18 / 18 / 17 / 9 / 94.44 / 50.00
1962 / 2 / 2 / 1 / 1 / 50 / 50
1963 / 1 / 1 / 1 / 1 / 100 / 100
1964 / 1 / 1 / 1 / 1 / 100 / 100
1965 / 5 / 4 / 2 / 0 / 50 / 0
1966 / 29 / 29 / 28 / 18 / 96.55 / 62.07
1967 / 17 / 17 / 16 / 11 / 94.12 / 64.71
1968 / 2 / 2 / 2 / 1 / 100 / 50
1969 / 3 / 2 / 2 / 1 / 100 / 50
1970 / 4 / 4 / 4 / 1 / 100 / 25
1971 / 7 / 7 / 5 / 0 / 71.43 / 0
1972 / 50 / 48 / 44 / 34 / 91.67 / 70.83
1973 / 17 / 17 / 16 / 12 / 94.12 / 70.59
1974 / 5 / 5 / 4 / 2 / 80 / 40
1975 / 3 / 3 / 3 / 1 / 100 / 33.33
1976 / 2 / 1 / 1 / 0 / 100 / 0
1977 / 37 / 32 / 31 / 23 / 96.88 / 71.88
1978 / 31 / 27 / 26 / 21 / 96.30 / 77.78
1979 / 1 / 1 / 0 / 0 / 0 / 0
1980 / 3 / 2 / 2 / 0 / 100 / 0
1981 / 0 / 0 / 0 / 0 / -- / --
1982 / 0 / 0 / 0 / 0 / -- / --
1983 / 11 / 9 / 9 / 6 / 100 / 66.67
1984 / 20 / 16 / 14 / 10 / 87.50 / 62.50
1985 / 0 / 0 / 0 / 0 / -- / --
1986 / 7 / 3 / 3 / 1 / 100 / 33.33
1987 / 11 / 7 / 6 / 3 / 85.71 / 42.86
1988 / 3 / 3 / 2 / 0 / 66.67 / 0
1989 / 171 / 159 / 114 / 71 / 71.70 / 44.65
1990 / 282 / 271 / 200 / 96 / 73.80 / 35.42
1991 / 14 / 12 / 10 / 4 / 83.33 / 33.33
1992 / 3 / 0 / 0 / 0 / -- / --
1993 / 8 / 4 / 3 / 1 / 75 / 25
1994 / 4 / 4 / 4 / 4 / 100 / 100
1995 / 12 / 4 / 2 / 0 / 50 / 0
1996 / 413 / 383 / 282 / 153 / 73.63 / 39.95
1997 / 25 / 24 / 21 / 10 / 87.50 / 41.67
1998 / 11 / 11 / 7 / 4 / 63.64 / 36.36
1999 / 3 / 3 / 3 / 1 / 100 / 33.33
2000 / 8 / 8 / 6 / 3 / 75.00 / 37.50
2001 / 84 / 57 / 34 / 21 / 59.65 / 36.84
2002 / 230 / 208 / 157 / 81 / 75.48 / 38.94
2003 / 14 / 13 / 8 / 1 / 61.54 / 7.69
2004 / 8 / 8 / 4 / 4 / 50 / 50
2005 / 2 / 2 / 2 / 0 / 100 / 0
2006 / 7 / 6 / 4 / 2 / 66.67 / 33.33

Note: General mayoral elections were held in 1959, 1965, 1971, 1977, 1983, 1989, 1995 and 2001. Because of the length of the appeal procedure, the CE only ruled on the validity of some elections one year, and in some cases even two years after election day.

1

Table SA3. Descriptive statistics: percentages of irregular and cancelled elections out of the total number of valid requests to the Conseil d’Etat

Observations / Mean / Std. Err / Min / Max
Cancelled Elections out of the total number of Valid Requests / 49 / 0.385 / 0.306 / 0 / 1
Elections with Irregularities out of the total number of Valid Requests / 49 / 0.746 / 0.301 / 0 / 1

Note: The detailed yearly data are given in Table SA2.

