A bold finding of a 2008 study by Russell J. Dalton is that--across the world--party fractionalization (roughly: how many parties there are in a country's legislative bodies, technically the Herfindahl index of the legislature) and party polarization (how far these parties are ideologically from each other)... are two measures almost completely uncorrelated (r=0.067).

A less way academic way to say this: across the world one can find political systems with many parties which are near-clone of each other, or two-party systems with significant ideological differences between the parties. (Or the other way around, just as likely.) I.e. trying to predict how polarized a party system is based on its degree of fractionalization is basically hopeless.

Measuring ideology is the trickiest part of this business, of course. Dalton used surveys in each country to measure ideology, as perceived by the electorate. Polarization is then computed as sort of standard deviation, but with the distance from the center weighted by each party's voter share (as not to allow a small extreme party to give a huge polarization score for the whole system).

Below is a table from Dalton's paper (which also measured polarization at different times, where that was possible from the database he used).

enter image description here

Dalton's paper is fairly cited; it has approximately 400 citations in Google Scholar. Dalton cites exactly one prior work that had found the same, Gross & Sigelman (1984), which has a lot fewer citations ~150, although Google is somewhat biased in counting citations for older publications. Gross & Sigelman relied on Encyclopedia Britannica's description of each party\s ideology, so I guess that was a bit less insightful (or exciting) than using the electorate's view of ideological position.

And since Dalton doesn't have a scatter plot for this... below is Gross & Sigelman's. The "center of gravity" axis means the center/average is more left-wing in the direction of the arrow. (Yeah, the political left is on the right in the graph, somewhat confusing.) Also, this is not your usual scatter plot because it's obtained by multi-dimensional scaling of a four-dimensional distance (which found however that there were mostly just two orthogonal dimensions: party fractionalization and party polarization) Some unusual country abbreviations: De is Denmark, Wg is West Germany, Ni is Nicaragua, Es = El Salvador, St is Switzerland, Se is Senegal, Sweden is Sw.

enter image description here

Has this finding (of independence of fractionalization from polarization) been confirmed/replicated in other studies (which potentially use a different methodology and/or data set)? I mean besides these two I found myself.


  • Russell J. Dalton, "The Quantity and the Quality of Party Systems Party System Polarization, Its Measurement, and Its Consequences", Comparative Political Studies Volume 41, issue 7 (July, 2008) page(s): 899-920

  • Donald A. Gross and Lee Sigelman, "Comparing Party Systems: A Multidimensional Approach", Comparative Politics Vol. 16, No. 4 (Jul., 1984), pp. 463-479

  • 1
    One idea that comes to my mind is that in a two party system, parties can grow by attracting the moderates from the other party, which pressures them into not drifting too far away ideologically. I did read somewhere that elections in the USA were usually characterized with some radical program in the primaries to get the support of the party, followed by a more moderate presidential campaign.
    – SJuan76
    Commented Aug 19, 2018 at 17:14
  • Another idea is a part of the political spectrum being taboo in a country (e.g. Socialism in the USA) restraining the available "ideological space" to "distribute" between the parties.
    – SJuan76
    Commented Aug 19, 2018 at 17:19
  • @SJuan76 While there may be restrictions of ideological space on the left, the U.S. "mainstream" political spectrum extends further to the right than all but a handful of Western countries. Pro-death penalty, anti-welfare state, anti-minimum wage, favoring large military expenditures, favoring repealing many civil rights, denying climate change, favoring public school teaching of evolution, etc.
    – ohwilleke
    Commented Aug 20, 2018 at 22:44

2 Answers 2


Has this finding (of independence of fractionalization from polarization) been confirmed/replicated in other studies (which potentially use a different methodology and/or data set)?

It seems plausible that it would be, although measuring political distance is inherently a problematic concept without better definition of what is meant by it. It is also complicated because not all researchers frame the issue in the same way.

  • As somewhat similar premise was explored in the article:

Paul Frymer, "Ideological Consensus within Divided Party Government" 109(2) Political Science Quarterly 287 (1994). DOI: 10.2307/2152626

This explored how there could be overlap in ideologies in U.S. politics in a time period in which the Democratic party as a whole was the "liberal" party, while the Republican party was the "conservative party" due to a de facto three party system with the Democrats divided between Northern and Southern Democrats. Since then, U.S. political parties have "full sorted" ideologically as a multi-decade process known as "realignment" has run its course.

  • The hypothesis of those papers also tends to be confirmed, although less quantitatively in this article:

Radoslaw Markowski, "Political parties and ideological spaces in East Central Europe" 30(3) Communist and Post-Communist Studies 221 (1997) doi.org/10.1016/S0967-067X(97)00006-8 ("The results indicate different levels of ideological structuration and political divisions of the party systems in Eastern Europe, which are explained not only by socio-economic factors, but mainly by varying experiences of pre-communist rule, communist governance and pathways to democracy.")

  • A pretty useful article critiquing the use of the left-right political dimension in assessing the issue is this one:

Michael J. Willis, "Political parties in the Maghrib: ideology and identification. A suggested typology." 7(3) Journal of North African Studies 1 (March 29, 2007) doi.org/10.1080/13629380208718471

The abstract of that article explains that:

Maghribi political parties and the party systems in which they operate are clearly different from those in the West. One difference is the weakness of traditional western‐style ideological divisions between parties. Ideological divisions do nevertheless exist between political parties in Algeria, Tunisia and Morocco but relate to different issues. They can be summarised as differing viewpoints over the role of religion in the state; the role of minority identities in the state; and most importantly, the continued centralisation of political power in the state. It is differences on these questions that provide a general typology of political parties that distinguish themselves from each other on these points. On this basis the paper examines firstly, parties based upon Islamist ideas, secondly, parties associated with the Berber population, and finally and most importantly, parties that are either supportive of or in opposition to the existing political order and rulers. It is argued that it is this final division between ‘regime’ and ‘opposition’ that is the most crucial and overrides all other distinctions. Having established this basic typology, the paper will also look at more general issues such as party structure and social and geographic bases of support of parties.

  • An article on Dutch political parties at the local government level also provides insight, although not a quantitative assessment about how these numerous parties elected by proportional representation differ ideologically from each other, in a country that is arguably more homogeneous ideologically than many:

Marcel Boogers and Gerrit Voerman, "Independent Local Political Parties in the Netherlands" 36(1) Local Government Studies 75 (2010).

Its abstract notes:

In the last 15 years, the Netherlands has witnessed the enormous growth of independent local parties and of their electoral support. In order to assess the success of independent locals in the Netherlands, this paper explores how independent local parties distinguish themselves ideologically, organisationally, and operationally. It presents the results of two online surveys of 1,800 independent and other local parties in the Netherlands. Compared to local party branches, independent local parties are more successful in performing the organisational, programmatic and nomination function of political parties. They are spearheading the change to a modern cadre party with politicians and a small number of active volunteers taking care of the party’s rootedness in society. Furthermore, they enrich local politics with new political dividing lines, smoothing the entry of new demands, themes, and issues onto the political agenda. Finally, they are proving to be more resourceful in recruiting citizens to stand as candidates for municipal councillorship.

  • A 2014 paper recounts how the amount of ideological devision in the U.K. has declined despite having a similar number of political parties:

Katharine Dommett "Ideological Quietism? Ideology and Party Politics in Britain" Political Studies (2014) doi.org/10.1111/1467-9248.12160

It's abstract notes that:

Ideology and political parties are frequently depicted as disparate entities, with scholars citing a range of exogenous and endogenous changes to demonstrate the decreasing relevance of ideology to party politics. This article moves away from such accounts by looking at the role of actors, and specifically party leaders, in contributing to perceptions of ideological decline. Through an examination of the rhetoric of Labour and Conservative Party leaders in Britain between 1946 and 1997, this article contends that politicians have engaged in, what is termed here, ‘ideological quietism’. In this sense, parties have not abandoned ideology but have made rhetorical shifts indicative of ideological decline.

This implicitly supports the hypothesis that ideological division is not closely linked to the number of political parties extant at any given time.

The number of political parties and their mix is a function of election laws.

A quite solid conclusion of political theory is that the number of political parties and mix of political parties by size in a legislative body are predominantly a function of the election system used and secondarily, in geographic based legislative election systems, by the extent to which there are nationalist movements giving rise to regional political parties (which is the main way that you get more than two parties in a first past the post single member district system), which in turn has more parties than a formally one party system or a dominant party system.

For example, a pure proportional representation system (e.g. Israel and for a substantial period of time Italy) has more parties and a greater mix of parties in the legislature, than one with an explicit minimum threshold (e.g. Germany's 5% floor for representation in the legislature) or an implicit one (e.g. by handling proportional representation in multi-member districts with 18 members each), which in turn has more parties than one with a ranked choice voting system from single member districts, which in turn has more than a first past the post single member district system.

