Parties can be viewed from numerous perspectives and measured in countless ways. The ICPP Project measured parties on a small number of abstract concepts. Most concepts were scored on multi-item scales composed of concrete party characteristics. Information for scoring parties on these concepts was gathered from library materials using a microfilm-computer system of information retrieval (Janda, 1982). The scoring procedures and scale construction were quite complex, and the interested reader is directed elsewhere for a complete explanation of the data, concepts, and scoring techniques (Janda, 1980). Even by limiting our focus to a few major party concepts, we are confronted with a large problem in data analysis and interpretation. Our initial approach to this problem is to test for regional effects on each party concept. Our basic test is a one-way analysis of variance, using the 10 regions as independent variables predicting to party characteristics.2
The null hypothesis being tested is that region has no effect on how parties score on each concept: that is, the characteristics of parties in one region should show the same central tendencies and variations (means and standard deviations) as parties in any other region. A maximum of 147 parties are involved in the analyses; the number varies somewhat depending on missing data.3 We review regional effects on each party characteristic in order of the strength of the effect, measured by the eta-squared statistic.4 All the relationships are signfficant at least at the 0.05 level. The strength and significance of the relationships are reported at the end of Table 1.
Because the data pertain to parties scored for characteristics in 1957-62, some findings might change in particulars if the study were replicated with 1980 data. The threat to contemporary validity comes mainly from the demise of old parties and the rise of new ones, for parties seldom change characteristics much over time. In any event, the reputed effects of region on parties are thought to be enduring and, if they exist, should not change in broad outline within one generation.
Parties may follow three
types of strategies in gaining governmental office: (1)
competing openly in elections with other parties, (2)
restricting competition from other parties, and (3)
subverting or overthrowing the existing political system.
Three scales were created to measure the extent to which a
party followed each of these strategies, or mix thereof
(Janda, 1980: 81). Each scale ranged from 0 (the party did
not rely on the strategy) to 4 (it relied exclusively on
that particular strategy). The correlations among these
items were high (about -0.80) but not perfect, indicating
some special mixes of party strategies. All 147 parties were
scored for 'restrictive orientation'. The mean score was
0.87 (out of 4.0), implying that most parties in the sample
did not follow a restrictive orientation. Region had a
pronounced effect on which did and did not.
Most parties in the sample relied primarily on winning elections to gain office. The mean score was 2.8 (out of 4.0) over all 147 parties, and the effect of region on competitive orientation was very strong, explaining 51 per cent of the variance.
Parties in the three Western regions--Anglo-America, West Central Europe, and Northern Europe--scored uniformly high on this concept. Except for four parties in West Central Europe, all Western parties obtained the highest score of 4. The exceptions were the Portuguese National Union, the Greek ERE, the Greek EDA, and the French Communist Party. The first two parties deviated toward restricting competition. The leftist EDA and the French Communists were primarily oriented toward electoral competition, but they also sought to subvert the political system through strikes at critical times. With these exceptions, parties in the three Western regions were distinguished by their orientation toward electoral competition. The only other region whose parties approached the Western orientation toward competition was South America.
The extent to which the party exists as a social organization, apart from its momentary leaders, and demonstrates recurring patterns of behavior (Janda, 1980: 19) is defined as institutionalization. The concept was measured by four items:
In measuring this concept, the parties' scores on the individual items were transformed into z scores, summed, and averaged (excluding missing data) to form a scale with an alpha reliability of 0.79. Although all 147 parties received scale scores because they were scored on at least one of the four items, the mean was 0.05 rather than 0.00 due to the incidence of missing data for individual items for some parties. Regional effects on institutionalizations were again very strong, explaining 51 per cent of the variance in the scale scores.
The most institutionalized parties were in Anglo-America, followed closely by parties in Scandinavia and Northern Europe. The mean institutionalization score for West Central Europe was lowered substantially by the four Greek parties in the study. The only other region whose parties stood above the world mean on institutionalization was Eastern Europe. The developing regions demonstrated a lack of party institutionalization that many (Coleman, 1960; Scott, 1966; Sartori, 1976) attributed to them. Again, there were substantial differences between South American and Central American parties. Parties in South America were slightly below average in institutionalization, while the Central American parties stood ninth, just ahead of those in West Africa.
Diversity of social support
Diversity of support refers to the extent to which the party attracts support from various social groups, reflects the social composition of society, and does not find its support concentrated in any particular social group (Janda, 1980: 42-43). These three ways of measuring social support were applied to six different social groups: occupation, religion, ethnicity, region, urban-rural, and education. The parties' scores for the three social support measures were summed over all groups, yielding a theoretical range from 0 (least diverse) to 3 (most diverse). The scale reliability is 0.84. A total of 146 parties received scale scores, but scores for about half of the parties were based on data for only one or two of the social groups. The mean support score was 1.07. Parties with high scores on the social support scale would usually be labelled 'heterogeneous or aggregative' of various social groups.
Region had a strong effect on the diversity of social support for parties, explaining 39 per cent of the variance in scale scores. In accordance with statements by Kirchheimer (1966) and Segal (1974), parties in Western countries, especially Anglo-America and West Central Europe, were among the most socially diverse regional groups (ranking first and third respectively). The mean diversity score for Northern Europe was substantially decreased by the Communist Parties in Luxembourg and The Netherlands. Although Anglo-American parties were the most heterogeneous, South American parties were virtually as diverse. This finding corresponds with assertions by Anderson et al. (1974), Lipset (1964) and Ranis (1971) that South American parties tended to be culturally plural and approximated the support patterns of European parties. In line with these statements was the finding that the only other region to fall about the world mean was Central America, whose mean was reduced by the low score of the Guatemalan Labor Party. The remaining regions demonstrated less diversity in support. Again, these findings support the literature. Many writers (Wallerstein, 1967; Anderson et al., 1974; Grove, 1975; Young, 1976) suggested that, in these remaining regions, parties tended to reflect social divisions, whether these divisions were based on class, religion, region, ethnicity, or urban-rural differences; and, as Grove (1975) asserted, East European parties demonstrated moderate levels of social cleavages in their membership.
Liberalism refers to the party's orientation toward a set of issues in liberal democratic government (Janda, 1980: 148):
The Liberalism scale was formed following the procedure used for Institutionalization-- which was used for all subsequent scales. The parties' scores on each item were transformed into standard scores, summed, and averaged. The scale reliability was 0.81. Due to missing data, only 133 parties received scale scores (the average party was scored on 3.1 issues in the scale).
Region explains 34 per cent of the variance in the parties' stance on the four liberal issues. The mean values for parties in the three 'Western' regions were all above the world mean. These parties existed in what writers call 'liberal democracies' (Roth and Wilson, 1976), so their high scores on this scale should not be surprising. The major outlier in these three regions was the Portuguese National Union. The South American parties were very similar to those in West Central Europe. The mean scores for parties in the remaining areas were below the world mean. Although these regions contained some parties above the mean, most tended to be of a non-liberal tradition. No parties from West Africa or East Europe scored above 0. In fact, all of the Eastern European parties scored lower than the least liberal parties in the Northern Europe and Anglo- American regions.
Centralization of power
The extent to which decision-making authority is located in the national organs conforms to the centralization of power (Janda, 1980: 108). The concept was measured with eight indicators of centralization:
These items formed a scale with a reliability of 0.83. All 147 parties were scored on this scale, but the average party was scored on only 6.2 items.
Region explains 24 per cent
of the variance in centralization of power. Unlike many of
the previous concepts, the 'Western' parties did not score
high on this scale. The West African parties tended to be
the most centralized. Only one West African party,
Marxism refers to the party's orientation toward a set of seven leftist issues (Janda, 1980: 148):
These items formed a scale with a reliability of 0.90. Due to missing data, only 142 parties received scale scores; the average party was scored on 6.1 issues in the scale.
Region had only a moderate effect on Marxist orientation, explaining 21 per cent of the variance on the Marxism scale. As expected, East European parties scored the highest on Marxism. In fact, all the East European parties scored above the mean values for the other nine regions. Asia was a distant second on Marxism, perhaps supporting Pye's assertion that 'the political parties of Southeast Asia tend to profess concern for ideological matters' (1960: 110). All five Asian countries also had a Communist Party in our sample. Most of the regions, however, showed substantial variation in the Marxist orientation of their parties. This would be expected given the dispersion of ideological preferences within the population, a rational strategy by political leaders, and permissive conditions for the formation of new parties. The only region in which a strongly leftist party did not exist was Anglo-America. The most extreme rightist parties were in South America (Peru and Ecuador) and Central America (Nicaragua--where all three parties were rightist in our time period).
This concept refers to the complexity of regularized procedures for mobilizing and coordinating the efforts of party supporters in executing the party's strategy and tactics (Janda, 1980: 98). The concept was measured with six indicators:
These items form a scale with a reliability of 0.82. Due to missing data, only 144 parties received scale scores, with the average party scored on 4.8 items.
Region explains 20 per cent of the variance in the complexity of party organization. Not surprisingly, the parties of East Europe tended to be the most complex in organization. The Scandinavian and Dutch parties also tended to be highly organized. Contrary to Weiner (1960: 207), Asian parties ranked third in organization. However, the region's high score was due to its Indian, Indonesian, North Korean, and Malayan Communist Parties. The remaining Asian parties were less well organized. The mean score for West Central Europe was drastically reduced by three of the four Greek parties, which were characterized by very low organizational complexity. There was considerable variation among parties throughout the other regions. The West African region had outliers on both extremes: the Ghanaian CPP and the Guinean Democratic Parties were more highly organized than the average, while the short-lived Ghanaian United Party and the Togolese Democratic Union were well below the average. All of the parties in Central and East Africa fell below the mean scores for the parties of Eastern Europe, Northern Europe, Asia, and Anglo-America. The lack of organization in African parties in general was noted by LaPalombara and Weiner (1966).
Involvement of party members
Involvement is defined as the intensity of psychological identification with the party and as the commitment to furthering its objectives by participating in party activities (Janda, 1980: 154). The concept was measured with five indicators:
These items formed a scale with reliability of 0.78; the average party was scored on four items.
Region explained only 16 per cent of the variance in the involvement of party members as we measured it. The region showing the greatest level of involvement was not East Europe, with its Communist Parties, but Asia. Only four Asian parties (two each in Indonesia and Malaya) scored below the world mean. However, Asia's ranking is again due mainly to its Communist Parties, whose members tended to be more involved in the parties' ideological purpose than members in East .European parties-- especially in the four auxiliary parties in East Germany.
This concept refers to the degree of congruence in attitudes and behavior among party members (Janda, 1980: 154). The concept was measured by five items:
These items formed a scale with a reliability of 0.72; the average party was scored on 4.2 items.
Region explained a small portion (13 per cent) of the variance in coherence of party behavior, which is primarily a measure of party factionalism. As with the involvement scale, there was almost as much variation within as between regions. The most coherent parties were found in East Europe. Even the least coherent party in Eastern Europe (which, ironically, was the CPSU) was p~iore coherent than the average party in South America. Except for these differences at the extremes, however, region has only a minor effect on party coherence.
This concept refers to the amount of access to the governmental structure enjoyed by the party (Janda, 1980: 144). It was measured by five items:
The parties' scores on these items were summed to form a scale with a reliability of 0.92.
Region had the least effect on governmental status, explaining only 11 per cent of the variance in parties' access to governmental structure. Upon consideration, there is little reason to expect that this concept would vary much by region, for it really measures party success and each region (indeed each country) should have both successful and unsuccessful parties. Regional differences arise, however, due to the clustering of non-competitive party systems, which produce parties that are extremely high in governmental status. Eastern parties in Eastern Europe, not surprisingly, rank highest on governmental status--despite the low status of the five minor parties (four in East Germany and one in Bulgaria). The party-states in West Africa at the time of independence qualified that region for second place. The minor role accorded to political parties in North Africa and the Middle East (Rustow, 1960) is reflected in that region's lowest rank on governmental status.
Summarizing regional effects
Region is consistently related to all party characteristics at the 0.05 of significance, but the strength of this relationship is highly variable. Region alone explains 52 per cent of the variance in party orientations toward restricting competition but only 11 per cent of the variance in parties' governmental status. Thus region, as a surrogate variable for political culture, does predict to the traits of political parties. Moreover, the nature of these regional effects is largely as claimed in the party literature. For example, AngloAmerican and European parties do tend to be more socially diverse than parties in most other regions, and parties outside of Europe and Anglo-America do tend to be less organizationally complex.
One general finding that has emerged from the review of regional differences in party politics is that parties in the three 'Western' regions (Anglo-America, West Central Europe, and Scandinavia) tend to be similar to one another, thus supporting Russett's delineation of a 'Western Community' region (1967). Similarly, parties in West Central Africa tend to be similar to those in Eastern Africa, justifying their combination into a broader 'Subsaharan Africa' category. Moreover, the other 'developing' areas-- especially Asia, North Africa and the Middle East, and Central America--often show similar tendencies in party characteristics. Finally, the parties in Eastern Europe often stand out from all other regions. We pursue these findings when conducting a more complex analysis in the next section of this paper.
Janda and Gillies: "How well does 'region' explain political party characteristics?" --> back to first page