Table SA4. Structural break tests on the percentages of irregular and cancelled elections out of the total number of valid requests to the Conseil d’Etat

Cancelled Elections out of the total number of Valid Requests / Elections with Irregularities out of the total number of Valid Requests
Partial Break Model / Pure Break Model / Partial Break Model / Pure Break Model
Break Date / 1980 / No break / 1980 / 1980
Confidence Interval (95%) / [1973-1985] / [1974-1981] / [1967-1982]
Parameters
α1 / -0.14 / -0.24* / 0.881***
[0.104] / [0.138] / [0.034]
α2 / 0.585*** / 1.083*** / 0.645***
[0.08] / [0.117] / [0.056]
α3 / 0.327*** / 0.803***
[0.06] / [0.115]
Bai-Perron Sequential Procedure / 1 / 0 / 1 / 1
BIC criterion / 1 / 1 / 2 / 1
LWZ criterion / 0 / 0 / 0 / 0
SupFT(1) / 8.81** / 3.93 / 13.34*** / 12.26**
SupFT(2) / 5.52 / 3.54 / 8.98** / 6.02**
SupFT(3) / 4.13 / 2.75 / 10.35*** / 8.16
SupFT(4) / 2.79 / 2.15 / 8.004*** / 7.29
SupFT(5) / 2.46 / 1.91 / 6.64*** / 5.90
SupFT(2|1) / 2.13 / 2.97 / 7.74* / 4.11
SupFT(3|2) / 2.13 / 2.02 / 7.74 / 4.11
SupFT(4|3) / 0.53 / 0.82 / 0.48 / 0.17
SupFT(5|4) / 0.23 / 0.28 / 0 / 1.35
UD max / 8.81* / 3.93 / 13.34** / 12.26**
WD max / 8.81 / 4.21 / 14.90** / 13.82**

Notes:

 This Table presents structural break tests using the Bai and Perron (1998, 2003, 2006) methodology over the shares of irregular and cancelled elections out of the total number of requests to the Conseil d’Etat between 1958 and 2007.

 *** indicates the event is significant at the 1%-level; ** indicates significance at the 5%- level; * indicates significance at the 10%-level. Standard errors are given in brackets.

1

Table SA5. Characteristics of irregular and cancelled mayoral elections: robustness check on the rulings where the CE did not use the vote differential to draw conclusions on the validity of the elections.

Characteristics of cancelled mayoral elections (second step of Heckman's two-step method)
Under right-wing governments / Under left-wing governments / Under right-wing governments
1958-1981 / 1981-2002 / 1986-2007
(1) / (2) / (3) / (4) / (5) / (6)
Winner Communist / 0.152 / 0.103 / -0.04 / -0.017 / -0.317 / -0.32
[0.234] / [0.267] / [0.173] / [0.179] / [0.148]** / [0.146]**
Winner Mainstream Left Wing / -0.089 / -0.055 / -0.26 / -0.265 / -0.126 / -0.165
[0.176] / [0.273] / [0.113]** / [0.120]** / [0.103] / [0.109]
Winner Left Wing / 0.404 / 0.395 / -0.031 / -0.139 / 0.152 / 0.124
[0.283] / [0.490] / [0.218] / [0.237] / [0.150] / [0.152]
Winner Mainstream Right Wing / -0.092 / -0.043 / -0.172 / -0.2 / -0.341 / -0.348
[0.135] / [0.228] / [0.108] / [0.113]* / [0.085]*** / [0.089]***
Winner Right Wing / -0.107 / -0.053 / -0.174 / -0.155 / -0.157 / -0.2
[0.367] / [0.526] / [0.150] / [0.152] / [0.114] / [0.117]*
Senat / -0.044 / -0.109
[0.115] / [0.211]
Winner Far Right / -0.622 / 0.355 / 0.323
[0.525] / [0.473] / [0.474]
Winner Regionalist / -0.015 / -0.036 / -0.002 / -0.018
[0.222] / [0.231] / [0.285] / [0.277]
Cohabitation / 0.104 / 0.079
[0.080] / [0.086]
Constant / 0.607 / 0.479 / 0.836 / 0.818 / 0.787 / 0.743
[0.140]*** / [0.285]* / [0.135]*** / [0.149]*** / [0.099]*** / [0.103]***
Characteristics of mayoral elections with irregularities (first step of Heckman's two-step method)
Winner Communist / 6.167 / -0.521 / -0.424 / -0.281 / -0.31
[0.000] / [0.374] / [0.394] / [0.440] / [0.435]
Winner Mainstream Left Wing / 1.205 / 0.348 / -0.03 / 0.029 / -0.856 / -0.947
[0.904] / [0.669] / [0.319] / [0.327] / [0.320]*** / [0.330]***
Winner Left Wing / 6.509 / -0.266 / -0.303 / -0.387
[0.000] / [0.363] / [0.374] / [0.340]
Winner Mainstream Right Wing / -0.63 / 0.108 / -0.141 / -0.111 / 0.278 / 0.286
[0.964] / [0.644] / [0.368] / [0.375] / [0.295] / [0.257]
Winner Right Wing / 3.683 / -0.177 / -0.152 / -0.241
[0.000] / [0.390] / [0.395] / [0.318]
Winner Regionalist / 0.628 / 0.63 / 0.044
[0.501] / [0.506] / [0.617]
Winner Far Right / -0.227
[0.983]
Request Prefet / 0.334 / 0.259 / 1.557 / 1.454 / 1.333 / 1.257
[0.461] / [0.464] / [0.388]*** / [0.401]*** / [0.396]*** / [0.409]***
Request Voter / 0.451 / 0.499 / 0.267 / 0.227 / -0.412 / -0.32
[0.685] / [0.658] / [0.953] / [0.946] / [0.567] / [0.574]
Request Communist / 0.078 / -0.114 / 0.515 / 0.399 / -0.04 / 0.396
[0.938] / [0.760] / [0.511] / [0.528] / [0.635] / [0.732]
Request Mainstream Left Wing / 1.32 / 0.631 / 0.34 / 0.281 / 0.115 / 0.161
[1.031] / [0.676] / [0.391] / [0.399] / [0.304] / [0.271]
Request Mainstream Right Wing / -0.897 / 0.078 / 0.578 / 0.421 / 0.987 / 1.12
[0.938] / [0.663] / [0.320]* / [0.335] / [0.330]*** / [0.350]***
Request Left Wing / 0.316 / 0.29 / 0.525 / 0.443
[0.422] / [0.426] / [0.404] / [0.352]
Request Far Right / 0.223 / 0.337
[0.682] / [0.681]
Request CNCCFP / 1.373 / 1.456
[0.319]*** / [0.325]***
Senat / -0.56 / -0.454
[0.343] / [0.375]
Cohabitation / -0.107 / -0.116
[0.150] / [0.159]
Constant / 1.032 / 1.086 / -0.006 / -0.015 / 0.066 / -0.007
[0.195]*** / [0.214]*** / [0.099] / [0.103] / [0.107] / [0.107]
Observations / 130 / 115 / 390 / 356 / 368 / 331
Non-censored observations / 19 / 15 / 176 / 166 / 139 / 125
λ / 0.149 / 0.769 / -0.399 / -0.324 / -0.207 / -0.095
0.467 / 1.155 / 0.172 / 0.19 / 0.136 / 0.136
Wald chi2 test / 4.339 / 1.448 / 10.333 / 8.568 / 22.809 / 23.166
Prob > chi2 / 0.631 / 0.963 / 0.242 / 0.285 / 0.002 / 0.002