Political distance reflects underlying political opinion, which doesn't have any necessary relationship with the the electoral system.

The amount of partisan distance, measured by any consistent reasonable measure, is going to be a function of how much of a range of political opinion exists in the country.

A country with a completely proportional representation system conducted at a national level is going to have the most political parties, but if there is a lot of homogeneity in political opinions on most issues, they may not have that substantial ideological differences from each other. But, a country that is deeply divided culturally, ethnically, and in political views, may be deeply divided ideologically from each other even in a two or three party system.

Methodologically Measuring Political Distance Is Inherently Hard, However.

One of the tricky parts of measuring differences in distance in different political ideology, however, is how to quantify those differences, not just in terms of true difficulties operationalizing and measuring those differences, but in terms of defining those terms in a meaningful way, which is one reason there is less consensus in the field on political distance as a concept.

For example, the Parti Quebecqois in Canada didn't differ that much in positions on many issues from the Labour party, but is diametrically opposed on the really critical and salient issue of whether Canada should continue to exist in the long run. In contrast, the Democratic Party and the Republican party in the U.S. are very deeply divided on economic and social issues, but are largely in consensus on the continued existence of the United States as a single country.

Which parties are more deeply divided?

One could define "political distance" to mean the extent to which two parties are "well sorted" in that members of one party rarely vote for bills supported by the other party. And, one could limit this to avoid trivial cases where all parties (or even most parties) support a bill.

But this has a lot to do with parliamentary procedure and the kind of issues that are undertaken at a legislative level rather than an administrative level. For example, there could be deep partisan differences over the question of assigning names to post offices and other public buildings that are nonetheless not reflective of deep policy differences.

If you are measuring policy, it pretty much has to be a relative measure.

For example, Democrats have favored less defense spending and stronger protections for civil rights for at least 38 years (since Reagan).

But, if you look at stances on particular issues, those have changed dramatically over time and adjusting the bars can determine if there is a difference or not. No federal elected official of whom I am ware in 1980 was advocating for gay marriage or civil unions or even non-discrimination against LGBT. But, most Democratic elected officials today support gay marriage, and there is a meaningful divide on non-discrimination against transgender individuals between the parties. Senator Strom Thurmond was strongly against interracial marriage at the start of his career and had someone in an interracial marriage on his campaign staff at the end of it. So, at any given time, the amount of partisan division on particular issues varies.

You can chart changed in political ideology over time between parties by using a chain that compares sessions over time to previous sessions in a single country on a single issue. But, it isn't nearly so possible to do so internationally where the issues differ dramatically between countries.

  • 1
    This sounds like an extended comment (on the methodological difficulties), not an answer. And your middle-to-last paragraphs seem to argue with examples that the finding could be true, but I wasn't asking for the claim to be exemplified with anecdotal evidence. There's some of that in Dalton's paper as well. The question was too long already without quoting any of that (only because some people here didn't understand what uncorrelated means, so I had to expand on that.). Commented Aug 20, 2018 at 20:55
  • 2
    @Fizz One legitimate way to answer a question is to assert that it is based on an inaccurate or unsound premise, or is ill defined.
    – ohwilleke
    Commented Aug 20, 2018 at 21:28
  • The paper you based your answer on was published 14 years before OP's. It's hard to say that it replicates or confirms a paper that it preceded. Commented Aug 20, 2018 at 22:16
  • @indigochild It is ten years after the oldest paper cited in the OP so it is in an overlapping time range. And, obviously, one paper is not a comprehensive survey of the literature, but is more than any other answer produced.
    – ohwilleke
    Commented Aug 20, 2018 at 22:19

I am not sure why you or anyone would call this bold, this is an expected result. In governments with multiple parties it is common that no single party can gain enough votes to have a majority and thus seize power and govern so coalition politics are the rule. Coalitions composed of polar opposites are highly unstable so all parties are going to orbit a central ideology so that they can create stable coalitions in those governments. By contrast in 2 party systems each party does have enough votes so that they can gain a majority from time to time, thus they are free to differentiate themselves.

  • You disproved your own point; your last sentence claims that there will be more polarization in two-party systems, which is not actually the case. Commented Aug 19, 2018 at 9:19
  • @Fizz, you wrote "across the world one can find political systems with many near-clone parties, or two-party systems with significant ideological differences between the parties" - isn't that very much "more polarization in two-party systems"?
    – janh
    Commented Aug 19, 2018 at 9:24
  • @janh: No, that's not what it means. Commented Aug 19, 2018 at 9:30

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .