SOCIAL MOBILIZATION. ECONOMIC DEVELOPMENT, AND LEFT VOTING: CHILE, 1958-1964 Thesis for the Degree of Ph. D. MICHIGAN STATE UNIVERSITY CHARLES LEE PRYSBY 1 973 I “Wt-w 6.7.- r . ' ' ,. " {is L IBIZA R Y Michigan State 1,- University This is to certify that the thesis entitled SOCIAL MOBILIZATION, ECONOMIC DEVELOPMENT, AND LEFT VOTING: CHILE, 1958-19614 presented by CHARLES LEE PRYSBY has been accepted towards fulfillment of the requirements for Ph. D. degree in POLITICAL SCIENCE Major professor é Dual/14,61 5,, /473 0-7639 ‘ I; 863'! BINDERY INB. I'M _ T ' BIRDING‘BY * DAB 8: SONS' LIBRARY BINDERS menu] ABSTRACT SOCIAL MOBILIZATION, ECONOMIC DEVELOPMENT, AND LEFT VOTING: CHILE, 1958-1964 BY Charles Lee Prysby The focus of this study is on specifying and testing some possible causal relationships between certain aspects of socio—economic change and electoral support for leftist political parties. The basic theme is that the process of socio—economic change affects electoral support for leftist political parties by affecting the existence, perception, and interpretation of socio-economic deprivations, principally for lower- strata groups in the society. A review of the literature on left voting leads to the hypothesis that working-class individuals with higher levels of politicization and relative economic deprivation will be more likely to support leftist political parties. Relative economic deprivation refers to perceived or subjective economic deprivation, not objective condition, while politicization refers to political awareness and involvement. A review of the Charles Lee Prysby literature on socio-economic modernization leads to the argument that changes in social mobilization and economic conditions for lower strata groups will affect levels of politicization and relative economic deprivation for these groups and will therefore affect the extent of electoral support for leftist political parties. The above arguments apply particularly to develop- ing societies with competitive elections and political party cleavages based largely on social class. Specific propositions derived from the above arguments are tested by using data pertaining to the Chilean presidential elections of 1958 and 1964. Sample survey data are used in an examination of individual-level relationships, while aggregate electoral and census data are used in an eco- logical analysis of electoral support for the Chilean left (the FRAP). Relying on both survey and aggregate data makes it possible to examine both micro-relationships and macro-relationships, which allows for a fuller testing of the theoretical relationships under study. The survey analysis supports the conclusion that politicization and relative economic deprivation are both important subjective factors related to working-class sup— port for the Chilean FRAP. The survey analysis also indi- cates that, as predicted, politicization is affected by the various social mobilization variables examined Charles Lee Prysby (education, organizational membership, and media participation). Contrary to expectations, the data indicate that working-class levels of relative economic deprivation are completely unrelated to the social mobili- zation variables and to objective economic condition. The ecological analysis uses multiple and partial correlation and regression analysis to examine the impact of working-class levels of social mobilization, economic conditions, and recent economic improvement upon the extent of electoral support for the Chilean FRAP. The analysis shows that the size of the FRAP vote in a province is related to the level of social mobilization for the working class in the province, even when other relevant factors are controlled for. However, very little connec— tion is found between the size of the FRAP vote in a province and either the level of economic conditions for the working class in the province or the extent of recent economic improvement for the working class in the province. The process of social mobilization is likely to heighten electoral support for leftist political parties in developing societies by increasing the extent of working-class politicization and, perhaps, class awareness. Social mobilization may also have an important indirect effect on electoral support for the left through potential effects on the organizational strength and activity of the Charles Lee Prysby left. Social mobilization does not, however, appear to have any systematic or consistent effects on relative economic deprivation, either by itself or in conjunction with changes in economic conditions, and thus does not affect electoral support for the left in this manner. This last finding contradicts some frequently appearing generalizations about the tendency for social change in developing societies to produce a revolution of rising frustrations, but these generalizations have rarely been subjected to an adequate empirical test. SOCIAL MOBILIZATION, ECONOMIC DEVELOPMENT, AND LEFT VOTING: CHILE, 1958-1964 BY‘ Charles Lee Prysby A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree Of DOCTOR OF PHILOSOPHY Department of Political Science 1973 65‘ Copyright by CHARLES LEE PRYSBY 1973 ACKNOWLEDGMENTS A number Of people have provided support and assistance to me in this project and deserve special acknowledgment. Professor E. Spencer Wellhofer, my committee chairman, continually assisted me along the way and Challenged me to rise above the data and write a dissertation that was more theoretically meaningful. Professor Joseph A. Schlesinger provided perceptive criticism of earlier drafts that improved the conceptual clarity of the final product. Professor Cleo H. Cherry— holmes provided valuable advice on the research design and helped me improve the methodological soundness of the study. Professor Thomas H. Greene encouraged me in the initial formulation Of the project and shared his own research ideas with me. 'Finally, my wife, Anita, provided continual moral support and generously supplied a cheap source of labor that I readily exploited. To her and all the others who helped me, I am grateful. Naturally, I am solely responsible for the errors or shortcomings that remain in this dissertation. iii TABLE OF CONTENTS Page CHAPTER I. SOCIAL SOURCES OF LEFT VOTING . . . . . 1 Introduction Economic Deprivation and Left Voting Relative Deprivation and Left Voting Psychological Stress and Left Voting Social Change and Left Voting Summary CHAPTER II. SETTING AND METHODS . . . . . . . . . . 33 Introduction The Chilean Political System The Socio-economic Setting The Research Design CHAPTER III. SURVEY ANALYSIS . . . . . . . . . . . . 57 Introduction Social Class and Political Preference Subjective Factors and Political Preference The Impact Of Objective Factors Conclusions CHAPTER IV 0 AGGREGATE ANALYSIS 0 O O O O O O O O O 97 Introduction Propositions Analysis Design Analysis Results Conclusions iv Page CHAPTER V. CONCLUSIONS 0 O O O O O O O O O O O O O 135 Introduction Summary and Interpretation Of Findings Generalizability of Findings Suggestions for Future Research APPENDIX 0 O O O O O O O O O O O O O O O O O O Q C O 166 BIBLIOGRAPHY O O O I O O O C O O O O O O O O O C O 0 l7]- Table 111-1. III-2. III-3. III-4. III-6. III-7 O III-8. III-9. IV-l. IV‘Z o IV-3. Social Class by Party Preference Candidate Preference by Social Class Party Preference by Politicization and Relative Economic Deprivation for LIST OF TABLES Working-Class Respondents . Politicization by Education for Working- Class Respondents Politicization by Organizational Member- ship for Working-Class Respondents Politicization by Media Participation and Social Class Relative Economic Deprivation by Educa- tion for Working-Class Respondents Relative Economic Deprivation by Organizational Membership for Working- Class Respondents Relative Economic Deprivation by Media Participation for Working-Class Respondents Correlation Matrix for Social Mobiliza- tion Index and Component Variables Correlation Matrix for Economic Conditions Index and Component Variables . Correlation Matrix for Independent and Dependent Variables . vi 0 Page 60 61 68 78 78 79 83 83 83 111 114 115 Table Page IV¥4. Provincial and Individual Level Variance for Selected Variables . . . . . . . . . . 119 IV-S. Statistics for Linear Regression Predic— tion of Per Cent Vote for Allende, 1958 O O O O O O O O O O O O O O O O O O O 122 IV-6. Statistics for Linear Regression Predic- tion of Per Cent Vote for Allende, 1964 O O O O O O O O O I O O O O O O O O O 128 A-l. Aggregate Voting Statistics for Chilean Provinces . . . . . . . . . . . . . . . . . 168 A-2. Social Mobilization Statistics for Chilean Provinces, 1960 . . . . . . . . . . 169 A-3. Economic Conditions Statistics for Chilean Provinces . . . . . . . . . . . . . 170 vii CHAPTER I SOCIAL SOURCES OF LEFT VOTING Introduction Left voting long has been a topic of interest for social scientists.1 There already exists a lengthy and distinguished list of publications concerned with identi- fying the socio-economic and/or social-psychological sources of left voting. Not only do these studies differ in the nature of their explanations, but also in terms of what they try to explain--from the vote for moderate majority parties to support for radical sectarian organizations. Since the specific subject of this study is the social base of support for the Chilean left, my examination of the literature emphasizes material primarily concerned with mass-based socialist parties of a fairly radical character. In particular, I attempt to identify some social and economic factors affecting left voting and to relate these factors to the process of socio—economic development. The propositions that I put forth in this chapter are scarcely original. As the review of the literature will show, the general ideas have been suggested, in one form or another, by several social scientists. But what they have rarely received is a systematic formulation and rigorous empirical test, which is what I attempt in this study. Because these relationships should be of fairly broad interest in the fields of political development, comparative political parties, and cross-national voting behavior, such a research task may be quite worthwhile. Economic Deprivation and Left Voting Left voting is commonly presumed to come dispro- portionately from the lower strata, who vote left in response to socio-economic needs. Formulations of this basic View generally relate left voting to occupation or income. Seymour Martin Lipset, for example, Observes that: . . . the most impressive single fact about political party support is that in virtually every economically developed country the lower-income groups vote mainly for parties of the left, while the higher-income groups vote mainly for parties of the right.2 The simplest explanation for this widespread pattern is simple economic self-interest. The leftist parties repre— sent themselves as inétruments of social change in the direction of equality; the lower-income groups support them in order to become ecOnomically better off, while the higher- income groups Oppose them in order to maintain their economic advantage. The above Observation is, of course, common knowledge. Voting behavior studies have invariably found this rough connection between social class and political party support. However, this simple connection between socio-economic position and left voting begins to break down when finer distinctions are made. Focusing on the relationship between income and left voting, we find that poorer-paid workers are Often no more left in their voting than their better-paid counterparts.“ Also, in multi—party systems with two significant socialist parties, the more radical one does not necessarily have a poorer clientele.S Finally, the relationship between left voting and income is far from a strong one. In many Western democracies, for example, a large proportion Of the low-income individuals fail to vote left and a sizable portion of the left vote Often comes from the relatively better-off. In sum, while the lower strata may be a greater source of left votes than the upper strata, there is no strong and consistent connec- tion between economic position and left voting, particularly within lower—strata groups. Of course, the lack Of a perfect correlation between income and voting behavior can mean nothing more than that there are other independent variables affecting the depen- dent variable. If left voting is viewed primarily as a response to socio—economic deprivation, then factors other than income might well be considered. Lipset suggests three deprivations, besides low income, strongly affecting the tendency to vote left: low status, low economic security, and unsatisfying work.6 Other factors, such as opportunities for economic improvement and/or social mobility, would also seem to be important.7 Thus, some Of those with low income may be relatively well Off in status and economic security, which may result in their failing to vote left; others may be fairly well paid but low in status and work satisfaction, causing them to cast their ballot for the left. Consider- ing several different types of socio-economic deprivation may allow better prediction of left voting. In fact, occu- pation is frequently used in relating socio—ecOnomic position to voting behavior because certain occupational distinctions, such as manual versus non-manual, are very good indicators of differences in wealth, status, economic security, and so on.8 One socio-economic deprivation, unemployment, seems particularly important for left voting. Several studies bear this out. Lipset finds that workers in occupations particularly susceptible to high unemployment rates generally have high rates Of left voting.9 A survey of Cuban workers shows that pre-revolutionary support for the Cuban Commu- nist Party was strongly related to unemployment experiences.1° Richard Hamilton's secondary analysis of survey data on French workers concludes that ". . . the experience and fear of unemployment are among the most important correlates of political attitudes to be discovered in the entire study."11 Moreover, unemployment leads to left voting or radicalism not primarily because of the lower income resulting from less work, but because Of the relatively permanent psychological impact of the unemployment expe- rience: . . . the "lesson" of unemployment, once learned, stays with the workers, and high wages does not suffice to make them forget it.12 . . . there is something about the experience of unemploy— ment, apart from the experience of economic deprivation, that is responsible for making the workers more amenable to radical politics.13 But even when the different aspects Of Objective socio-economic deprivation are considered, inconsistencies in the empirical evidence still exist. In many countries the skilled workers are more radical in their voting behavior than the unskilled."+ Yet skilled workers tend to be better Off in every socio-economic category, from income to work satisfaction, suggesting a fundamental weak- ness in the original formulation Of a relationship between Objective socio-economic deprivation and left voting. Of course, these inconsistencies can always be regarded as mere aberrations of the true relationship, due perhaps to the unusual cross-cutting of economic and non-economic cleavages or to special circumstances Of the specific situa- tion. More likely, the frequency Of these inconsistencies, along with the general lack Of any strong connection between economic position and voting behavior within the working class, points to the inadequacy Of this simple economic- deprivation explanation. Lipset suggests that the greater likelihood of skilled, rather than unskilled, workers in Northern Europe to vote left can be explained by the rigidity Of the status hierarchy in these countries: in the more status-differentiated societies (e.g., Germany, Sweden) better-Off working-class members experience the frustra- tion Of being rejected by the middle class despite their economic success.15 This is certainly an interesting explanation of what, from the standpoint of the relation— ship between economic position and voting behavior, appear to be deviant cases. However, even granting the perhaps unwarranted assumption that skilled workers seek middle-class accep- tance rather than working-class admiration, the limited empirical evidence does not support the Lipset hypothesis. First Of all, it is not clear that the Northern European countries are significantly more status-differentiated than other industrial nations: there is great similarity among France, Japan, Sweden, the United States, and West Germany in the rate of inter-generational mobility from manual to non-manual occupations and in the proportion Of inter-class marriages.16 More damaging, Maurice Zeitlin's study found that in Cuba (a relatively more status-differentiated society) the skilled workers were more likely than the unskilled to have supported the Cuban Communist Party prior to the 1959 revolution, but his evidence directly contradicts the status-rejection explanation of this phenomenon: . . . those workers who think of themselves (subjectively per— ceive themselves) as enjoying a relatively higher economic position in the working class should experience more status- frustration than those who think their position generally is on a par with other workers. After all, the fact that they think they are better Off than other workers and still get rejected by middle-class individuals ought to be particularly galling to them, whatever their actual Objective economic position. They should, therefore, if the Lipset-Bendix hypothesis was correct, be more likely to develop social resentment and consequent pro-Communist views than other workers. . . . we find that, contrary to what we should expect on the basis of the Lipset—Bendix hypothesis . . . the workers who think their wage is "higher" than that of other workers are less likely to have supported the Communists than those who think their wage is the "same" as that earned by other workers.17 Although the Lipset status—rejection hypothesis may be lacking in empirical confirmation, it does suggest some— thing very important: the deprivations an individual feels may be much more important than the ones he is actually subject to in determining his political behavior. Relative Deprivation and Left Voting Whatever connection may exist between Objective socio-economic deprivation and the propensity to vote for the left, it is clear that objective deprivation in itself explains nothing--e.g., . . . income changes per se clearly exert no pressure or influence on the wage earner. The pay check, after all, is an inert object. Obviously, it is how the individual perceives and inter- prets his socio-economic condition that affects his political attitudes and political behavior, and any stated connection between Objective economic deprivation and vot- ing behavior must, at least implicitly, make some assump- tions about the perception and interpretation Of these deprivations. Put this way, it would seem that what should be looked at is perceived or relative deprivation and the factors that affect this. Relative deprivation is defined by Ted R. Gurr as: . . . a perceived discrepancy between men's value expectations and their value capabilities. Value expectations are the goods and conditionscxflrfe to which people believe they are rightfully entitled. Value capabilities are the goods and conditions they think they are capable of attaining or main— taining, given the social means available to them. 9 A similar definition is presented by W. G. Runciman: . . we can roughly say that A is relatively deprived of X when (i) he does not have X, (ii) he sees some other person or persons, which may include himself at some previous or expected time, as having X (whether or not this is or will be in fact the case), (iii) he wants X, and (iv) he sees it as feasible that he should have x."-° Relative deprivation can vary in both the intensity and the nature of the deprivation, although this study will be limited to economic deprivations. Relative economic deprivation is purely subjective in nature and need not correspond to Objective economic deprivation. The visibility Of the Objective economic deprivation may be suggested as an important factor here, which would indicate the relevance of communications- related variables. But as a source of relative economic deprivation, simple awareness of others who are better Off is insufficient unless the better-Off are taken as a reference group. Individuals must not only realize that they are worse Off than others but must also feel that they should not be so deprived. Because Of this, there fre- quently is little connection between Objective economic deprivation and relative economic deprivation. Runciman Claims that for objectively deprived groups in a highly static social system reference groups are generally lim- ited.21 Gurr indicates that improvement in the economic position Of similar individuals has a much greater effect on one's relative deprivation than the condition of highly dissimilar groups.22 Zeitlin suggests that skilled workers, even though they are better paid than the less skilled, are not less likely to be high in relative deprivation for they probably feel that their skill entitles them to better pay.23 All of this points to the difficulty in determin- ing the degree of relative deprivation in a group from the objective condition Of the group. 10 Relative economic deprivation does not automatically affect voting behavior. Where individuals feel that their own deprivation stems largely from their membership in an unjustly rewarded group and that action designed to improve the situation of the group as a whole is an effective way Of relieving their deprivation, the potential for that deprivation to affect political behavior is quite high. Where opposite feelings concerning deprivation are present, the potential for affecting political behavior is low. Thus, whether or not higher relative economic deprivation increases working-class support for leftist political parties depends on several factors. First of all, for relative economic deprivation to affect voting behavior, individuals must perceive the efficacy of political means for reducing their deprivation. This makes the level of politicization a crucial factor here. Politicization, as used in this study, refers to the extent of political awareness and involvement and can be thought of as ". . . a continuum ranging from lack of perception of the relevance Of government to one's life, through perception of it, to active involvement in poli- tiCS fl 2"! More politicized individuals are more likely to respond to their relative economic deprivation in political terms, so the more politicized workers should be a greater source of support for leftist parties. Specifically, the 11 more politicized workers should be: (a) more likely to understand the importance Of political action designed to improve the condition Of the working class for improving their own situation, and (b) more likely to see leftist political parties as representing the interests of the working class. Related ideas are expressed by Lipset, who suggests that some social aspects of the working-class environment may affect the probability Of workers becoming aware of the possibilities for collective action to improve their eco- nomic situation: A large plant makes for a higher degree of intra-class communi— cation and less personal contact with people on higher economic levels. In large cities social interaction is also more likely to be within economic classes. In certain cases the working: class districts of large cities have been so thoroughly organ- ized by working-class political movements that the workers live in a virtual world of their own, and it is in these centers that the workers are the most solidly behind leftist candidates, and, as we have already seen, vote most heavily.25 Although more attention has been devoted to the factors affecting the perception Of and response to economic deprivation on the part Of industrial workers, the same principles can be applied to rural groups as well. For example, Lipset argues that socialist strength in Sas- katchewan is attributable to the high economic vulnera- bility of the wheat farmers.26 Similarly, Donald Zagoria identifies some economic organization factors influencing political radicalism among the rural poor in his ecological 12 examination of the social base of peasant communism in India.27 An emphasis on how socio-economic conditions are perceived and interpreted helps to explain why unemploy- ment has such a strong influence on left voting. Clearly, the experience of being without work is a highly visible deprivation. Having been exposed to a period of steady work and pay, the worker suddenly finds himself without either, probably through no real fault of his own. Furthermore, he can easily compare his situation with that of some more fortunate reference group, either other workers who are still employed or himself at some previous point in time. Where unemployment is a recur- ring phenomenon, the likelihood of the workers blaming the existing social and economic order for the problem increases.28 In a situation of cyclical unemployment there are likely to be strong feelings Of common interest within the working-class, so the effect of the fear of unemployment on political attitudes will be strengthened. Also, the effect of unemployment on left radicalism is sharply accentuated by the presence of a trade union, indicating the importance of working-class communications— related variables.29 At this point it may be useful to summarize some arguments derived from the above review of the literature. 13 First, in a great many countries there is a strong connection between social class and political party sup- port, with those in manual occupations consistently more left in their voting behavior than those in non~manual occupations. This, as Robert Alford points out, is ". . . natural and expected . . . given the character of the stratification order and the way political parties compete for support from various groups."30 Second, within the working class there is little or no connection between Objective economic position and support for leftist political parties. The working-class members who are more likely to vote left include those who have higher levels of relative economic deprivation and who feel that their deprivation will be significantly reduced by political action designed to improve the condi- tion of the working class or some segment of it. Runciman presents an example of such a worker: Consider, by contrast, a factory worker who feels that he is grossly underpaid. He is conscious, and even militantly conscious, of belonging to the working class. He has no ambition to rise above his fellows. But he feels that he and all those like him are insufficiently rewarded both in money and status by the society to whose welfare they are contributing by their work. He feels relatively deprived as one Of a class whose members all share the same conditions of life and employment.31 Since the level Of politicization is likely to affect per- ceptions concerning the efficacy of, or necessity for, class-related political action for reducing deprivation, 14 the workers more likely to support leftist political parties include those with higher levels Of both relative economic deprivation and politicization. Workers with lower levels of politicization and relative economic deprivation should be substantially less likely to sup- port leftist political parties. Third, objective socio-economic variables can probably be used to specify the segments of the working class that are more likely to provide electoral support for leftist political parties if conditions affecting the perception and interpretation of deprivation are con- sidered. Relevant factors include those that increase the visibility of deprivations, that encourage compari— sons with more fortunate groups, that influence awareness of the possibilities for political solutions, and that stimulate feelings for the need for collective action on the part of the working class or some segment Of it. Some specific factors will be suggested later in this chapter. Psychological Stress and Left Voting The approach presented thus far has been largely in what might be termed the "Marxian" tradition: left voting is presumed to result primarily from socio- economic deprivation, particularly as it relates to 15 social class. Also legitimately part Of this approach are appropriate modifications to account for the percep— tion and interpretation of such deprivation, especially in terms of the awareness of common interests and possi- bilities for collective action within the working class.32 There is an entirely different approach to the relationship between social class and left voting, which assumes that political radicalism results from the lack Of a clear class identification. Derived from Emile Durkheim's idea that the lack of effective and Clear norm regulation leads to anomie, this approach focuses on the degree of status crystallization (i.e., the degree to which the individual occupies the same rank on different status variables), with the hypothesis being that low crystallization leads to greater political radicalism.33 Very briefly, the reasoning is that individuals with low) status crystallization suffer from social-psychological stress, due to their poor integration into society, and support parties of social change in the hope that a new social order will ameliorate their situation. There are a diversity Of formulations of this basic view, but the empirical evidence concerning the possible effects of low crystallization on voting behavior or political party preference is at best contradictory.3” Highly related to the status-inconsistency expla- nation of left radicalism are those approaches attributing 16 radical-left voting to various forms of social-psychological stress resulting from a wide variety of causes. Several writers emphasize that rapid social change frequently leads to social disorganization, resulting in alienated or anomic masses having a marked tendency to support extremist political movements. Erik Allardt, for example, finds that in Finland: Increase in Communist strength is nowhere related to factors which would reflect a high degree of stability in social conditions . . . increase in Communist strength is associated with changes which in one way or another are likely to uproot individuals.35 Industrialization and urbanization are generally cited as the two social processes most responsible for uprooting individuals and causing social disorganization: Although it is undoubtedly true that there are other sources of mass tendencies, the very rapid expansion of cities and industries has constituted perhaps the most general source of social atomization in the modern world, insofar as they have inhibited the growth of new forms of group life to replace the village community, extended family, and guild which they destroyed. More concretely, industrialization, when it is not accompanied by the evolution and legitimation of trade unions, . . . favors the atomization of the working class and the formation of mass movements. In Western Europe it appears that areas where industri- alization occurred rapidly were more likely to support extremist working-class political movements (e.g., Com- munist parties).37 Similarly, the literature on Latin American urbanization suggests a relationship between rapid urbanization and political radicalism.38 17 Many explanations of Communist appeal focus on social-psychological stress of one sort or another, with Gabriel Almond's study being one of the fullest expres- 9 Thomas Greene's critique sions of this point of view.3 of this "protest hypothesis" nicely points out the vari- eties Of psychological distress that can be suggested as the sources of Communist appeal: . . . communism becomes an expression of every real and imagined affliction that ails society--socio—economic dislo- cation (the peasant transplanted to the city), psychological guilt (the intellectual who feels himself an economic para- site), negative affect and political alienation (the individual vis—a-vis society and vis-a-vis himself), cultural cleavage (the society at large), not to mention the general distress stimulated by conditions of economic misery. Outside Of some ecological correlations between various indices Of social disorganization and left radicalism there is little empirical evidence to support the view that the Communist voter is an alienated or anomic individual. In fact, the available survey data strongly contradict such a view.1+1 Most of the "protest" explanations have relied on the most impressionistic evidence, with Almond's semi-structured interviews with a non-random sample Of ex-Communists in four countries being one of the more rigorous attempts. Commenting on the "protest" explanation Of Communist voting, Greene concludes: . . . almost all students of communism appear to have initiated their inquiries with a presumption in favor of the protest characteristics of the communist clientele; and their research methodologies have been too carelessly devised to reveal the deficiencies Of their hypotheses."2 18 The explanations of radical-left voting in terms of various forms Of psychological stress are largely characterized by conceptual inadequacies and empirical inconsistencies. However, there is some evidence to sug- gest that rapid social change does have implications for the electoral support Of leftist political parties, al- though for different reasons. Social Change and Left Voting There are several aspects Of the general process Of socio-economic modernization likely to have an impact on electoral support for the left. What I attempt is to identify some of these aspects, with the general orienta— tion being that socio-economic modernization or Change affects left voting insofar as it affects the existence, perception, and interpretation of socio-economic depriva- tions. Of course, these remarks are applicable primarily to those developing countries having a competitive party system with one or more distinct leftist parties, although some of the ideas have more general implications. First of all, socio-economic modernization involves the social mobilization of substantial parts of the popu- lation. Social mobilization is defined by Karl Deutsch as: . . . the process in which major clusters of old social, economic, and psychological commitments are eroded or broken and people become available for new patterns of socializa- tion and behavior."3 19 Above all, social mobilization heightens awareness of the social and political environment. Daniel Lerner discusses this in terms of empathy or psychic mobility--i.e., the capacity to identify with new aspects of the environment.“ Prime movers in this aspect of modernization include mass media participation, literacy and education, organizational membership, industrialization, and urbanization. In this study the term social mobilization will be used to refer specifically to this set Of interrelated socio-economic factors. Thus, the process Of social mobilization will refer to increases in these factors, and the level of social mobilization will refer to the magnitude of these factors. The mass media are seen by Lerner as particularly important in present developing societies, as the com- munications revolution ". . . Opened to the large masses "”5 Education Of mankind the infinite vicarious universe. hastens the spread of modern values as well as increasing the level of literacy, which is described by Lerner as ". . . the basic personal skill that underlies the whole modernizing sequence.“6 Increases in organizational membership are also involved in the spreading of modern values, the heightening Of awareness, and the creation Of new patterns of socialization and behavior.”7 Highly related to all of this is the general process of 20 urbanization and industrialization, as these geographical and occupational shifts in the population are at the heart of the modernization process."8 Although these various aspects of social mobilization are interrelated in complicated patterns, it is not necessary to spell out these connections here. For the purposes of this study, it will be convenient and sufficient to simply consider these social mobilization factors as generally occurring together and as generally having a substantial impact, both together and individually, on psychological orienta— tions and dispositions. It would seem that the process of social mobiliza- tion makes social and economic deprivations more visible. As individuals move to the city, take jobs in factories, obtain literacy, join organizations, and become exposed to the mass media, they become more aware Of the socio— economic condition Of others. More importantly, they are less likely to perceive their own situation as a natural given and more likely to make comparisons with more for- tunate groups. The more modernized person is aware of the possibilities of change and is likely to seek alterations in his own situation. One Of the most important effects of the process of social mobilization, according to Lerner, is that it almost invariably increases aspirations and expectations.“9 21 But this "revolution of rising expectations" caused by the process of social mobilization need not result in widespread dissatisfaction. Possibly, these increased aspirations, or at least most of them, will be met. As was pointed out previously, it is the unfulfilled expectations that are Of particular significance for political behavior: . . . an individual's level of satisfaction is always, at any moment of his life, a ratio between what he wants and what he gets, i.e., between his aspirations and his achieve— ments . . . relative deprivation, as has been shown, is the effective measure of satisfaction among individuals and groups.so While the effect of social mobilization is to increase aspirations and expectations, which if unfulfilled increase the extent of relative deprivation or dissatis- faction in the society, a concomitant of social mobiliza— tion is economic development, and it might seem that economic development would increase the capacity of a society to satisfy the increased expectations, thus balancing the effects of social mobilization. However, in the short run this is not necessarily true. In the long run, economic development does produce higher living standards, more economic opportunities, and perhaps some narrowing of the rich-poor gap. But the more immediate effect of economic growth frequently is a worsening of the relative position of thelower strata and no real improvement in living conditions for much Of the 22 population.51 Thus, even if economic growth matches the rate Of social mobilization, and few of the present developing nations have experienced the sustained increases in the economy necessary to match the usually rapid rates of social mobilization, the level of unfilled aspirations will still probably increase in the short run. Similar ideas are expressed by Samuel Huntington in his attempt to identify the conditions causing politi- cal instability in developing countries.52 Specifically, he argues that the degree of social frustration in a society is a function of the ratio of social mobilization to economic development: Social mobilization is much more destabilizing than economic development. The gap between these two forms of change furnishes some measure of the impact of modernization on political stability. Urbanization, literacy, education, mass media, all expose the traditional man to new forms of life, new standards Of enjoyment, new possibilities of satis- faction. These experiences break the cognitive and attitudinal barriers of the traditional culture and promote new levels Of aspirations and wants. The ability of a transitional society to satisfy these new aspirations, however, increases much more slowly than the aspirations themselves. Conse- quently, a gap develops between aspiration and expectation, want formation and want satisfaction, or the aspirations function and the level-of-living function. ThlS gap generates social frustration and dissatisfaction.5d Social mobilization also results in increased political awareness and interest. This is a well estab- lished point in the literature on political development. Deutsch claims that ". . . in whatever country it occurs, social mobilization brings with it an expansion of the 23 "5“ Lerner politically relevant strata of the population. finds that political participation is strongly tied to other aSpects Of modernization, particularly urbanization, literacy, and media participation.55 As individuals become more politically aware and involved, they become more likely to place demands on the political system if their aspirations and expectations are not otherwise ful- filled, although the nature Of these demands may take a variety of forms. Thus, although even the most dissatis— fied individual may not blame the political or social order for his deprivation, the increased politicization accompanying social mobilization makes it likely that much of the relative deprivation resulting from the unfulfilled expectations generated by social change will be translated into discontent with existing policy outputs or even the existing social order. Even where social change does not produce higher levels of relative deprivation, the increased politicization will still result in many individuals desir- ing political solutions to problems that previously would have elicited a different response. Much of the relevant literature, then, suggests that the process Of social change will increase levels of relative deprivation and politicization. I argue that in developing countries having a competitive electoral process and one or more viable leftist parties these effects may 24 be translated into electoral support for the left. Even in such countries, however, high rates of social mobili- zation and low rates of economic development do not auto- matically result in a high left vote. Since leftist parties, and particularly socialist ones, appeal specifi— cally to lower-strata groups, the important question is what the levels of relative deprivation and political awareness are for the lower-strata groups, not for the population as a whole. Increases in electoral support for leftist parties from the lower-strata groups should be related to the rates Of change in social mobilization and economic conditions for these groups. Where increases in social mobilization are high and improvements in economic conditions low, support for leftist parties will be much greater than where the opposite is the case. The reason- ing behind these statements is that these objective condi- tions affect the previously specified subjective individual Characteristics, relative deprivation and politicization, that influence lower—strata electoral support for leftist parties. Certainly there is little reason to expect a perfect correspondence between Objective conditions and subjective factors, but if the previous arguments have been correct, there should be a relatively strong connec- tion between these Objective socio-economic conditions and electoral support for leftist parties. 25 Summary It has been argued that the process of socio- economic change affects electoral support for leftist political movements by affecting the existence, perception, and interpretation of socio—economic deprivations, prin- cipally for lower-strata groups in the society. These arguments may be briefly expressed as follows: 1. The process of the social mobilization of lower- strata groups has a positive effect on the strength Of their electoral support for leftist parties by increasing both the level of relative economic deprivation and the level of politicization for these groups. Improvements in the economic conditions for lower- strata groups has a negative effect on the strength of their electoral support for leftist parties by decreasing the level of relative economic depriva- tion for these groups. Since increases in social mobilization and improve- ments in economic conditions are likely to be strongly associated in many cases, any determina— tion Of the net effect of socio-economic change on support for leftist parties in a specific case must consider the relative rates of change and/or levels of both factors. Following a description of the research setting and methods in chapter two, specific propositions derived from the above general statements will be tested in chap- ters three and four. Several factors that may affect electoral support for leftist political parties have not been discussed. One of these factors is the organizational strength Of 26 the left parties and their related organizations. A strong correlation between the size of the Communist vote and Communist Party membership is found in France by Thomas Greene and in Italy by Sidney Tarrow.56 However, it is not easy to explaifi what these correlations repre- sent. The organizational strength Of the left may be considered as an aspect of the process of social mobiliza- tion, resulting from increased social mobilization and, in turn, affecting the perception and interpretation of socio-economic deprivation and channelling discontent into left votes.57 This possibility is consistent with and adds to the relationships I have put forth. Another pos- sibility is that the left may attempt to organize where it knows it already has support.58 Finally, the organiza— tional strength of the left may be a separate variable affecting electoral support completely independent of the above socio-economic relationships.59 More research is certainly needed in this area, but this study devotes only limited attention to these factors. Also not considered is the possibility of a left vote being largely a "traditional" vote for some groups --i.e., the normal and expected way of voting for the group, with party preference simply handed down from father to son, and perhaps reinforced by the existence of leftist organizations that isolate the group from other 27 social and political influences.6° The impact of cleavages other than economic ones has similarly been ignored, although ideological and cultural cleavages com- monly affect voting behavior and might significantly contribute to the understanding Of left voting in some cases. My failure to consider many of these factors in this study should not be interpreted as a dismissal of their importance. Rather, I simply attempt to specify and test some possible causal relationships between cer- tain aspects of socio-economic change and electoral support for leftist political parties. More than this I do not claim. 28 Notes for Chapter I 1I am aware of the difficulties of classifying political parties on a left-right continuum that have been pointed out by Donald E. Stokes in "Spatial Models of Party Com— petition," American Political_§cience Review, LVII (June, 1963), pp. 368—377. However, I am concerned with the socio-economic programs of political parties and am focusing primarily on socialist parties, which can generally be considered more left in their socio-economic programs than the non-socialist parties in the same system.. 2Seymour Martin Lipset, Political Man (New York: Double- day, 1960), p. 234. 3Ibid., p. 239. .‘Thomas H. Greene, "Communist Electorate in France: Testing the Protest Hypothesis" (unpublished paper, University of Southern California, 1969), p. 24; Morris Janowitz and David R. Segal, "Social Cleavage and Party Affiliation: Germany, Great Britain, and the United States," American Journal of Sociology: LXXII (May, 1967), p. 608; Robert T. McKenzie and Allan Silver, "The Delicate Experiment: pIndustrialism, Conservatism, and Working-Class Tories in England," in Party Systems and Voter Alignments, ed. by Seymour Martin Lipset and Stein Rokkan (New York: The Free Press, 1967), p. 119. sGreene, "Communist Electorate in France," p. 24. 6Lipset, Political Man, pp. 2434252. 7Ibid., pp. 267-273. 8Robert R. Alford, Party and Society (Chicago: Rand McNally, 1963). pp. 73-79. 9Lipset, Political Man, pp. 243-248. 1°Maurice Zeitlin, Revolutionary Politics and the Cuban Workin Class (Princeton: Princeton UniversIEy Press, 1 7 I p. 5 0 11Richard Hamilton, Affluence and the French Worker in the Fourth Republic (Princeton: Princeton University Press, 1967), p. 187. 29 12Zeitlin, Revolutionary Politics and the Cuban Working Class, p. 64. 13Ibid. 1"Juan J. Linz, "Cleavage and Consensus in West German Politics: The Early Fifties," in Lipset and Rokkan, PartygSystems and Voter Alignments, p. 287; Lipset, Political Man, p. 253; Zeitlin, Revolutionary Politics and the Cuban Working Class, p. 97. Even in the United States, where the conservatism Of skilled workers is commonly cited, Richard Hamilton, "Skill Level and Politics," Public Opinion Quarterly, XXIX (Fall, 1965), pp. 390-399, finds that With region, age, and race con- trolled there is little or no difference between skilled and unskilled workers in terms of political attitudes. lsLipset, Political Man, p. 254. 16Seymour Martin Lipset and Reinhardt Bendix, Social Mobility in Industrial Society (Berkeley: University of California Press, 1958), pp. 42-47. 17Zeitlin, Revolutionary Politics and the Cuban Working Class, p. 111. 18Hamilton, Affluence and the French Worker in the Fourth Republic, p. 111. 19Ted Robert Gurr, Why Men Rebel (Princeton: Princeton University Press, 1970), p. 13. 2°W. G. Runciman, Relative Deprivation and Social Justice (Berkeley: University of California Press, 1966i, p. 10. 21Ibid., pp. 19-35. 22Gurr, Why Men Rebel, pp. 105-107. 23Zeitlin, Revolutionary Politics and the Cuban WOrking 2"Daniel Goldrich, "Toward the Comparative Study of Politi- cization in Latin America," in Contemporary_Cultures and Societies of Latin America, ed. by Dwight B. Heath and RiChard N. Adams (New York: Random House, 1965), p. 361. 25Lipset, Political Man, p. 262. 30 26Seymour Martin Lipset, Agrarian Socialism (Berkeley: University of California Press, 1950), pp. 21-71. 27Donald S. Zagoria, "The Ecology of Peasant Communism in India," American Political Science Review, LXV (March, 1971), PP. 144—160. 28Zeitlin, Revolutionary Politics and the Cuban Working Class, p. 81. 29Hamilton, Affluence and the French Worker in the Fourth Republic, p. 200. 3°Alford, Party and Society, p. 37. 31Runciman, Relative Deprivation and Social Justice, p. 32. 32Such considerations are at best only implicit in the writings of Marx and Engels. Lenin is more explicit on these factors. See his "What Is to Be Done?" in Essen- tial Works of Lenin, ed. by Henry M. Christman (New York: Bantam Books, 1966). 33Gerhard Lenski, "Status Crystallization: A Non-Vertical Dimension of Social Status," American Sociological Review, XIX (August, 1954), pp. 406-413; Gerhard Lenski, 1'iStatus Inconsistency and the Vote: A Four-Nation Test," Ameri- can Socological Review, XXXII (April, 1967), pp. 288—3OI. 3"Leonard Brown and F. Lancaster Jones, "Status Consistency and Political Preference: The Australian Case," Ameri- can Sociological Review, XXXV (December, 1970), pp. 989- 1001; William Kenkel, "The Relationship Between Status Consistency and Political-Economic Attitudes," American Sociological Review, XXI (June, 1956), pp. 365-368; Alejandro Portes, "Left Radicalism in Chile: A Test of Three Hypotheses," Comparative Politics, II (January, 1970), PP. 251-274. 35Erik Allardt, Social Sources of Finnish Communism: Tra- ditional and Emerging Radicalism, research report, Insti- tute of Sociology (Helsinki: University of Helsinki, 1963), pp. 10-12. ‘ ' 36William Kornhauser, The Politics of Mass Society (New York: The Free Press, 1959), p. 157. 37Lipset, Political Man, p. 54. 31 38See Richard Morse, "Recent Research on Latin American Urbanization," Latin American Research Review, I (Fall, 1965), pp. 35-74, for a summary of the literature. 39Gabriel Almond, The Appeals of Communism (Princeton: Princeton University Press, 1954). l‘I’Greene, "Communist Electorate in France," p. 8. “lGreene, "Communist Electorate in France," pp. 24-31; Hamilton, Affluence and the French Worker in the Fourth Republic, pp. 113-115; Sidney G..Tarrow, Peasant Communism in Southern Italy (New Haven: Yale University Press, 1967), pp. 158-161; Gabriel Almond and Sidney Verba, The Civic Culture (Boston: Little, Brown and Company, 1985T, pp. 115-116. “ZGreene, "Communist Electorate in France," p. 9. “3Karl W. Deutsch, "Social Mobilization and Political Development," American Political Science Review, LV (September, 1961), p. 494. I"’Daniel Lerner, The Passing of Traditional Society (New York: The Free Press, 1958), pp. 47—51. “51bid., p. 53. “‘Ibid., p. 64. “7Deutsch, "Social Mobilization and Political Development," p. 495. “BLerner, The Passing of Traditional Sociepy, pp. 43-68. “sDaniel Lerner, "Toward a Communications Theory of Moderni- zation," in Communications and Political Development, ed. by Lucian W. Pye (Princeton: Princeton University Press, 1963), pp. 330-335. 5°Ibid., p. 333. 51Mancur Olson, Jr., "Rapid Economic Growth as a Destabil- izing Force," Journal of Economic History, XXIII (December, 1963), pp. 529-552. 52Samuel P. Huntington, Political Order in Changing Soci- eties (New Haven: Yale University Press, 1968). 531bid., pp. 53-54. 32 5"Deutsch, "Social Mobilization and Political Development," p. 498. 55Lerner, The Passing of Traditional Society, pp. 60-65. 56Greene, "Communist Electorate in France," p. 34; Tarrow, Peasant Communism in Southern Italy, p. 203. 57Tarrow, Peasant Communism in Southern Italy, pp. 204—209. 58Greene, "Communist Electorate in France," p. 35. 59Giovanni Sartori, "European Political Parties: The Case of Polarized Pluralism," in Political Parties and Poli- tical Development, ed. by Joseph LaPalombara and Myron Weiner (Princeton: Princeton University Press, 1966), pp. 145-147. 6°Allardt, Social Sources of Finnish Communism, p. 35. CHAPTER II SETTING AND METHODS Introduction The propositions developed in the first chapter are tested by using both survey and aggregate data on Chile during the period 1958-1964. Chile, at least during the period under study, presents an excellent setting for examining the relationships under study. Long cited as unique in Latin America because of its history Of democ- racy and its competitive multi—party system, it more recently has been considered noteworthy because of the 1970 electoral victory of the Marxist presidential candi- date, Salvador Allende. Thus, any empirical investigation Of the electoral support for the Chilean left is likely to be of interest to many. But whatever the popular interest in contemporary Chilean politics, the most compelling reason for the selection of this setting is that many of the factors that could complicate an empirical analysis of the theoretical relationships are Of minimal importance. First of all, strong and electorally-oriented leftist political parties, 33 34 Operating in a system of regular and free elections, existed throughout the period under study. Given the difficulty of accurately ascertaining support for illegal leftist movements, the absence of significant violent revolutionary activity during this period in Chile is also important. Furthermore, the cultural homogeneity of the Chilean population eliminates the problems that would be introduced if party preference followed racial, religious, or ethnic lines. From a practical pOint of View, the availability and relatively high reliability of survey and aggregate data on Chile also influenced the choice of research setting. In order to give the reader some familiarity with the research setting, a brief description of Chilean politics and society during the period under study is pre- sented. This is followed by a discussion of the data and methods employed in the study. The Chilean Political System The Chilean governmental institutions are specified in the 1925 Constitution, which provides for a unitary system with separation of powers. The key institutions are the executive and the bicameral legislature. Sub-national governmental institutions are not loci of important decision-making; provincial assemblies, called for in the 35 Constitution, have yet to meet, and municipal councils, although regularly elected, have minimal authority and responsibility. The dominant institution is the presidency, which possesses broad appointive, administrative, and legisla- tive powers.1 As chief executive, the president names his cabinet members, the governors of the various administra- tive divisions (the 25 provinces being the most important), the mayors of the large cities, and other top administra- tors. Vested with considerable administrative authority, he can issue a variety of executive decrees and orders, particularly important because much of the legislation provides for considerable executive discretion in imple- mentation. The president's role in the legislative process is also important; the ability to initiate legislation, the power to force Congress to immediately consider "urgent" legislation, the item veto, and special powers concerning financial legislation are all part of executive authority. Finally, special emergency powers may be granted to the president by the Congress. Presidential elections occur every six years.2 If no candidate receives an absolute majority of the votes cast, a situation that frequently occurs, the Congress selects from the top two candidates, with both houses sitting jointly and voting by secret ballot. Although 36 the Congress may legally select the runner-up, tradition specifies that the candidate receiving a plurality should be chosen, and there never has been an instance to the contrary. Prior to the 1964 elections there was specula- tion that if Salvador Allende achieved a plurality but lacked a majority, the Congress would pass him and select Eduardo Frei, but Frei publicly stated he would not accept the presidency under such circumstances. Similar specula- tion occurred in 1970 when Allende actually did receive a plurality but fell short of a majority. The Chilean Congress consists of a Senate and a Chamber of Deputies.3 Given the considerable role accorded the president in the rule-making process, the Congress must be considered the weaker partner. Nevertheless, the Chilean Congress is by no means a "rubber—stamp." The essentially negative power of the Congress has frequently been used to block or alter important proposed legislation, such as land- reform bills. Congressional elections are held every four years, with the entire Chamber of Deputies and about one-half of the Senate up for election each time.“ The d'Hondt system of proportional representation is used, with the voters determining the order of the party lists in each district by casting ballots for individual candidates on party lists rather than for the lists as a whole. Municipal elections 37 are also held every four years, and although municipal councils are not important policy-making institutions, the parties take the elections seriously because of the large number of Offices up for election and because municipal elections are carefully looked at as indicators of party strength.s Presidential, congressional, and municipal elections are all held separately, generating frequent electoral activity. All literate adult citizens are legally required to register and vote, but these laws are not rigidly en- forCed.6 In 1964 somewhat over 80% of the estimated eligible electorate was registered, and 87% of those registered actually participated in the presidential election that year. This represents a high rate of elec— toral participation for Chile. There was a large increase in voter registration from 1958 to 1964, due largely to electoral reforms, and the abstention rate Of those registered was particularly low in the 1964 election. Taking into account the illiteracy rate for adults, it can be estimated that about 40% of the entire adult popu- lation voted in the 1958 presidential election and about 60% did so in 1964. During the period under study, the Chilean multi- party system consisted of six major political parties, semi-organized into three of four blocs. While it is 38 always a simplification to classify parties along a left-right continuum, it is not particularly misleading to do so here, since this is a party system in which the major parties have a strong and stable ideological focus and in which party conflict is focused on the economic issue domain. In fact, there is no disagreement in the relevant literature over how the parties should be clas- sified for the period under study: the left includes the Communists and Socialists; the right consists Of the Liberals and Conservatives; and the Christian Democrats and Radicals comprise the center.7 The early 1950's were years of weakness and frag- mentation for the Chilean left, with the Communist Party outlawed and the Socialists seriously split. But by 1958 the left was strong and well organized, with a legalized Communist Party, a unified Socialist Party, and two minor leftist parties all banded together in the Popular Action Front (FRAP).8 The FRAP presidential candidate in 1958, Allende, nearly won the election. Since then, the Chilean left has been a major electoral force. The FRAP maintained its strength throughout the 1960's, receiving 39% of the vote in the 1964 presidential election, in which Allende again was the runner-up. Supported by a broader coalition of leftist organizations in 1970, Allende won a narrow victory to become president. 39 Both the Chilean Socialists and Communists claim a strong adherence to Marxist ideology, and although there has been ideological friction between the two parties, it is difficult to identify one as significantly more radical than the other.9 Both strongly supported the FRAP electoral program, which emphasized the need for thoroughgoing socio-economic reform. Specifically called for were: (a) nationalization of foreign owned companies, particularly the 0.8. copper companies; (b) socialization of key sectors of the economy, such as banks and public utilities; (c) direct government promotion of industrial- ization; (d) extensive agrarian reform; (e) a variety of income redistribution and social welfare measures; and (f) improvement of educational opportunities for the lower classes.1° The Liberal Party and the Conservative Party were the two important parties of the right during the period 1958-1964.11 Together they supported the successful presidential candidacy of Jorge Alessandri in 1958, and shortly afterward they allied with the Radical Party to form the Democratic Front, which controlled a majority of seats in both houses of the legislature during the Alessandri administration. Both the Conservatives and Liberals maintained a conservative stance on socio- economic issues, favoring measures to strength free 40 enterprise, encourage foreign investment, protect large landowners, and aid business and professional groups. The two center parties were more heterogeneous than the other major political parties. The Radical Party had a left wing in favor of broad social reform and a right wing very close to the Liberals and Conserva- tives, with this internal diversity resulting in a gener- ally vague program on socio-economic issues, considerable vacillation in policy and behavior, and a high turnover of leadership.12 During the period under study, the Radicals should be considered center-right, as they sup- ported the Alessandri administratiOn and were allied with the Liberals and Conservatives. The other center party, the Christian Democrats, rose rapidly from Obscurity in the mid-1950's to dominance in the mid-1960's.” Formed in 1957 when the small National Falange absorbed new elements, it Obtained only 9% of the vote in the congressional elections that year. But in the presidential election the following year the Christian Democratic candidate, Eduardo Frei, received 21% of the vote. The growth Of the Christian Democratic Party culminated several years later in Frei's 1964 presidential victory and the party's surprising success in the 1965 congressional elections, when it became the first party in the modern history of Chile to hold a majority of seats in the Chamber of Deputies. 41 Generally classified as center-left, the Christian Democrats were quite heterogeneous during this period, with a left wing that advocated a program of socio-economic reform almost as extensive as that of the FRAP.”’ The dominant faction, which Frei was identified with, was more moderate in its approach. Although the Christian Demo- cratic program, like that of the FRAP, stressed the need for economic development and social reform, there were important differences between the two programs. The Christian Democrats emphasized the need for measures to increase economic productivity and attempted to appeal across class lines with an ideology that rested on a social-Christian value system. The FRAP relied primarily on an appeal to the working class and, in comparison with the Christian Democrats, favored stronger economic redis- tribution measures and greater government intervention in the economy. A high degree of competitiveness characterized the Chilean political party system during the period under study. For example, in the 1958 presidential election there were four major candidates, all of whom received less than one-third of the vote.15 The candidates, along with their supporting parties and proportion of the vote received, were: (1) Alessandri, Conservative and Liberal Parties, 32%; (2) Allende, Socialist and Communist Parties, 42 29%; (4) Frei, Christian Democratic Party, 21%; (4) Bossay, Radical Party, 15%. The 1964 presidential election was much less competitive.16 Fearing a FRAP victory, the Conservatives and Liberals threw their support behind Frei. This re- sulted in Frei winning easily with 55% of the vote, while Allende finished a distant second with 39% (the Radical candidate received the remaining 5%). Such a one-sided contest was atypical. The more familiar form reappeared in the 1970 presidential election: the three major can- didates divided the vote quite evenly, each drawing between 37% and 27%. This high degree Of competitiveness did not exist only in presidential contests. Congressional and municipal elections were equally competitive. There were three such elections between 1958 and 1964, and in each case no party received more than 22% or less than 9% of the vote. Such patterns of party competition were not limited to the national level, but extended down to the local com- munities as well. Although each party had its electoral strongholds, an analysis of the Chilean communes reveals that most had a high degree of party competitiveness.17 Furthermore, within-commune competitiveness was substan- tially unrelated to socio-economic characteristics; the less-developed communes were almost as competitive as the 43 more-developed ones.18 Thus, the Chilean political party system was truly national in its scope. The Socio-economic Setting A brief discussion of socio-economic conditions during the period under study is also in order, as several key points need to be made. With a 1960 per-capita GNP Of about $400, Chile was among the more developed Latin American countries. Despite this favorable ranking, the Chilean economy suffered from serious problems. Rampant inflation and economic stagnation existed throughout the 1950's, key sectors of the economy were plagued by structural problems, and acute inequalities prevailed throughout the society. Chronic inflation has long characterized the Chilean economy, and the 1950's were no exception.19 Uncontrollable inflation, totaling almost 2,000% from 1950 to 1960, was coupled with minimal economic growth. This decade of economic stagnation and inflation hit the wage-earner the hardest; real wages actually declined during this period, while unemployment remained at rela- tively high levels. Quite possibly, the inflation of the 1950's contributed to the rise of the FRAP: Persistent inflation will arouse or strengthen demands for basic social and economic reforms; and a society that is unable to make the relatively small intergroup adjustments required to end inflation is likely to find itself faced 44 with strong and persuasive demands for much more fundamental social change.2° The economic problems of the 1950's did stimulate claims that basic structural changes in key areas of the economy were necessary for future economic growth. Min- ing, even though it employed only 5% of the labor force, was considered a key sector of the economy, as one- quarter of all governmental revenues and two-thirds of the value of all exports came directly from mining.21 In particular, foreign ownership of the major copper mining companies was a salient political issue.22 The FRAP con- sistently argued for the nationalization of the U.S. copper companies, while more moderate groups pushed for greater Chilean control over the operation of the compa- nies. . An equally important political issue concerned the need for change in the agricultural sector.23 Agri- culture employed over 30% of the economically active population in 1960, but it accounted for only 12% of the GNP. This low productivity of the agricultural sector was seen as an impediment to further economic growth and as a major cause of the inflation of the 1950's. Spe- cifically, the organization of agriculture and the pattern of land ownership were held to be inefficient and unre- sponsive to economic incentives. 45 By 1960 it was widely accepted that a re-organization of Chilean agriculture was economically necessary, but the question remained as to how agricul- ture should be reformed. The Chilean right favored governmental assistance for farmers and resisted land reform, pressing for minimum redistribution of land and maximum payment for expropriated property.”’ The Chris- tian Democrats advocated the expropriation of large and inefficient holdings, with repayment by a 10% down pay- ment plus the remainder in 25-year bonds carrying 5% interest.25 The FRAP went much further in proposing extensive redistribution of land, payment in the form of 30-year bonds bearing 4% interest, and the creation of a mixed state and private agricultural system.26 Persistent demands for land reform were not sur- prising in light of the high concentration of landholding. Over 80% of the total farm land was controlled by about 7% of the landowners, while 75% of the farms accounted for but 8% of the total farm land.27 Furthermore, only about one-half of the agricultural labor force were owners, the rest being wage-laborers, tenant farmers, or sharecrop- pers.28 In general, the system was one of a small number of vary large holdings, a large number of very small hold-t ings, and an increasing number of landless laborers. Inequalities in landholding were coupled with serious 46 serious socio-economic disparities, as the rural masses were discriminated against in terms Of minimum wages, collective bargaining possibilities, educational oppor- tunities, and governmental services. Although socio-economic inequalities in the rural areas were the more notorious, those in urban areas were also high. In part, this was due to the Chilean labor code, which covered all phases of labor relations. The labor code: (a) drew a legal distinction between the empleado and the Obrero--i.e., between the salaried white- collar employee and the wage-earning blue-collar worker; and (b) severely weakened the strength of organized labor with strict governmental regulations concerning trade unions.29 The distinction between white-collar and blue- collar workers is of considerable importance. Besides the social superiority of non-manual work over manual labor, there were important differences in remuneration and social welfare provisions. For example, during the period under study, the minimum wage for non-manual work was twice that for manual labor. This disparity in minimum wages resulted in sizable differences in actual wages. Peter Gregory reports data from a survey of Chilean firms to show occupational wage differentials for the early 1960's: differences between manual and 47 non-manual were great (the lowest-paid category of Office worker had an average wage 25% higher than the highest- paid skilled worker category), while differences within the manual class were relatively small (as compared to the unskilled, income levels for the semi—skilled averaged 25% more, and for the skilled, 50% more).30 The economic condition Of the Chilean working class reflected the weakness of organized labor.31 Only 12% of the labor force was effectively organized in 1960. Labor code regulations making it almost impossible for the agricultural labor force to organize were partially responsible for this low figure, but even in the non- agricultural sector less than one—fifty of the labor force belonged to a trade union. Furthermore, labor code regulations limited the power of trade unions (e.g., paid union Officials and strike funds were prohibited) and fragmented organized labor (industry—wide collective bar— gaining was sharply limited). Given these conditions, it is probably not surprising that the Communists and Social- ists were quite strong in the trade union movement. In sum, there were several key socio-economic issues that the parties divided on during the late 1950's and early 1960's. These issues generally related to either: (a) the measures necessary to achieve a healthy economy and promote economic development; or (b) the desirability of improving the relative socio-economic 48 position of the lower-strata groups. The Chilean FRAP differentiated itself from the other parties by its stress on the need for a fundamental transformation of the economic structure and a radical alteration in the condition Of the lower class. The Research Design Although the methodological details of this study are presented in the succeeding chapters, a brief outline of the research design may be useful at this point. Basically, this study utilizes both survey and aggregate data to test the relationships suggested in chapter one at both the individual and the ecological level. The survey data are from a 1958 sample survey Of the greater Santiago area.32 An area probability method was used to sample the adult population of the area, with the sample size being slightly over 800 respondents.33 Interviews were administered just prior to the 1958 presi- dential election, and those indicating an intention to vote in the election were reinterviewed shortly after the election. The fact that the survey was not a recent national one presents no real problem in this study, since the concern is with testing some theoretical propositions, rather than with describing the contemporary behavior of 49 the Chilean electorate. Of course, it is possible that the relationships found to exist in the greater Santiago area in 1958 do not hold for all of Chile in 1970, but the problem Of the generalizability of the findings remains regardless of the scope of the study. It would be equally difficult, if not more so, to generalize from a recent national survey study of voting in Chile to electoral behavior in other countries. This is not to say that the problem of whether the findings of this study can be generalized to other situations and settings will be ignored. I am merely pointing out that this sur- vey constitutes an adequate data base for testing the propositions; the sample is representative of a clearly definable and fairly broad population, and the 1958 presi- dential election in Chile is excellent for analyzing electoral support for leftist political parties. The survey analysis focuses on testing relation— ships specifying some determinants of voting behavior, particularly within the working class. The effects of relative economic deprivation and politicization on politi- cal preference are examined, as are the effects of some objective socio-economic characteristics on these two sub- jective factors. The aim is to establish some causal relationships between objective socio-economic variables and electoral support for the left, with individual 50 attitudes and dispositions considered as intervening variables. These micro-relationships provide a basis for interpreting the aggregate analysis. The aggregate data used in this study consist of socio-economic figures, from the 1952 and 1960 censuses, and electoral statistics, from the 1958 and 1964 presi- dential elections. These data are collected primarily from a variety of publications, but some of the socio- economic figures had to be calculated from a sample of the 1960 census interview schedules.31+ The data and the data sources are contained in the appendix. The aggregate data are used in an ecological analysis of the FRAP vote. Ecological analysis is fre— quently criticized because the correlations found at the ecological level may not exist at the individual level. However, the purpose of social science research is not primarily to specify individual—level correlations, but to determine causal relationships. In an ecological analysis, as in a survey analysis, this involves using the appropriate statistical analysis, including control- ling for potential extraneous variables. It is true that more caution is warranted in an ecological analysis, so considerable attention is paid in chapter four to the design and method of the ecological analysis. The major reason for using an ecological analysis in this study is to specify some macro-relationships 51 between certain aspects Of socio-economic development and electoral support for leftist political parties. Ideally, a time-series analysis of these relationships would be used, but this was not possible in this case because Of significant variations in important political factors.35 Instead, an analysis Of the relationships at the provincial level during a particular time period is used. Thus, longitudinal or developmental relationships will be inferred from a cross-sectional analysis, and this pro- cedure also requires some caution. Although an ecological analysis is well-suited for the investigation of macro-relationships, a prime weakness of any ecological analysis is that for a given pattern of aggregate relationships, a number of possible underlying processes are generally possible, and it usually is necessary to rely on survey data to determine the under- lying micro-relationships. Where survey and ecological analyses are combined, as they are in this study, the methods complement each other. The micro-relationships that produce the macro-relationships can be analyzed in the survey analysis, while the ecological analysis allows for an investigation of macro-relationships that could not be safely inferred from the survey analysis alone. By themselves, neither the survey analysis nor the eco- logical analysis would constitute an adequate empirical 52 test of the theoretical relationships examined in this study. 53 Notes for Chapter II 1On the presidency see Federico G. Gil, The Political System of Chile (Boston: Houghton Mifflin CO., 1966), pp. 92—106; and KalmanIL.Silvert, Chile: Yesterday and Today (New York: Holt, Rinehart, and Winston, 1965), pp. 91-93. 2For presidential elections see Chile: Election Factbook (Washington: Institute for the Comparative Study of Political Systems, 1963), pp. 32-36; and Gil, The Politi- cal System of Chile, pp. 226-233. 3Congress is discussed in Gil, The Political System of Chile, pp. 106-122; and Silvert, Chile: Yesterday and Today, pp. 93-95. l’For congressional elections see Chile: Election Factbook, pp. 8-11, 30-31, 42-44; Gil, The PoIitiCal System of Chile, pp. 215-226; and Charles J. Parrish, Arpad J. von Lazar, and Jorge Tapia Videla, The Chilean Congres- sional Election of March 7, 1965, Election Analysis Series, NO. 4 (Washington: Institute for the Comparative Study of Political Systems, 1967). sFor municipal elections see Chile: Election Factbook, pp. 8-11, 38-42; Michael Francis and Eldon Lannihg, "Chile's 1967 Municipal Elections," Inter-American Economic Affairs, XXI (Autumn, 1967),!pp. 23e3E; and Gil, The Political System of Chile, pp. 236—243. 6Electoral participation is disCussed in Chile: Election Factbook, pp. 13-15; and Gil, The PolitiCEl System of Chile, pp. 206-212. 7Literature on the political party system includes Ben G. Burnett, Political Groups in Chile, Latin American Monographs, No. 21 (Austin: University of Texas Press, 1970), pp. 160-232; Gil, The Political System of Chile, pp. 244-297; Ernst Halperin, Nationalism and Communism in Chile (Cambridge: The MIT Press, 1965), pp. 36-41; James Petras, Politics and Social Forces in Chilean Sevelppment (Berkeley: The University of California Press, 1969), pp. 73-255; Silvert, Chile: Yesterday and Today, pp. 98-109; and Peter Snow, WThe Political Party Spectrum in Chile," South Atlantic Quarterly, LXII (Autumn, 1963), PP. 474-487. 54 6On the Chilean left see Robert J. Alexander, Communism in Latin America (New Brunswick: Rutgers University Press, 1957), chpt. 10; Burnett, Political Groups in Chile, pp. 170-177, 188-195, 215-225; Gil, The Political System of Chile, pp. 276-294; Halperin, NatibnaliSm and CommuniSm in Chile, pp. 42-177; and Petras, PolitiCs and Socihl Forces in Chilean Development, pp. 158-196. 9For ideological friction within the FRAP during this period see Halperin, Nationalism and Communism in Chile, pp. 42- 177. < 1°Burnett, Political Groups in Chile, pp. 188-195; Gil, The Political System Of Chile, pp. 276-294. 11Onthe Chilean right see Burnett, Political Groups in Chile, pp. 161-170, 178—181, 196-204; Gil, The POlitical System of Chile, pp. 245-256; and Petras, Politics and Social Forces in Chilean Development, pp. 73-113. 12The Radical Party is discussed in Burnett, Political Groups in Chile, pp. 161-170, 181-183, 204-210; and Gil, The Political System of Chile, pp. 257-266. 13For the Christian Democratic Party see Burnett, Political Groups in Chile, pp. 177-178, 183—188, 210—215; Gil, The Political System of Chile, pp. 266-276; Halperin, NatiOnal- ism and Communism in Chile, pp. 178-205; and Petras, Politics and Social Forces in Chilean DevelOpment, pp. 197- 255. 1"Ideological friction within the Christian Democratic Party is discussed in George W. Grayson, Jr., "Chile's Christian Democratic Party: Power, Factions, and Ideology," Review of Politics, XXXI (April, 1969), pp. 147-171. 15For the 1958 election see Chile: Election Factbook, pp. 32-36. 16The 1964 election is discussed in Orville G. Cope, "The 1964 Presidential Election in Chile: The Politics of Change and Access," Inter-American Economic Affairs, XIX (Spring, 1966), pp. 3-29; and Federico G. Gil and Charles J. Parrish, The Chilean Presidential Election of September 4, 1964: Part I, Election Analysis Series, No. 3 TWashington: Institute for the Comparative Study of Political Systems, 1965). 17Arturo Valenzuela, "The Scope of the Chilean Party System," Comparative Politics, IV (January, 1972), pp. 179-199. 18 19 20 21 23 28 25 26 27 28 29 30 55 Ibid. Inflation is discussed in Albert 0. Hirschman, Journeys Toward Progress (New York: The Twentieth Century Fund, 1963), chpt. 3; and U.S. Department of Labor, Labor in Chile, Bureau Of Labor Statistics Report NO. 224 (I962), p. 21. Hirschman, Journeys Toward Progress, p. 216. For a discussion of mining see Gil, The Political System of Chile, pp. 156-161; and U.S. Department of Labor, Labor in Chile, pp. 5-7. Gil and Parrish, The Chilean Presidential Election of September 4, 1964, pp. 27—29. On agriculture in Chile see William C. Thiesenhusen, Chile's Experiments in Agrarian Reform, Land Economic Monographs, No. 1 (Madison: The University of Wisconsin Press, 1966), pp. 3-54. Robert R. Kaufman, The Chilean Political Right and Agrarian Reform (Washington: Institute for the Comparative Study of Political Systems, 1967). Gil and Parrish, The Chilean Presidential Election of September 4, 1964, pp. 28-30. ' Ibid. Solon L. Barraclough and Arthur L. Domike, "Agrarian Structure in Seven Latin American Countries," in Agrarian Problems and Peasant Movements, ed. by Rodolfo Stavenhagen (Garden City, New York: Doubleday & Co., 1970): P. 48. Thiesenhusen, Chile's Experiments in Agrarian Reform, pp. 10-260 For a discussion of the labor code and its effects see Robert J. Alexander, Labor Relations in Argentina, Brazil, and Chile (New York: McGraw Hill Book Co., 1962), part 3; and U.S. Department of Labor, Labor in Chile, pp. 16-50. Peter Gregory, Industrial Wages in Chile, Cornell Inter- national Industrial and Labor Relations Report, No. 8 (Ithaca: New York State School of Industrial and Labor Relations, 1967), pp. 85-101. 56 31On organized labor see Alexander, Labor Relations in Argentina, Brazil, and Chile, part 3; U.S. Department of Labor, Labor in ChiIe, pp. 35-50; and James 0. Morris, Elites, Intellectuals, and Concensus: A Study Of the Social QuestiOn and the Industrial Relations System in Chile, Cornell International Industrial and Labor Relations Reports, No. 7 (Ithaca: New York State School of Industrial and Labor Relations, 1966), pp. 9-19. 32These survey data were Obtained from the Political Science Data Archive, Department of Political Science, Michigan State University. The study was carried out under the direction of Eduardo Hamuy, University of Chile, and is available from the International Data Library and Reference Service, University of California at Berkeley, which originally provided the data to the Political Science Data Archive. 33The details of the sampling plan are as follows: Each district in every zone in the area was divided into seg- ments of about 30 dwellings each. A stratified random sample of 1000 dwelling units was selected; the selected units were visited and a list of all adults living in each selected unit was prepared. The person to be inter- viewed in each dwelling was randomly selected, and a total of 807 interviews were Obtained. A comparison of the sample with some available census data on the area indicates that females and middle—class members are some— what overrepresented. The proportion of the sample claiming to have voted for Allende is reasonably close to the percentage he actually received in the area, sug- gesting that the sample is fairly representative of the area population in terms of political behavior. More importantly, the sample does not appear to contain any systematic biases that would make specific sub-sample groups unrepresentative of the larger population groups that they correspond to. A computer tape containing a sample of coded interview schedules for the 1960 Chilean census was provided by the Latin American Data Bank, Department of Political Science, University of Florida. 35The nature of the Chilean left has fluctuated greatly over time. During the period under study here, the Com- munists and socialists formed a strong and electorally oriented political left. But in the early 1950's, as pointed out earlier in this chapter, the Chilean left was virtually nonexistent as an electoral force. This changing nature of the Chilean left makes a time-series analysis of how changes in socio-economic conditions have affected left voting an unwise research strategy. CHAPTER III SURVEY ANALYSIS Introduction The basic relationshipsix>be examined in this study have been presented in chapter one. Chapter two briefly outlined the research design and described some aspects Of Chilean politics and society in order to familiarize the reader with the research setting. In this chapter survey data are analyzed to empirically test some propositions relating the variables in this study to each other at the individual level. The data are from the 1958 sample survey described/in chapter two. The focus of the survey analysis is on establishing rela- tionships between some socio-economic characteristics, certain subjective factors, and electoral support for leftist parties within the lower class. The survey analysis is intended as a means of establishing necessary groundwork for the testing of macro-developmental rela- tionships in chapter four by providing a basis for the interpretation Of the ecological analysis. 57 58 Social Class and Political Preference As a starting point the relationship between socio-economic position and political preference is con— sidered. In this study political preference refers to whether or not the respondent supported the Chilean left --i.e., the FRAP parties and candidates. Two variables are used to measure this concept: (1) party preference and (2) candidate preference for the 1958 presidential election. For party preference respondents were classi- fied as left, non-left, or none.1 For candidate prefer- ence respondents were classified either as Allende sup- porters Or as supporters Of one of the three major candi- dates.2 The association between party preference and candidate preference is extremely strong. For respon- dents with a preference for the FRAP parties, 94% Of those with a candidate preference favored Allende. For respondents with a non-left party preference, 7% of those with a candidate preference favored Allende. It should also be pointed out that those with no expressed party preference do not differ from the sample as a whole in terms of the proportion favoring Allende, indicating that the "none" category for party preference does not particu- larly harbor leftists reluctant to openly admit their preference for Marxist parties to an interviewer. 59 Social class in this study refers to objective socio-economic position, with the major emphasis placed on the manual/non-manual distinction, since chapter two noted this as particularly important in Chile.3 Two indicators were used to categorize respondents: the occupation of the respondent (or the head of the respon- dent's family) and the interviewer's estimation of the respondent's economic position. These two measures were strongly associated, and the sample clustered into three groups, which I labeled upper—middle class, lower—middle class, and working class.“ This Objective classification Of respondents does correspond to their self—categorization quite well, except that there is a slight tendency for self-ratings to be higher than Objective placement.5 Thus, while social class has been defined and Operational— ized in objective terms, it appears that most Of the respondents have some subjective identification with the class I have placed them in. Table III-1 shows the social class composition Of supporters of the major political parties, while Table III-2 presents candidate preference by social class, and it is clear that the expected connection between social class and political preference exists. More spe- cifically, the two FRAP parties have a base of support that is predominantly working-class, while the other major parties all have a base of support that is primarily 60 “human owumuooEOo Chapmauno .muumm O>Hum>uwmsou u Dom “mouse HMHOQHA mum “women unflamfioom N an “abuse Hmowonm n Mm mm “auumm umflcofifioo u 00m Aemav lam V Aeoac Amoav AHNHV Ana a 1mm c n z woos eooa woos woos eoea woes woos shape as mm mm as am as me essence me om we be mm am am mappazuumsoq me mm mm ea we a a mappaznummep mzoz pom he ma one we uoa «mocmummmhm human mmmao defloom mozmmmmmmm semen mm mmeqo AeHoom HIHHH mqmdfi 61 .moflpasou 0» wow mooa Op annoy no: NOE mommucmoumm Ammmv Aaomv Aooav n z wooa mooa wooa «Hmuoe IMII IMMW. .IMII mommom ma em em floum em me mm flupcnmmmam me ma ma mocoaam mcflxuoz Odoowzlnwsoq OHCUHSIHOQQD mocOHOMOHm mumofiocmu mmmao Hmwoom mmdflv AdHUOm Mm muzmmmmmmm HB4QHQZdU NIHHH Wflmdfi 62 middle-class.6 Although the FRAP parties can be considered working-class parties in this sense, a majority of the working-class respOndents preferred a presidential candidate other than Allende. To use Robert Alford's terminology, the Chilean situation is one of high class distinctiveness of the left parties and low political distinctiveness of the working class.7 Given this, the focus Of the analysis is on identifying those segments of the working class most likely to pro— vide electoral support for the Chilean left. The relatively high level of class voting and high class distinctiveness of the FRAP should not be surprising. As chapter two explained, the Chilean multi— party system is composed Of parties that aim their appeals at fairly distinct clienteles. Party distinctive- ness is probably reinforced by organizational and com- munication patterns, as economic organizations and the media are frequently connected with a specific party. The Communists and Socialists aim their appeal at the working and lower classes, are strongly involved in the trade union movement, and publish several newspapers and magazines. The class distinctiveness of the FRAP appeal is reflected in the survey respondents' perceptions: three-fourths felt Allende would obtain a majority of the‘ workers' vote and two-thirds saw him winning a majority 63 Of the votes of the poor, whereas only one in seven believed Allende would collect a majority of the middle- class vote and one in six claimed he would receive a majority of the votes of white-collar employees. Subjective Factors and Political Preference Chapter one hypothesized that the process of socio-economic change would increase potential support for leftist political movements by increasing both the level of relative economic deprivation and the extent of politicization in the lower strata. As defined in chap- ter one, relative economic deprivation refers to perceived or subjective economic deprivation, not to Objective economic condition, while politicization refers to politi- cal awareness and involvement. I argue that within the lower strata those indi- viduals with higher levels of relative economic depriva- tion will be more likely to express a preference for leftist parties. Since parties of the left, and par- ticularly socialist ones, propose policies more-or-less explicitly aimed at improving the economic condition of the lower classes, those segments of the lower classes most concerned about economic improvement should be most receptive to such appeals. 64 I also argue that greater politicization increases the likelihood of lower—strata individuals voting for the left. Higher politicization makes lower-strata members more conscious Of the appropriateness of using political means to relieve their economic deprivation and thus more likely to respond to deprivation in political terms. Lower-class individuals with higher levels of politiciza- tion are also more likely to perceive political parties as representing certain class interests and thus more likely to vote for the parties that most attempt to appeal to their class.8 From the above, it is expected that within the Chilean working class: (a) the higher the level of relative economic depri- vation, the more likely there will be support for the FRAP. (b) the higher the level of politicization, the more likely there will be support for the FRAP. Four variables were used to measure the politici— zation of respondents: (a) whether or not the respondent 9 was a registered voter; (b) how frequently the respondent 0 talked about politics;1 (c) how interested the respondent 1 and (d) whether or not the respon- was in the election;1 dent participated in some way in the election campaign.12 These four variables were so highly related that it would have been impossible to separate out the influence of each or to consider different dimensions of the phenomenon, so 65 a summated index was constructed (which I termed the politicization index), and respondents were grouped into four categories of roughly equal size: high, medium, fair, and low politicization.13 Conceptually, this variable is extremely similar to Lester Milbrath's idea of political involvement, which is specified as ranging along an active-inactive dimension, and to Daniel Goldrich's con- cept of politicization, which ranges from a lack of per- ceived relevance of government to active involvement in politics.‘“ Relative economic deprivation proved somewhat more difficult to Operationalize, as the survey did not contain extensive data on what is a rather complex psycho- logical disposition. Chapter one presented Ted R. Gurr's definition of relative deprivation as a perceived dis— crepancy between value expectations and value capabilities. The intensity of relative deprivation is specified by Gurr as a function of the degree of perCeived discrepancy and of the "salience of the value class with respect to which discrepancy is experienced."15 Thus, for relative eco- nomic deprivation: (a) the greater the perceived dis- crepancy between the respondent's economic expectations and his perceived capabilities, the more intense the respondent's deprivation; and (b) the more intense the respondent's desire to actually fulfill his economic 66 expectations, the more intense the respondent's deprivation for any given level of perceived discrepancy. In an attempt to come as close to the above con- ceptualization as possible, responses to two questions were used to Operationalize relative economic deprivation: (a) what the respondent claimed his major goal in life was; and (b) how difficult the respondent claimed it would be to achieve that goal. Respondents identifying improvement of their economic condition as their major goal in life and also claiming that it would not be easy to achieve this goal were classified as "high" in relative economic depri- vation; the remainder were classified as "low" in relative economic deprivation, except for a small number who could not be categorized due to incomplete information.16 Admittedly, this simple dichotomized variable is not a highly precise measure of relative economic depri- vation. However, the adequacy of this measure is sug- gested by its connection with several other factors. Those respondents who claimed to be dissatisfied with their present occupational situation (or who claimed that their family head was dissatisfied) were considerably higher in relative economic deprivation.l7 Also, those high in relative economic deprivation were more likely to claim that they had not enjoyed the opportunities they desired in their life.18 Turning to Objective factors, 67 respondents who had some recent experience with unemployment were more likely to be high in relative economic deprivation.19 Finally, there is an association between social class and relative economic deprivation among the survey respondents.z° All of this suggests a reasonable degree of validity for the Operationalization of relative economic deprivation used here. Table III—3 presents party preference by level of politicization and relative economic deprivation for working-class respondents. Politicization is simply dichotomized (high and medium versus fair and low) for ease of analysis. The strong connection between politici— zation and party preference supports the previous con— tention that within the working class the more politicized will be more likely to support the Chilean left.21 When candidate preference is used as the dependent variable similar results are Obtained, although the strength of the relationship is not as great.22 Thus, there clearly appears to be a connection between politicization and political preference within the working class. Possibly this does not represent a causal or asymmetrical relation- ship but is simply a spurious association. Several potential extraneous variables may be suggested, and it is necessary to briefly consider these possibilities. 68 A.u:moncoo oams OHQOHHM> unmocommocfi “capo Ono sues manneun> unmocmmmo may one mannanm> unmooommocfl Ono consume mwachHunHOH on» mono Iwocw hon» acowumuwmo mcwaoom m an omoflmuno Ohm moaumflunum Ommnav mm. H Teena “ca. v a .Hm.m u A.m.o my mumoqmlwno ":OHum>wumOo Owfiocoom O>flumamu mo uomwmm unmocmmmocfl Mom mm. H mfifimm «Ho. v m .ve.~a u A.m.o NV mnmoonlfiso "cowunuflowuflaom mo uommmo unmocmmmosfi you n OHOZ .annu was» EOHM ooooaoxm one mouse Oflmwoomm man How OOGOHOMOHQ n mcwmmwumxm poo mucoocommmm .mmflunmm ummarcoc esp mo moo Mom OOCOHOHONQ manna m O>mn wuommumnu mmv mcwcfimaou Ono ammwuumm puma on» NO OCO How OOGOHOMOHQ mouse m O>mn wmm anommpmo umnwm mop cw .OHQEOxO Mom .Oocwummmum women m £HH3 whoomumo use» me mucoocommwu mo Hones: Hmuou may no ommmo we ommucmouom mesa « Ammo Anqv Amvv Acme n ma be mm mm apnea m 30a Enema: 30A EOAoOz HO “Hem no seam HO mem no seam cowumNHOHuwaom OOCOHOHOHA mouse 33 seem cowum>wumoo Owaoooom O>HumHQm mBZMQZOmmmm mmflAUIGZHMMO3 mOm ZOH84>Hmme UHEOZOUfl W>HB€AMM 02¢ ZOHBflNHUHBHAOm Nm mUmemmmmm NBMdm mIHHH mqmdfi 69 One variable that very possibly could produce a spurious relationship between politicization and political preference is the sex Of the respondent, since females are both less politically aware and more conservative in political choice than males.23 Of course, it may be that the lower level of politicization among females simply accounts for their reluctance to support the FRAP. In fact, when the relationship between politicization and party preference within the working class is re-examined controlling for sex, the result is not only that the rela— tionship persists but also that working-class females are no less likely to support the FRAP than their male counter- parts when the level of politicization is held constant. The possibility also exists that organizational membership produces a spurious relationship between politicization and political preference. Many of the organizations the working-class members belong to (such as trade unions) are connected with the FRAP, and organi- zational membership might increase political awareness on the one hand and stimulate radicalism on the other.% But if we look at the relationship between politicization and party preference within the working class while con- trolling for organizational membership, we still find the strong association between these two variables.25 Routine controls for other factors, such as age, religion, and length of residency in the Santiago area, 70 were also run, and in no case does the relationship between politicization and party preference diminish.26 Although it is conceivable that there is some other unconsidered explanation for the connection between politicization and political preference within the work- ing class, it does appear that these data strongly sup- port the conclusion that there is a causal or asymmetrical relationship. Table III—3 excludes respondents with "none" as their party preference. What this table shows is that for those working-class respondents with a party pref- erence, the higher the level of politicization the more likely that preference will be for the FRAP. It is also true that the higher the level Of politicization, the more likely it is that respondents will actually have a pref— erence for some political party. Therefore, increasing the level of politicization for the working class would increase preference for the FRAP in two ways: (1) it would increase the proportion of the working class with a party preference, and since working-class members are more likely to support the FRAP than middle-class mem- bers, this would tend to benefit the FRAP; (2) it would increase the proportion of working-class party preferers that favored the FRAP. In terms of actual voting behavior, increasing the level of politicization for the working class would tend to increase both the proportion of 71 working-class members actually voting and the proportion of working-class voters casting their ballot for the FRAP.27 Thus, working-class politicization levels, both absolute and relative to those for other social classes, have a very important impact on electoral support for the FRAP. Table III-3 also presents the data for the rela- tionship between relative economic deprivation and party preference for working-class respondents. There is a moderate association between these two variables, but it is not statistically significant at the .05 level.2° Similar results are obtained when candidate preference is used as the dependent variable. These results are diffi- cult to interpret. Although the lack of significance at the .05 level might indicate that the association Obtained was simply due to chance, the strength of the association (gamma = .29) and the level of significance obtained (.09) suggest that rejecting the proposition may be committing a serious Type II error. I would argue that it is legitimate to relax the acceptable level of statistical significance from the con- ventional .05 up to .10 in this case. As was pointed out previously, there is probably a considerable amount of random error in the measure of relative economic depriva- tion, and such error would tend to deflate the strength and 72 statistical significance of the relationship. Given the number of respondents involved and the probable measure- ment error present, it does not seem unreasonable to cautiously accept findings at the .10 level of signifi— cance. Of course, this depends on the relative importance placed on avoiding a Type I or a Type II error. When testing propositions, the primary consideration should be to avoid a Type I error, but some attention must be devoted to avoiding a Type II error. Where the power of the sta- tistical test is low, as is the case here, the .10 level Of significance may be more reasonable than the conven- tional .05; if there is no real association between the variables, the likelihood of Obtaining a significant asso— ciation by chance alone is still fairly low under this more relaxed decision rule. While any interpretation of these results must be hihgly tentative, the best inference appears to be that there is a relationship between relative economic depri- vation and support for the FRAP among working-class respondents. If the association is accepted as not being accidental, strong evidence can be presented to support the conclusion that it is not spurious. No matter what controls are run--sex, age, religion, length of residence, occupational skill level--the assOciation between relative economic deprivation and party preference persists.29 73 Thus, the conclusion is that relative economic deprivation affects support for the FRAP for working-class members. It may not be surprising to find a relationship between relative economic deprivation and support for leftist parties within the working class, as chapter one showed this to be a common contention in the literature. But in fact this relationship has rarely been subjected to an adequate empirical test; more commonly, it is simply offered as an explanation of some empirical associations between objective factors and left voting. The possibility that greater politicization increases support for leftist parties within the working class has received very little consideration in the relevant literature, so the finding of a strong relationship here is quite interesting. The importance of this relationship will be clearer later in this study when the impact of socio-economic factors on electoral support for the left is examined. The Impact of Objective Factors Having established the effects of politicization and relative economic deprivation on support for leftist political parties, it is now necessary to examine the impact of certain socio-economic characteristics on these two variables. Chapter one paid particular attention to the potential impact of social mobilization on politiciza- tion and relative economic deprivation. Social 74 mobilization, as defined in chapter one, denotes a set of highly interrelated socio-economic factors, with education, organizational membership, mass media exposure, urbaniza- tion, and industrialization specified as particularly important. It was suggested that each Of these components of social mobilization affects both politicization and relative economic deprivation. In this section the survey data are analyzed to determine if causal or asymmetrical relationships exist at the individual level between three of the social mobilization variables (education, organi— zational membership, and mass media participation) and the two subjective factors previously examined (politicization and relative economic deprivation). Several empirical studies have investigated the individual-level effects of these social mobilization variables on factors similar or highly related to politici- zation. Gabriel Almond and Sidney Verba, in a cross— national study of the political culture in five democra— cies, emphasize the strong effeCts of education and organizational membership on political participation and political competence.3° A more thorough analysis of the Almond and Verba data by Norman Nie, G. Bingham Powell, Jr., and Kenneth Prewitt concludes that organizational membership and social status (which includes education) have a strong impact on political participation and that 75 urbanization, in and of itself, does not have any significant effect on political participation.31 Alex Inkeles' cross-national survey study of six developing countries (including Chile) finds that education, mass media exposure, and industrial employment all affect "participant citizenship" (a concept including interest in, information about, and participation in pOlitics) and that urban residence has no such effect when these other factors are held constant.32 These studies are the most relevant because they investigate the impact of several variables in a variety Of different countries; similar findings are reported in many studies having a narrower focus.33 From the above, it is expected that for working— class respondents: (a) education has a positive effect on politicization; (b) organizational membership has a positive effect on politicization; (c) mass media participation has a positive effect on politicization. Since this analysis fOCuses specifically on the Chilean working class, it would also be desirable to examine the impact of industrial employment to see if results similar to Inkeles' would be obtained. Unfortunately, the neces- sary data were not collected in the survey. Also, the effects of urban residence cannot be investigated because 76 the sample is entirely from the Santiago area. However, the previously cited studies indicate that urban resi- dence has no effect on politicization, except insofar as it affects organizational involvement, mass media expo- sure, etc., which are examined here. As chapter one showed, there is considerable literature suggesting that the process of social mobili- zation tends to heighten economic aspirations and thus contribute to relative deprivation. Urbanization, the expansion of literacy and education, the spread of mass media, and the general process of social change itself, it is argued, are all capable of raising economic expecta— tions far beyond the capacity of the society to fulfill them-~thus transforming the revolution of rising expecta- tions into a revolution of rising frustrations. Sur— prisingly, there has been very limited empirical investi— gation into this phenomenon, and what little research does exist is contradictory.3“ On the basis Of the generali- zations of much of the literature, it is expected that within the Chilean working class: (a) education has a positive effect on relative economic deprivation; (b) organizational membership has a positive effect on relative economic deprivation; (c) mass media participation has a positive effect on relative economic deprivation. 77 While it would also be desirable to examine the effect of urbanization and industrialization, this cannot be done for previously mentioned reasons. Education was measured by the number Of years of formal education the respondent claimed he had.35 Organi- zational membership was Operationalized in terms of whether or not the respondent claimed he belonged to some organization.36 Mass media participation was measured by an index that combined two variables: newspaper reading and radio listening.37 Data on the relationships between the social mobilization variables and politicization are presented in Tables III-4, III-5, and III-6. Table III-4 shows the connection between education and politicization for working-class respondents; as expected, the more educated working-class respondents are higher in politicization. Similarly, Table III-5 clearly indicates that the working- class respondents who belong to some organization are higher in politicization than those who do not. Table III-6 contains data on the relationship between media participation and politicization while controlling for social class. Since data on media participation was available for only a portion of the respondents, the relationship was examined within each social class in order to have a sufficient number of respondents in the 78 TABLE III-4 POLITICIZATION BY EDUCATION FOR WORKING-CLASS RESPONDENTS Education Politicization 6 Years 4-6 Years 0-3 Years % High or Medium 78 46 33 % Fair or Low 22 54 67 Total 100% 100% 100% N = ( 37) (137) (120) .001; gamma = .43 Chi-square (2 d.f.) = 24.0, p < TABLE III-5 POLITICIZATION BY ORGANIZATIONAL MEMBERSHIP FOR WORKING-CLASS RESPONDENTS Organizational Membership Politicization Yes No % High or Medium 62 38 % Fair or Low 38 62 Total 100% 100% N = ( 86) (210) = 14.2, p < .001; gamma = .45 Chi-square (l d.f.) 79 .AMMMHO Hnfloomv Ofinnfium> Houucoo can no muommuwo some Eonm mowumaumum on» ocflaoom an oocflmpoo Ohm Mosumaumum Onona at .mcwocoon on moo wooa on Hmuou uoc use mommuoooumm « u eemssma umo. v m .me.~a u A.m.p we «emnauemuaso Amm V Ape V Amm v Aooav Amm V Amm v n z wooa mooH wooa wooa mooa wooa «Hmuoa IMMI ma mm ma Hm ma Hash w me em we mm mm me Enema: w am am mm mm om me team a son swam 309 seem 30A seam :OflummHOfluHmm meow: :Oflumnwowpwaom mcwxuo3 OHooHZIHO3OA OHoowzlummmD mmmao Hmwoom mmdAU AflHUOm 02¢ ZOHBGmHUHBm¢m dHsz Mm ZOHEdNHUHBHAOm mIHHH mflmdfi 80 analysis.38 These data show the association between media participation and politicization to be moderately strong and statistically significant; if just working- class respondents are looked at, the connection between media participation and politicization still exists. Routine controls--age, sex, and so on--were introduced and the associations between the social mobilization variables and politicization persisted. In sum, the results here are consistent with the findings of other empirical studies previously cited. Of course, the primary aim here has been to examine the impact of the social mobili- zation variables on politicization specifically within the working class, and it is clear that these variables do have a strong influence on politicization levels for working-class members.39 So far, causal or asymmetrical relationships have been established, for the working—class respondents, between: (a) the social mobilization variables and politicization; and (b) politicization and support for the Chilean FRAP. However, the direction of causality or influence may be unclear in some cases. In order to more firmly establish that the chain of causality runs from the social mobilization variables, through politici- zation, to support for the FRAP, an additional step in the analysis was taken. The association between the social 81 mobilization variables and party preference was examined while controlling for politicization. When the level of politicization is held constant, there is absolutely no association between education or media participation and party preference, although some connection between organi- zational membership and party preference does remain. This indicates that politicization is an intervening variable between the social mobilization variables and party preference.“° In other words, the impact of the social mobilization variables on support for the Chilean left within the working class is almost entirely through the impact these variables have on politicization levels. These data do suggest that some types of organiza- tional membership have a direct influence on political orientations for working-class members, since some asso— ciation between organizational membership and party pref- erence remains when politicization is held constant. Specifically, membership in a trade union may have a direct effect on the political preference of workers. A sample survey of heads of local blue-collar trade unions in three large Chilean cities showed that one-half of the union leaders favored the FRAP, so the possibility of union influence seems well-founded."l In order to further examine this possibility, I compared working-class re- spondents who belonged, or whose family head belonged, to 82 a trade union to those working-class respondents without any trade union connection. The unionized working-class members were more left in their party preference than the non-unionized even when the level of politicization is held constant. While this supports the above conclu- sion, it should be realized that the number of respondents involved here is small, and thus any inference must be highly tentative. It was also hypothesized that higher levels of the social mobilization variables would produce higher levels of relative economic deprivation for working-class members, thus influencing support for the Chilean left. These data do not Support that contention. Tables III-7, III-8, and III-9 show that virtually no association exists between any of the social mobilization variables and relative economic deprivation. Only in Table III—7 does the proportion of respondents who are high in relative economic deprivation vary across the categories of the independent variable, and here the results are not in the predicted direction. Working-class respondents with more than six years of education tend to be lower in relative economic deprivation than those with six or fewer years of education, whereas it was predicted that they would be higher. However, only a small number of working-class respondents have more than six years of education, so even 83 TABLE III-7 RELATIVE ECONOMIC DEPRIVATION BY EDUCATION FOR WORKING-CLASS RESPONDENTS Relative Economic Education Deprivation 6 Years 4-6 Years 0-3 Years % High 45 67 64 N = (35) (132) (115) TABLE III-8 RELATIVE ECONOMIC DEPRIVATION BY ORGANIZATIONAL MEMBERSHIP FOR WORKING-CLASS RESPONDENTS Relative Economic Organizational Membership Deprivation Yes No N = (86) (198) TABLE III-9 RELATIVE ECONOMIC DEPRIVATION BY MEDIA PARTICIPATION FOR WORKING-CLASS RESPONDENTS Media Participation Relative Economic Deprivation High Low % High 60 61 N = (47) (54) 84 this finding is not statistically significant. The expected associations simply do not emerge in these tables. On the possibility that some suppressor vari- able was masking the true relationship between these variables, I controlled for a variety of factors-—age, sex, religion, occupational skill level, and length of residency. In no case does a significant relationship emerge between relative economic deprivation and any one of the three social mobilization variables. Clearly, education, organizational membership, and media partici- pation do not have any individual-level effects on rela- tive economic deprivation. One might argue that social mobilization does heighten expectations but that the more socially mobilized workers are also better-off in their economic condition, and their higher expectations are therefore balanced by greater achievement and do not result in higher relative economic deprivation. But again these data indicate that the social mobilization variables do not have any syste- matic or consistent effect on economic expectations: when economic condition is held constant (by controlling for occupational skill level), there is no connection between relative economic deprivation and the social mobilization variables for the working-class respondents. It is also true that the skilled workers do not differ 85 from the unskilled in terms of their level of relative economic deprivation. This is not surprising, though, as chapter one pointed out that there frequently is little or no connection between Objective economic condition and relative economic deprivation. It would be hasty to conclude on the basis of the findings of this one study that the process of social mobilization is without importance for relative economic deprivation. Given the contradictory findings of the extremely limited empirical work in this area, the safest conclusion is probably that effects may occur under certain circumstances. It is clear, nevertheless, that much of the literature is far too simplistic in specifying a general relationship between social mobilization and relative deprivation. Conclusions These survey data support the conclusion that both politicization and relative economic deprivation affect working-class support for the Chilean FRAP and that social mobilization affects politicization but not relative economic deprivation. However, the data are from a sample survey of the greater Santiago area, so there may be some question about whether these relationships hold true for the country as a whole. The absence of a national sample 86 survey makes it impossible to give a definitive answer to this question, but some points may be discussed. It was found that the FRAP support in 1958 came predominantly from the working class in the analysis of the Santiago survey. Sufficient evidence exists to war- rant concluding that this is the case nationally. James Petras and Maurice Zeitlin have carried out fairly care— ful ecological analyses of the FRAP vote in both the 1958 and 1965 elections in order to determine the occupational base of support for the Chilean left.“2 They conclude that in urban areas Allende's support came largely from the working class and in rural areas it came primarily from agricultural wage-laborers. They also point out that the FRAP, for both theoretical and practical reasons, focused its appeals and organizational efforts on the urban and rural proletariat. Appropriate survey data do not exist to conclu— sively show that the relationships between the two sub- jective factors, politicization and relative economic deprivation, and electoral support for the FRAP within the working class hold true nationally. Inkeles' survey study does note a tendency for those who claim the economy and society need a total and immediate change to have higher scores on his participant citizenship scale."3 This is consistent with my finding of a relationship 87 between politicization and support for the FRAP, but unfortunately Inkeles does not relate these variables to party preference. Alejandro Portes finds a relationship between occupational dissatisfaction and his index of leftist radicalism, which includes party preference as one of the three components.”' While this is consistent with my finding of a relationship between relative economic deprivation and support for the FRAP, Portes' survey data are also limited to the Santiago area. Furthermore, Portes does not bother to control for social class, leav- ing himself open to the charge that the association is spurious. Thus, the results of other analyses of Chilean survey data do not contradict the findings of this chapter, although they do not necessarily indicate that the findings hold true nationally. There should be little question about the general influence of the social mobilization variables on politici— zation levels. As was pointed out earlier, these relation- ships have consistently been confirmed in empirical cross- national studies, one of which included Chile among the countries studied. Education, organizational membership, and media participation should be consistently related to politicization levels throughout Chile. Somewhat surprising is the finding that the social _mobilization variables do not affect relative economic 88 deprivation, as this contradicts much of the literature on development. However, the connection between social mobilization and relative deprivation has usually been simply inferred from observed connections between high rates of social mobilization and high levels of political violence or radicalism at the societal level. The find- ings of this chapter suggest that the effects of social mobilization on political behavior are primarily through its effects on politicization, not relative deprivation. Unfortunately, there has been very little direct empirical investigation Of the possible effects Of social mobiliza- tion on relative economic deprivation, and it is not possible to cite other empirical work to confirm the idea that there is no systematic effect. But in the absence of any positive evidence for a connection between these two factors, the results Of this analysis shall be assumed to hold true across Chile, not just in the Santiago area. In sum, this Chapter has provided a basis for the interpretation of the aggregate analysis in the following chapter by (a) determining the occupation base of the FRAP vote; (b) specifying certain subjective factors that affect working-class support for the Chilean left; and (c) examin— ing the impact of certain objective variables on these subjective factors. 89 Notes for Chapter III 1Those expressing a preference for the Communist or Socialist parties were classified as "left." Supporters of the Christian Democrats, Radicals, Liberals, or Con- servatives were classifed as "non-left." Those indicating that they had no party preference were classified as "none." The NA category includes a few respondents who expressed a preference for some minor party. The number of respondents falling into each category is as follows: left, 110; non-left, 421; none, 184; NA, 92. 2The other three major candidates were: Alessandri (Con- servatives & Liberals); Frei (Christian Democrats); Bossay (Radicals). The NA category includes respondents that expressed a preference for some other candidate or claimed they had no candidate preference. The distribution of respondents for this variable is: Allende, 169; other three candidates, 473; NA, 165. 3Robert Alford, Partyfand Society (Chicago: Rand McNally, 1963), pp. 73-79, uses a similar conceptualization of social class. Alford relies on Max Weber's definition-- see Hans Gerth and C. Wright Mills (eds.), From Max Weber: Essa s in Sociolo (New York: Oxford University Press, I958}, p. I81. For other conceptualizations of social class see: Reinhardt Bendix and Seymour Martin Lipset (eds.), Class, Status, and Power (Glencoe: The Free Press, 1953), pp. 17-128; RiChard Centers, The Psychology of Social Classes (PrinCeton: Princeton University Press, 1949), pp. 12—29; Ralf Dahrendorf, Class and Class Conflict in Industrial Society (Stanford: Stanford University Press, 19597, pp. 3-35. I'The categories for occupation were: unskilled laborer, skilled laborer, independent or service worker, non— managerial employee, manager, businessman, professional. The categories for the interviewer's estimation of the respondent's economic condition were: very good, good, fair, poor, very poor. Those classified as "working- class" were primarily skilled and unskilled workers who were fair, poor, or very poor in economic condition, but included were those in service, independent, or non- managerial work if they were poor or very poor in economic condition. The "lower-middle-class" group was comprised almost exclusively of those in independent or non- management work and also good or fair in economic condi- tion. The "upper-middle-class" group consists primarily 90 Of managers, businessmen, and professionals who were also very good or good in economic condition, but also includes some in independent or non-management work if they were very good in economic condition. 5Respondents were asked whether they were upper—class, upper-middle-class, lower-middle-class, working-class, or lower-class. I classified 296 respondents as working- class, but only 260 self-categorized themselves as working- Or lower-class. A cross-tabulation reveals that 75% of those I classified as working-class indicated that they belonged to the working or lower class. 6It must be remembered that this is not a national survey, but simply a sample of the Santiago area, which includes about one-third Of the national population. It would, of course, be unwise to directly generalize from this sample to the Chilean electorate, but the general tendencies indicated by Tables 111-1 and III-2 are fairly accurate. This will be discussed in more detail later. 7Alford, Party and Society, pp. 79-86. 8The claim here is simply that the more politicized members of the working class are more likely to perceive social class as an important political cleavage and more likely to perceive leftist political parties as best representing their class interests. It is not argued that more so- phisticated views of class conflict and politics are neces— sarily present or that Marxist ideology is necessarily internalized. 9For those not registered, the reason for this was asked. Most admitted that it was simply a lack of interest, not legal barriers, that prevented them from registering, although some were not legally eligible because they were an illiterate, a foreigner, or a recent migrant to the area. For the most part, this variable can be considered a measure Of the individual's interest in and desire to participate in politics. The frequency of talking about politics was inferred from the respondent's indication of whom he talked about politics with. Those Classified as "high" are those indicating they talked about politics with friends, neighbors, co—workers, etc. Those categorized as "medium" talked about politics with their spouse or relatives only. Those classified as "low" talked with no one. 11 12 13 11b 15 16 91 This was a closed item with the possible responses being high, medium, and low. Campaign participation was defined broadly to include such things as attending a political rally, listening to a speech, or attempting to convince others how to vote. The variable was simply dichotomized so that respondents were classified as either participants or non-participants. This index was a simple summated index in which each of the four component variables had equal weight. For voter registration and campaign participation the "yes" response was given a score of 3 points and the "no" response a score of 1 point. For campaign interest and frequency of talking about politics, the "high" response was given a score of 3 points, the "medium" response received 2 points, and the "low" response 1 point. For each respondent the . scores for all four component variables were totaled. Those with 8 or more total points were classified as "high" on the politicization index. Those with 6 or 7 points were classified as "medium." Those with 4 or 5 points were classified as "fair," and those with 3 or fewer points were classified as "low." The distribution of respondents for the politicization index is: high, 188; medium, 209; fair, 224; low, 186. The gammas expressing the association with the total index for each of the four component vari- ables are: voter registration, .89; campaign interest, .78; campaign participation, .93; frequency of talking about politics, .70. Since all the gammas are equal to or greater than .70, there is ample justification for con- sidering these four component variables together in a single cummated index. . Lester Milbrath, Political Participation (Chicago: Rand McNally, 1965), p. 18; Daniel Goldrich, "Toward the Comparative Study of Politicization in Latin America,“ in Contemporary Cultures and Societies of Latin America, ed. By Dwight E. Heath and Richard N. Adams TNew YorE: Random House, 1965), p. 361. Ted Robert Gurr, Why Men Rebel (Princeton: Princeton University Press, 1970), pp. 60-71. For major goal in life, the following were cOnsidered as "improvement of economic condition" responses: to improve economically; not to lack a job; to ensure the financial future of one's family; to own a home. The responses not considered as "improvement of economic condition" included: to be a good professional; to live 17 18 19 92 tranquilly; to educate one's children; to have good health; other. The distribution for this variable was: major goal in life the improvement of one's economic condition, 452 respondents; major goal in life something else, 328 respondents. Respondents were also asked whether it would be easy, fair, or difficult to have the opportunities to reach their major goal in life. Of those respondents with improvement of their economic condition as their major goal, 397 claimed it would be "fair" or "difficult" to have the opportunities to reach their goal; these respondents were classified as "high" in relative economic deprivation. All other respondents were classi- fied as “low" (369 respondents) or as "NA" when incomplete information existed (41 respondents). Respondents who were employed were asked if they were satisfied or dissatisfied with their occupational situa- tion. Respondents who were not the head of their family were asked (when the family head was employed) if the family head was satisfied or dissatisfied with his occu- pational situation. Those respondents indicating either that they were dissatisfied or that their family head was dissatisfied were classified as “high" in occupational dissatisfaction (162 respondents). The rest were classi- fied as "low" (495 respondents) or as "NA" if incomplete information existed (150 respondents). The gamma for the association of this variable with relative economic depri- vation is .38. Respondents were asked if they felt they had the Oppor- tunities they desired in their life. This was a closed item, with the possible response being either yes or no. The gamma for the association of this variable with relative economic deprivation is .30. Classified as having recent unemployment experience were those respondents who (a) were unemployed; (b) were employed but had been unemployed at least once in the past five years; or (c) had an unemployed head of the family. It would have been desirable to also include those respon- dents whose family head was employed but had been unem- ployed in the recent past, but this information was not Obtained. The distribution of this variable is: had recent unemployment experience, 158 respondents; no recent unemployment experience, 625; NA, 24. The gamma for the association between this variable and relative economic deprivation is .13. 20 21 23 2k 25 26 27 93 The gamma for this relationship is .35, with working- class respondents highest in relative economic deprivation and upper-middle-class respondents lowest in relative economic deprivation. There is some tendency for these two variables to be associated for middle-class respondents as well, although this finding should not be taken too seriously given the small number of FRAP supporters in the middle class. ' The gamma for the relationship between politicization and candidate preference is .22. Men and women vote separately in Chile and electoral statistics are calculated for each sex, so the voting behavior of men and women can easily be compared. In 1958 32% of the male voters cast their ballots for Allende, but only 22% of the female voters did so. Respondents were classified as either "yes" or "no" for organizational membership. Details on this variable are presented in the following section Of this chapter. For those working-class respondents with organizational membership, 63% Of those that were high or medium on politicization expressed a preference for the FRAP, but only 20% Of those with fair or low politicization did so. For those working-class respondents without organizational membership, the respective figures are 40% and 27%. For age respondents were categorized as: under 35 years of age, 35-55 years of age, over 55 years of age. For length of residency in the area respondents were classi— fied as: 0-10 years, 11-19 years, 20 or more years. For religion respondents were classified as: regularly practicing Catholics, occasionally practicing Catholics, non-practicing Catholics, other religion, no religious affiliation. In each category of each of these three control variables there is a strong association between politicization and party preference, with the gammas for this association ranging from .31 to 1.00. To illustrate this point a simple hypothetical situation can be considered. Perhaps 40% of Chilean working-class adults voted in 1958, and one—half of those voting cast a ballot for the FRAP. Therefore, one-fifth of all Chilean working-class adults voted for the FRAP candidate and one-fifth voted for other candidates (the remaining 94 three-fifths not voting). If the level of politicization for the working class had been higher (and everything else had remained constant), then a higher proportion of those voting would have cast a ballot for the FRAP candi— date and a higher percentage would have voted. Perhaps 60% of all working-class adults would have voted and 70% of them would have voted for the FRAP candidate. Thus, about two—fifths of all working-class adults would have voted for the FRAP and about one-fifth for other candi- dates. This would have been a substantial increase in the total vote for the FRAP candidate, but no increase at all in the total vote for the other candidates. 28If middle-class respondents are looked at, there is little connection between relative economic deprivation and party preference. If anything, these data suggest that within the upper-middle class lower levels of deprivation are related to support for the FRAP. This certainly follows from the reasoning behind expecting higher deprivation to be connected with support for the FRAP for working-class respondents. Among higher SES individuals, those with higher relative deprivation should be more likely to sup- port parties proposing policies to benefit upper-class groups. 29When the association between relative economic deprivation and party preference is examined while controlling for any one of these five variables, a gamma of at least .20 is found. 30Gabriel Almond and Sidney Verba, The Civic Culture (Boston: Little, Brown & Co., 1965), pp. 244-265, 315-324. 31Norman H. Nie, G. Bingham Powell, Jr., and Kenneth Prewitt, "Social Structure and Political Participation: Develop- mental Relationships," American Political Science Review, LXIII (June, 1969). PP. 361-378. 32Alex Inkeles, "Participant Citizenship in Six Developing Countries," American Political Science Review, LXIII (December, 1969), pp. 1120-1141. 33Milbrath, Political Participation, pp. 38-89 and 110-141, contains an inventory of propositions drawn from other studies. 3"Gurr, Why Men Rebel, pp. 92-121, reviews the contradictory findings of empirical studies. 35 37 38 39 95 For education, respondents were classified as: 0-3 years; 4-6 years (six years being complete primary education); more than 6 years. Education is highly associated with social class, so most working-class respondents fall into the first two categories on education. About 31% of the respondents indicated that they belonged to some economic, social, political, or cultural organiza- tion. For working-class respondents this figure is 29%. Media participation is an index constructed from two variables, frequency Of newspaper reading and frequency of radio listening. Frequency of newspaper reading was constructed from the respondent's indication of how Often he read specific newspapers, with the categories being high, medium, and low. The categories for radio listening were also high, medium, and low. The index was a summated one, constructed by (a) scoring 3 points for a high response, 2 points for a medium response, and 1 point for a low response; (b) summing the scores for both variables; and (c) dichotomizing the scores into "high" and "low." This information was Obtained for the re-interviewed portion of the sample (those indicating that they planned to vote in the election). When education or organizational membership is the inde- pendent variable, politicization is simply dichotomized (high and medium versus fair and low). When media par— ticipation is the independent variable this would not be advisable, since this variable is available only for those who were re—interviewed. Those who were re—interviewed were the ones indicating that they intended to vote, so they tend to fall largely into the high and medium cate- gories for politicization (none were low on politicization). In order to obtain more variance for the dependent vari- able I considered high, medium, and fair as separate categories for politicization in the cross—tabulation between media participation and politicization. The social mobilization variables also have a strong impact on politicization for the middle-class respondents as well. In fact, although the middle-class respondents have a higher level of politicization than the working- class respondents, this is simply due to the fact that the middle-class respondents are higher in education, organizational membership, and media participation. For example, if education is held constant, there is no con- nection between social class and politicization. 96 “°Actua11y, two other interpretations would be consistent with the results. First of all, party preference could be the independent variable, politicization the inter- vening variable, and education, organizational membership, and media participation the dependent variables. The other possibility is that politicization is the indepen- dent variable and all of the other variables are dependent variables influenced by politicization. Given the rela- tive alterability of the variables in this analysis, and some inferences about the likely time order, these two possibilities seem remote. There may be some effects in these directions, but the dominant direction of influence should be from the social mobilization factors to politi- cization to political preference. “lHenry A. Landsberger, Manuel Barrera, and Abel Toro, "The Chilean Labor Union Leader: A Preliminary Report on His Background and Attitudes," Industrial and Labor Relations Review, XVII (April, 1964), pp. 399-420. “zJames Petras and Maurice Zeitlin, "Agrarian Radicalism in Chile," British Journal of Sociology, XIX (September, 1968), pp. 254-270; Maurice Zeitlin and James Petras, "The Working-class Vote in Chile: Christian Democracy Versus Marxism," British Journal of Sociology, XXI (March, 1970), pp. 16-19. “3Inkeles, "Participant Citizenship in Six Developing Countries," p. 1134. ”'Alejandro Portes, "Leftist Radicalism in Chile," Compara— tive Politics, II (January, 1970), p. 265. CHAPTER IV AGGREGATE ANALYSIS Introduction In this chapter the relationships presented in chapter one are tested by using aggregate socio-economic and electoral data in an ecological analysis of the FRAP vote. The socio-economic data are derived from the 1960 census, and the electoral data inClude the results of the 1958 and 1964 presidential elections. The data col- lection procedures have been discussed in chapter two, while the data and data sources are contained in the appendix. Partial correlation and regression analysis is the statistical method employed, with the province being the unit of analysis. Building on the results of the survey analysis in the previous chapter, this chapter further examines the potential impact socio-economic change may have on electoral support for leftist political parties. This ecological analysis is a necessary part of the examination of the macro-developmental relationships, as inferring these from individual-level relationships alone would be extremely risky. There is, of course, still 97 98 a problem in generalizing to longitudinal relationships from any cross-sectional relationships, regardless of the unit of analysis, and attention must be paid to this fact. Propositions Chapter one argued that the process of social change affects electoral support for leftist politiCal parties insofar as it affects levels of politicization and relative economic deprivation for the working class. Specifically, it was argued that politicization is affected by social mobilization and that relative eco— nomic deprivation is affected by both social mobilization and economic conditions. The survey data analyzed in chapter three do show that both politicization and rela— tive economic deprivation affect support for the FRAP among working-class respondents. However, the results of the survey analysis also indicate that the argument con- cerning the influence Of objective socio-economic variables on these two subjective factors needs to be reconsidered. First of all, chapter one stated that the extent of social mobilization of the working class would affeCt electoral support for leftist political parties through its impact on working-class politicization levels. The survey data clearly show that FRAP support among working-class 99 respondents is determined by, among other things, their politicization, which in turn is related to their educa— tion, organizational participation, and media exposure. More politicized working-class members provide greater electoral support for the FRAP because they are both more likely to prefer leftist candidates over others and more likely to actually vote. The influence of the social mobilization variables on politicization is clearly demon- strated not only in the preceding survey analysis, but also in several cross-national empirical investigations cited earlier.1 Thus, this part of the argument presented in the first chapter appears sound, and it leads to the following proposition: The level of social mobilization of the working class in a province has a positive effect on the size of the FRAP vote in the province. The corresponding developmental relationship would be: the greater the recent increase in the social mobilization of the working class in a province, the greater will be the recent increase in the FRAP vote for the province (everything else being equal). This proposition is not directly tested in this chapter. However desirable it might be to actually look at change over time when testing for causal relationships, it was not possible to do this for reasons mentioned in chapter two, so the analysis is 100 limited to comparing provinces in terms of differences in social mobilization and FRAP strength at the same point in time. Chapter one also stated that social change would affect electoral support for leftist political parties by affecting levels of relative economic deprivation fOr the working class. A common contention in the relevant lit- erature is that the process of social mobilization raises economic expectations, which if unfulfilled will heighten levels of relative economic deprivation; therefore, the objective socio-economic determinants of relative economic deprivation can be expressed in terms of a ratio between social mobilization and economic conditions.2 The greater the gap between the level of social mobilization and the quality of economic conditions, the greater the discrepancy between economic expectations and economic achievements, and thus the higher level of relative economic deprivation. However, the relationship between social mobiliza- tion and relative economic deprivation is not supported by the survey data analyzed in chapter three. Even for working- class respondents, relative economic deprivation is com- pletely unrelated to any of the social mobilization variables examined. While this finding contradicts some generaliza- tions commonly found in the literature, these generaliza- tions have rarely been subjected to a direct empirical test; 101 as chapter three pointed out, the limited empirical evidence that does exist in this area is contradictory. The lack of a connection between social mobiliza- tion and relative economic deprivation means that, contrary to the argument in chapter one, the Objective determinants of relative economic deprivation cannot be expressed as a ratio of social mobilization to economic conditions. There is simply no systematic or consistent tendency for social mobilization to heighten economic expectations. Similar questions can be raised about the other half of the ratio that supposedly determines the level of relative deprivation. Objective economic conditions probably have as little impact on relative economic depri- vation as social mobilization does. Although this rela— tionship, between objective economic conditions and rela- tive economic deprivation for working-class individuals, was not thoroughly investigated in chapter three, the data do show that there is no difference between skilled and unskilled workers in terms of their level of relative economic deprivation. Literature reviewed in chapter one also pointed to the frequent lack of connection between Objective conditions and relative deprivation. There does not seem to be a particularly convincing reason to believe that Objective economic conditions will be significantly related to relative economic deprivation, but on the 102 possibility that there might be some very general connection between these two factors, the following pro- position will be tested: The level of economic conditions for the working class in a province has a negative effect on the size of the FRAP vote in the province. Perhaps a better argument can be made that the extent of recent improvement, or lack of such, in Objec- tive economic conditions is an important factor affecting relative economic deprivation. Ted Gurr argues that past experiences have an important effect on the value capa- bilities of a group: Perceptions of value capabilities are affected . . . by men's recollections of how well or badly things were for them in the past. If they have been unable to improve their condi- tions, or worse, if their value position has steadily dete- riorated, either absolutely or relative to other groups, they are likely to see their future prospects in static or declin- ing terms. The worse things have been in the past, the lower their value capabilities are likely to be.3 Since Gurr's formulation of relative deprivation, presented in chapter one of this study, is in terms of a ". . . per- ceived discrepancy between men's value expectations and their value capabilities,"“ anything that lowers value expectations will increase relative deprivation. A high rate of economic improvement will tend to raise economic value capabilities and thus lower relative economic depri- vation. This reasoning leads to the following proposition: 103 The recent rate Of improvement in economic conditions for the working class in a province has a negative effect on the size of the FRAP vote in the province. Unfortunately, the impact of recent economic improvement on relative economic deprivation was not investigated in chapter three, due to the lack of appropriate data, so the above proposition is not backed up by supporting survey data in this study. Chapter three clearly shows that the FRAP vote, at least during the period under study, came primarily from the urban and rural proletariat. Where leftist political parties draw heavily from the working class for their vote, socio-economic change may affect electoral support for such parties simply by increasing or decreasing the relative size of the working class. This is an Obvious relationship of little theoretical interest in this study, but the size of the working class should be considered in this analysis as a control variable. The final proposition to be tested is: The greater the relative size of the working class in a province, the greater will be the size of the FRAP vote in the province. In sum, the propositions tested in this chapter indicate potential effects of socio-economic change on electoral support for leftist political parties. By 104 altering levels of politicization and relative economic deprivation for the working class, socio-economic change may have a substantial impact on voting patterns. The influence of politicization and relative economic depriva- tion on working-class support for the left seems fairly clear, as does the effect of social mobilization on politicization. Unfortunately, when it comes to the Ob- jective socio-economic determinants of relative economic deprivation, considerable uncertainty remains. Although propositions specifying effects for economic conditions and recent economic improvement on electoral support for the left are tested, direct evidence on the relationship between these two objective factors and relative economic depriva- tion is lacking. Analysis Design Considerable attention must be paid to several important theoretical and methodological problems involved in the design of the analysis. There are, first of all, the problems usually involved in any ecological analysis. Additionally, there are important questions about the form of the relationships to be expected. Also requiring con- sideration here are certain difficulties in adequately measuring some of the variables. 105 The measurement Of electoral support for the FRAP is quite straightforward. Two separate variables are used here: the per cent of the vote, by province, for the FRAP candidate, Salvador Allende, in both the 1958 and 1964 presidential elections. The use Of the 1958 election results to test the propositions in this chapter is espe- cially advisable, since the survey data analyzed in the previous chapter pertain to this election. The 1964 elec- toral statistics are used to retest the propositions to see if similar results are Obtained at a different point in time and in a slighly different context. Background information on the 1958 and 1964 presi- dential elections is presented in chapter two. In 1958 there were four major candidates, allowing Allende to finish a very close second with 29% of the vote. The 1964 contest was essentially between Allende and Eduardo Frei, with Allende winning 39% of the vote while finishing a dis— tant second. Allende's support over the 25 provinces ranged from 49% to 16% in 1958, while the comparable range in 1964 was from 60% to 30%. More difficulty is involved in adequately opera- tionalizing the socio-economic concepts under consideration. The social mobilization, economic condition, and improve- ment in economic condition of the working class in a prov— ince are all probably too complex to be measured by any 106 one single indicator. A better procedure, and one frequently used, is to construct an index from several relevant socio-economic variables. A related problem here is that of multicollinearity.5 Data for a large number of socio-economic variables were obtained, and a simple correlation matrix showed a high degree of association among many of the variables. Where this is the case, any ecological analysis using only a few of these socio-economic variables to predict voting patterns would be inadvisable. There would be too many other socio-economic variables highly related to the ones selected, and it would be impossible to statistically con— trol for all of them. Obviously, in such a case as many of the variables as possible must be brought into the analysis, and in such a manner that they are manageable. The above problems characterize an ecological analysis of the 1952 vote for Allende by Glaucio Soares and Robert Hamblin.6 For example, in attempting to measure Objective economic deprivation they use the percentage of the total provincial population not born in the province as a negative indicator of economic deprivation, reasoning that migration usually goes from the more deprived areas to the less deprived areas. One must certainly be skeptical about this indicator being a good measure of economic deprivation, particularly since the authors do not show 107 that it is correlated with any other indicator of economic conditions. The authors also maintain that the percentage of the population born out of the province is a positive indicator of anomie, again without showing that it relates to any other potential indicator of anomie. When such Operationalization procedures are relied upon, it would appear very difficult to conclude anything meaningful about the effects of either anomie or economic deprivation on left voting. Soares and Hamblin also fail to consider problems of multicollinearity. They start with twelve different socio-economic variables, find that these twelve account for 60% of the variance in the dependent variable (the 1952 vote for Allende), and then use a stepwise regression procedure to reduce this down to only four independent variables, which account for 45% of the variance. Soares and Hamblin do not present a correlation matrix for the twelve independent variables, but the multicollinearity is probably extremely high. Given this, it is certainly not ' possible to claim that these four variables represent the major influences on the dependent variable. It is likely that four different variables could be chosen without suffering that much Of a loss in terms of explained vari- ance . 108 The procedure used in the design of this analysis was to combine several variables together into a single index, which helps to control for problems of multicol- linearity as well as insure a more adequate operationaliza- tion of the socio-economic concepts under consideration. First of all, data for a variety of socio-economic aspects of Chilean provinces in 1960 were collected.7 A correla- tion matrix showed many of these variables to be highly interrelated, so an attempt was made to find sets of vari- ables that could be combined together because of their empirical and conceptual connections. Specifically, the aim was to form a social mobilization index, an economic conditions index, and an economic improvement index. In forming these indexes three considerations were used: (1) the index components should thoroughly sample the domain of the larger concept; (2) the index components should all be highly interrelated; and (3) when any one of several combinations of components would yield about the same result, components should be selected on the basis of general availability and comparability.8 Previous discussions of social mobilization indi- cated the major aspects to be urbanization, literacy or education, organizational membership, media participation, and industrialization. The survey analysis illustrated the impact of three of these factors on politicization. 109 Other evidence cited suggested the importance of industrialization. It was argued that urbanization, by itself, has little impact on politicization, but urbaniza- tion is so highly connected with these other factors that it is probably worthwhile to include it. Of course, what is desired here is a measure of the social mobilization of the working class in a province, not the social mobilization of the province as a whole. While data specifically on the working class are not available by province, it is possible to select indicators that best reflect the concept under consideration. For example, unionization is focused on as the best indicator of levels Of organizational membership for the working class. Guided by the above criteria, five variables were selected to form the social mobilization index: 1. per cent urban: the percentage of the population living in urban areas. The census definition of urban (towns over 2500 population) was used simply because of availability. If other definitions of urban are used (e.g., cities over 20,000 popula- tion) virtually identical results will be obtained for the index as a whole. 2. per cent literate: the percentage of the adult population that is literate, again by census definition. This variable was chosen over mea- sures of the education level because of greater availability and comparability and because it is more.relevant for the working class. It does cor- relate quite well with such measures Of education levels as the proportion of adults with a primary education. 110 3. per cent non-agricultural: the percentage of the labor force employed in non-agricultural occupations. This variable was chosen over other measures of industrialization (e.g., proportion of the labor force in manufacturing, mining, and construction) for reasons Of general availability and comparability. It does correlate with the other measures quite well. 4. per cent unionized: the percentage of the labor force that is unionized. This was selected as the best measure of working class organizational membership, despite the presence of a fair amount of error in this measure.9 5. per cent homes with radio: the percentage of housing units having a radio was selected to tap media participation. Unfortunately, other mea- sures Of media consumption, such as newspaper circulation, were not available by province. The usual procedure for forming an index from several interval-level variables was follows: the raw scores were transformed into standardized scores, which were then summed for each province. The correlation matrix for the social mobilization index and its component vari— ables is presented in Table IV—l. This index appears quite satisfactory in that the major dimensions of social mobili- zation have been tapped and all five component variables are highly interrelated. Furthermore, these components correspond to those used by other social scientists attempt- ing to measure social mobilization at the societal level for purposes of cross-national research.10 In forming the economic conditions index it was also possible to refer to similar attempts by others.11 lll oo.e «a. as. me. me. we. xmeaa aoaumnaaanoz Hmeoom .e oo.a He. me. as. me. oases sues page: a .m oo.e as. as. me. essences: e .e oo.H as. ea. Hapaaaaoaumeueoz e .m oo.H om. «swamped a .m oe.e than: a .H .e .m .e .m .~ .H mmqmfimfir ezmzomzoo oze xmozH 20333302 .238 mom x232 zeaefimmmoo HI>H mnmdfi 112 Suggested aspects include: levels of income and security of income; housing conditions; health conditions; and nutritional levels. Information on average wages, unem- ployment levels, housing conditions, and health care conditions was obtained, but figures on nutritional levels, such as per capita caloric intake, were not available by province. Also, a preliminary examination showed the unemployment figures to be unusable.12 Four specific com- ponents were used to form the economic conditions index: 1. daily wages: average daily wages for the non- agricultural sector. This variable was selected because it refers directly to the urban working class. Of course, possible differences in the cost of living for the provinces are not taken into account here. 2. agricultural wages: average wages paid in the agricultural sector. This variable was selected because it refers directly to the rural working class. It is subject to the same qualification as the above indicator. 3. infant death ratio: the number of infant deaths per 1000 live births was selected as a measure of health care conditions. It correlates highly with other measures Of health care conditions, such as doctors per capita. This indicator may also reflect nutritional levels to a certain extent. 4. per cent homes substandard: the percentages of housing units classified as substandard in the census. This indicator correlates very highly with other measures of housing conditions, such as the percentage of homes with piped potable water, and was chosen largely because it is a more general evaluation of the condition of housing units. The economic conditions index was also formed by summing the standardized scores of the component variables. 113 Table IV-2 gives the correlation matrix for the economic conditions index and its component variables. This index may not be as satisfactory as the social mobilization index because the domain of the concept is not as thoroughly sampled (specifically, measures of nutritional levels and security of income were not included) and because the inter- correlations among the components are not as high. In order to measure recent economic improvement an index was formed by (a) constructing an economic conditions index from 1952 data, using the same components and pro- cedures as for the above economic conditions index, and (b) subtracting the 1952 economic conditions index score from the 1960 economic conditions index score for each province. Thus, the economic improvement index indicates the improvement in economic conditions in a province, rela- tive to the improvement that occurred in other provinces, from 1952 to 1960. It was previously indicated that it would be desir- able to include the size of the working class in a province as a control variable. The measure used here is the propor- tion of the labor force that are workers, including rural agricultural workers. In most of the provinces, workers comprise forty to sixty per cent of the labor force. Table IV-3 presents the correlation matrix for the independent and dependent variables to be used in the 114 .cofiuwocoo owEocoom mo HO>OH umuoom m mumowocfi mmuoom “women Omanowo moanmwnn> ones» How oomum>mu mucOHOHmmooo cowumamuuoo mo mcmwmt oo.e mm. mm. mm. om. xmscH maoauaccou oaaoeoom .m oo.H Hm. so. me. «pumeamumnsm mason a .e oo.H mm. me. «oases gamma pummeH .m oo.H as. means smashesoeume .N oo.H mamas edema .H .m .e .m .N .H mmqm BZMZOQZOU 02¢ xMQZH mZOHBHQZOU UHSOZOUN mom XHMBflS ZOHBdAmMMOU NI>H mflmdfi 115 oo.H «m. ma. vm. Hv. we. coma .Oocmaad w .m oo.H ha. mm. om. ma. mmmH .Oonmafld a .m oo.H be. we. mm. xmocH ucOEO>OHmEH owfiocoom .v oo.H Hm. oa.| xmocH mgofluflocoo Owaocoom .m oo.H eo.- xmeeH coaumNaHanoz Hmaoom .N oo.a mumxnoz w .H .m .m .w .m .N .H mmAm Bzmnzmmma 92¢ BZWDmemQZH mom ”Hm8¢z ZOHadqmmmoo MI>H mqmdfi 116 regression analysis. It is apparent that the per cent workers variable is very poorly related to the other socio- economic variables. While one might think that the more economically developed provinces should have a higher pro- portion of workers in their labor force, this is not the case in Chile because many of the more poorly developed provinces have large numbers of agricultural wage—laborers. The social mobilization index and economic conditions index are strongly correlated, as would be expected, and the economic improvement index is moderately correlated with both of the other indexes. Thus, the problem of multi- collinearity has not been eliminated, but it has been brought down to manageable proportions, as it is possible to obtain meaningful partial correlation and regression coefficients with these data. Of course, multicollinearity still exists within each index. It is not possible, on the basis of the aggregate analysis alone, to determine which specific com- ponents of a given index are primarily responsible for a relationship between the index and the dependent variable. This is one reason that survey data are also relied on in this study. There are problems with the above Operationalization procedures. The fact that 1960 data are used has already been discussed. Also, the components used in the indexes generally do not refer specifically to the socio—economic 117 characteristics Of the working class. Although an attempt was made to select variables that would best tap the social mobilization and economic condition of the working class, there must certainly be a fair amount of error involved here. Hopefully, all measurement errors are random rather than systematic. At best, these indexes are only rough measures of the concepts they correspond to. The potential problems of an ecological regression analysis also need to be considered here. Perhaps a con- venient starting point is a consideration of why ecological correlations may be much larger than the corresponding individual-level correlations.13 One reason is that when geographical units of analysis are used the individuals are generally not randomly grouped with respect to the relevant variables. Where the grouping has the effect of maximizing the variation Of the independent variable, relative to that of the dependent variable, the ecological correlation coef- ficient will be larger than the corresponding individual- level correlation coefficient, but the ecological regression coefficient will not be similarly inflated. A much greater problem occurs when the grouping maximizes the variation in the dependent variable, relative to that in the independent variable, as both the correlation and regression coefficients are adversely affected. As long as the grouping is of the first type, causal relationships existing at the individual 118 level are discernible at the ecological level if attention is focused on the regression coefficients, rather than the correlation coefficients, and if potential extraneous and nuisance variables are controlled for--although it is seldom possible to control for all of these factors. In this specific analysis the grouping appears to be of the first type. Some simple calculations can be made to demonstrate this. Most of the variables employed in this analysis are dichotomous at the individual level, so the individual-level standard deviation can be calculated from the mean value for the population as a whole. The two dependent variables, the per cent workers variable, and the five social mobilization variables all fall into this cate- gory.”' The individual-level and provincial-level standard deviations can then be compared to see if the geographical grouping maximizes the variation in the independent vari- ables relative to that in the dependent variables. Table IV-4 presents these figures, and it is clear that the grouping is of the desired type. Another difficulty with ecological analysis is the inability to distinguish between individual and contextual effects.15 The likelihood of workers, or others, voting for the FRAP may be a function not only of their own socio- economic characteristics but also of the socio-economic characteristics of the area they are in. For example, James 119 «hm. oma. nae. Owonm apes meson Hon. omo. omm. wmuficowco mmv. mma. «me. HmuouaoowumMIcoz cam. «mo. emm. mumnouflq mmv. eam. owe. gonna mow. ooa. mme. muoxno3 eva. one. mme. emma .Oocoaad «ma. mmo. eme. mmma .Oocoaad coauma>mo coauma>Oo O O I no a>a c . . . . \mm meowuwmaweawmum cementum cementum magmaum> . . HO>OHanwocw>oum HO>OHIHmsoH>fiocH mmqdeM¢> Qmfivmflmm mom NUZ Qm>mfl A¢50H>HQZH QZ< A4HUZH>Omm ¢I>H mnmflfi 120 Petras and Maurice Zeitlin argue that the rural labor force in areas adjacent to mining communities (which are centers of FRAP strength) will provide greater electoral support for the FRAP than the rural labor force in areas nonadjacent to mining communities.15 Individual and contextual effects are merged together in an ecological analysis, which poses problems of interpretation, and this is another reason that survey data were also used in this study. Adam Przeworski and Glaucio Soares attempt to formu- late several mathematical models that include contextual effects in relating together the size of the working class and the size of the left vote.r7 Although their work con- tains serious mathematical flaws, the general idea that contextual effects may result in curvilinear ecological relationships is correct. However, unless the contextual effects are quite strong, they are not likely to be dis- cernible in this analysis, given the nature of the data. Other reasons might also be considered for expect- ing a curvilinear relationship. Soares and Hamblin find that a multivariate power relationship better fits their data than a multivariate linear relationship, but their explana- tion relies on the questionable assumption that left voting is a non-voluntary aggressive response to frustration.18 Other authors employing variables similar to the social mobilization and economic conditions indexes use the ratio. 121 of the two as a predictor of political behavior, arguing that the gap between social mobilization and economic con- ditions determines the level of relative deprivation.19 However, this argument has already been dismissed. In the absence of compelling reasons to the con— trary, a linear regression equation is used to predict left voting. The equation has the following form: Y = a + bx + cX + dX + ex 1 2 a u where: Y vote for Allende per cent workers X X = social mobilization index X = economic conditions index X economic improvement index The possibility that the relationship is not a linear addi- tive one can still be considered by examining the residuals, as any strong departure from a linear additive relationship will create a noticeable pattern among the residuals. Analysis Results The result of the multiple linear regression predic- tion of the 1958 vote for Allende is presented in Table IV-S. By focusing on the partial correlation and regression coef- ficients, it is possible to examine the relationship between any one of the independent variables and the dependent .5. usonm we OOGMOflmecmHm mo HO>OH ecu .OHON scum ucouom than we ucoaoammmoo we» we easements on some we saw» smaamuuosu m we «coeuomuap oeuOAOOMQ may Eonm ouwmommo we ucmwowmmmoo c0flmmmnmmu Omsnoon omucwmoum uoce 122 NH. m~.r no.m| xOocH OGOEO>OHQEH oesocoom a HH. mv. xmocH mcowuwocou Oflfiocoom vv. mo. hm. NHJH MOUGH GOflumeHHnOS Hmwoom no. mm. mm. nuoxuos w i 1 c ucmflowmmooo omnwoumocmumco _m mocmowwwcmflm cowumamuuou ucOflOflmmooo OHomwum> ucoocommocH N m H on Hmwuumm GOflmmOHmmm mmmH emnzmflfld mom MBO> BZNU mmm m0 ZOHBUHQmmm ZOmemmwmm “dmZHQ mom mUHBmHamfim mI>H mnmmfi 123 variable with the remaining independent variables statistically controlled for. Given the intercorrelation among the independent variables, it is absolutely neces- sary to do this in order to determine the separate impact of each of the four independent variables. It is also necessary to carefully examine the regression coefficients, as ecological correlation coefficients can be misleading. These data strongly support the hypothesis that social mobilization has a positive effect on electoral sup- port for the FRAP. The partial correlation and regression coefficients are statistically significant, and the regres- sion coefficient can easily be interpreted. The regression coefficient indicates that an increase of one unit in the social mobilization index results in an increase of slightly over one percentage point in the vote for Allende. An increase of one unit in the social mobilization index cor- responds to a simultaneous increase Of about three percentage points in each of the index components (the exact figure varies for each component variable), so increases in social mobilization do appear to have a substantial impact on the FRAP vote.” The regression analysis shows the economic conditions index to be unrelated to the vote for Allende. The partial correlation and regression coefficients are small in size, opposite from the predicted direction, and statistically 124 insignificant. Clearly, the hypothesis that the economic condition of the working class in a province will be nega- tively related to the size of the FRAP vote shOuld be com- pletely rejected. Table IV-S contains ambiguous findings concerning the relationship between the economic improvement index and the FRAP vote. Although there is a modest partial correla- tion coefficient in the correct direction, it is not sta- tistically significant (even at a more relaxed .10 level). Moreover, interpreting the regression coefficient for the economic improvement index is difficult. An increase of one unit in the economic improvement index produces an increase of about two percentage points in the vote for iAllende, but it is not possible to specify in absolute terms what one unit of the index corresponds to. This is because two components refer to wage levels. To state, for example, _that provinces with a score of one on this index had an increase Of some certain amount in average daily wages Sybetween 1952 and 1960 would be meaningless because the 1960 . figures are not corrected for inflation. This index simply y measures the change in the position of a province, relative to other provinces, on the economic conditions index from 1952 to 1960.21 These data obviously do not justify accepting the proposition that the recent rate of economic improvement for 125 the working class in a province has a negative effect on the size Of the FRAP vote in the province. At the same time, the data do not warrant concluding that there is no connec- tion between these two variables. The association obtained in the analysis is too weak to support the original propo- sition and too strong to really confirm the idea that there is no relationship between the variables. In other words, the data are inconclusive on this point. The final independent variable in the analysis, the per cent of the labor force that are workers, was included primarily for control purposes. Although the regression coefficientgdoes not quite reach the .05 level of signifi- cance, it is still interesting to interpret this coefficient. Assuming no contextual effects, the regression coefficient should equal the difference between the proportion of workers who voted for Allende and the proportion of non-workers who voted for Allende. Referring back to Table III-2, and averaging upper-middle and lower-middle groups together, the difference between the proportion of working-class respondents favoring Allende and the proportion of middle- class respondents favoring Allende comes to about .27. This ‘value is very close to that of the regression coefficient, so the survey and aggregate data seem to correspond nicely on this point . The possibility that a linear additive equation might loot be the best predictor of left voting was previously 126 discussed. However, a brief examination of the residuals from the above regression indicates that they are fairly randomly distributed with respect to the independent vari- ables. It is possible that some equation other than a linear additive one might result in a slight improvement in explained variance, but it would be inappropriate to test a variety of different equations and then choose the one that best fits the data. In the absence of a strong tendency for the data to depart from a linear additive relationship and a good theoretical reason for expecting a particular type of departure, the linear additive form of the equation is accepted. The four independent variables used in the regres- sion analysis collectively explain 44% of the variance in the 1958 vote for Allende. If the economic conditions index and the economic improvement index are deleted from the regression equation, there is only a small drop in explained variance. The social mobilization index and the per cent workers variable together explain 40% of the variance in the dependent variable. In fact, the social mobilization index alone accounts for 36% of the variance in the FRAP vote (i.e., the zero-order correlation between these two variables is .60).22 In order to provide a check on the above findings, the regression analysis was repeated using the 1964 vote for 127 Allende as the dependent variable. Table IV-6 contains the summary statistics for this multiple linear regression. Very similar results are obtained for two of the variables: the economic conditions index is unrelated to the FRAP vote, and the relationship between the economic improvement index and the vote for Allende is ambiguous. The major difference between the results of the two regression analyses concerns the regression coefficients for the other two independent variables: the social mobilization index is less strongly related to the FRAP vote in 1964, whereas the opposite is the case for the per cent workers variable.' The most likely explanation for the difference that exists between the 1958 and 1964 results involves the behavior of the FRAP itself. In 1964 the FRAP made a much more concerted effort to obtain the vote of the rural prole- tariat, and was quite successful in doing so.23 This organizational activity on the part of the Chilean left raised the proportion of workers voting for Allende, which accounts for the increased size of the regression coefficient for the per cent workers variable in 1964. Also, this activity resulted in a considerable increase in left voting in the more rural, and therefore less socially mobilized, provinces. This accounts for the decreased size of the regression coefficient for the social mobilization in 1964. 128 .m. ozone we mocmowmacmwn mo HO>OH on» .OHON Eoum ucmumu Imwo ma ucOfiOflmmmoo on» ma mcflEumuoo on own: me ummu omdwnulosu n we acowuomuwo omuowoOHm ma» scum mnemommo we u:O«OflmuOOO cowmmmummn museums omucmmoum poet mo. om.| Hm.HI xoocH mucosm>oueEH owaocoom . a ma. hm. soocH mcoaueocoo Ofleocoom me ca. mm. we. xoocH cowunuwawnoz anfioom Ho. mm. mm. numxuoz w «M OOMMONWMMMHm wmwwmmwwwmw I Aoflncwmfluomwflwmmuomocov masseuse, ucoocmomocH Hmwuumm cownmmumom inf IIHJ some .mnzmqqm mom meo> azmo see so oneOHommm onmmmmOmm memqu mom mOHemHeasm ml>H MAmGB 129 Conclusions The FRAP appeal in these elections was directed toward the Chilean working class, so one reason why Allende did better in some provinces than in others is simply because of differences in the size of the working class. But it is apparent that other factors are also important. This ecological analysis shows that the level of social mobilization of the working class in a province has an impact on the FRAP vote in the province. It is highly unlikely that the connection between these two variables is spurious, since it remains even when several other rele- vant socio-economic factors are controlled for. Little evidence exists to support a relationship between the vote for Allende and either the objective economic condition of the working class or the recent rate of economic improvement for the working class. At best, there is a possibility that the recent rate of economic improvement has a negative impact on the FRAP vote, but the data are simply inconclusive here. The conclusions of this chapter are based on a cross- sectional analysis of Chilean provinces. For determining causal relationships, such an analysis is superior to a cross-national comparison. By comparing Chilean provinces, differences in culture, historical experiences, and institu- tional arrangements were held to a minimum. Other differences, 130 of course, were controlled for by statistical procedures. Thus, most potential extraneous variables have been accounted for in this analysis, at least so far as this is possible in non-experimental research. The cross-sectional relationships examined in this chapter are of little interest in themselves. Their primary value, so far as this study is concerned, lies in their ability to provide some information concerning the effects of socio-economic change on support for leftist political parties in developing societies. Quite obviously, it is not possible to automatically generalize from these cross- sectional findings to developmental relationships. For example, the analysis shows that the'larger the size of the working class in a province, the greater the size of the FRAP vote. It certainly does not necessarily follow that if the size of the working class increases in Chile over time, the vote for the FRAP will similarly increase. The final chapter will discuss this problem, along with the question of whether or not these results can be generalized to other situations and settings. 131 Notes for Chapter IV 1Gabriel Almond and Sidney Verba, The Civic Culture (Boston: Little, Brown & Co., 1965); Alex InKeIes, "Par- ticipant Citizenship in Six Developing Countries," American Political Science Revigy, LXIII (December, 1969), pp. 1120-1141; Norman H. Nie, G. Bingham Powell, Jr., and Kenneth Prewitt, "Social Structure and Political Participa- tion: Developmental Relationships," American Political Science Review, LXII (June, 1969), pp. 361—378. 2Samuel P. Huntington, Political Order in Changin Societies (New Haven: Yale University Press, 1968), pp. 5 -55; Ernest A. Duff and John F. McCamant, "Measuring Social and Political Requirements for System Stability in Latin America," American Political Science Review, LXII (December, 1968), pp. 1125-1143; Ivo K. and RosaIind L. Feierabend, “Aggres— sive Behaviors Within Polities, 1948-1962: A Cross-National Study," Journal of Conflict Resolution, X (September, 1966), pp. 249-271. 3Ted Robert Gurr, Why Men Rebel (Princeton: Princeton Uni- versity Press, 1970), p. 153. “Ibid., p. 13. 5For a discussion of multicollinearity see Hubert M. Blalock, Jr., "Correlated Independent Variables: The Problems of Multicollinearity," Social Forces, XLII (December, 1963), pp. 233-237. eGlaucio Soares and Robert L. Hamblin, "Socio-Economic Variables and Voting for the Radical Left: Chile, 1952," American Political Science Review, LXI (December, 1967), pp. 1053-1065. 7It would have been best to use 1958 socio-economic data to predict the 1958 vote for Allende and 1964 socio- economic data to predict the 1964 vote for Allende, but detailed information by province is available only for census years, which is why 1960 data were used. This introduces a certain amount of error because the position of a province, relative to other provinces, on these socio- economic variables in 1960 may not correspond perfectly to its position in 1958 or 1964. 132 8By general availability I refer to the ease with which others could obtain these data. By comparability I refer to the degree to which data on the same variable are available for other countries. All other things being equal, it is preferable to use components high in avail- ability and comparability, as it adds to the value of the study by making replication and further examination of the findings easier. 9The error present in this variable results from the fact that: (a) only figures on non—agricultural unions (which most unions were in 1960) were need, and (b) precise figures on the number Of different people who were union members are not possible because of some double—membership. 10Karl W. Deutsch, "Social Mobilization and Political Develop- ment," American Political Science Review, LV (September, 1961), pp. 493-514; Peter R. and Anne L. Schneider, "Social Mobilization, Political Institutions, and Political Violence: A Cross—National Analysis," Com arative Politi- cal Studies, IV (April, 1971), pp. 69-90; Duff and McCamant, “Measuring Social and Political Requirements for System Stability in Latin America"; Feierabend and Feierabend, "Aggressive Behaviors Within Polities, 1948-1962." 11Duff and McCamant, "Measuring Social and Political Require- ments for System Stability in Latin America"; Feierabend and Feierabend, "Aggressive Behaviors Within Polities, 1948-1962." 12The unemployment figures show unemployment to be higher in those provinces that are otherwise better-Off in economic conditions. The problem is that unemployment in rural areas goes unreported, so reported unemployment is higher in the more urbanized and industrialized provinces, which tend to rank higher on the other indicators of eco- nomic conditions. Thus, these unemployment figures cannot be used as a valid measure Of real unemployment. 13The following discussion is based on Hubert M. Blalock, Jr., Causgl Inferences in Nonexperimental Research (Chapel Hill: The University of North Carolina Press, 1961), pp. 95-114. 1"Some of the components of the economic conditions index, such as average wages, are not dichotomous at the individual level, so it is not possible to calculate the standard deviation at the individual level without knowing the dis- tribution of the variable over the whole population. 133 15Discussions of contextual effects are contained in several of the articles in Mattei Dogan and Stein Rokkan (eds.), Quantitative Ecological Analysis in the Social Sciences TChmbridge: MIT Press, 1969). 16James Petras and Maurice Zeitlin, "Miners and Agrarian Radicalism," American Sociological Review, XXXII (August, 1967), pp. 578-586. 17Adam Przeworski and Glaucio A. D. Soares, "Theories in Search of a Curve: A Contextual Interpretation of Left Vote," American Political Science Review, LXV (March, 1971), pp. 51-680 18Soares and Hamblin, "Socio-Economic Variables and Voting for the Radical Left." 19Duff and McCamant, "Measuring Social and Political Require— ments for System Stability in Latin America"; Feierabend and Feierabend, "Aggressive Behaviors Within Polities, 1948- 1962." 2°It should be realized that any conclusion about the amount of change in the social mobilization index that produces some specified increase in the FRAP vote may apply only within certain ranges Of the social mobilization variables, particularly since all of the components are "ceiling" indicators--i.e., they have a maximum value Of 100 because they are percentage figures. 21For example, if a given province was exactly at the mean level (for all provinces) on each of the economic condi— tions index components in 1952, but by 1960 had dropped to one standard deviation unit below the 1960 level on each of the components, then that province would have a score of -4.0 on the economic improvement index. If another province had been at the mean level on each economic conditions index component in both 1952 and 1960, then its score on the economic improvement index would be zero. Of these two examples, the regression equation would predict that the first province would have a FRAP vote eight percentage points higher than the second province, everything else being equal. Unfortunately, this is not that informative for interpreting the regression coefficient. It would be much better if a score on the economic improvement index could be translated into a specific rate of increase, but this is not possible. 134 22Perhaps more importantly, the regression coefficient for the social mobilization index is not greatly altered when the other independent variables are deleted from the analysis. Regardless of which other variables are included in the regression equation, this coefficient remains above 1.0 and below 1.4. This stability of the regression coef- ficient further supports the inferences made about the impact of social mobilization on the FRAP vote. 23James Petras and Maurice Zeitlin, "Agrarian Radicalism in Chile," British Journal of Sociology, XIX (September, 1968), pp. 254-270. CHAPTER V CONCLUSIONS Introduction The previous two chapters have presented empirical tests of several propositions in order to provide a basis for statements about the potential effects of socio- economic change on support for leftist political parties in developing countries. In this chapter the results of the survey and aggregate analyses are summarized and given a theoretical interpretation that supports arguments for some macro-developmental relationships. Additionally, other implications of these findings for politics in devel— oping societies are considered. Also, several aspects of the research setting are examined as possible factors limiting the generalizability Of the findings to other situations and settings. Finally, some suggestions for further research are made. Summary and Interpretation of Findings The results of the survey analysis in chapter three showed that: (a) FRAP support was located primarily in the 135 136 working class, although working-class electoral support did not go predominantly to the FRAP; (b) within the work- ing class, higher levels of politicization and relative economic deprivation led to greater support for the FRAP; and (c) education, organizational membership, and media participation had the expected effect on politicization, but were unrelated to relative economic deprivation. The sources of working-class support for the Chilean FRAP can be placed in a more general theoretical context. Consider a hypothetical society with: (a) a competitive party system; (b) at least one significant leftist party appealing primarily to the working class; and (c) party cleavages that are largely related to social class. In such a situation, why some workers support the leftist party while others do not can be explained in terms of three subjective factors: subjective class placement, perceived political relevance of social class, and salience of class-related political issues. First of all, whether or not working-class members (objectively defined) subjectively place themselves in the working class is an important determinant of their likeli- hod of supporting the leftist party. Not all working-class members will think of themselves as such; some may feel they are middle-class and others may simply lack a class identi— fication altogether. Working-class members who fail to see 137 themselves as working class should be less receptive to the appeals of the leftist party than those working—class members who do place themselves in the working class, and the more the leftist party aims its appeal explicitly at the working class, the stronger this tendency should be. This lower likelihood of favoring the leftist party should apply to both misidentifiers and non-identifiers within the working class, although workers without any class identifi- cation may be more likely to support the leftist party than workers who think of themselves as middle class. Focusing on subjective class placement does not eliminate the importance of objective social class. Those working-class members who see themselves as middle-class will still be more likely to support the leftist party than middle-class members who see themselves as such, for the working-class misidentifiers will be exposed to cross- pressures that the middle-class members will avoid.1 Thus, objective social class will have an impact on party pref- erence even after subjective class placement is taken into account. Where the leftist party relies primarily on the working class for electoral support, subjective class placement is best considered as one of the subjective fac- tors influencing the likelihood of working—class members supporting the leftist party. The influence of subjective class placement on party preference within the working class was not examined in 138 chapter three simply because few of the working-class respondents lacked a subjective identification with their social class. In part, this may reflect inadequacies in the survey data. A certain number of the working-class respondents probably did not normally think of themselves as belonging to any social class but were able to place themselves in a social class at the request of an inter— viewer. Further questioning could have distinguished these respondents, who are really non-identifiers, from those who normally thought of themselves as belonging to the working class. But even if low levels of class non- identification and misidentification really do exist within the survey sample analyzed in chapter three, the general theoretical importance of subjective class placement should still be recognized. Perceived political relevance of social class is a second subjective factor that should be considered in explaining why some working—class members vote left while others do not. Working-class members may clearly see themselves as such, but this does not necessarily indicate that social class has any political relevance for them. Some may draw a strong connection between social class, political parties, and politics in general; others may fail to grasp any relationship at all. David Butler and Donald Stokes, in their examination of such perceptions among the British electorate, distinguish 139 two common views of the connection between social class and politics.2 One view interprets politics in terms of conflict between class interests that are necessarily opposed, with different political parties appealing to different social classes. The other view simply recog- nizes one party as generally representing the interest of one's social class. Close to 90% of the working-class Labour supporters examined in the Butler and Stokes study held one of these two views that relate class to party. The importance of politicization levels within the working class should now be clear. Poorly politicized working-class members will be more unaware of what the different political parties stand for, and therefore they will have more difficulty in seeing one party as best representing the interests of their social class. Even when they have some recognition of one party as the repre— sentative of working-class interests, poorly politicized working-class members will be more unaware of the political importance of class-related issues or how such issues affect them. The more highly politicized working-class members will be more likely to see the leftist party as representing their class interests, more likely to see how class-related issues are involved in conflict over public policy, and more likely to see how class-related issues directly affect them. For these reasons, the more politicized working-class 140 members should be more likely to support the leftist party. This relationship is clearly supported by the survey data analyzed in chapter three. The authors of The American VOter rely on concepts similar to politicization and subjective class placement in explaining class voting among the American electorate.3 Their results indicate the separate and independent effect of both awareness of class location and level of conceptuali- zation of politics on voting behavior: In short, when external conditions warrant, status polarization occurs in politics primarily among those who are sophisticated and for whom class position is salient. we have introduced these two intervening psychological dimensions separately because as an empirical matter they are independent of one another. That is, knowledge of a respondent's level of con- ceptualization of politics does not help us predict whether he will report awareness of his class location. At all levels of conceptualization the probability of class voting increases if there is some sensitivity to social class location, although these differences are not large save among the sophisticated. But political sophistication and class awareness vary indepen- dently.“ A third subjective factor needs to be added to the two discussed so far. WOrking-class members may have both a clear class identification and a high level of politiciza- tion, but they may not perceive the most important political issues to be class-related ones. Some working-class members will see the most important political issues to be those involving the need for improvement in the economic situation of the working class, and these individuals should be the most likely to support the leftist party. Less likely to 141 cast a left vote are those working-class members who see other political issues as more important. Of course, working-class members who are primarily concerned with non-economic issues may still prefer the position of the leftist party on these issues, but some will probably favor the policies of other parties. Thus, some working-class members may fail to vote left even though they clearly recognize the leftist party as best representing the eco- nomic interests of their social class. The importance that working-class members place on class-related economic issues should be strongly related to their relative economic deprivation. Those with high levels of relative economic deprivation should be much more con- cerned about class-related economic issues than those with low levels of relative economic deprivation. This is not necessarily the case in all situations, for some working— class members with high levels of relative economic depriva- tion may not feel that they will benefit from governmental action designed to improve the economic situation of the working class. But except for those with extremely low levels of politicization and class awareness, the working- class members who are most concerned about improving their own economic situation should be the most receptive to political party appeals that focus on the need for improve- ment in the working-class situation. Therefore, working-class 142 members with higher levels of relative economic deprivation will be more likely to support the leftist party than those with lower levels of relative economic deprivation. This proposition is supported by the survey data analyzed in chapter three. Similar ideas are also contained in some of the literature reviewed in chapter one, although the empirical evidence presented in these studies is generally of an indirect nature. For purposes of generality, the above explanation has been presented for a hypothetical society, characterized by a competitive party system, a significant leftist party appealing primarily to the working class, and political cleavages largely related to social class. Three subjective factors--class placement, politicization, and relative economic deprivation--have been used to specify the segments of the working class that are more likely to support the leftist party. Obviously, the usefulness of these theoreti- cal propositions in explaining working-class voting behavior in any actual society depends on two things: (1) how close the characteristics of the society are to those assumed in the explanation; and (2) how important other subjective factors, besides the three used here, are in the voting behavior of working-class members. In the case of Chile, the material presented in chapter two shows that the assumptions are met fairly well, and the survey analysis in 143 chapter three indicates that the subjective variables used in this study are important determinants of working-class support for the FRAP, although all of the variance is certainly not accounted for by these variables. Although these theoretical propositions simply reflect ideas contained elsewhere, it has been necessary to systematically present the relationships here. All macro-relationships must rely on some conceptualization of individual behavior, and the propositions presented above provide the base from which further propositions will be constructed. But before proceding to that task, a few points should be made clear. First of all, the three subjective variables described above should be recognized as continuous variables. Empirical analyses may dichotomize these variables, but this is simply a matter of convenience or a problem of measure- ment. Theoretically, it is the extent of politicization, the intensity of relative economic deprivation, and the strength of class identification that affect the likelihood of a working-class member supporting the leftist party. It is important to think in terms of continuums so that, for example, low relative economic deprivation is not confused with the absence of relative economic deprivation. Second, it is unlikely that the previous propositions can be extended to cover middle-class voting behavior. It 144 might appear that the subjective factors relied on in this study could be used to construct a general explanation of why some individuals vote for the party associated with their social class while others do not. However, some evidence indicates that middle-class individuals are less prone to see politics in class terms. Butler and Stokes, in their examination of the British electorate, are struck by ". . . how very much more salient to the working class are the ideas of class interest and class conflict."5 Others have noted the same divergence between working-class and middle-class conceptions of the social order and its relation to politics.6 The survey analysis in chapter three confirms such ideas, as politicization and relative economic deprivation were only weakly related to party preference among middle-class respondents. Third, the above propositions are intended to apply only within a society. That is, the three subjective variables can be used, given certain assumptions, to identify the segments of the working class in a particular society that will provide greater support for the leftist party; but these variables probably will not adequately account for cross-national differences in the level of class voting, the extent of left voting within the working class, or the degree of radicalism of the leftist parties. It would be unwise, for example, to infer that in societies where a non-ruling 145 Communist party attracts a high proportion of the workers' vote, the workers in those societies are necessarily characterized by unusually high levels of relative economic deprivation, as compared with workers in societies lacking significant Marxist parties. Other factors will undoubtedly have to be considered in order to explain such cross- national differences. Robert Alford, for example, relies on historical and institutional differences in explaining dif- ferences in the level of class voting among the Anglo- American democracies.7 For the purposes of this study, it is not necessary to consider the conditions that produce these cross-national differences, as these differences can simply be considered parameters. The theoretical propositions that have been presented so far can now be used to develop propositions concerning the short-run effects of socio-economic change on support for leftist parties, at least in certain types of societies. The concern here is with specifying potential short—run effects, not the long-term results of socio-economic develop- ment. Over a long period of time, the process of socio- economic development may drastically alter the nature of political cleavages and thoroughly transform the nature of the political party system.6 But in the short run, socio- economic change may alter the electoral strength of leftist parties while the basic nature of political cleavages and 146 the party system remains fundamentally intact, and the focus of this study is precisely on these more limited effects. Of course, this distinction between short-run and long-term effects is somewhat artificial; long-term trans- formations may result from a series of short-run shifts. Indeed, an understanding of the dynamics of the short-run effects may be invaluable in explaining long-term develop- ments. Thus, although the focus of this study is on the potential effects that socio—economic change may have on the electoral strength of leftist parties during a period in which the fundamental nature of political cleavages and the party system remain stable, the relationships described below may be of interest to those more concerned about broad historical patterns of development. For purposes of generality, a hypothetical society shall again be assumed. Consider a society that: (a) is at a middle level of economic development; (b) has a competitive party system with a fairly significant leftist party that appeals primarily to the working class; and (c) has political cleavages that are largely related to social class. In such a society, socio-economic change may produce short-run changes in the electoral strength of the leftist party through its effects on the subjective variables previously described. By altering working-class levels of relative economic deprivation, politicization, and class identification, 147 socio-economic change may cause increases or decreases in the size of the left vote. For politicization, the most important aspect of socio-economic change is social mobilization. Increases in working-class politicization levels should result from increases in the level of social mobilization for the work- ing class, with the key determinants being education (or literacy), organizational membership, and media participa- tion. The literature in this area is consistent in its findings: education, organizational membership, and media participation all have a positive effect on politicization, while urbanization per se has little or no effect.9 The survey analysis in chapter three also supports the rela- tionships between politicization and education, organiza- tional membership, and media participation. Since more politicized workers are more likely to support the leftist party, increases in the level of social mobilization for the working class should produce increases in the size of the left vote. The results of the aggregate analysis in chapter four are consistent with this develop— mental relationship: the provinces with higher levels of social mobilization had higher levels of FRAP voting, even when other factors were controlled for. However, this was a cross-sectional test of the relationship, so that while the findings are consistent with the proposition, they do not constitute a direct longitudinal test. 148 Although it is possible that a direct longitudinal test of the effects of increased social mobilization on electoral support for the left might disprove the proposi- tion, such a possibility does seem remote in this case. This is so because the concern is with causal relationships, not correlations or their equivalents, and because these causal relationships have been tested at two different levels of analysis, the individual and the provincial. It would not be possible to generalize to macro- developmental relationships from the results of the survey analysis alone. Although the effects of the social mobili- zation variables on working-class support for the FRAP is clearly demonstrated by the survey data, it is possible that the process of social mobilization consistently has other effects as well. These other effects, either individual-level or contextual-level, could tend to cancel out the effects described above, thus resulting in no net effect on the size of the left vote. The aggergate analysis results can be used to dismiss this possibility, as the data support a causal relationship between the level of social mobilization and the size of the FRAP vote, which would not be the case if the process of social mobilization consis- tently had no net effect on left voting. It also would not be possible to generalize to developmental relationships from the results of the aggregate 149 analysis alone. Several different interpretations could be given for the results of the ecological analysis, and some of these interpretations might not logically support developmental relationships. By using survey data it is possible to determine the individual-level causal relation- ships that produced the aggregate results. By combining empirical tests at two different levels of analysis, as well as findings from other studies, it has been possible to construct an explanation that logically supports the macro-developmental proposition. Thus, increases in the level of social mobilization for the working class will, at least in the short run, produce increased support for the leftist party, everything else being equal. A frequent claim in the literature is that the process of socio-economic modernization in developing countries produces short-run increases in relative economic deprivation. These arguments, which were reviewed in chapter one, are based on the hypothesis that the process of social mobilization heightens economic expectations, which if unfulfilled result in increased relative depriva- tion. Samuel Huntington, for example, argues that the larger the gap between the level of social mobilization and the level of economic development for a society, the greater the degree of "social frustration" in the society.10 This hypothesis has been the basis for several quantitative 150 cross-national studies of political instability, violence, or radicalism.n If such arguments are correct, they indicate another way that social change may affect electoral support for the left. Despite the popularity of the above view, there is little empirical evidence to support it. Increases in urbanization, education, media participation, organizational membership, etc. are assumed to heighten expectations by exposing individuals to new consumption patterns and new ways of life. But the survey analysis in chapter three found that levels of relative economic deprivation were completely unrelated to education, organizational member— ship, or media participation. The possible effects of urbanization on relative economic deprivation were not examined in this study, but other studies have found that urban migrants do not suffer the heightened relative deprivation that the above View implies.12 It is unlikely that the structural determinants of relative economic deprivation can be expressed in terms of a ratio between social mobilization and economic conditions or development. There is little direct empirical evidence to suggest that social mobilization has the effects on relative economic deprivation that so commonly have been assumed. The impact of objective economic conditions on relative economic deprivation is equally questionable. The 151 survey analysis in chapter three found that skilled workers did not differ from unskilled ones in terms of their level of relative economic deprivation, and the aggregate analysis in chapter four found very little connection between objective economic conditions and the size of the FRAP vote when other factors were controlled for. In sum, there is no reason to believe that the general process of social change has any systematic and consistent effect on relative economic deprivation: in some situations it may increase levels of relative economic deprivation, while it may lower such levels in other situations. Where social change does increase working-class levels of relative economic depriva- tion, it will tend to increase left voting. Unfortunately, this study is unable to provide any indication as to where social change would tend to increase levels of relative eco- nomic deprivation. Perhaps all of this points to the importance of relying on survey data to interpret aggregate patterns of mass political behavior. If only the aggregate data had been used, the ecological connection between social mobilization and FRAP voting might have been erroneously interpreted. The inference could have been that social mobilization affects relative economic deprivation, which then affects left voting. But the survey data show that it is politici- zation, not relative economic deprivation, that forms the 152 link between social mobilization and left voting. It is, then, extremely risky to draw conclusions about the impact of social change on relative deprivation solely from cross- national aggregate comparisons, as so many studies have done.13 Social change may also affect support for the leftist party by altering levels of class awareness within the work- ing class. The key aspects of social change here are probably urbanization and industrialization. Urbanization, by concentrating workers together in cities, and industriali- zation, by concentrating workers together on the job, should lead to increased communication within the working class and decreased communication between classes. Where workers interact primarily with other workers, rather than with middle-class members, they should be more likely to think of themselves as working-class members and thus more likely to support the leftist party.”‘ The measure of social mobilization used in chapter four included urbanization and industrialization as two of the components, so the ecologi- cal connection between social mobilization and FRAP voting may be partially due to the effects of social mobilization on levels of class identification within the working class. Unfortunately, this interpretation cannot be substantiated by the survey data used in chapter three. It was not pos- sible to examine either the determinants or effects of 153 subjective class placement within the working class because most of the working-class respondents identified with their class. Of course, this high level of class awareness may have existed simply because the survey was limited to an urban area. So far, for the hypothetical society under discus- sion, the potential effects of social change on electoral support for the leftist party have been examined in terms of potential effects on three subjective factors within the working class (politicization, relative economic deprivation, and subjective class placement). Social change appears to have an impact on working-class preference for the left not through any systematic or consistent effect on relative economic deprivation, but through effects on levels of politicization and class awareness. Increases in various aspects of social mobilization--particularly education, organizational membership, and media participation--produce increases in working-class levels of politicization, while increases in other aspects of social mobilization--particularly urbanization and industrialization--produce increases in working-class levels of class awareness. Up to this point it has been assumed that all working- class members who are eligible to vote will do so. When the likelihood of voting is considered, another possible effect of social change on left voting is introduced. The 154 more socially mobilized segments of the working class are not only more likely to prefer leftist candidates and parties, but are also more likely to actually vote. In- creases in social mobilization may, then, increase the size of the left vote simply by increasing the turnout rate for the working class (of course, if the turnout rates for other segments of the population increase in a similar manner, then this effect will not occur). If illiterates are ineligible to vote, then a decrease in the working-class illiteracy rate, which is a likely result of increased social mobilization, may increase the size of the left vote by increasing the proportion of the eligible electorate that are from the working class. Social change may also have an important indirect effect on electoral support for the left through potential effects on the organizational strength and activity of the leftist party and its related organizations, such as trade unions. There are several ways that the organizational strength and activity of the left may affect left voting. First of all, simple increases in organizational membership should have some obvious effects. As workers become :members of leftist dominated trade unions, they become exposed to the political influence of trade union leaders, and this should increase support for the leftist party.15 The survey analysis in chapter three supports this hypothesis, 155 as trade union membership appeared to have a direct effect on working-class support for the FRAP, apart from the effect due to increased politicization. Naturally, increased membership in the leftist party organization itself will not involve any political conversion, since those who join the party will undoubtedly already be sup- porters; but those supporters who become party members will then be subjected to influences that may inhibit them from changing their political allegiance. Thus, increases in the proportion of the working class who are members of leftist organizations should produce increases in electoral support for the leftist party through these obvious effects on the organizational membership. More importantly, increased organizational strength allows for greater organizational activity, which can reach out far beyond the membership in its effects. Several studies of political party activity in the United States indicate that local organizational activity during a cam- paign has an effect on the division of the vote.“5 More- over, organizational activity need not be restricted to election campaigns; party organizations and trade unions can engage in political proselytizing on a year-round basis. While the effects of campaign activity are probably limited to the immediate election with little carry over to future elections, constant and persistent organizational activity ma‘ OI Ci 156 may have more fundamental effects. Over time, leftist organizational activity may be capable of altering working- class levels of class awareness, politicization, and relative economic deprivation, thus altering basic politi- cal orientations and preferences. In sum, the potential impact of organizational activity may be quite substantial. In support of this claim, James Petras and Maurice Zeitlin, in an examination of the 1958 and 1964 Chilean presidential elections, show that agricultural areas adjacent to mining communities (which are center of FRAP organizational strength) had a much higher vote for Allende than agricultural areas that were not mining satellites.17 Organizational activity by leftist political parties is likely to be more difficult in some social settings than in others. Sidney Tarrow, for example, argues that the Italian Communist Party has difficulty in the less developed South because ”. . . the organizational incapacity of a developing society places an objective limitation on the "1° Increases in social party's vote getting ability. mobilization are likely to make it easier for a leftist party to develop and sustain organization strength and activity. Increases in urbanization and industrialization, by concentrating workers together, make it easier to enroll them into party organizations and trade unions. Increases in media participation, along with this greater geographical 157 concentration, make it easier for organizational activity to reach its intended targets. In general, the more socially mobilized segments of the working class are both more likely to join leftist organizations and more likely to be reached by organizational activity. Thus, social change may affect the electoral strength of the leftist party by creating a more favorable environment for the leftist organizations to operate in. It is possible that social change could increase support for the leftist party by increasing the size of the working class. However, this is not likely to be a sig- nificant short-run effect, as the proportion of the labor force that are manual workers will probably not change very much in a short period of time. Even where rapid indus- trialization does have a substantial short-run impact on the size of the working class, this may not have that much of an effect on left voting. New members of the working class may be very low in class awareness, politicization, and relative economic deprivation, particularly if these new members were formerly part of the peasantry, and thus they may provide very little support for the left. The primary short-run effect of social change on the size of the left vote will come from changes in the rate of left voting within the working class, not from changes in the size of the working class. 158 For purposes of generality, the potential effects of social change on electoral support for leftist parties have been examined in a hypothetical society with certain assumed characteristics. The next section of this chapter will discuss the applicability of these theoretical propo- sitions in situations that do not meet the assumptions put forth here. It should be realized that the propositions contained in this section are based on the assumption that the leftist party draws its vote predominantly from the working class. No consideration is given to possible sources of middle-class support.19 It also should be realized that the theoretical propositions put forth simply indicate some potential effects of social change on electoral support for leftist parties, particularly in developing countries. There are a great many other factors that also affect voting behavior, and any examination of actual changes in left voting in a specific society will have to consider these other factors as well. A given society may undergo social changes that tend to produce a greater left vote, but other factors may also be changing in such a way as to counteract the effects of the social changes. For example, Miles WOlpin has described some of the institutional barriers to the Chilean FRAP during the 1960's (e.g., external inputs by the U.S., conservative control over mass media).2° If the strength of such institutional barriers had increased 159 greatly during the 1960's, then the size of the FRAP vote might have decreased even if the level of social mobiliza- tion for the working class increased during the decade. Thus, nothing in this chapter should be interpreted as claiming that increases in the level of social mobilization for the working class in a given society will necessarily result in increases in the size of the left vote. This will be the case only if other factors remain constant. Generalizability of the Findings The arguments contained in the preceding section assumed a developing society with a competitive party system and party cleavages that are largely related to social class. These are fairly restrictive assumptions, likely to be met in only a limited number of cases, so it may be worthwhile to consider the applicability of the propositions to other situations and settings. Since all the empirical evidence analyzed in this study comes from a research setting that does meet the above assumptions, no definitive answer to the question of generalizability can be supplied, but some speculation can be presented. Party cleavages in many developing societies have not been based primarily on social class, but on cultural factors, such as race, religion, or ethnicity. In Guyana, for example, the Marxist People's Progressive Party does 160 not appeal to the working class as a whole, but to the East Indian Guyanese.21 Although this kind of situation does not fit the above assumptions, the ideas put forth in this study may be extended, with appropriate modifications, to cover such cases. Specifically, where a party appeals primarily to a given cultural group, the individuals in that group who will be the most likely to support the party will be those who are the highest in politicization, relative deprivation, and group identification. Also, increases in the level of social mobilization of the group will increase the extent of group support for that party by increasing levels of politicization and group identification (and, perhaps, relative deprivation in some cases).22 Developing societies frequently lack competitive party systems altogether. Naturally, propositions about the impact of social change on support for leftist parties have no direct applicability in societies lacking competitive elections, although they may indicate some of the reasons for the absence of a competitive electoral process.23 How- ever, the basic ideas behind the propositions might be extended to cover the sources of support for revolutionary or other illegal mass movements. To put it more generally, a group characterized by relatively high levels of politi- cization, relative deprivation, and group identification can respond primarily in a nonviolent and legal manner or 161 a violent and illegal manner, and the type of response that occurs is likely to be largely a function of the existing political situation. In some situations, then, the process of social change may increase the potential of a specific group to engage in collective political violence by increas- ing group levels of politicization, relative deprivation, and group identification.” Thus, the findings of this study may well have some broader implications that apply to politics in developing countries, including those lacking either party cleavages based on social class or a competitive party system of any sort. It is less likely that the propositions developed in this study have any applicability in highly developed societies. Ronald Inglehart's description of intergenera— tional change in Western Europe indicates that a basic realignment is taking place, and working-class members with traditional concerns over economic security are becoming less likely to cast their ballot for the left.25 In western industrial democracies, therefore, the process of social change probably will not increase electoral support for leftist parties by increasing working-class levels of politicization, relative economic deprivation, or class awarene S S o 162 Suggestions for Further Research This study, like many others, raises as many questions as it answers. Therefore, it is appropriate to end with some brief suggestions for further research. First, the relationships tested in this study need to be retested in other settings, as it is possible that the relationships do not generally hold outside Chile. Second, more research is needed on the way that social and economic factors affect relative deprivation. The effect of relative deprivation on certain forms of political behavior seems fairly clear, but the determinants of relative deprivation are only poorly understood at present. And third, more attention might be paid to the importance of politicization as a factor affecting political preference. While voting studies frequently examine the impact of political interest and awareness on the likelihood of voting, these studies rarely investigate the possible effects of politicization on partisan choice or other aspects of political preference, and the findings of this study indicate that such effects may be present. 163 Notes for Chapter V 1Angus Campbell, et al., The American Voter (New York: John Wiley & Sons, 1964), pp. 204-206. 2David Butler and Donald Stokes, Political Change in Britain (New York: St. Martin' 3 Press, 1971), pp. 47- 70. 3Campbell, et al., The American Voter, pp. 193-196. “Ibid., p. 196. 5Butler and Stokes, Political Change in Britain, p. 68. 6Ralf Dahrendorf, Class and Class Conflict in Industrial Societ (Stanford: Stanford University Press, 1959), pp. 80-289, reviews some of the work in this area. 7Robert Alford, Party and Society (Chicago: Rand McNally & C00, 1963), pp. 94-1220 °For a discussion of these effects see Seymour Martin Lipset and Stein Rokkan, "Cleavage Structures, Party Systems, and Voter Alignments: An Introduction," in Party Systems and Voter Alignments, ed. by Lipset and Rokkan (New York: The Free Press, 1967). 9See, for example, Gabriel Almond and Sidney Verba, The Civic Culture (Boston: Little, Brown & Co., 1965),— pp. 244-265 and 315-324; Norman H. Nie, G. Bingham Powell, Jr., and Kenneth Prewitt, "Social Structure and Political Participation: Developmental Relationships," American Political Science Review, LXIII (June, 1969), pp. 361-378; and Alex inkeles, "Participant Citizenship in Six Develop— ing Countries," American Political Science Review, LXIII (December, 1969), pp. 1120-1141. Lester Milbrath, Political Participation (Chicago: Rand McNally, 1965), pp. 39-89 and 110-141, contains an inventory of proposi- tions drawn from other studies. 1"Samuel Huntington, Political Order in Changing Societies (New Haven: Yale Univer31ty Press, 1968), pp. 53- 55. llSee Ernest A. Duff and John F. McCamant, "Measuring Social and Political Requirements for System Stability in Latin America," American Political Science Review, LXII (December, 1968), pp. 1155-1143; Ivo K. and;§3salindfif. Feierabend, 164 "Aggressive Behaviors Within Polities, 1948-1962: A Cross- National Study," Journal of Conflict Resolution, X (September, 1966), pp. 249-271; andfiieter R. and Anne L. Schneider, "Social Mobilization, Political Institutions, and Political Violence," Comparative Political Studies, IV (April, 1971), PP. 69-90. 12Wayne A. Cornelius, Jr., "Urbanization as an Agent in Latin American Political Instability: The Case of Mexico," American Political Science Review, LXIII (September, 1969), pp. 833-857, uses survey data to disprove some of the con- ventional assumptions about urban migrants and cites other studies that concur with his findings. 13See the studies cited in note eleven of this chapter, all of which make precisely such inferences. 1"See Adam Przeworski and Glaucio A. D. Soares, "Theories in Search of a Curve: A Contextual Interpretation of Left Vote," American Political Science Review, LXV (March, 1971), pp. 51-68, for a discussion ofhhow left voting is affected by patterns of communication and interaction involving workers. 15See, for example, Richard Hamilton, Affluenge and the French Worker in the Fourth Republic (Princeton: Princeton University Press, 1967), pp. 228-244. 16See Daniel Katz and Samuel J. Eldersveld, "The Impact of Local Party Activity upon the Electorate," Public Opinion Quarterly, XXV (Spring, 1961), pp. 1-15; and Phillips Cutright, "Measuring the Impact of Local Party Activity on the General Election Vote," Public Opinion Quarterly, XXVII (Fall, 1963), pp. 372-386. 1"'James Petras and Maurice Zeitlin, "Miners and Agrarian Radicalism," American Sociological Review, XXXII (August, 1967), pp. 578-586. 18Sidney Tarrow, Peasant Cgmmunism in Southern Italy (New Haven: Yale University Press, 1967), p. 204. 19For a discussion of the appeals of Communism to intellec- tuals in developing societies see John H. Kautsky, Com- munism and the Politics of Development (New York: John Wiley & Sons, 1968). 165 ‘mMiles D. Wolpin, "Chile's Left: Structural Factors Inhibiting an Electoral Victory in 1970," Journal of DevelopinggAreas, III (January, 1969), pp. 257—235. 21See Alvin Rabushka and Kenneth A. Shepsle, Politics in Plural Societies (Columbus: Charles E. Merrill, I972), pp. 95-105, for a brief description of the PPP and its base of support. 22See Robert Melson and Howard Wolpe, "Modernization and the Politics of Communalism: A Theoretical Perspective," American Political Science Review, LXIV (December, 1970), pp. 1112-1130, for a fuller treatment of this. 23Martin C. Needler, "Political Development and Military ' Intervention in Latin America," American Political Science Review, LX (September, 1966), pp. 616-626, indicates that as social and economic development takes place in Latin America, military interventions in politics increasingly occur to prevent leftist or reformist groups from coming to power. 2"For a thorough discussion of the sources of collective political violence see Ted Robert Gurr, Why Men Rebel (Princeton: Princeton University Press, 1970). ‘gRonald Inglehart, "The Silent Revolution in Europe: Intergenerational Change in Post-Industrial Societies," American Political Science Review, LXV (December, 1971), pp. 99141617. ' :5». W1” APPENDIX APPENDIX Tables A-l, A-2, and A-3 contain the aggregate electoral and socio-economic data for the Chilean provinces. The sources of these data are listed below. For a more detailed explanation of the variables refer to chapter four. Data Sources Table A-1 l. 1958 electoral statistics: Chile: Election Factbook (Washington: Institute for the Comparative Study of Political Systems, 1963). 2. 1964 electoral statistics: The Chilean Presidential Election of September 4, 1964: Part II,§lection Analysis Series, No. 3 (Washington: Institute for the Comparative Study of Political Systems, 1965). Tables A-2 and A-3 l. % urban, % non-agricultural: Armand Mattelart, Atlas Social de las Communas de Chile (Santiago: Editorial deI Pacifico, 1966). 166 167 2. % homes with radio, % homes sub-standard: calculated from a computer tape record of coded interview schedules from the 1960 Chilean census; the tape was obtained from the Latin American Data Bank at the Uni- versity of Florida. 3. literacy rate, infant death ratio: F. S. Weaver, "Backwash, Spread and the Chilean State," Studies in Comparative International Development, Volume V (1969- 1970), PP. 239-251. 4. % workers, % unionized, wage indices: F. S. weaver, "Regional Patterns of Economic Change in Chile, 1950-1964" (unpublished Ph.D. dissertation, Cornell University, 1968). 168 TABLE A-l AGGREGATE VOTING STATISTICS FOR CHILEAN PROVINCES . % Voting for % Voting for P r °V1n°e Allende , 1958 Allende , 1964 Tarapacé 39.2 47.1 Antofagasta 43.8 48.3 Atacama 34.6 44.6 Coquimbo 33.3 45.1 Aconcagua 25.1 40.4 Valparaiso 23.1 36.1 Santiago 28.4 35.7 O'Higgins 31.3 45.6 Colchagua 22.1 36.5 Curic6 32.2 41.4 Talca 23.8 45.3 Maule 15.9 38.5 Linares 24.2 41.7 Nuble 23.7 34.2 Concepcién 41.2 49.3 Arauco 48.2 60.3 Bio-Bio 31.3 38.4 Malleco 26.4 36.5 Cautin 21.4 31.2 Valdivia 29.7 43.3 Osorno 24.8 38.2 Llanquihue 18.4 29.8 Chiloé 22:8 32.7 Aysén 28.3 '34.7 Magallanes 48.6 49.9 SOCIAL MOBILIZATION STATISTICS 169 TABLE A-2 FOR CHILEAN PROVINCES, 1960 H m 'o m m H m o u o N -H m m u -H 'o H Province g 3", , '3 5 8g '3 Q u at) -H E H H -H owq c or: o D r-il Z M D :1: JJ 3 m -a w w w’m w wtz w Tarapacé 87.0 91.6 86.0 13.6 61.5 38.5 Antofagasta 95.0 91.2 97.0 38.5 65.8 52.2 Atacama 74.0 85.2 88.1 20.9 50.4 53.7 Coquimbo 51.0 76.6 63.4 9.5 36.2 44.7 Aconcagua 56.0 79.8 54.5 7.4 49.5 53.0 Valparaiso 89.0 91.0 87.4 15.0 72.5 40.0 Santiago 90.0 89.8 81.6 12.1 76.4 38.9 O'Higgins 53.0 77.1 51.5 22.1 53.9 61.9 Colchagua 33.0 67.8 36.9 3.8 30.6 62.2 Curic6 41.0 70.6 41.8 2.8 38.2 55.8 Talca 44.0 72.4 49.1 14.0 41.5 55.5 Maule 40.0 71.4 39.6 3.5 23.9 45.6 Linares 36.0 71.2 41.5 2.6 29.6 56.0 Nuble 40.0 70.2 40.0 3.9 25.6 48.8 Concepcién 82.0 '82.9 83.4 19.2 55.5 48.8 Arauco 36.0 67.9 53.2 37.0 23.9 55.3 Bio-Bio 36.0 69.2 44.3 5.9 27.5 53.6 Malleco 45.0 69.6 46.4 5.2 27.6 44.2 Cautin 39.0 72.2 44.2 2.3 27.0 31.4 Valdivia 44.0 76.5 54.4 13.3 27.7 50.2 Osorno 46.0 78.6 53.6 4.3 34.1 49.9 Llanquihue 42.0 77.9 52.2 5.1 24.8 40.2 Chiloé 22.0 78.2 32.6 4.1 11.3 17.4 Aysén 53.0 78.4 57.6 9.0 27.5 37.2 Magallanes 83.0 93.1 83.0 19.3 76.9 41.7 170 TABLE A-3 ECONOMIC CONDITIONS STATISTICS FOR CHILEAN PROVINCES Index of Index of Infant % Homes Daily Agric. Death Sub- Province Wages Wages Ratio standard 1952 1960 1952 1960 1952 1962 1952 1960 Tarapacé 110 141 180 182 87 77 15.8 17.6 Antofagasta 147 183 208 171 111 111 12.5 11.2 Atacama 88 120 121 109 115 113 29.8 21.5 Coquimbo 85 88 91 101 135 126 41.3 27.5 Aconcagua 63 80 94 105 106 92 21.7 19.5 Valparaiso 111 118 99 117 110 90 13.9 12.4 Santiago 106 115 112 107 100 82 15.9 16.6 O'Higgins 124 176 92 95 142 116 15.2 17.5 Colchagua 40 44 67 91 142 108 17.1 23.2 Curic6 46 49 76 85 152 124 16.6 26.5 Talca 78 105 83 87 153 140 17.6 23.4 Maule 43 47 70 70 115 115 18.1 17.3 Linares 36 57 62 71 160 132 20.7 34.0 Nuble 36 54 58 84 167 151 23.6 31.3 Concepcién 113 146 123 110 147 140 20.9 26.0 Arauco 91 116 100 96 162 177 26.5 32.7 Bio-Bio 43 65 65 88 177 159 27.3 35.2 Malleco 44 56 98 87 170 153 27.7 34.1 Cautin 45 63 91 90 172 154 27.6 31.6 Valdivia 68 75 122 102 181 142 21.1 23.3 Osorno 54 65 94 93 204 169 17.7 25.9 Llanquihue 55 73 101 99 171 147 15.8 22.5 Chiloé 70 54 270 127 181 147 17.5 27.7 Aysén 93 99 141 185 135 137 ‘ 20.6 20.7 Magallanes 153 156 394 242 69 58 12.6 4.8 BIBLIOGRAPHY BIBLIOGRAPHY Alexander, Robert J. Communism in Latin America. New Brunswick: Rutgers University Press, 1957. . Labor Relations in Ar entina, Brazil, and Chile. New York: McGraw Hill Book Company, 1962. Alford, Robert. Party and Society. Chicago: Rand McNally, 1963. Allardt, Erik. Social Sources of!Finnish Communism: Traditional and Emerging Radicalism. Research Report, Institute of Sociology. Helsinki: Uni- versity of Helsinki, 1963. Almond, Gabriel. The Appeals_of Communism. Princeton: Princeton University Press, 1954. , and Verba, Sidney. The Civic Culture. Boston: Little, Brown and Company, 1965. Barraclough, Solon L., and Domike, Arthur L. "Agrarian Structure in Seven Latin American Countries." Agrarian Prgblems and Peasant Movements. Edited by Rodolfo Stavenhagan. Garden City, New York: Doubleday & Co., 1970. Bendix, Reinhardt, and Lipset, Seymour Martin, eds. Class, Status, and Power. Glencoe: The Free Press, 1953. Blalock, Hubert M., Jr. Social Statistics. New York: McGraw Hill Book Co., 1960. . Causal Inferences in Nonexperimenta1 Research. ChapeI H111: The University of North Carolina Press, 1961. "Correlated Independent Variables: The Problems of Multicollinearity." Social Forces, XLII (December, 1963), 233-237. 171 172 Brown, Leonard, and Jones, F. Lancaster. "Status Consistency and Political Preference: The Australian Case.“ American Sociological Review, XXXV (December, 1970), 989-1001. Burnett, Ben G. Political Groups in Chige. Latin American Monographs, No. 21. Austin: University of Texas Press, 1970. Butler, David, and Stokes, Donald. Political Chan e in Britain. New York: St. Martin's Press, 1 71. Campbell, Angus; Converse, Philip E.; Miller, Warren; and Stokes, Donald E. The American Voter. New York: John Wiley & Sons, 1964. Centers, Richard. The Psycholggy of Social Classes. Princeton: Princeton UniversityPress, 1949. Cope, Orville G. "The 1964 Presidential Election in Chile: The Politics of Change and Access." Inter—American Economic Affairs, XIX (Spring, 1966), 3—29. Cornelius, Wayne A. "Urbanization as an Agent in Latin American Political Instability: The Case of Mexico." American Political Science Review, LXIII (September, 1969), 833-857. Cutright, Phillips. "Measuring the Impact of Local Party Activity on the General Election Vote." Public Opinion gparterly, XXVII (Fall, 1963), 372-386. Dahrendorf, Ralf. Class and_§lass Conflict in Industrial Society. Stanford: Stanford University Press, 1959. Deutsch, Karl W. "Social Mobilization and Political Develop- ment." American Political Science Review, LV (September, 1961?, 493-514. Dogan, Mattei, and Rokkan, Stein, eds. Qpantitative Eco- logical Anal sis in the Social Sciences. Cambridge: MIT Press, 1 69. Duff, Ernest A., and McCamant, John F. "Measuring Social and Political Requirements for System Stability in Latin America." American Political Science Review, LXII (December, 1968), 1125-1143. 173 Feierabend, Ivo K., and Rosalind L. "Aggressive Behaviors Within Politics, 1948-1962: A Cross-National Study." Journgl of Conflict Resolution, X (September, 1966), Francis, Michael, and Lanning, Eldon. "Chile's 1967 Muni- cipal Elections." Inter-American Economic Affairs, XXI (Autumn, 1967), 23-36. Gil, Frederico. The Political System of Chile. Boston: Houghton Mifflin Co., 1966. , and Parrish, Charles J. The Chilean Presidential Electionpf September 4, 1964: Part I. Election Analysis Series, No. 3. Washington: Institute for the Comparative Study of Political Systems, 1965. Greene, Thomas H. "Communist Electorate in France: Test— ing the Protest Hypothesis." Unpublished paper, University of Southern California, 1969. Goldrich, Daniel. “Toward the Comparative Study of Politicization in Latin America." .Contem orar Cultures_§pd Societiesggf Latin America. Edited by Dwight D. Heath and Richard N. Adams. New York: Random House, 1965. Grayson, George W., Jr. "Chile's Christian Democratic Party: Power, Factions, and Ideology." Review of Politics, XXXI (April, 1969), 147-171. Gregory, Peter. Industrial Wages in Chile. Cornell International Industrial and Labor Relations Report, No. 8. Ithaca: New York State School of Industrial and Labor Relations, 1967. Gurr, Ted Robert. Thy Men Rebel. Princeton: Princeton University Press, 1970. Halperin, Ernst. Nationalism and Communism in Chile. Cambridge: The MIT Press, 1965. Hamilton, Richard. Affluence gpd the Frengh Worker in the FourEh Republic. Princeton: Princeton Uni- versity Press, 1967. . "Skill Level and Politics." Public Opinion Quarterly, XXIX (Fall, 1965), 390-399. 174 Hirschman, Albert 0. Journeys Toward Progress. New York: The Twentieth Century Fund, 1963. Huntington, Samuel P. Political Order in Chan ing Societies. New Haven: Yale University Press,'196 . Inglehart, Ronald. "The Silent Revolution in Europe: Intergenerational Change in Post-Industrial Societies." American Political Science Review, LXV (December, 1971), 991-1017. Inkeles, Alex. "Participant Citizenship in Six Developing Countries." American Political Science Review, LXIII (December, 1969), 1136—1141. Institute for the Comparative Study of Political Systems. Chile: Election Facpppok. Washington: Institute fOr the Comparative Study of Political Systems, 1963. . The Chileap Presidential Electiop of September 4, 1964: Part IT. ElectiOn Analysis Series, No. 3. Washington: Institute for the Comparative Study of Political Systems, 1965. Janowitz, Morris, and Segal, David R. "Social Cleavage and Party Affiliation: Germany, Great Britain, and the United States." American Journal of Sociology, LXXII (May, 1967), 601-618: Katz, Daniel, and Eldersveld, Samuel J. "The Impact of Local Party Activity Upon the Electorate." Public Opinion Quarterly, XXV (Spring, 1961), 1-15. Kaufman, Robert R. The Chileap Political Rightand Agrarian Reform. Washington: Institute for the Comparative Study of Political Systems, 1967. Kautsky, John H. Communism andphe Politics of Development. New York: John Wiley & Sons, 1968. Kenkel, William. "The Relationship Between Status Con- sistency and Political Economic Attitudes." American Sociological Review, XXI (June, 1956), 365-368. Kornhauser, William. The Politics of Mass Society. New York: The Free Press, 1959: . 175 Landsberger, Henry A.; Berrera, Manuel; and Toro, Abel. "The Chilean Labor Union Leader: A Preliminary Report of His Background and Attitudes." Industrial %%%Labor Relations Review, XVII (April, 1964), 399- LaPalombara, Joseph, and weiner, Myron, eds. Political Parties and Political Development. Princeton: Princeton University Press, 1966. Lenin, V. I. "What Is to Be Done?" Essential Works of Lenin. Edited by Henry M. Christman. New York: Bantam Books, 1966. Lenski, Gerhard. "Status Crystallization: A Non-Vertical Dimension of Social Status." American Sociological Review, XIX (August, 1954), 406-413. "Status Inconsistency and the Vote: A Four— Nation Test." American Sociological Review, XXXII (April, 1967) , ml. Lerner, Daniel. The Passing of Traditional Society. New York: The Free Press, 1958. "Toward a Communications Theory of Modernization." Communications and PoliticaTlDevelopment. Edited by Lucian W. Pye. Princeton: Princeton University Press, 1963. Lipset, Seymour Martin. Agrarian_Socialism. Berkeley: University of California Press, i950. . Political Man. New York: Doubleday, 1960. , and Bendix, Reinhardt. Social Mobilipy in Industrial Society. Berkeley: University of Cali- fornia Press, 1958. , and Rokkan, Stein, eds. Part S stems and Voter Alignments. New York: The Free Press, 1967. Mattelart, Armand. Atlas Social de las Communas de Chile. Santiago: Editoriai’dél Paciiico, 1966. Melson, Robert, and Wolpe, Howard. "Modernization and the Politics of Communalism: A Theoretical Perspective." American Political Science Review, LXIV (December, 1976), 1112-1130. 176 Milbrath, Lester. Political Participation. Chicago: Rand McNally, 1965. Morris, James O. Elites, Intellectuals and Consensus A Study oftphe Social Question and the Industrial Relations System in Chile. Cornell International Industrial and Labor Relations Reports, No. 7. Ithaca: New York State School of Industrial and Labor Relations, 1966. Morse, Richard. "Recent Research on Latin American Urbanization." Latin American Research Review, I (Fall, 1965), 35-74. Needler, Martin C. "Political Development and Military Intervention in Latin America." American Political Science Review, LX (September, 1966), 616-626. Nie, Norman H.; Powell, G. Bingham, Jr.; and Prewitt, Kenneth. “Social Structure and Political Participation: Developmental Relations." American Political Science Review, LXII (June, 1969), 36l-378. Olson, Mancur, Jr. "Rapid Economic Growth as a Destabiliz— ing Force." Journal of Economic History, XXIII (December, 1963), 529-552. . Parrish, Charles J.; von Lazar, Arpad J.; and Videla, Jorge Tapie. The Chilean Congressional Election of March 7, 1965. Election Analysis Series, No. 4. Washington: Institute for the Comparative Study of Political Systems, 1967. Petras, James. Politics and Social Forces in Chilean v Development. Berkeley: The University of Cali— fornia Press, 1969. , and Zeitlin, Maurice. "Miners and Agrarian Radicalism." American Sociological Review, XXXII —¥ (August, 1967), 578-586. . "Agrarian Radicalism in Chile." British Journal of Sociology, XIX (September, 1968), 254-270. Portes, Alejandro. "Left Radicalism in Chile: A Test of Three Hypotheses." Comparative Politics, II (January, 1970), 251-274. 177 Przeworski, Adam, and Soares, Glaucio A. D. "Theories in Search of a Curve: A Contextual Interpretation of Left Vote." American Political Science Review, LXV (March, 1971), 51-68. Rabushka, Alvin, and Shepsle, Kenneth A. Po11§ics in Plural Societies. Columbus: Charles E. Merrill, 1972. Runciman, W. G. Relative Deprivation and Sogial Justice. Berkeley: University of California Press, 1966. Schneider, Peter R., and Anne L. "Social Mobilization, Political Institutions, and Political Violence: A Cross-National Analysis." Comparative Political Studies, IV (April, 1971), 69-90. Silvert, Kalman H. Chile: Yesterday and Today. New York: Holt, Rinehart and Winston, 1965. Snow, Peter. "The Political Party Spectrum in Chile." South Atlanticiguarterly, LXII (Autumn, 1963), 474-487. Soares, Glaucio, and Hamblin, Robert L. "Socio—Economic Variables and Voting for the Radical Left: Chile, 1952." American Political Science Review, LXI (December, 1967), 1053—1065. Stokes, Donald E. "Spatial Models of Party Competition." American Political Science Review, LVII (June, 1963), 368-377. Tarrow, Sidney G. Peasant Communism in Southern Italy. New Haven: Yale University Press, 1967. Thiesenhusen, William C. Chile's Experiments in Agrarian Reform. Land Economic Monographs, No. 1. Madison: The University of Wisconsin Press, 1966. U.S. Department of Labor. Labor in Chile. Bureau of Labor Statistics, Report No. 224 (1962). Valenzuela, Arturo. "The Scope of the Chilean Party System." Comparative Politics, IV (January, 1972), 179- 199. Weaver, F. S. "Regional Patterns of Economic Change in Chile: 1950—1964." Unpublished Ph.D. dissertation, Cornell University, 1968. 178 . "Backwash, Spread, and the Chilean State." Studies in Comparative International Development, V (1969-1970), 239-251. Weber, Max. "Class, Status, Party." From Max Weber. Edited by H. H. Gerth and C. Wright Mills. New York: Oxford University Press, 1958. Wolpin, Miles D. "Chile's Left: Structural Factors Inhibit— ing an Electoral Victory in 1970." Journal of Developing Areas, III (January, 1969), 207-230. Zagoria, Donald S. "The Ecology of Peasant Communism in India." American Political Science Review, LXV (March, 1971), 144-160. Zeitlin, Maurice. Revolutionapy Politics and the Cuban Working Class. Princeton: Princeton University Press, 1967. , and Petras, James. "The Working-Class Vote in Chile: Christian Democracy Versus Marxism." British Journal of Sociology, XXI (March, 1970), 16-29 0