Path: Janda's Home Page > Workshop > Canonical Analysis > Return to Opening Page
Political Studies (1998), Vol. 46, pp. 61l-632
Effects of Party Organization on Performance during the 'Golden Age' of Parties
KENNETH JANDA and TYLER COLMAN
Organizational Effects on Each Aspect of Performance

In theorizing about the relationship of party organization and performance, we must be clear on our assumption of causality. For example, do we assume that high complexity, low centralization, and low involvement cause electoral success? Or, do successful parties develop more complex organizations, become more decentralized through expansion, and sacrifice the psychological involvement of their members? Clearly, there is feedback in the relationship, but like most scholars we assume that organization causes performance, not the other way around.[43] Inquiring further into the direction of causality is the task for another study with a different design. Recall also that we are dealing with party organization and performance at national, not sub-national, units. Up to now, there has been no firm evidence that organizational traits and party performance are significantly related at the national level.

Assessing Effects on Electoral Success

Any theoretically complete explanation of party success in winning votes must involve such critical factors as the parties' positions on issues, the voters' attitudes toward party policies, the state of the economy, and the parties' traditional bases of social support. Our analysis includes none of these factors, so our explanatory model will certainly be incomplete." Seeking the causally correct model of political phenomena is, like pursuing the Holy Grail, inherently elusive. Causal models improve by a process of correcting theoretical errors and omissions through continuing research. Our goal is not to provide a complete explanation of electoral success but to determine whether organizational factors alone have significant effects--and whether their apparent empirical effects make theoretical sense. Later, however, we will include institutional factors to tease out additional effects of organization.

The theory that guides our inquiry comes from several sources. The effect of complexity on electoral success is treated in the empirical research literature cited above. Studies of party organizational activity to mobilize voters were concerned closely with what we have called 'organizational complexity'. This yields our first proposition:

H 1: The greater the complexity, the greater the electoral success.

There is no scholarly consensus about the effect of the next concept, centralization of power, on electoral success. Some major non-quantitative studies theorize that centralized parties are also more successful in mobilizing voters. Certainly this figured in Duverger's explanation of the superiority of 'modern' mass-membership organizations, adopted by leftist parties, over the loose caucus-type organizations of older, more conservative parties.[45] In contrast, Epstein argued that modern technology favored rightist parties that could afford to campaign via television. The counter-organizational tendencies he saw in the 'new' modern party rejected only the complexity or mass-membership aspect of leftist organization, not centralization of power. Indeed, Epstein believed that the small membership and less complex parties 'made it easier to impose a central and an efficient direction of campaigns by professionals'.[46]

On the other hand, some scholars have stressed campaign advantages from the decentralization of power, which enables parties to capitalize on local environmental conditions.[47] This argument appeared in Agranoff's early analysis of the 'new style' in election campaigns and is echoed in some modern explanations of the 'decline of parties'.[48] Reflecting this more recent argument, our second proposition is:

H2: The less the centralization, the greater the electoral success.

Conventional parties' wisdom says that 'pragmatic' parties are more successful in elections than 'doctrinaire' parties, which are reluctant to bend principles to win votes.[49] In 1964, many Republicans worried that nominating the right-wing candidate, Barry Goldwater, would cost the presidential election. In 1972, many Democrats opposed nominating the left-wing favorite, George McGovern, for the same reason. For both groups, the folklore was vindicated. Kirchheimer saw a tendency to abandon doctrinaire involvement in favor of electoral rewards in the 'catch-all' party, which collected votes from socially diverse groups by adopting policies to fit their interests.[50] In our terminology, the relevant proposition becomes:

H3:The less the involvement, the greater the electoral success.

The data in Table 2 support the argument that variations in party organization do indeed affect electoral success. About 20% of the variance (R square) in relative electoral strength can be attributed simply to differences in complexity, centralization, and involvement--without taking into account the state of the economy, political personalities, or other important electoral factors. If we assume that these organizational variables pertain mainly to 'the party on ground' that helps mobilize the electorate, we would expect such results.[51] However, the organizational effects are not entirely as hypothesized. While complexity increases electoral success and involvement decreases it as predicted, contrary to expectation centralization also increases success. This finding supports Duverger-style arguments for party centralization over later arguments for decentralization in campaigning. But it may only hold for the 1957-62 period, which predates the impact of television on elections in most of the nations covered. In sum, the manner in which parties are organized does relate to party performance at the polls. Involvement of party members in the broader purposes of the party has a significant negative effect on electoral success, meaning that pragmatic parties do indeed win more votes. Well-organized (i.e., more complex and more centralized) parties also do better in elections. The beta coefficients in the regression analysis demonstrate that, when the other two factors were controlled, each organizational variable had a stronger effect on electoral strength than suggested by their simple correlations. These findings shed no light on questions about alternative forms of party performance. For example, do parties with greater involvement of their members do better on other aspects of performance? Do they spend their members' energies in activities beyond electioneering? We turn to the concept of 'Breadth of Activities' for a partial answer.

TABLE 2. Effects of Organization on Electoral Success

Intercorrelation matrix
Regression equation
Variables
Elec. St.
Complexity
Centralization
B
Std. err. B
Beta
T*
Elect. Strength

Complexity
0.15

0.07
0.02
0.42
0
Centralization
0.13
-0.10

0.04
0.02
0.2
0.037
Involvement
-0.28
0.50
0.06
-0.09
0.02
-0.5
0

(Constant)

0.01
0.01

0.492
N of parties = 92.
Adjusted R2 = 0.20.

*Two-tailed significance test.

Assessing Effects on Breadth of Activities

To construct a 'complete' explanation of party efforts in propagandizing their ideas and programs and providing for members' welfare, one might cite the type and intensity of the party ideology, the economic condition of the party's supporters, the nature of the competition the party faced from other parties, and the social welfare role assumed by the government itself. In attempting to explain parties' reliance on propaganda and welfare activities using only organizational characteristics, we again rely on only a few factors theoretically important. As before, we seek only to determine what proportion of the variance in breadth of party activities can be attributed to organizational characteristics in theoretically sensible ways.

The basic theory underlying this analysis was expressed in different terms by several scholars during or soon after the 'golden age' of parties. Duverger wrote of variations in the ~nature' of participation within parties, some of which were communities' or even 'orders' instead of mere 'associations'.[52] Neumann distinguished between the parties of 'individual representation' and those of social integration', which take over a good part of their members' social existence.[53] Blondel contrasted 'representative' and 'mobilizing' parties.[54] All these authors separated parties which were exclusively vehicles for electing candidates to government office from those which did not confine their activities to election campaigns but conducted continuous campaigns of political education and attended to the social needs of their supporters. The general argument was that the broader the scope of party activities, the greater the need for 'strong' party organization and the more involved members are in party life. Limitations on organizational complexity, centralization, and involvement constituted limits to party activities. Translated into concepts in the ICPP Project, the proposition to be tested is 

H4:The greater the complexity, centralization, and involvement, the broader the scope of party activities.

The data in Table 3 generally support the proposition. Complexity, centralization. and involvement all display simple correlations with breadth of activities from 0.42 to 0.58. Moreover, the multiple regression analysis explains 44% of the variance, with all of the Betas in the predicted direction (0.41, 0.31, and 0.15, respectively). However, only complexity and centralization have significant effects at the customary 0.05 level.

TABLE 3. Effects of Organization on Breadth of Activities

Intercorrelation Matrix
Regression Equation
Variables
Breadth of Activities
Complexity
Centralization
B
Std. Err.
Sig.
T*
Complexity
0.58

0.48
0.12
0.47
0
Centralization
0.42
0.10
0.32
0.12
0.31
0.009

Involvement
0.49
0.48
0.36

0.18
0.14
0.243

(Constant)

0.1
0.08

0.212
N of Parties = 50.

Adjusted R2 = 0.44.

*Two-tailed significance test.

Assessing Effects on Legislative Cohesion

Ozbudun carefully studied the factors in party organization thought to be Important for explaining the voting cohesion of parties in parliament. He cites 'strong party organization' (our 'complexity'), 'central control' of discipline and nominations (our centralization), and the party's 'social integrationist character' (our 'involvement').[55] Other studies of voting behaviour in the US Congress and in other countries have also cited high factionalism.[56] These considerations lead to the hypothesis: 

H5: Legislative cohesion is increased by complexity, centralization, and involvement--but decreased by factionalism.

To test H5, legislative cohesion was regressed on all four organizational characteristics. The results (not shown here) are unsatisfying. For all 70 parties scored on this variable, the adjusted R-square is only 0.19, and complexity and factionalism are the only significant variables. This is the time when other conditions need to be included in the model for the organizational variables to produce substantial effects.

There are several possible system-level causes of cohesion in parliamentary voting. Kornberg's comparison of party cohesion in the US and Canada also attributes fundamental importance to the legislative structure, confirming Ozbudun's contention that parliamentary systems elicit more cohesive behavior than presidential systems.57 Accordingly, the parliamentary status of the party system will be included as an institutional variable, supplementing our organizational factors. Moreover, trying to explain legislative cohesion makes sense only in countries that have effective legislatures, and 7 of our 28 countries did not have effective legislatures during the period of the data.58 The revised model holds that party cohesion in effective legislatures is a positive function of one environmental variable, parliamentarism, and three organizational variables: complexity, centralization, and involvement. Only factionalism is expected to predict negatively to cohesion:  

H6: Legislative cohesion in effective legislatures is increased by parliamentarism, complexity, centralization, and involvement--but decreased by factionalism.

The data reported in Table 4 demonstrate the effects of organizational characteristics after features of the parliamentary system are taken into account. Eliminating systems lacking an effective legislature reduces the sample size from 70 to only 53 parties. However, the adjusted R-square is 0.39 and all of the variables are significant at the 0.05 level using a one-tailed test. As hypothesized, party cohesion in legislative voting is positively related to parliamentary government, organizational complexity, and centralization--and negatively related to factionalism. But, contrary to theory, cohesion is negatively related to 'involvement'.

TABLE 4. Effects of Organization on Legislative Cohesion in Effective Legislatures

Intercorrelation matrix
Regression equation

Variables

Cohes.
Complx.
Central.
Involv.
Faction.
B
Std. Error
Beta
Sig. T*

Legis. Cohes.

Complexity

0.22

0.14
0.04
0.44
0.001

Centraliz.

0.36
-0.11

0.09
0.05
0.28
0.049

Involvement

0.08
0.33
0.4

-0.08
0.04
-0.26
0.053

Fact'ism

-0.43
0.17
-0.25
0

-0.10
0.03
-0.38
0.002

Parliament**

0.38
-0.12
0.51
0.28
-0.20
0.19
0.08
0.29
0.028

(Constant)

0.69
0.08

0
N of parties = 53
Adj. R-square = 0.39.
*Two-tailed significance test; given a one-tailed test, involvement is significant <0.05.
**Scored 1 if the parties operated in a parliamentary system; scored 0 otherwise.

Involvement has an insignificant, but positive, simple correlation with cohesion (0.08). Yet its effect is significant and negative once the other conditions are taken into account. Controlling for the parliamentary system, complexity, centralization, and factionalism, the analysis shows that high citizen involvement in the party actually decreases cohesion. Ex post facto explanations are suspect, of course, but parties with high levels of member involvement in party purposes may be more apt to have legislators who deviate from the majority on matters of principle in voting. In contrast, going along with the majority may be easy for legislators when there is little involvement by party activists.

Path: Janda's Home Page > Workshop > Canonical Analysis > Return to Opening Page