Content uploaded by Serkan Ünal
Author content
All content in this area was uploaded by Serkan Ünal on Apr 12, 2025
Content may be subject to copyright.
Form of government, electoral
system, and party system
fragmentation: A global
comparison
Serkan Ünal
Cankiri Karatekin University, Turkey
Abdullah Metin
Cankiri Karatekin University, Turkey
Carsten Anckar
Åbo Akademi University, Finland
Bilgehan Tekin
Cankiri Karatekin University, Turkey
Abstract
This study aims to determine the characteristics of party systems within the context of electoral
systems, forms of government, and continents. There is no study in the literature that quantita-
tively reveals the relationship between the party system and forms of government.
Furthermore, this study differs from other studies in that it deals with the relationship between
the electoral system and the party system on a global scale. In the study, the effective number
of parties (ENEP and ENPP) was calculated for the last three legislative elections of the countries
governed by presidential, president–parliamentary, premier–presidential and parliamentary forms
of government, using the Laakso–Taagepera Index. The dataset was then analyzed with ANOVA
and post-hoc tests. The study revealed that party systems do not differ significantly from forms
of government, that is, forms of government do not determine party systems. Similarly, party sys-
tems do not significantly differ in terms of the continents, that is, there are no continental party
Corresponding author:
Serkan Ünal, Faculty of Economics and Administrative Sciences, Department of International Relations, Cankiri Karatekin
University, UluyazıCampus, IIBF, Cankiri 18100, Turkey.
Email: serkanunal33@yahoo.com
Original Research Article
Asian Journal of Comparative Politics
2024, Vol. 9(2) 308–323
© The Author(s) 2023
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/20578911231166689
journals.sagepub.com/home/acp
system characteristics of the forms of government. Also, the study confirmed that it is the elect-
oral system that determines the party systems.
Keywords
effective number of parties, electoral system, form of government, Laakso–Taagepera index, party
system
Introduction
The first major theoretical work on the classification of party systems is generally attributed to
Duverger (1964), who divided parties into single-party, two-party, and multi-party categories,
taking into account the rate of votes. Following Duverger, a number of authors, such as Blondel
(1968; Dahl, 1966: 332; Golosov, 2010b; Sartori, 1976; Siaroff, 2000; Ware, 1996), developed cat-
egories based on the number of parties. Rokkan (1970) shifted the basis of the party system from the
percentage of votes to the number of seats. This means that the electoral system becomes very
important in the emergence of the party system.
The argument that plurality electoral systems generate two-party systems and proportional elect-
oral systems multi-party systems is old (e.g. Riker, 1986: 22–23). Duverger (1964: 206–255) pro-
vided two explanations for why two-party systems should emerge in units with plurality elections.
The mechanical effect refers to the rules that apply when shares of votes are transferred to shares of
seats. In a plurality system, only the candidate receiving the largest share of the votes receives a seat
in parliament, which means that smaller parties have no real chance of gaining representation. The
psychological effect, again, refers to the fact that as voters are aware of how the electoral system
operates, they refrain from voting for candidates for smaller parties and instead concentrate their
votes on the two largest parties.
The relationship between electoral system characteristics and party system fragmentation has been
assessed in numerous studies (e.g. Anckar, 2002; Croissant and Völkel, 2012; Ferland, 2014; Laakso
and Taagepera, 1981). Space does not allow a thorough exposition of findings from these studies;
suffice it here to say that in general, the claim that plurality electoral systems have fewer parties than
proportional systems wins support. However, this is far from a rule without exceptions, as demonstrated
notably by countries like India, Canada, and, nowadays, Britain.
The researchers examining the party system with regard to the effective number of parties (ENP)
have provided clearer data in order to understand the party systems and party system changes of the
countries. To reach this aim, Kesselman (1966), Rae and Taylor (1970), Wildgen (1971), Laakso
and Taagepera (1979), Molinar (1991), Dunleavy and Boucek (2003), and Golosov (2010a) formulated
ENP as a party system fragmentation index. The Laakso–Taagapera Index (L-T Index) has become the
most widely used index over time. This study is based on this index that counts the political parties in a
country according to their relative strength instead of their actual numbers.
The use of the L-T Index in the literature show a great variety when used to determine the party
system competitiveness (Brambor et al., 2007; Chhibber and Nooruddin, 2004; Kuenzi and
Lambright, 2005; Laakso and Taagepera, 1979; Lijphart, 1994; Mozaffar and Scarrit, 2005;
Neto and Cox, 1997; Paskhina and Telin, 2017; Schleiter and Voznoya, 2014; Taagepera and
Shugart, 1989) comparison of party systems (Lijphart et al., 1999). The index has also been
employed in the following contexts: changes in a party system (Quinn, 2013); the effects of elect-
oral misconduct on a party system (Donno and Roussias, 2012); the relationship between a party
Ünal et al. 309
system and an electoral system (Croissant and Völkel, 2012; Ferland, 2014; Laakso and
Taagepera, 1981); the relationship between ENP in the previous election and the subsequent
emergence of new parties (Kselman et al., 2016); coalitions in elections (Cox and Schoppa,
2002); presidential elections (ENPRES) (Hicken and Stoll, 2017; Ordeshook and Shvetsova,
1994); and the effect of presidential elections on legislative elections, and thus on the electoral
system in certain countries (Ferrara, 2011). As in this study, L-T Index has regularly been
employed for conducting comparative analyses of party systems. As we can see from the
above, this index has come to the fore as a measure that is used across a wide array of situations.
ENP and also L-T Index have been criticized in several ways. Sartori (1994: 35) criticizes
Taagepera and Shugart’s ENP measure for ignoring the impact of small parties on the formation
of governments; for him, some small parties such as the Free Democrats in Germany may be “pol-
itically pivotal”, meaning that some small parties may have power beyond their votes and seats.
Golosov (2010a) states that the L-T Index yields unrealistic results, in particular, when the
largest parties exceed 50%. For Blau (2008: 170), this potentially results in an underestimation con-
cerning the degree of variability in party systems. Taagepera (1999) accepts this insufficiency and
suggests a supplementary index, especially in those cases where a party exceeds 50% and therefore
dominates a large number of smaller parties. As squaring each vote or share of seat results in small
parties being ineffective compared with larger parties, the index is naturally more sensitive to larger
parties. A number of authors (Dunleavy and Boucek, 2003; Golosov, 2010a; Kline, 2009; Molinar,
1991; Taagepera, 1999) tried to modify ENP formula or proposed new measures in time and ENP
has more or less become a standard measure of party system fragmentation in the literature.
This study aims to examine the characteristics of party systems within the context of electoral
systems, forms of government and continents. The party system was accepted as the dependent
variable and the others as independent variables. Although a considerable number of studies
have been conducted on the formation of party systems (e.g. Golosov, 2010b; Sartori, 1976;
Siaroff, 2000) and the classification of the forms of government (e.g. Elgie, 1998; Metin and
Ünal, 2022; Shugart and Carey, 1992; Verney, 1959), the relationship between these two variables
has not yet been studied. There are no studies that calculate party systems via ENP and analyse their
relationship with the electoral system and its continental distribution on a global scale. Accordingly,
this study aims to fill these gaps in the literature.
Null and research hypotheses
The effective number of electoral parties (ENEP) is determined according to the vote shares of the
parties competing in an election, and the effective number of parliamentary parties (ENPP) is deter-
mined according to the legislative seats that the parties win. The number of parties represented in
the parliament cannot naturally be more than the number of parties competing the elections, and the
electoral systems fail to provide absolute fair representation for all parties. For all these reasons, our
first null (zero-test) and research (alternative) hypotheses are as follows:
H1
0
There is no significant difference between ENPP and ENEP.
H1
1
There is significant difference between ENPP and ENEP.
The study also examines the relationship between party system and form of government. The
seminal study on this subject in the literature is the one by Shugart and Carey (1992: 206–258).
310 Asian Journal of Comparative Politics 9(2)
They claim that party system fragmentation should be lower in presidential systems than in parlia-
mentary ones. The authors use Laakso and Taagepera’s (1979) “effective number of parties”as an
indicator of party system fragmentation and find that party system fragmentation is low particularly
in presidential systems with concurrent presidential and parliamentary elections, where propor-
tional elections are used in order to elect the parliament (Shugart and Carey, 1992: 220).
In this study, we acted on the most common forms of government in the world, namely presi-
dential, president–parliamentary, premier–presidential and parliamentary. The first two forms of
government are generally preferred to speed up the executive and decision-making process in deco-
lonized countries. For these forms of government, the number of political posts was either reduced
or their power was weakened. We expect lower ENP in these forms of government. For the parlia-
mentary form, on the other hand, we expect a higher ENP because the parliament is more active; we
also expect the number of parties to be high, as the parliamentary system broadly allows political
representation of social cleavages. Among these forms, the president–parliamentary system is gen-
erally adapted by those countries in which authoritarian regimes or hegemonic parties are dominant.
Of the 28 president–parliamentary countries present worldwide, 16 are located in Africa, seven in
Asia, four in Europe and one in Americas. As is commonly known, election boycotts are frequently
encountered in Africa. The authoritarian tendencies and the greater number of military coups in
Africa are likely to reduce both ENP and ENEP–ENPP in president–parliamentary forms. We
also predict ENPP to be lower in president–parliamentary and higher in parliamentary forms.
Considering all these factors, our null and research hypotheses as follows:
H2
0
∶There is no significant difference among the four different forms of government in terms of ENP.
H2
1
∶There is a significant difference among the four different forms of government in terms of ENP.
As noted above, any study on the requisites of party system fragmentation must also take into
account electoral system characteristics. Although both majoritarian and proportional systems can be
split up into many subcategories, too fine-graded categorizations should be avoided owing to the
limited number of cases in many of these categories. We have therefore opted to make use of three elect-
oral system categories: majoritarian systems, proportional systems, and mixed systems.
H3
0
∶There is no significant difference among the three different electoral systems in terms of ENP.
H3
1
∶There is significant difference among the three different electoral systems in terms of ENP.
Another intriguing question is whether there is a continental party characteristic of the forms of
government. Countries on one continent tend to be more similar to one another economically,
sociologically, and historically than they are to countries on other continents. These commonal-
ities—such as parliamentary monarchies in Europe, Cold War-Era military coups in Latin
America, similar periods and coups in Africa in the post-colonial period, similar political
systems in those countries that gained their independence from the USSR—cause political cultures
to resemble one another. We have stated that we expect lower ENP for the presidential and presi-
dent–parliamentary forms. Considering the fact that 17 out 45 countries in Africa are presidential,
16 are president–parliamentary, seven are premier–presidential, and five are parliamentary and also
considering the weakness of electoral democracy in Africa, and the multitude of political systems in
which the president is more active, we expect the lowest ENPP to be in this continent. As for
Ünal et al. 311
Europe, the continent hosts 18 parliamentary, 18 premier–presidential, four president–parliamen-
tary, and two presidential countries. Considering the existence of a large number of parliamentary
and premier–presidential forms in which parliaments are more active, we expect the highest ENP in
Europe. We hypothesize that party systems do not have continental characteristics, since electoral
systems also affect the party system in the Americas, Africa and Europe, where partial forms of
government characteristics are observed.
H4
0
∶There is no significant difference among the five different continents in terms of ENP.
H4
1
∶There is significant difference among the five different continents in terms of ENP.
Methods
Data collection method
We examined 159 countries, including 43 presidential, 21 president–parliamentary, 38 premier–
presidential, and 57 parliamentary forms (Metin and Ünal, 2022). Following Shugart and Carey
(1992), we divided semi-presidentialism into the categories of premier–presidential and presi-
dent–parliamentary and accepted them as main forms.
As single election result cannot provide statistical reliability, we took the last three election
results for each country into account. We calculated the effective number of parties for each election
in two ways: ENEP (Eff Nv) and ENPP (Eff Ns). We employed the following formula from Laakso
and Taagepera (1979), in the calculation:
N=1
n
i=1Pi2
where Nis effective number of parties and P
i
stands for the number of seats or votes for the ith
party. Although this formula gives reliable results in elections in which there are no independent
candidates or members of parliament and for which the data is complete, it gets away from reliabil-
ity in adverse situations. Realizing this problem, Taagepera suggested the following formula, in
which independents and missing data are not squared.
N=P2
[f(R)+Pi2]
In this equation, Pis the total number of seats or valid votes and Ris the residue of votes or seats
lumped as ‘Other’; we did not consider the independents and the missing votes as a single party and
used the second formula for such situations.
By processing the data, we considered the following principles:
▪The calculations were based on legislative elections.
▪In the case of a bicameral parliament, only lower house results were taken into account.
▪In the case of a two-round election, only the vote rates in the first round were taken into account.
▪In those countries where the elections are held at the constituency and proportional level, only
the latter was taken into account.
312 Asian Journal of Comparative Politics 9(2)
▪In the case of electoral alliances, if the voting rate of individual parties is unclear, the total vote
of the alliance was accepted as a single party.
▪Against all, or NOTA (None of the Above) options in the ballot paper were also accepted as a
single party.
▪Vacant seats and those seats that do not correspond to a certain percentage of votes or that are
specially reserved—such as for presidential appointees, and reserved seats for minorities, indi-
genous populations, ex-presidents and presidents’running mates—were not included in the
calculation.
▪In those countries that changed their forms of government, the form at the time of the election
was taken into account.
We had some difficulties in collecting the study data for several reasons. First, elections were not held
regularly or were postponed or canceled, which prevented us from obtaining data for seats and votes,
especially among countries in Africa and Asia. The missing data totals are as follows: 37 in Africa, three
in the Americas, 14 in Asia, three in Europe, and nine in Oceania. Second, in Africa, opposition parties
sometimes boycotted the elections, which results in a lack of data. Third, owing to electoral alliances, we
could not detect individual party votes in certain elections. Finally, countries with high independent
votes, such as Belarus, and those countries without political parties, such as Palau and Tuvalu, were
not included in the calculations because this makes the party system impossible to identify.
Data analysis method
In the study, we employed one-way analysis of variance (ANOVA), one of the parametric hypoth-
esis tests, in order to understand whether the party systems differ according to the forms of govern-
ment in the relevant country. ANOVA is a statistical tool used to determine differences between
experimental group means. It is used in research with more than two experimental groups for
one dependent variable and one or more independent (categorical) variables, which is a continuous
parametric numerical outcome measure. In ANOVA terminology, independent variables are called
“factors”and groups within each factor are called “levels”(Sawyer, 2009). The following hypoth-
eses are tested in the ANOVA analysis:
H
0
(null hypothesis): The means of all the groups are equal.
H
1
Not all the means are equal.
•Does not say how or which ones differ.
•Can follow up with “multiple comparisons”.
In the ANOVA analysis, a test statistic is calculated over the sample related to the hypothesis
being tested. Then, this value is compared with the distribution of test statistics under the null
hypothesis and converted into a p-value. This p-value is a measure of the probability that the
test statistic value is under the null hypothesis.
p-Value ≤α⇒Reject H
0
at level α(αcan be 1, 5 or 10%);
p-Value > α⇒Do not reject H
0
at level α.
Ünal et al. 313
The first step of the ANOVA analysis is to give descriptive statistics. In the second stage of the
ANOVA test, the homogeneity test of the intergroup variance is performed. The homogeneity con-
dition of the variances of the observed values in the ANOVA analysis must be met. Failure to meet
this requirement will result in incorrect results and interpretations.
As a result of the test, it is decided which of the intergroup difference tests will be performed,
depending on whether the significance value is greater than 0.05 or less. If this value is less than
0.05, ANOVA test results can be interpreted with appropriate post-hoc tests. As a result of the
ANOVA analysis, post-hoc tests are carried out in order to determine where the significant differ-
ences between the group averages originate. These tests differ according to whether the variances
between groups are homogeneous or not.
Statistical comparison of multiple algorithms on multiple datasets is performed with the
Friedman (ANOVA) test (Demšar, 2008). When the Friedman test rejects the null hypothesis
that there is no difference between algorithms, post-hoc analysis is performed to evaluate which
differences are significant (Benavoli et al., 2016).
We also employed Tukey HSD and Games-Howell tests, two types of the post-hoc tests, to test
the differences between the groups. Tukey HSD test is a test in which comparisons between vari-
ables are made after data is collected. Tukey test aims to test whether the relationship between two
data sets is statistically significant. In other words, the Tukey test is a way of testing an experimental
hypothesis (Nanda et al., 2021). As for the GamesHowell test, it is a non-parametric approach used
to compare group combinations. It also gives the most reliable results when the variances are not
homogeneous (Dunnet, 1980; Games and Howell, 1976; Keselman and Rogan, 1978; Osborne,
2008).
Findings
The results in Table 1 showed that H1
0
was rejected. That is, there is a significant difference
between ENEP and ENPP, as expected. Considering the descriptive statistics in Table 1, ENPP
with an average of 3.67 is lower than ENEP with an average of 4.95. The standard deviation (S)
of ENPP is also lower than that of ENEP. A low SD indicates that the deviation from the mean
is low, that is, the ENPP data are spread over a narrower range. The second part of Table 1
shows the correlation between ENEP and ENPP. The Pearson correlation coefficient is ,829.
This indicates that the relationship between the two variables is significant, positive and high.
Table 1. Paired samples t-test.
Descriptive statistics
χNS
ENEP 4,95 406 3,958
ENPP 3,67 406 2,549
Nr p
ENEP and ENPP 406 ,829 ,000
Paired samples test
χStp
ENEP –ENPP 1,281 2,332 11,070 ,000
314 Asian Journal of Comparative Politics 9(2)
Afterwards, the results of the ANOVA test were interpreted in the study. Before the interpret-
ation of the ANOVA test, descriptive statistics and homogeneity test results are included.
Sample number (N) and mean and standard deviation of descriptive statistics are given in
Table 2. A high standard deviation indicates that the deviation from the mean is high, so there
are data that are considerably lower or higher than the sample mean in the dataset. The results of
ANOVA analysis and post-hoc tests are interpreted according to the means.
In Table 3, the test results of homogeneity of variances are given. As a result of the tests per-
formed, it was seen that the intergroup variance was homogeneous in the government and electoral
systems (for this reason, the Tukey test was used as the multiple comparison test), while it was not
homogeneous in the continental factor (for this reason, the Games–Howell test was used as the mul-
tiple comparison test).
Table 2. Descriptive statistics.
Factor Dependent groups NχS
Form of government ENEP Presidential 112 4,66 3,201
President–Parliamentary 57 4,58 3,784
Premier–Presidential 82 5,34 2,838
Parliamentary 156 5,15 4,985
Total 407 4,98 3,995
ENPP Presidential 120 3,38 2,293
President–Parliamentary 78 3,31 3,477
Premier–Presidential 93 4,51 3,894
Parliamentary 162 4,05 3,212
Total 453 3,84 3,227
Electoral system ENEP Majoritarian 123 4,69 5,256
Proportional 199 5,38 3,518
Mixed 85 4,45 2,652
Total 407 4,98 3,995
ENPP Majoritarian 158 3,26 3,227
Proportional 204 4,24 2,529
Mixed 91 3,94 4,332
Total 453 3,84 3,227
Continent ENEP Africa 98 4,64 3,871
Americas 81 4,20 3,440
Asia 86 4,56 2,566
Europe 123 5,14 1,918
Oceania 19 10,85 11,393
Total 407 4,98 3,995
ENPP Africa 130 3,62 4,271
Americas 84 3,37 2,579
Asia 95 3,45 2,186
Europe 123 4,15 1,653
Oceania 21 7,07 6,120
Total 453 3,84 3,227
Ünal et al. 315
Tukey HSD and Games–Howell tests were carried out in order to determine which forms of
government caused significant differences as a result of the ANOVA analysis. The reason for
using the Tukey test is that the variances between groups are homogeneous and this test is
more sensitive to alpha error (Abdi and Williams, 2010; Rusticus and Lovato, 2014). The
Tukey test keeps the level of Type I error (i.e. finding a difference when it is absent) equal
to the chosen alpha level (for example, α=0.05 or α=0.01). The Tukey test also allows the
calculation of confidence intervals for differences between means (Abdi and Williams,
2010). As stated in the data analysis method part, the Games-Howell test, a nonparametric
approach used to compare group combinations, gives the most reliable results when the var-
iances are not homogeneous. Although its formulation is quite similar to the Tukey test, the
Games–Howell test does not take into account homogeneity of variances and sample sizes.
According to the results of the Tukey and Games–Howell tests in Table 4, forms of government
with significant differences in terms of ENPP are between presidential and premier–presiden-
tial, and between president–parliamentary and premier–presidential; electoral systems with sig-
nificant differences are majoritarian and proportional, and the continent significant difference is
Oceania.The value of these differences can be seen in the Mean Difference column.
According to the results given in Table 3, ENPP differs statistically significantly according to the
forms of government, electoral systems and continents. However, although ENEP does not differ sig-
nificantly according to the forms of government and electoral systems, there is a significant difference
between the averages according to the continents (F, 0.746; p, 0.525). So, the first H
0
hypothesis of the
study is rejected and the H
1
alternative hypothesis is accepted.
Table 4. ANOVA results.
Factor Dependent Sum of squares d.ff χ
2
Fp
Form of government ENEP 35,756 3 11,919 0.746 0.525
ENPP 95,734 3 31,911 3,107 0.026
Electoral system ENEP 66,751 2 33,376 2,103 0.123
ENPP 86,233 2 43,117 4,198 0.016
Continent ENEP 733,247 4 183,312 12,827 0.000
ENPP 270,925 4 67,731 6,839 0.000
Table 3. Homogeneity of variances.
Factor Dependent variable Levene statistic
Form of government ENEP 1,251
ENPP 1,513
Electoral system ENEP 2,505*
ENPP 0.529
Continent ENEP 53,895***
ENPP 13,283***
*, **, *** Significant at less than 10, 5 and 1%, respectively.
316 Asian Journal of Comparative Politics 9(2)
Also in the study, a dependent (paired) samples t-test was used between ENEP and ENPP and
significant results were obtained. This test compares the mean of two paired (or related) groups of
cases studied at two different points in time, or the mean of a single group. If the same group is
retested to the same extent, the t-test is called as a repeated measures t-test (Ross and Willson,
2017).
Evaluatian of findings
The aim of this study was to examine the characteristics of party systems according to the forms of
government, electoral systems and continents. The first null hypothesis (H1
0
) of this study, which
examines the party system over the effective number of parties, was that there is no significant dif-
ference between ENPP and ENEP. This hypothesis has been rejected, so the differences between
the variables has been detected, as expected. According to Table 1, the ENPP average is 3.67
and the ENEP average is 4.95. ENEP differs significantly from ENPP (significance, 000). These
results on a global scale show that absolute fair representation is not aimed at in any form of gov-
ernment or electoral system, on any continent. On the other hand, the ENEP/ENPP ratio of 1.28
indicates that the reasonable balance between governmental stability and fair representation is
maintained.
That the H2
0
null hypothesis regarding the relationship between the party system and the form of
government has been rejected means that there is significant difference among the four different
forms of government in terms of ENP for ENPP.This can be easily seen from the descriptive sta-
tistics data (see Table 1). Also, the post-hoc test showed that there was no significant difference
between forms of government in terms of ENEP (see Table 5). This means that the party system
Table 5. Multiple comparisons (post-hoc tests).
Factor Dependent (I)(J) Mean difference (I−J) Standard error
Form of government ENPP 1 2 0.066 0.466
3−1,130* 0.443
4−0.668 0.386
21 −0.066 0.466
3−1,196* 0.492
4−0.735 0.442
3 1 1,130 0.443
2 1,196* 0.492
4 0.461 0.417
4 1 0.668 0.386
2 0.735 0.442
3−0.461 0.417
Electoral system ENPP 1 2 −0.978* 0.340
3−0.675 0.422
2 1 0.978* 0.340
3 0.303 0.404
3 1 0.675 0.422
2−0.303 0.404
*10% significance level.
Ünal et al. 317
is not determined by forms of government. In other words, it is not possible to find party system
characteristics in forms of government. The following two conditions must be met for a character-
istic to be determined: first, values of each form of government diverge significantly from one
another; second, values of countries that adopt a particular form of government have little or no
overlap with those that adopt another form of government. On the basis of these conditions,
Table 2 shows that “countries with the same form of government do not have similar party
systems”. For four forms of government, the mean of ENEP varies between 4.66 and 5.33,
while the mean of ENPP varies between 3.31 and 4.50. The median ENEP for these four forms
ranges between 3.20 and 4.43 for ENEP and between 2.25 and 3.58 for ENPP. Considering the
closeness of the values, as well as the intertwining of the maximum and minimum values of
each form of government, no characteristics could be determined.
Despite the aforementioned analysis, the findings enable us to assess the relationship between the
form of government and the party system, albeit partially. The mean ENEP and ENPP for the presi-
dential and president–parliamentary forms are lower than for the premier–presidential and parliamen-
tary forms. In other words, the mean is low in those forms in which the president is more active, and
the mean is high in those forms in which the prime minister is more active. We hadpredicted the mean
to be lower for the president–parliamentary form and highest for the parliamentary form. According to
Table 2, the first part of our prediction came true. However, the highest mean appeared in the premier–
presidential form, contrary to our expectations. We do not know whether the reason for this is elect-
oral systems or social cleavages, but the results show that the form with the highest party fragmen-
tation is the premier–presidential and, according to the post-hoc test, it differs significantly from the
president–parliamentary and presidential forms in terms of ENPP.
While there is no significant difference among forms of government on the basis of ENEP, the
significant results on the basis of ENPP suggest that the electoral systems may cause separation
rather than the voting behavior of the voters. In this case we need to look at electoral systems
implemented in forms of government (Table 6).
Considering the fact that the majoritarian system causes low ENPP and the proportional system
causes high ENPP (see Table 1), when we look at the electoral system applied in forms of govern-
ment, while the majoritarian and proportional distributions are very close to each other in presiden-
tial, president–parliamentary and parliamentary forms, the proportional system is clearly dominant
in the premier–presidential form, which increases the ENPP average.
The fact that party fragmentation is low in president–parliamentary and presidential forms can be
interpreted as these systems being less encouraging in terms of political participation and party
organization.
Table 6. The relation between forms of government and electoral systems.
Electoral system
TotalMajoritarian Proportional Mixed
Form Presidential 15 17 10 42
President–parliamentary 12 12 4 28
Premier–presidential 5 20 6 31
Parliamentary 24 20 11 55
Total 56 69 31 156
318 Asian Journal of Comparative Politics 9(2)
The H4
0
hypothesis has been rejected and the relationship between the party system and contin-
ent is statistically confirmed. Therefore there is significant difference among the five different con-
tinents in terms of ENP, for both ENEP and ENPP.According to post-hoc analyses, the only
meaningful result on the basis of ENEP and ENPP is the situation of Oceania compared with
other continents. Only nine of the 14 countries in Oceania (eight parliamentary and one presiden-
tial) were included in this study. The scarcity of data, the immaturity of the party system and there-
fore the excess of independent votes, the missing data and the high standard deviation caused
Oceania to differ from other continents. For this reason, when we exclude Oceania, there is no
meaningful result on the basis of ENEP or ENPP among the continents. While formulating the
hypothesis, we stated that we expect the highest ENP in Europe owing to the high number of
premier–presidential and parliamentary forms (36 out 42), and the lowest ENP in Africa owing
to the high number of presidential and president–parliamentary forms (33 out 45). Although
Table 1 confirms our expectations for Europe, the continent with the highest average on ENEP
and ENPP is the Americas. It is quite surprising that in the Americas, where the number of parlia-
mentary and premier–presidential forms is relatively high (13 out 30), the ENP is lower than in
Africa, whose electoral system is mostly majoritarian. At that rate, the median becomes an import-
ant indicator for evaluating the African party system. In Africa, the median is 2.25 based on ENPP,
the lowest rate in the entire dataset. In this case, it is the highly uneven distribution of the data, that
is, the high S, which puts it behind the Americas in terms of mean (Table 7).
It can be said that the overwhelming preference for the proportional system in Europe has
increased the ENP of the continent. Europe has the highest level of ENPP in regard to both its
mean and median. Europe also comes to the fore with its relatively regular distribution and low
standard deviation (S) in regard to ENEP and ENPP. As Oceania has a small number of countries,
its values fluctuate considerably and its S, in regard to both ENEP and ENPP, is quite high. In par-
ticular, the huge difference between mean and median concerning ENEP indicates a large number
of small parties in Oceania.
As for the relationship between the party system and the electoral system, H3
0
is rejected and it
is concluded that there is a significant difference among the three different electoral systems in
terms of ENP for ENPP. While the differentiation of the majoritarian system at the ENEP level
according to the proportional system and the mixed system is not significant, it is significant at
the ENPP level compared with the proportional system (significance 012; see Table 5).
Interestingly, the mean of ENEP was lowest in the mixed system, not in the majoritarian
Table 7. The relation between continents and electoral system.
Electoral system
Majoritarian Proportonal Mixed Total
Continent Africa 21 13 10 44
Americas 13 12 3 28
Asia 13 9 11 33
Europe 3 32 7 42
Oceania 6 3 0 9
Total 56 69 31 156
Ünal et al. 319
system, contrary to our expectations. However, the ENPP mean, as expected, was lowest in the
majoritarian system and highest in the proportional (see Table 2). This shows that the party
system is determined by the electoral system rather than the voting behavior of the voters. In
other words, whether the electoral system provides fair representation does not definitely/neces-
sarily guide the voting behavior.
Discussion and conclusion
The purpose of this study was to determine the party system characteristics within the context of
electoral systems, forms of government, and continents. Using the Laakso–Taagepera Index, we
calculated ENEP and ENPP results of the last three lparliamentary elections of the countries gov-
erned by presidential, president–parliamentary, premier–presidential and parliamentary forms of
govenment, and then analyzed them with ANOVA and post-hoc tests.
It was seen that the party system did not have a meaningful competition with the form of gov-
ernment and continent. However, it was mentioned above that the electoral system is an important
factor in small divergences that allow interpretation. It has been revealed that the electoral system
significantly determines the party system. This finding is well in line with previous research in the
field.
The main point of interest of the study concerned the relationship between the party system and
the form of government. The findings revealed that no party system characteristic is specific to any
form of government; that is, it is not possible to consider the form of government as an independent
variable and the party system as a dependent variable. Thus, the main hypothesis of the study is
rejected. This finding constitutes an important contribution to the research on regime forms.
During recent decades, authors have discussed positive and negative consequences of presidential-
ism in particular. The present contribution clearly lays out that variations in regime forms do not
affect party system fragmentation.
We have underlined that there was no study in the literature that quantitatively reveals the rela-
tionship between party systems and forms of government. While not addressing this relationship
directly, Mainwaring (1993) and Linz (1994) tried to link presidentialism with stable democracy,
not party systems. Linz claimed that presidentialism was inimical for democratic stability while
Mainwaring argued that it was the combination of presidentialism and multipartism that was inimi-
cal. The findings of this study are not similar to those of Mainwaring and Linz. Since there is no
significant relationship between regime form and party system fragmentation according to the
ANOVA and post-hoc tests, our study indicates that the form of government is not connected to
party system fragmentation.
During the collection of the data, we learned more about political culture factors affecting the
emergence, shaping, and change of party systems. As we see, many factors affect, either directly
or indirectly, the party system, such as the desire of leaders to hold on to power, voters’willingness
to participate in politics, the availability of procedural democracy tools, the effectiveness of the
electoral security institutions, the recognition of political rights, and the participation of the free
or state-controlled press in this process. However, it was not possible to digitize and include
these factors in the numerical results of this study. Therefore, we state that although political
culture has a direct impact on electoral system, none of the numerical results in this study can be
interpreted as determining a level of development of political culture. The same numerical
results do not correspond to the same level of political culture, and it is for this reason that we
refrained from making assertive interpretations on the aforementioned continental results.
320 Asian Journal of Comparative Politics 9(2)
We experienced some difficulties in obtaining the study data, such as those arising from the
ambiguity of the party votes, electoral protests and the excessive number of independent votes.
Although these shortcomings are likely to change the study findings somewhat, we think that it
does not affect the veracity of the results.
This study is based on data from the past three elections in 159 countries. Further studies may
only focus on those countries that have changed their form of government, and so may reveal
whether this change also results in them changing their party system. This study dealt with the
party system through investigating data on legislative elections. A further study to be conducted
on presidential elections may yield significant results regarding the comparison of the party
system in presidential, president–parliamentary, and premier–presidential forms of government.
Data availability
The dataset was generated by the authors.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publi-
cation of this article.
Ethics
This manuscript has not been published or presented elsewhere in part or in entirety and is not under consid-
eration by another journal. We have read and understood your journal’s policies, and we believe that neither the
manuscript nor the study violates any of these.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
ORCID iDs
Serkan Ünal https://orcid.org/0000-0003-0235-5780
Abdullah Metin https://orcid.org/0000-0003-4426-6380
References
Abdi H and Williams LJ (2010) Newman–Keuls test and Tukey test. Encyclopedia of Research Design 2: 897–
902.
Anckar C (2002) Effekter av valsystem. En studie av 80 stater. Stockholm: SNS förlag.
Benavoli A, Corani G and Mangili F (2016) Should we really use post-hoc tests based on mean-ranks? The
Journal of Machine Learning Research 17(1): 152–161.
Blau A (2008) The effective number of parties at four scales: Votes, seats, legislative power and cabinet power.
Party Politics 14(2): 167–187.
Blondel J (1968) Party systems and patterns of government in western democracies. Canadian Journal of
Political Science 1(2): 180–203.
Brambor T, Clark WR and Golder M (2007) Are African party systems different? Electoral Studies 26:
315–323.
Chhibber P and Nooruddin I (2004) Do party systems count? The number of parties and government perform-
ance in the Indian states. Comparative Political Studies 37(2): 152–187.
Cox KE and Schoppa LJ (2002) Interaction effects in mixed-member electoral systems. Comparative Political
Systems 35(9): 1027–1053.
Ünal et al. 321
Croissant A and Völkel P (2012) Party system types and party system institutionalization: Comparing new
democracies in east and Southeast Asia. Party Politics 18(2): 235–265.
Dahl RA (1966) Patterns of opposition. In: Dahl RA (ed) Political Oppositions in Western Democracies. New
Haven, CT: Yale University Press, pp. 332–347.
Demšar J. 2008. On the appropriateness of statistical tests in machine learning. In: Workshop on Evaluation
Methods for Machine Learning in Conjunction with ICML, p.65.
Donno D and Roussias N (2012) Does cheating pay? The effect of electoral misconduct on party systems.
Comparative Political Studies 45(5): 575–605.
Dunleavy P and Boucek F (2003) Constructing the number of parties. Party Politics 9(3): 291–315.
Dunnet CW (1980) Pairwise multiple comparison in the homogenous variance, unequal sample size case.
Journal of the American Statistical Association 75: 789–795.
Duverger M (1964). Political Parties (trans. Barbara North and Robert North). New York: Methuen.
Elgie R (1998) The classification of democratic regime types: Conceptual ambiguity and contestable assump-
tions. European Journal of Political Research 33: 219–238.
Ferland B (2014) How do voters’strategic behaviors mediate the impact of electoral systems on the effective number
of electoral parties? An experimental study. Journal of Elections Public Opinion and Parties 24(3): 265–290.
Ferrara F (2011) Cleavages, institutions and the number of parties: A study of third wave democracies. Journal
of Elections, Public Opinion and Parties 21(1): 1–27.
Games PA and Howell JF (1976) Pairwise multiple comparison procedures with unequal n’s and/or variances:
A Monte Carlo study. Journal of Educational Statistics 1(2): 113–125.
Golosov G (2010a) Effective number of parties: A new approach. Party Politics 16(2): 171–192.
Golosov G (2010b) Party system classification: A methodological inquiry. Party Politics 17(5): 539–560.
Hicken A and Stoll H (2017) Legislative policy-making authority, party system size, and party system nation-
alization. Electoral Studies 47: 113–124.
Keselman HJ and Rogan JC (1978) A comparison of the modified-Tukey and Scheffe methods of multiple
comparisons for pairwise contrasts. Journal of the American Statistical Association 73(361): 47–52.
Kesselman M (1966) French local politics: A statistical examination of grass-roots consensus. American
Political Science Review 60(4): 963–973.
Kline R (2009) How we count counts: The empirical effects of using coalitional potential to measure the effect-
ive number of parties. Electoral Studies 28(2): 261–269.
Kselman DM, Powell EN and Tucker JA (2016) Crowded space, fertile ground: Party entry and the effective
number of parties. Political Science Research and Methods 4(2): 317–342.
Kuenzi M and Lambright G (2005) Party systems and democratic consolidation in Africa’s electoral regimes.
Party Politics 11(4): 423–446.
Laakso M and Taagepera R (1979) Effective number of parties: A measure with application to west Europe.
Comparative Political Studies 12(1): 3–27.
Laakso M and Taagepera R (1981) Proportional representation and effective number of parties in Finland. In:
Holler MJ (ed) Power, Voting, and Voting Power. Würzburg: Physica, pp. 107–120.
Lijphart A (1994) Electoral Systems and Party Systems. Oxford: Oxford University Press.
Lijphart A, Bowman PJ and Hazan RY (1999) Party systems and issue dimensions: Israel and thirty-five other
old and new democracies compared. Israel Affairs 6(2): 29–51.
Linz J (1994) Presidential or parliamentary democracy: Does it make a difference? In: Linz J and Valenzuela A
(eds) The Failure of Presidential Democracy, Vol. I. Baltimore, MD: Johns Hopkins University Press,
pp. 3–87.
Mainwaring S (1993) Presidentialism, multipartism and democracy: The difficult combination. Comparative
Political Studies 26(2): 198–228.
Metin A and Ünal S (2022) Classifying forms of government on a global scale. Asian Journal of Comparative
Politics: 20578911221127176.
Molinar J (1991) Counting the number of parties: An alternative index. The American Political Science Review
85(4): 1383–1391.
322 Asian Journal of Comparative Politics 9(2)
Mozaffar S and Scarrit JR (2005) The puzzle of African party systems. Party Politics 11(4): 399–421.
Nanda A, Mohapatra BB, Mahapatra APK, et al. (2021) Multiple comparison test by Tukey’s honestly signifi-
cant difference (HSD): Do the confident level control type I error. International Journal of Statistics and
Applied Mathematics 6: 59–65.
Neto OA and Cox GW (1997) Electoral institutions, cleavage structures and the number of parties. American
Journal of Political Science 41(1): 149–174.
Ordeshook PC and Shvetsova OV (1994) Ethnic heterogeneity, district magnitude, and the number of parties.
American Journal of Political Science 38(1): 100–123.
Osborne JW (ed) (2008) Best Practices in Quantitative Methods. the United States of America: Sage.
Paskhina IS and Telin KO (2017) “Party zero”: Russian elections through the prism of the effective number of
parties. Polis, Politicheskie Issledovaniya–Political Studies 5: 43–53.
Quinn T (2013) From two-partism to alternating predominance: The changing UK party system. 1950–2010.
Political Studies 61(2): 378–400.
Rae DW and Taylor M (1970) The Analysis of Political Cleavages. New Haven, CT: Yale University Press.
Riker W (1986) Duverger’s law revisited. In: Grofman B and Lijphart A (eds) Electoral Laws and Their
Political Consequences. New York: Agathon Press, pp. 19–42.
Rokkan S (1970) Citizens, Elections, Parties: Approaches to the Comparative Study of Political Development.
Oslo: Universitetsforlaget.
Ross A and Willson VL (2017) Paired samples T-test. In: Ross A and Wilson VL (eds) Basic and Advanced
Statistical Tests. Rotterdam: SenseP, pp. 17–19.
Rusticus SA and Lovato CY (2014) Impact of sample size and variability on the power and type I error rates of
equivalence tests: A simulation study. Practical Assessment, Research, and Evaluation 19(1): 11.
Sartori G (1976) Parties and Party Systems. Cambridge: Cambridge University.
Sartori G (1994) Comparative Constitutional Engineering: An Inquiry into Structures, Incentives and
Outcomes. Hampshire: Macmillan.
Sawyer SF (2009) Analysis of variance: The fundamental concepts. Journal of Manual & Manipulative
Therapy 17(2): 27E–38E.
Schleiter P and Voznaya AM (2014) Party system competitiveness and corruption. Party Politics 20(5): 675–
686.
Shugart MS and Carey JM (1992) Presidents and Assemblies. Constitutional Design and Electoral Dynamics.
Cambridge: Cambridge University Press.
Siaroff A (2000) Comparative European Party Systems: An Analysis of Parliamentary Elections Since 1945.
New York: Garland.
Taagepera R (1999) Supplementing the effective number of parties. Electoral Studies 18(4): 497–504.
Taagepera R and Shugart MS (1989) Seats and Votes: The Effects and Determinants of Electoral Systems. New
Haven, CT: Yale University Press.
Verney D (1959) The Analysis of Political System. Michigan: Compton Printing Works.
Ware A (1996) Political Parties and Party Systems. New York: Oxford University.
Wildgen J (1971) The measurement of hyperfractionalization. Comparative Political Studies 4(2): 233–243.
Ünal et al. 323
