Selected Research Papers in Social Change, Education, Labour Market, and Criminology Volume II

2 Approaches in the analysis of ideological profiles of political parties

There have been several different research approaches established to date. They differ mainly in terms of source data used to determine a party’s position – experts’ opinions, voters' opinions, political programs and speeches. Use of a given source forces us to adopt a specific theoretical approach and specific data analysis methods. Therefore, it is worth discussing the advantages and disadvantages of the most important approaches.

2.1 Political party affiliation

The first approach used to determine a party’s position was simple ranking of political groups based on their political identity and affiliation with one of the party families. Political parties are grouped along the left-wing – right-wing continuum, from radical left-wing (communist) parties to radical right-wing (authoritarian and nationalistic) parties. Mair (2011) calls such an approach a priori judgments, in other words: positioning of parties based on their origins. Because such a method is relatively simple, it was widely implemented (Taylor, Herman 1971; Sigelman, Yough 1978). Its simplicity is also its main limitation, and can be broken down to four issues (Mair 2001):

Marcin Kocór 159

1. only one, general positioning dimension is used; 2. measurements are taken at the ordinal level, making it impossible to compare distances between parties; 3. a simplified assumption is made that there are no differences between parties from one family; 4. parties that cannot be assigned to one of the identified families cannot be positioned.

2.2 Positioning based on party’s documents The second method of determining a party’s position consists of referring to the 1existing party documents and determining its position on such a basis. It is different from the first approach as it uses a lot more information, on the basis of which the researchers perform the positioning. Such information can include parties’ programs, press releases, transcripts of parliamentary debates etc. Such positioning is also based on more precise scales than the a priori   approach. It is not an intuitive continuum   of political families, but rather a concrete point of reference. Janda (1980) established a synthetic left-wing – right-wing division comprising 13 issues measured on 11-point scales. Each of the analysed parties was evaluated by competent experts.

2Together with the development of digital data analysis, use of experts’ knowledge gave place to automatic work while coding text data. Currently, two competing methods are the most popular: the approach developed by Manifesto Research Project/Comparative Manifesto Group and the Wordscores approach proposed by Laver and Garry (2000, 2003). In these developmental methods of determining position on the basis of party documents, in order to avoid (at least partially) subjective results and achieve better accuracy, content of the political programs is analysed in an advanced way. Use of cutting-edge digital content analysis allows for quantitative analysis of parties’ programs (other documents, such as parliamentary speeches, press releases etc. can also be analysed) and determination of not only a given party’s position, but the dimensions of the ideological space existing in a specific party system as well.

 

1  Janda’s research entitled: International Comparative Political Parties Project covered political parties in 53 countries from 1950 to 1962, but the obtained results were rarely used by other researchers (Janda 1980).

2  Manifesto Research Group (MRG) Project was established in 1979. In 1989 it was transformed into an institution under the name Comparative Manifestos Project (CMP) at Wissenschaftzentrum Berlin für Sozialforschung (Volkens 2002).

Marcin Kocór 160

2.3 Positioning based on the voters’ attitudes

An attempt to determine a party’s position and refer it to the political reality is to use electoral surveys. In such a case, voters’ preferences and evaluations indicate the party’s position. The most important issue is proper selection of the voters’ category, on the basis of which the party’s ideological profile is identified. Depending on the above, only electorates of the positioned parties, voters and society in general can be analysed. Moreover, referring to the voters’ preferences, we may create a map of the ideological profile of the whole party system using the ratings of all the parties existing in a given political arena. The research approach using electorates’ preferences is used in two ways (e.g. in Comparative Study of Electoral Systems research). The studied group may be asked to place all the analysed parties on the left-wing – right-wing scale, resulting in a picture of the whole party system. The second method is to ask the respondents to define themselves on the right-wing – left-wing scale and the aggregated preferences of their electorates are then the indicators of a party’s position.

While using the approach based on the voters’ opinions, an arbitrary assumption is made that individuals are well-oriented in the issues determining the ideological space of the political arena and can tell the difference between the positions of various parties. However, the early studies by Campbell, Converse, Miller and Stokes (1964) showed that only one-third of the respondents could identify a series of political issues and show the differences between the parties with regard to such issues. The voters may evaluate various parties on the basis of issues crucial to the voters only and ignore other aspects. Electoral decisions and parties’ images are often established on the basis of non-professional factors, such as emotional attitudes towards some politicians.

2.4 Studies on elites

Another method of analysing a party’s position eliminating at least some of the flaws of the previous methods are studies on political elites. It is assumed that politicians are the group directly responsible for policy-making. Therefore, determining their attitudes, preferences and concrete actions can help recreate images of parties existing in the political arena (Kitschelt, Masfeldova, Markowski, Tóka 1999; Laver Hunt 1992). The approach referring to the analysis of political elites was established relatively early (Taylor, Laver 1973) and has been followed by its several variations. In the simplest form, the subject of analysis are politicians from various groups, whose attitudes are considered to be the indicators of the parties’ position. Such research is often narrowed down to the members of the Parliament only, which limits the analysis of ideological profiles only to those parties that made it to the Parliament. Instead, the research sometimes includes the representatives of regional elites that, in Poland, are the “waiting room” for the central level (Wasilewski, Kocór 2001) and may complete the image of a party’s position at the lower level of the political arena.

   

Marcin Kocór 161

2.5 Research on legislation processes An alternative to the direct surveys of political elites are studies on the roll-call behaviour, i.e. analyses of acts, resolutions and regulations (Lipski 2003). Behaviour of the representatives of various parties during voting is considered the indicator of attitudes towards specific issues for voting. Such an approach entails at least three problems. Firstly, it is difficult to choose suitable issues as reference points for determining a party’s position. Secondly, parties in the Parliament are connected with each other. Mair (2001) points out that the members of Parliament do not act on their own, but their behaviour depends on the decisions made by other members. Sometimes, the result of voting does not reflect the actual attitude of the party, but is a result of tactical voting. Finally, the results of such studies concern only the parties in the Parliament.

An alternative to the roll-call studies of some kind is the analysis of the budget spending structure established by the members of Parliament. Allocating specific sums of money for some purposes defines the attitudes of the governing parties. However, in my opinion, such studies (Klingemann, Hofferbert, Budge 1994), apart from lacking a theoretical connection between the positions occupied by the parties and budget spending, make it difficult to choose indicators suitable for determining the parties’ ideological profiles.

2.6 Experts’ opinions The last type of studies on parties’ positions is based on using experts’ knowledge. Such analyses require the conducting of interviews or surveys with experts in politics. This method, used for the first time in the 1970s (Morgan 1976 after: Mair 2001), gained recognition among the researchers dealing with party systems. The largest studies of that kind – conducted by Castels and Mair (1984), Huber and Inglehart (1995), Laver and Hunt (1992) and Ray (1999) – are often used as some kind of standard for determining the position of a party in several dozen countries. Budge (2000), however, warns about using the results of such studies without reservations. Firstly, the main problem is subjectivism of experts’ opinions. Secondly, it is unclear what is used to determine the ideological image of the analysed parties – evaluation of the party leaders, their members, electorate or all of the above. Thirdly, parties are placed on the left-wing – right wing scale by the experts, while such a scale is usually defined by a number of dimensions. However, for various experts the meaning of the scale itself may be different. The meaning of the left and the right wing also changes with time, which may be reflected in the differences between the positions occupied by political parties. Fourthly, we are unsure whether the behaviour or the declarations of the parties is analysed. If experts are used to evaluate the actions of the party, it may lead to tautology – the parties’ behaviour is determined on the basis of its symptoms. Eventually, the time perspective for evaluation criteria is not clearly defined. It is unclear whether the evaluating persons position the parties on the basis of their current profiles or refer to the past. We do not know what period of time is taken into consideration while evaluating a given party – the last year, the last ten years

Marcin Kocór 162

or the period since the establishment of the party. Taking into consideration the above, the results of expert interviews devoted to political parties’ profiles should be treated sceptically.

Of the presented research approaches, three of them are actually used today: interviews with experts, analysis of parties’ programs developed by CMP and Wordscores  program analysis method (Keman 2007, Klemmensen et al. 2007; Whitefield et al. 2007; Volkens 2007). Those three methods are the most accurate ones and – contrary to the disputes – the most reliable ones as well. They are also relatively easy to apply. In the next part of my study, I am going to focus on two approaches based on the analysis of the existing data in the form of party programs. Those two approaches seem the most practical to me, giving a possibility to conduct a thorough analysis of the collected results. Expert interviews aimed at determining the positions of political parties, although still used, usually play the role of a reference point – a specific benchmark – focused on confirming the accuracy of analyses of the content of the parties’ program documents (Klemmensen et al. 2007; Volkens 2007).

3 Analysis of the political parties’ programs

Using political programs as the source of information we have the relatively easiest and cheapest access to data. Moreover, programs of political parties may be treated as reliable sources of information on the parties’ ideological image. Political programs are created by all the parties in the elections, they present the position of the whole party, not only its leaders, and refer to the most important political issues. Of course, such data has its faults (e.g. different methods of formulating the party’s attitude, often making the analysis more difficult), but the use of a proper method of analysis enables us to obtain fairly accurate ideological profiles of political parties on the basis of program materials.

While analysing a party’s documents, one of two methods are usually used: the content analysis method / coder-based method (CMP approach) or the fully automated Wordscor es approach utilising computer software only. To quantify the obtained results, it is necessary to justify the analysis of the content of the parties’ documents theoretically. There are two competing theories underlying the positioning of parties based on their political programs.  

The first one is the saliency theory proposed by Robertson (1976). According to that theory, political parties do not compete with each other in the ideological space created by various political issues, but strive to “take over” specific issues. Such questions become the basis for electoral statements and attracting voters. The parties try to monopolise specific issues close to them. In such situations, political opponents stop referring to those issues and look for other issues, not “taken over” yet, on the basis of which they could formulate their program. Studies on the principles of political discourse driven by the dominance principle confirm the reasonableness of such an approach (Rikker 1993). The studies on the role of the mass media in the political process also stress the importance of some issues taken over by specific parties (Rabinowitz et al. 1998). Consider the problems of tax policy and environmental policy as an example. In their programs, neo-liberal

Marcin Kocór 163

parties would stress the importance of lowering taxes, reducing impositions on entrepreneurs etc. whereas the environmental issues would be rather marginalised, if not ignored. The environmentally-friendly parties are quite the contrary – they try to attract voters by focusing on environmental problems, healthy lifestyles etc., ignoring (or marginalising) the tax policy.

3Such a way of perceiving the ideological space is also adopted by the voters, who have specific attitudes towards the most important political issues. Relations between different issues are insignificant. For example, on the one hand, every reasonable man would vote for a reduction in taxes, on the other hand, however, many people want their country to care for them. Those two issues are rarely linked as it would inevitably lead to contradictions.  

3As a result, parties build their position in the political arena on the basis of specific sets of issues characteristic of a given party and offer them to the voters, who take into consideration their own preferences while making decisions. A party’s place in the realm of politics is determined on the basis of analysis of importance attached to various issues. The more an issue is stressed by a party in its program (or other documents or politicians’ speeches), the more important that issue is for that party. In other words, putting emphasis on specific political issues determines the party’s attitude towards such issues. The saliency theory is a point of reference in the CMP approach. 

The opponents of the saliency theory claim that the position of a party cannot be determined on the basis of importance attached to some issues only. Political parties, apart from other ways of emphasising political problems, may propose various ways of solving such problems. The politicians do not actually speak of certain issues but rather express various opinions on such issues, either positive, neutral or negative. Therefore, instead of focusing on the importance of given issues, a direction of the proposed changes should be determined (Janda, Harmel, Edens, Goff 1995; Laver, Garry 2000). For example, one political party speaks of the need to lower taxes, meaning reduction of progressive tax and liquidation of tax reliefs, the second one suggests that a flat tax should be introduced. Political parties do not “take over” some issues, they rather compete with each other, proposing concrete solutions to such issues to their voters. The confrontational approach assumes that political parties compete with each other by presenting different attitudes towards the most crucial political issues. The voters also express their opinions on specific problems – agreeing with the political parties or not. In such an approach, a party’s position does not depend on the importance of specific problems, but on the differences between the parties in respect of suggested solutions. Therefore, in order to determine a party’s attitude, we do not have to be familiar with its whole program, we only have to choose several most important issues. The researchers may decide what dimensions or political issues should be the basis for positioning (Gabel, Huber 2000; Laver, Garry 2000). Such a confrontational approach is the basis for Wordscores analyses. 

The MRG/CMP research team, using the saliency theory, has collected and analysed, according to its own code scheme, an impressive number of party

3 It is the result of Stokes’ (1966) observations of the decision characterised by a single alternative towards main political issues, common among the voters.

Marcin Kocór 164

programs – 2347 documents related to 632 parties in 52 countries in the post-war period (Volkens 2001). The scheme includes 56 code categories that enabled the team to identify the most important issues in political programs. Twelve of them are presented confrontationally (using opposites), but the existing analyses have shown that parties do not compete on the basis of such categories, but stress only one of the suggested solutions. For example, some parties speak of the need to strengthen the welfare state, ignoring the issues of its limitations or support for business. Application of MRG/CMP scheme enables us to calculate what percentage of the whole document has been covered by specific issues.

4On the basis of MRG/CMP coding, we may position parties in the ideological space in two ways, referring to the multi-dimensional left-wing – right-wing division. The differences are found in the method of determining the dimensions of the leftwing – right-wing continuum. According to the first solution, the dimensions in which political parties are located are determined a  pri ori,   on the basis of previous studies or intuition. Depending on the above, the left-wing – right-wing division is divided into several dimensions. Each of those dimensions is characterised by specific code categories (left-wing, for example, by raising the issues of progressive taxes, legalisation of abortion etc., the right wing by referring to privatisation, participation of the Church in public life etc.). Then, the number of categories attributable to the left and right wing is counted. A party’s position on the left-wing – right-wing continuum is usually defined as the difference between the sum of left-wing and right-wing issues (Klingemann 1995; Laver, Budge 1992; Volkens 1996).  

The second method has no previous assumptions concerning the shape of the ideological space, therefore we may call it the ad hoc   method.  Parties are positioned on the basis of an analysis of all the issues raised in all the considered programs of a given party. Usually, one of the statistical space reduction techniques, e.g. factor analysis, correspondence analysis or cluster analysis, is used. In such space, a party’s position is determined on the basis of frequency of occurrence of categories creating singled-out dimensions in the programs (Gabel, Huber 2000). 

The confrontational approach taking into consideration direction and level of parties’ attitudes towards political issues has probably been operationalised best by Laver and Garry (2000). They created their own, enhanced coding scheme including 300 categories. Units of text – groups of ten words – were classified using a three-grade scale: negative, neutral and positive grades.. Parties’ positions in the left-wing – right-wing continuum are determined on two dimensions determined a  priori: economic dimension and social dimension. Laver and Garry used two simple summary indexes that may be expressed using the following symbols:

  

=E +EE EELR LRLR

4 =S +SS SSLR LRL

R 

  The following dimensions are usually defined: economic dimension, ideological (social and cultural) dimension, materialism-post-materialism. More: Elff 2002.

Marcin Kocór 165

L

5

and EL

R


5

ER respectively mean: number of units of text falling into category expressing “limitation of the role of the state in the economy” and “increase of the role of the state in the economy” and while determining social dimension, S and S are calculated as the number of units classified in the following categories: “conservative, limiting or traditional social values” and “liberal, permissive or progressive social values” (Laver, Garry 2000: 628-629). Another feature of Laver’s and Garry’s method is the method of coding party documents used. The authors, in order to ensure better objectivism and reliability of the analysis, used CACA (computer-assisted content analysis) utilising specially prepared word dictionaries. In this method coders are not used and the material is classified by a computer application. The text is divided into groups of ten words that are later compared with special databases including words (or some phrases) typical for left-wing or right-wing parties – hence the name of the method: Wordscores. According to the authors, digital content analysis ensures high accuracy of the results together with high reliability of the measurement, but only with correct dictionary definitions. The main problem is the selection of suitable referential texts, to which the analysed materials are referred. Such texts should be typical of a given political system, not encumbered with any specific issues (e.g. such texts cannot be programs from elections changing the shape of the political arena dramatically).

Both approaches have their advantages and disadvantages presented in Table 1. As far as the CMP approach is concerned, the main advantage is high accuracy of the results (Klemmensen et al. 2007; Volkens 2007). Moreover, the structure of the categorisation key allows for relatively deeper analyses than in the Wordscores approach. The obtained results make it possible not only to determine ideological profiles of political parties based on the left-wing – right-wing division, but to recreate the content underlining such division as well. In the case of Wordscores it is impossible directly and only an analysis of referential texts would help recreate such content, but it would require using a different research approach (e.g. CMP or expert interviews).

In turn, one of the main advantages of Wordscores   is the high reliability of the results resulting from the automatisation of the analysis and omittance of human influence (encoders). The application of the CACA method for coding materials makes the Wordscores  approach relatively quick and cheap, although requiring more advanced IT skills to create applications for making comparisons. An advantage of this approach is that it is possible to use documents other than electoral programs for coding, e.g. parliamentary speeches (Klemmensen et al. 2007), while maintaining relatively high result accuracy. Although the use of CMP encoding would also be possible in relation to such materials, it would probably require the categorisation key to be modified and surely require lots of effort. Another advantage of the Wordscores approach is the possibility to calculate the confidence range of the estimated results that indicate the quality of the analyses. It is impossible to obtain such indicators using the CMP approach. 

The Wordscores method can be easily applied to analysis of documents prepared in other languages without knowing the language – it is enough to select suitable referential texts .  

Marcin Kocór 166