Notes:

• The regressions in Columns (1) and (2) focus on the rulings on the validity of requests to the Conseil d’Etat handed out between 1958 and 1981 under right-wing governments. As for the 1981-2007 period, Columns (3) and (4) focus on the rulings under left-wing governments (1981-1986, 1988-1993, 1997-2002) while Columns (5) and (6) focus on the rulings under right-wing governments (1986-1988, 1993-1997, 2002-2007). Columns (1), (3) and (5) show the regressions on the whole sample of general and by-elections, while Columns (2), (4) and (6) show the regressions on the sample restricted to the sole general elections.

• The number of observations corresponds to the number of rulings while the number of non-censored observations provides the number of rulings where the CE acknowledged the existence of irregularities.

• Standard errors are given in brackets. *** indicates significance at the 1%-level, ** indicates significance at the 5%-level, * indicates significance at the 10%-level.

Table SA6. Characteristics of irregular and cancelled mayoral elections: robustness checks including the interaction variables between the vote differential and the affiliation of election winners from the mainstream left-wing and mainstream right-wing parties.

Characteristics of cancelled mayoral elections (second step of Heckman's two-step method)
Under right-wing governments / Under left-wing governments / Under right-wing governments
1958-1981 / 1981-2002 / 1986-2007
(1) / (2) / (3) / (4) / (5) / (6)
Vote differential / -3.40E-04 / -3.20E-04 / -7.37E-05 / -6.92E-05 / -6.34E-05 / -0.0000923
[0.0001]*** / [0.0001]*** / [0.00005] / [.00005] / [0.0001] / [0.0001]
Winner Mainstream Left Wing * Vote Differential / 0.001 / 0.001 / -0.0001 / 0.0001 / -5.86E-06 / 0.00003
[0.001] / [0.001] / [0.0001] / [0.0001] / [0.0001] / [0.0001]
Winner Mainstream Right Wing * Vote Differential / 0.0001 / 0.0001 / -0.0001 / 0.0001 / -0.0001 / -0.0001
[0.0001] / [0.0001] / [0.0001] / [0.0001] / [0.0001] / [0.0001]
Winner Communist / 0.139 / 0.027 / -0.132 / -0.095 / -0.0004 / 0.066
[0.215] / [0.216] / [0.134] / [0.132] / [0.194] / [0.208]
Winner Mainstream Left Wing / 0.15 / -0.016 / -0.201 / -0.187 / -0.343 / -0.356
[0.187] / [0.199] / [0.112]* / [0.112]* / [0.112]*** / [0.120]***
Winner Left Wing / 0.081 / -0.085 / -0.329 / -0.381 / -0.133 / -0.045
[0.246] / [0.278] / [0.173]* / [0.174]** / [0.164] / [0.183]
Winner Mainstream Right Wing / -0.057 / -0.065 / -0.169 / -0.148 / -0.014 / -0.031
[0.119] / [0.126] / [0.102]* / [0.103] / [0.105] / [0.118]
Winner Right Wing / 0.275 / 0.187 / -0.151 / -0.141 / 0.173 / 0.21
[0.250] / [0.244] / [0.140] / [0.135] / [0.126] / [0.153]
Winner Regionalist / -0.382 / -0.364 / -0.076 / -0.067
[0.315] / [0.302] / [0.342] / [0.345]
Winner Far Right / 0.539 / 0.445
[0.361] / [0.360]
Senat / -0.141 / -0.229
[0.090] / [0.097]**
Constant / 0.739 / 0.946 / 0.931 / 0.868 / 0.453 / 0.584
[0.169]*** / [0.200]*** / [0.168]*** / [0.181]*** / [0.150]*** / [0.231]**
Characteristics of mayoral elections with irregularities (first step of Heckman's two-step method)
Winner Communist / -0.314 / -0.245 / -0.517 / -0.157
[0.325] / [0.341] / [0.452] / [0.436]
Winner Mainstream Left Wing / -0.359 / -0.411 / -0.165 / -0.112 / -0.741 / -0.259
[0.763] / [0.812] / [0.289] / [0.298] / [0.314]** / [0.299]
Winner Left Wing / -0.076 / -0.117 / -0.43
[0.313] / [0.317] / [0.338]
Winner Mainstream Right Wing / 0.911 / 0.773 / 0.062 / -0.005 / -0.835 / -0.322
[0.791] / [0.829] / [0.316] / [0.326] / [0.347]** / [0.286]
Winner Regionalist / -0.123 / -0.124 / -0.997
[0.618] / [0.614] / [0.704]
Winner Right Wing / -0.071 / -0.111 / -0.852
[0.342] / [0.348] / [0.347]**
Senat / 0.101 / 0.194
[0.310] / [0.338]
Request Prefet / -0.228 / -0.328 / 1.09 / 1.024 / 0.122 / 0.222
[0.521] / [0.530] / [0.432]** / [0.436]** / [0.577] / [0.576]
Request Voter / -1.101 / -1.174 / 0.651 / 0.627 / -0.702 / -6.084
[0.977] / [0.986] / [0.713] / [0.710] / [0.670] / [0.000]
Request Communist / -1.313 / -0.896 / 0.086 / 0.136 / 1.436 / 1.411
[0.951] / [1.045] / [0.484] / [0.489] / [0.568]** / [0.633]**
Request Mainstream Left Wing / -0.078 / -0.144 / 0.219 / 0.315 / 1.046 / 0.667
[0.762] / [0.797] / [0.326] / [0.342] / [0.326]*** / [0.269]**
Request Mainstream Right Wing / 0.829 / 0.763 / 0.697 / 0.618 / 0.916 / 0.574
[0.710] / [0.766] / [0.280]** / [0.292]** / [0.317]*** / [0.309]*
Request Left Wing / 0.139 / 0.135 / 1.117 / 0.533
[0.400] / [0.404] / [0.413]*** / [0.350]
Request Far Right / 0.922 / 0.461
[0.620] / [0.605]
Cohabitation / -0.179 / -0.123 / 0.404
[0.137] / [0.144] / [0.495]
Constant / 0.893 / 0.954 / 0.159 / 0.156 / 0.131 / 0.031
[0.207]*** / [0.229]*** / [0.093]* / [0.096] / [0.107] / [0.102]
Observations / 122 / 108 / 444 / 419 / 328 / 300
Non-censored observations / 19 / 15 / 179 / 168 / 141 / 127
λ / 0.206 / -0.267 / -0.421 / -0.326 / 0.166 / -0.027
0.479 / 0.613 / 0.232 / 0.254 / 0.202 / 0.295
Wald chi2 test / 19.524 / 18.859 / 19.94 / 20.167 / 24.972 / 26.732
Prob > chi2 / 0.021 / 0.026 / 0.018 / 0.017 / 0.005 / 0.003

Notes: