SPATIAL mmwmn or POLITICAL PARTICIPATION: GUATEMALA, A CASE STUDY , - Thesis for the Degree of M. A. ' . MOWER"? 3mm UNVERSQY ’ JGHN }. FORD 1971 , : ; .. -‘ L ---A ‘ Jar E . Michigan State l; ' k Universxty {A 'n -. ' A . k 3 5m. “1.--...;.»--—--- n J...— ._ 1.... .. . ’u‘ 3 I "rd-.5. hi SPATIAL INTEGRATION OF POLITICAL PARTICIPATION: GUATEMALA, A CASE STUDY By . "a, v John J. Ford A MASTERS THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Geography 1971 ACKNOWLEDGEMENTS I wish to thank my thesis advisor, Dr. Stanley D. Brunn and my reader, Dr. Robert N. Thomas for the opportunity to work and learn from them and for their helpful criticism of this research. Thanks are also due to Dr. Lawrence M. Sommers, Chairman of the Geography Department for giving me financial assistance and for making available to me the resources of the Geography Department. Finally, I wish to express my deepest appreciation to my wife, Eleanor, who provided encouragement and compansionship throughout this research. ii TABLE OF CONTENTS Page ACKNOWLEDGEMENTS ........................ ii LIST OF TABLES ......................... v LIST OF FIGURES ........................ vi Chapter I. INTRODUCTION ..................... l Statement of Problem ................ l Development of Electoral Geography ......... 2 Overview ..................... 2 Political Sociology of Participation ....... 2 Electoral Research in Latin America ........ 6 Spatial Concepts of Political Participation . . . . ll II. GUATEMALA'S POSITION IN LATIN AMERICA ......... l4 A Survey of Participation in Latin America ..... l4 A Social and Economic Perspective on Guatemala . . . l9 III. PERSPECTIVE ON THE MAJOR GUATEMALAN POLITICAL PARTIES AND THE l966 PRESIDENTIAL ELECTION .......... 34 IV. RESEARCH DESIGN .................... 43 Statement pf Hypotheses ............... 43 Description of Data ................. 43 Basic Outline of Factor Analysis .......... 45 Basic Outline of Regression Analysis ........ 49 Chapter Page Assessment of the Data Normality .......... 5l V. ANALYSIS OF PARTICIPATION IN GUATEMALA ........ 55 Delineation of Aspects of the Socio—Economic Environment ......... . ......... 55 Identification of the Dynamic Correlates of Political Participation ............. 72 Analysis of Overall Participation ........ 72 Analysis of Party Participation ......... 76 VI. CONCLUSIONS AND SUGGESTIONS FOR FUTURE RESEARCH . . . . 84 Discussion ..................... 84 Limitations of the Study .............. 86 Suggestions for Future Research .......... 87 BIBLIOGRAPHY ......................... 89 iv Tables l0. ll. l2. l3. l4. l5. l6. l7. LIST OF TABLES Class Structure of Latin American Countries ...... Social Change Within Guatemala. . ........... Rankings of Social, Economic and Participation ..... Rank Order Coefficient for the Social, Economic and Participation Indices ................ Classification of Selected Countries in Latin America on the Basis of Participation Correlates ...... Percent of Department Vote Given to the Three Major Parties ....................... Rank Orders of Department for Urban Population, Population Change and Party Vote .......... Spearmen's Rank OrderCCoefficient for Parties, Urbanization and Population Change ......... Description of Data .................. Normality of Variables Employed in Study ........ Factor Loading and Communalities ............ Social Statistics and Factor Scores for Department Cabecera ...................... Sum and Sum of Squares of the Regression Model ..... Regression Coefficients for Overall Participation . . . . Regression Coefficient for Political Participation by Party ........................ Regression Coefficient for Political Participation by Party in Selected Areas ............... Averave Coefficients for the Correlates of Political Participation in Guatemala ............. Pages 4 l0 l5, TB 20, 38 39, 42 44, 53 58, 70, 73 74 78, 82 85 21 4O 71 LIST OF FIGURES Figure Page 1 Location Map of Guatemala ............... 26 2 Population Map of Guatemala .............. 28 3 Population Change in Guatemala (l950—l964) ....... 3O 4 FACTOR l ........................ 57 5 FACTOR 2 ........................ 62 6 FACTOR 3 ........................ 65 7 FACTOR 4 ........................ 67 8 FACTOR 5 ........................ 69 vi CHAPTER I INTRODUCTION STATEMENT OF PROBLEM Voting represents the interplay of social and psychological forces both exhibiting spatial characteristics (Kirkpatrick, 1964). In electoral research,problem solving can proceed in one of two ways: a systematic analysis of the psychological and social constraints on voting or an analysis of the spatial variations of these constraints. This paper focuses on the latter while recognizing the importance of the former. Proceeding in a problematic way, this paper attempts to answer the following question: how do aspects of the social and economic en- vironment affect political participation and party preference? In order to systematically answer this question, one basic assumption is made. In a country requiring all eligible citizens to vote, it is assumed that the number of votes represents an index of participation. In the following chapters, political participation is analyzed on a macro-level in all of Latin America and on a micro-level within Guatemala. At both of these levels political participation acts as a function of the social, and economic interactions between the voter and his environment. It is hoped that knowledge of these interactions may help social scientists explain-and predict electoral behavior. The social and economic environment is defined and consolidated by factor analysis. Regression coefficients are calculated to assess the nature and magnitude of the relationships between the environment and participation within Guatemala. Rank order correlation coefficients are used to determine the direction of the relationship between parti- cipation and its correlates on the macro—level. Before an analysis of participation on these two levels effec- tively begins, some prerequisites must be met. One such requirement is a review of relevant literature in political behavior. Two additional demands are brief accounts of the general Latin America and Guatemalan social and economic environments. The former is covered in the following chapter while the latter is treated in subsequent chapters. DEVELOPMENT OF ELECTORAL GEOGRAPHY Overview The first section of the review deals with the political and social literature which is most relevant to political participation. The second section redefines within the Latin American setting many of these political and social concepts. This second section also covers some political aspects of urbanization in Latin America. Finally, the last section places many of the above concepts dealing with participa- tion into a spatial context including location, distance and clustering patterns. Political Sociology of Participation Group belongingness, interest groups and alienation are aspects of social interaction affecting participation. Behavioral scientists have found that the feeling of group membership plays a crucial role in determining individual behavior (Festinger, l947; Cox, l970b; Cox, l97l). In a political context, this desire for group membership is fulfilled by the political party. If the group members are aware of other members of their party, then the group is classified as associa- tional. However, on a large scale such as an urban area, membership in political parties is mainly non-associational (Blankenstein, l968). Interest groups and political parties tend to vary with social class (Burnett et al” l968). Lower classes tend to hold different political attitudes and the members of a class who possess intra-class mobility often hold conflicting attitudes (Berelson, l966; Converse, l958; Beney et al” l956). If the social classes appear to be unevenly distributed and if social status is rigidly defined, for instance by possession of land, then class oriented politics are likely to emerge (Burnett, l968). Where this type of political climate prevails, intra- class movement and interactions are few (Lipset, l959). From Table 1 it is apparent that the social classes within Latin American countries are unevenly distributed. The most popular classes are the urban and rural workers. As the middle class increases, it is expected that they will bridge the chasm between the workers and the entrepreneurs. It is inferred from Table l and the above concepts that the Latin American political environment is indeed class oriented. Political interest groups are spatially dispersed. In Guatemala and some other parts of Latin America, the lower class interest groups find support in areas of rural as well as urban de- 4 TABLE I CLASS STRUCTURE OF LATIN AMERICAN COUNTRIES URBAN URBAN LANDOWNERS RURAL COUNTRY MIDDLE CLASS WORKERS ENTREPRENEURS WORKERS (%) (%) (%) (%) Argentina 28 45 8 l9 Chile 21 50 l 28 Venezuela l6 45 2 37 Cuba 2l 38 - 4l Ecuador l0 38 l 5l Panama l5 3l l 53 Costa Rica l4 3l 8 47 Paraguay l2 33 2 53 Colombia l2 32 lO 46 Brazil l3 24 2 6l Guatemala 6 3l 2 6l Bolivia 7 20 l 72 El Salvador 9 27 2 62 Honduras 4 l2 - 84 Haiti 2 12 l 85 SOURCE: Gino Germani, l965, “Estrategia para estimular la movili ad social," in J. A. Kahl (ed.) La Industrializacion en America Latina, Mexico-Buenos Aires: pp. 294-295. Fondo de Cultura Economica, privation (Byron, l968). The concept of political support arising from economic and social determinants indicates the non-applicability of much American political science postulates within the Latin American setting.1 Thus the electoral researcher possesses a valuable research goal, for he must assess which concepts and theories proposed within the American context explain political behavior within the Latin setting. In general, the political interest groups within an area in- fluence an individual's participation rates (Billet, l968). This type of group transmits and filters information about the candidate to the individual (Lazarsfield et ald l948). The individual response in a political context represents a decision along a continuum. Along one part is a set of institutionally derived alternatives arising from the interest groups and the other is a set of alternatives derived from individual experience (Rokkan, 1962). When political alternatives and interest group pressure confuse the voter, the individual may respond negatively. One of the most common reactions in such a situation is a feeling of political inefficacy. Political alienation may then soon result (Thompson et ald 1960). Other factors contributing to alienation are powerlessness, norm conflict and social isolation (Dean, l960). Alienation is the most common among the lower social classes (Thompson et aL, l960). Protest voting or invalidating a ballot are symptoms of political apathy or alienation 1The reader is referred to two of V. 0. Key's major works. One of the dominant themes of his research is the idea that political groups are independent of any social determinants. The two works in which this theme is developed are: Southern Politics in State and Nation (l949), New York: A. A. Knopf, and "Social Determinism and Electoral Decision: The Case of Indiana," in E. Burdick (ed.), American Voting Behavior (l959), Glencoe, Ill.: Free Press, pp. 220-232, within many Latin America countries. Many times this is done so that the victory is considered hollow for the opposition (Gomez, l962; Thompson et al” l960). Another form of vote protest is voting for a 'dark-horse' candidate (McDonald, l967a). In a very real sense then, there are segments within an electorate who are confused and unable to make a decision; a situation which results in their being politically alienated (McDonald, l967a). At the opposite end of the political spectrum are the politically aware individuals who become more involved in the governmental affairs and easily identify their goals and atti— tudes with the national government. Uneven participation patterns among individuals and their interest groups serve to increase the political autonomy of the interest groups (McDonald, l969c). If one assumes that the alienated voter is symptomatic of an 'unhealthy' social and economic climate within a country, then an aware voter exists in a 'healthy' environment. Some agreement exists among political scientists who have worked in Latin America that awareness and alienation are dependent upon factors such as literacy, education, status and income (Gomez, l962; Adams, l968; Needler, l966). As a rule, the more urban, literate, educated and wealthier areas are more politically aware and consequently more politically active (Lerner, l958; Horowitz, 1967). Thus economic development is closely connected to the political integration within a country and to a lessening of political alienation (Lipset, l959; Fitzgibbon, 1967; McDonald, l969a). Electoral Research in Latin America Historically, voting analysts in Latin America have questioned two themes which presently continue to be debated within their electoral research. The first theme relates political development and economic integration. That is the political sphere needs to expand as rapidly as the economic sphere and if it does not, political inequities such as minority disenfranchisement may arise (Lipset, l959; Needler, l968). This dissimilarity may solidify into major issues inhibiting the estab- lishment of effective communication between the parties. Likewise econo- mic growth promotes the politics of 'rising expectation'. Within Latin America the failure of certain segments of the population to realize these expectations has prompted many to express their frustration as overt political dissent (McDonald, l969a; Horowitz, l967). The second theme of electoral research in Latin America empha- sizes the change in political behavior occurring in the migrant. The exact nature of the migrants' political metamorphosis is often disputed. Some studies claim that the migrants change their political attitudes and increase their political interest once they live in an urban area (McDonald, l969c). Here the migrants develop a political consciousness and increase their awareness (Flinn et al” l969). As a result, the political participation of the area is increased (Mangin, l967; Alschuler, 1967; Micklin, l969; McDonald, l969a). In addition, this politicalization of the migrant induces political activity which may lead to the overthrow of the incumbent powers (Needler, 1968). Other studies dealing with this same topic are less optimistic and stress the political alienation and anomie confronting the migrant. Within the squatter settlements of Lima, the migrants are unaware of their political potential and view a solution to their economic problems as coming from outside of their interest group (Mangin, 1967). Political anomie and alienation tend to discourage immediate political participa- tion (McDonald, l967a; Byron, 1968; Flinn et a1“ 1969). Most studies dealing with this theme do tend to aggree that some form of increased participation does occur in areas undergoing strong in—migration (Horowitz, 1967), especially in those countries of Latin America exhi- biting a pronounced urbanization like Guatemala and Brazil (Wingo, 1967). This politicalization occurring in the migrants is most pronounced in the second generation urban residents (Mangin, 1967). If the poli- ticalization of the urban migrant takes the form of a radicalization, then the urban area may become a focus of national political dissent (McDonald, l969a). The rise of industrialization and urbanization in Latin America within the past twenty years has challenged the national political parties of many countries. Within this region, the political parties are interest aggregated and not areally aggregated (Blankenstein, 1959). Due to their class oriented politics, these traditional political parties, composed of a coalition of landed and commercial interests, find them- selves politically ineffective (Burnett, 1968; Horowitz, 1967; Adams, 1968). The rise of the national revolutionary parties, stressing a rural-urban lower class coalition, is proof of the frustration arising from mis- directed and ineffective political leadership of the traditional poli- tical parties (Alexander, 1965; Burnett, 1968). The rise of the urban- rural proletariat coalition in Chile emphasized the fact that no one class or region can dominate the Chilean political process (Cope, 1966). Within Guatemala, the urban-rural coalition exists but a lack of a common goal hinders effective political action (Gillin, 1956). Within Colombia the urban-rural coalition had as its goal a feeling of political dis- enfranchisement (McDonald, l967b). If the political parties adopt revolutionary schemes, political change may be peacefully effected (Horowitz, 1967). In addition to this adoption, political change depends also on the rate of social change occurring in society. In Latin America, this type of change normally progresses very slowly (Micklin, 1969). One reason for this tardiness is the military and the Catholic church, two interest groups pledged to forestall radical change and maintain the status quo (Needler, 1966; Alexander, 1965; Blankenstein, 1959). Table 2 presents an assessment of social change in Guatemala. The two above themes in Latin American electoral politics under- lie research analyzing recent elections. Alienation in Colombia was evident in the protest votes cast in a recent election (McDonald, 1967a). Participation in Chile has been retarded due to the ambivalence of the political elites (Cope, 1966; 1968). The alienation of the masses in Chile is evident in the votes cast for radical candidates (Soares et a1, 1967). In Guatemala, ideological conditioning has led the people to expect social and economic reforms. Numerous minor parties are formed as a response to the need for reforms. However, historically their hopes have been betrayed and as a result the political system is frowned upon by the masses (Johnson, 1967; Sloan, 1968). 10 TABLE 2 SOCIAL CHANGE WITHIN GUATEMALA YEAR 1921 1 YEAR 1950 YEAR 1964 INDICATORS OF ABSOLUTE RELATIVE ABSOLUTE RELATIVE ABSOLUTE RELATIVE SOCIAL CHANGE CHANGE CHANGE CHANGE CHANGE CHANGE CHANGE (7°) (23) W (7°) (74) (%) Urbanization 7.4 1.7 11.1 2.8 15.5 2.5 (Number of people living in urban areas with popu- lation greater than 20,000) Industrialization - - 25.0 .9 28.0 2.3 (Percent of males not engaged in agriculture) Education 13.0 4.5 30.0 1.7 37.0 4.2 (Percent of population seven or older who are literate) SOURCE: (Micklin, 1969) 1The relative change is calculated on the basis of the population growth during the period in question. 11 Spatial Concepts of Political Participation Some electoral research relates variation in political attitudes to changes of location and information (Cox et a1, l968b; Cox, l969a). The flow of information is affected by an absolute change in location, for instance, migration as well as a relative change in location which may occur after a person receives higher education (Cox, 1968a). Within the city of London, an absolute change in location produced among some migrants higher participation rates (Cox, 1968c). The spatial structure of an area's communication network affects the formation of politically meaningful information for the individual or the group (Cox, l968a).1 Often the network is poorly constructed and the area received outdated information. In many areas of Guatemala, as well as parts of Latin America, there are villages situated on a partially ineffective communication network. This position encourages the formation of political localism and it is most prevalent in those areas where the individual is exposed to a limited amount of politically important infor- mation (Cox, l969b). Distance, in addition to location, is a factor of importance when analyzing participation. An individual's sphere of influence is most pronounced around his immediate surrounding area and it decreases as dis- tance increases (Reynolds,l969). The votes cast for a candidate often exhibit a distance-decay function from the candidate's home (McCarty, 1960; Reynolds, 1969). The decay function may vary particularly when political 1Communication network as used in this paper refers to the flow of information due to the spatial spread of newspapers, radio, people and products. 12 boundaries are crossed for regional candidates (Birdsall, 1969). Un- doubtedly, the distance-decay of votes changes as its socioeconomic determinants vary (Rumage and Roberts, 1965). Areal variations in social, demographic and economic factors give impetus to regional voting patterns. Some of the most valuable demographic and economic indices used to study these patterns are: population density, sex and age ratios, occupations, income, education, religion, home owner- ship and value of property (Dean, 1949; Simmons, 1962; Van Duzer, 1962; Burghardt, 1964; Brunn et a1, 1969; Brunn et aL, 1970a; Ingalls, 1970). One social indicator influencing political vote is membership in voluntary organizations (Cox, 1970b). In most of these studies the votes for a par- ticular candidate or issue are viewed as a function of a group's socio- economic characteristics. The level of aggregation of the above studies often is crucial. If the data are on a highly aggregated level, conclusions must be confined merely to generalizations concerning the underlying associations (Kasperson, 1965). For this reason it is best to analyze voting data on the lowest procurable level of aggregation (Nelson, 1952; Prescott, 1959). A cartographic analysis of voting patterns and their associations rmaveals many underlying facets of the voting response (Prescott, 1969). lhwalysis of the variation and the clustering of these patterns may proceed frcun one of two possible methods. First, the clustering may be viewed froni a differential appeal of political parties in particular elections. In SLJCh a case, the variation may occur along territorial lines which are crften most pronounced in developing countries (Lipset, 1964). 13 Second, the clustering may occur due to the recent emergence of stage migration, industry and transportation development. In this situation, the variations are functionally defined and they are most prevalent in more developed nations (Cox, 1970b). The second approach is assumed in this paper and it is hoped that the factor analytical model provides the functional definition. 14 CHAPTER II GUATEMALA‘S POSITION IN LATIN AMERICA A SURVEY OF PARTICIPATION IN LATIN AMERICA The basic purpose of this chapter is to provide an overview of political participation in the Latin countries. In addition, this chapter develops some underlying relationships which are subsequently integrated into the analysis of political participation in Guatemala. It is assumed for this analysis that voter registration figures represent an index of participation. Although every registered voter does not vote, in most cases it is assumed that the majority do. Therefore, for the lack of a better index, registration is chosen. The final objective of this chapter is to answer the following research question: in regards to participation and its correlates, does Guatemala represent the typical Latin American country? In order to sufficiently answer this question, all of the countries were ranked on the basis of ten social and economic indices. These ranks are then compared with a rank of countries based on participation. The comparison is accomplished through the use of Spearmen's rank-order correlation coefficient.1 The basic advantage of the rank-order coef- ficient is its distribution-free tendency. Table 3 lists the variables and the ranks for each country. 1The Spearmen's "rho” coefficient is equal to the squared dif- ference between the ranks which are then summed and divided by the number of observations. The value is then standardized so that it has a range between +1.00 and -1.00. RANKINGS ON SOCIAL, ECONOMIC AND PARTICIPATION INDICES 15 TABLE 3 COUNTRY VARIABLES 1 2 3 4 5 6 7 8 9 10 ll ARGENTINA l9 17 18 l 2 2 2 2 22 2 4 BOLIVIA 16 12 20 8 13 18 5 4 22 6 2 BRAZIL 4 9 6 5 12 11 3 3 22 12 22 CHILE l7 18 14 2 3 3 6 4 4- 9 22 COLUMBIA 9 14 2 2 10 7 14 l 22 7 3 COSTA RICA 2 4 4 2 9 3 4 2 2 4 11 CUBA 18 16 12 2 4 4 7 22 22 22 l DOM. REP. l l 8 22 15 12 13 22 22 22 8 ECUADOR 7 8 7 7 5 9 l7 6 22 15 12 EL SALV. 15 5 ll 7 14 14 18 3 22 13 22 GUATEMALA 5 7 9 8 17 17 16 4 22 16 7 FMITI 13 10 19 10 19 19 20 1 22 22 13 HONDURAS 6 2 l3 9 18 16 19 22 22 10 5 MEXICO ll 13 5 3 8 8 15 22 5 11 22 NICARAGUA 8 3 10 22 16 15 10 22 22 22 13 PANAMA 14 6 15 5 7 5 8 7 3 5 10 PARAGUAY 12 ll 11 6 6 6 12 3 22 14 6 PERU 10 15 3 22 ll 13 ll 22 22 8 9 URUGUAY 20 20 16 22 l 1 1 22 l 3 22 VENEZUELA 3 19 1 4 4 10 9 3 22 1 5 SOURCE: Germani, 1965; Horowitz, 1967. 16 TABLE 3 (cont'd.) VARIABLE NUMBER DEFINITION 1 Urban growth 2 Rural growth 3 Percent internal migration 4. Percent of urban population in the middle class 5 Percent of population engaged in secondary or tertiary economic activities 6 Percent of population who are literate 7 Percent of population who live in metropolitan area 8 Percent of population who are landowners 9 Voters registered in a democratic country1 10 Voters registered in all countries 11 Number of civil disorders 22 Missing or incomplete data 1 The following countries were excluded from the rank in variable 10: Cuba, Dominican Republican, and Haiti. This was done due to the totalitarian nature of their political system which tended to inflate figures. 17 Based upon the pertinent studies treated and the above voting data and methodology, the following hypotheses are formulated: (1) If the ranks of percent voters and percent internal migra- tion are in close agreement, then those countries ranking in the top quartile have had migrants participating in elections. (2) If the ranks of percent voters and percent metropolitan population in urban areas are in close agreement, then those countries in the top quartile have politically primate metro- politan areas. (3) If the ranks of percent voters and percent land-owners are in close agreement, then those countries in the top quartile are labelled as politically stable. (4) If the ranks of percent voters in a democratic country is in close agreement with the ranks of percent literate, percent of population in secondary and tertiary activities and the percent middle class, then those countries ranking in the top quartile have a democracy based upon a well integrated, viable middle class. (5) If the ranks of percent voters and the countries experiencing a high percentage of civil disorders are in close agreement, then those countries ranking in the top quartile have had increased participation associated with a high degree of civil disturbances. From Table 4 it is possible to infer the direction of the associa- tion between the various ranks. Thus, the rank internal migration is in close disagreement with voter registration. Research mentioned in the previous chapter supports this negative relationship existing at least in the initial years. The negative relationship between the ranks percent 1The literature on political activity in Latin America supports the idea that people with fixed social mobility participate more than those with fluctuating mobility patterns (Lipset, 1959; Nash, 1958). It is assumed that people who own land have fixed social mobility. At least the patterns of these people are more stationary than migrants who have recently moved into the same area. 2A democratic country is defined as a country having a moderate degree of elected political succession. The classification was used by Horowitz (1967). 18 TABLE 4 RANK ORDER COEFFICIENT FOR THE SOCIAL, ECONOMIC AND PARTICIPATION INDICES VARIABLE RANK "RHO" NUMBER OF SIGNIFICANCE COEFFICIENT COUNTRIES LEVEL Percent Internal -.O9 16 .35 Migration Percent Urban -.19 16 .23 Percent Landowners .25 12 .20 Percent Population .47 (.82)* 16 (5) .03 (.04) Literate Percent Employed in .47 (.30)* 16 (5) .03 (.31) Secondary and Tertiary Activities Percent Population .56 (.21)* 14 (4) .01 (.39) in Middle Class Percent Population .65 16 .001 in Metropolitan Area Deaths from Group .30 16 .12 Violence *For the figures in parenthesis the dependent variable was voters registered in a democratic country. SOURCE: calculated by the author, 19 urban and voter participation or registration is misleading and there- fore nothing is inferred. The historical influence of the landed aristocracy in Latin politics remains dominant in at least twelve countries as an element favoring increased participation. Finally, the ranks of participation are in agreement with the ranks measuring educa- tional level and economic activity. Also the most cosmopolite city favors increased participation. However, these increases in partici- pation are costly as evidenced by the subsequent increase in deaths from group violence occurring in many of these countries. As evident from Table 5, Guatemala is far from a typical Latin American country. On four of the six indices, Guatemala and El Salvador placed the lowest. It appears that Guatemala and El Salvador are unique countries politically, and that generalizations concerning participation within the typical Latin setting may not apply. Thus the transferability of inferences about participation is almost non-existant. A SOCIAL AND ECONOMIC PERSPECTIVE ON GUATEMALA For this study the following criteria are deemed as important in the choice of Guatemala as a test case: (1) The country must legislate voter participation, thus forcing the voter to become involved; (2) The country has had numerous social and political up- heavals within the past twenty years which demand a nation- wide adjustment to new claims of status and power (Lipset, 1964); (3) The country must have recognizable class cleavages based on economic or ethnic lines (Nash, 1958; Benney et a1, 1956); (4) The country must be undergoing economic development (Benney et al., 1956); CLASSIFICATION OF SELECTED COUNTRIES IN LATIN AMERICA 20 TABLE 5 ON THE BASIS PARTICIPATION CORRELATES RELATIONSHIP DIRECTION COUNTRIES COUNTRIES COUNTRIES WHERE THE WHERE THE WHERE THE (CONCEPT IT RELATIONSHIP RELATIONSHIP RELATIONSHIP SUPPORTS) IS THE 1 IS THE 2 IS THE 3 STRONGEST AVERAGE WEAKEST Landowners and positive Costa Rica Brazil Guatemala participation (higher Argentina participation among pe0p1e who are economically stable) Literacy and par- positive Argentina Columbia El Salvador ticipation (higher Uruguay Mexico Guatemala participation re- Costa Rica Brazil quires increased Panama literacy) Secondary and terti- positive Venezuela Panama El Salvador ary activities and Argentina Columbia Guatemala participation (higher Uruguay Peru participation re- Mexico quires economic Brazil integration) Percent of nationals positive Argentina Panama El Salvador living in primate Uruguay Bolivia Guatemala city and partici- Costa Rica Peru pation (higher par- Chile ticipation is found in the most cosmo- politan areas) Deaths from group positive Argentina Panama ----------- violence and par- Peru ticipation (the Honduras urban area is the heart of political dissent) 21 TABLE 5 (cont'd) Percent of popu- positive Argentina Panama --------- lation in middle Costa Rica Bolivia class (higher par- Mexico ticipation encour- Brazil ages the growth of a middle class) 1Countries in the upper one fourth of both ranks. 2Countries in the middle one-half of the ranks. 3Countries in the bottom one fourth of both ranks 4Represents a borderline case 5If a country is not listed in one of the three columns, then it belonged to two different ranks or data were unavailable. Therefore that country was not classified. SOURCE: calculated by author. 22 (5) The country must have readily available and reliable data on social and voting activities; in addition, the elected officials must have a degree of political legitimacy (Lipset, 1959; Cope, 1966), and, (6) The country must possess a limited degree of national integration; the latter is judged by the political and economic cleavages which are recognizable within the country (Sloan, 1970). In Guatemala economic development and national integration are closely related. The Indian population is a good example of this relationship. They represent the largest cultural group resistant to the forces of national integration and economic advancement (Sloan, 1970). They reside primarily in the Northwest Highlamrswhere they are able to provide for themselves and their families a subsistence level of income (James, 1969). The Indians are integrated into the economic system principally through the growing of fresh foods which they market in the urban areas. Their position in the labor force is generally below that of a lggjng_ who may perform a similar task (Nash, 1958).1 Employment opportunities for the Indian have political signifi- cance in that these often bring him into direct contact with individuals from a variety of socio-economic levels and above all into contact with the government and its regulations. The selling of fresh produce and handicrafts in the large urban centers requires that they pay a form of sales tax. The employment opportunities of the Indian males often center around the commercial farms on the south coast. Here their 1In most Central American countries the term ladino rather than mestizo is used to describe the Indian-Caucasian mixture. In general, the ladino is far more urban oriented than his Indian counterpart. See text for more explanations. 23 working conditions, likened to the pre-1900 factory conditions in the United States, represent abuses by the farm owners. Thus, the economic activities of the Indian illustrates how the power and the influence of the national government varies among social classes and between the rural and urban areas of the country (Converse, 1958). This variation of influence and activity fosters the growth of political sectionalism which in turn affects the participation rates especially on the party level. Furthermore, the economic status of the Indian in Guatemala tends to substantiate earlier claims that Guatemala in microcosm exemplifies the characteristics of the developing world (Gillin and Silvert, 1958). The lagjng_on the otherhand represents a somewhat different socio—economic system. Although the ladinos are quite ubiquitous, they tend to concentrate in the urban areas. These descendants of former Spanish settlers or Indians have abandoned their traditional life styhs. This change of life styles is called ladinization and it is well documented in each ensuing census in Guatemala. Fewer and fewer people are freely admitting that they are Indian. One major requirement for this successful transformation of life styles is a change not only of language but also of location (Van de Berghe, 1968). Thus this internal migration plays a crucial role in the politicalization of the Indians. The Indian and the lagj§9_perception of the national government differs (Nash, 1958). The lagjng_is a national in a very broad sense, frequently participating in governmental functions. In addition, the lggiflg_has historically maintained his allegiance to the national government. A form of sentimentalism directed towards the incumbent 24 leader cements this allegiance (Nash, 1958). At the other extreme is the Indian who reluctantly supports only the local government which he views as a compromise between the national entity and no formal structure at all. Making the Indian an active member of larger political organization is a difficult conversion due to his provincial cultural values (Nash, 1958). The foreign born population is normally found in the larger urban areas. However, this discussion of political participation ignores the contributions of the foreign born population since they represent such a small fraction of the total population, less than three percent in 1964. Also, not being indigenous, they are unable to vote in the national elections. The highest population densities in the country are found in the western and central sections of the country (Figuragl and 2). Here major urban areas occupy the numerous valleys found within the highlands. Lower population densities are found in the northeast, southeast and coastal lowlands. Two exceptions to the above generalizations are the departments of Izabal and Escuintla which have undergone rapid population change (Figures 2 and 3). In general, the migration to Escuintla represents a rural-rural movement while the migration to Izabal is both a rural—rural and a rural-urban variety which reflects the increased importance of Puerto Barrios (Thomas, 1968). The migration to urban areas in developing countries carries with it some meaningful political ramifications. Information channels, social class consciousness and the social organization patterns of the 25 Figure 1. Location Map of Guatemala. SOURCE: Thomas, 1969. MEXICO OUICIK [ W. V!“ l I) / (“00.le OCEAN 27 Figure 2. Population map of Guatemala. SOURCE: Thomas, 1969. 28 ./ Ono Dot Roproutm Ono Thousand Puma, 29 Figure 3. Population Change in Guatemala, 1950-1969. SOURCE: Thomas, 1969. 31 migrants are some of these political considerations. Urban migrants tend to lack a familiarity with the new life styles, and frequently find themselves possessing limited skills which are already a surplus in an urban area. Normally migrants perceive only a limited portion of politically important issues, often neglecting national issues while supporting local issues (Byron, 1968). In urban areas with high in-migration, social and political adjust- ments to the new status, power and prestige are frequently required so that the migrants can become politically effective (Lipset, 1964). Political reformers may capitalize on the migrant's disenfranchisement by developing ideas and platforms localized for a particular locality. New migrants to the city mainly relocate in urban areas for increased economic opportunity. Hopefully, their social mobility will increase with this movement. However, a Finnish study recently discovered that people aspiring for higher social status have lower political partici- pation rates than their contemporaries with a more stable social status (Pertti, 1964). Such may be the case in Guatemala. Major receiving areas for the migrants are the urban centers. A hierarchy of foci exists with the large departmental capitals function- ing as the receiving points. Thus, Guatemala City and the other large department capitals especially serve as political core areas for the country (McDonald, 1969a). Although the core areas are technologically advanced and possess higher standards of living than the rural areas, they often remain areas of political instability, thus contributing to intra- party hostility and competition (McDonald, l969a; Johnson, 1967). 32 A tenet of political theory in developing countries states that the urban environment fosters political organization, particularly among the lower income groups. Political organizations are fostered when “re-occurring daily situations arise which place the lower class in a position to perceive of themselves as a group in opposition to others" (Byron, 1968). Within the urban slums of Guatemala City, the weak and informal political organization discourages the formation of this perception (Byron, 1968). One of the most pervasive forces of urban political activity in Guatemala is the politics of dissent found among the young people of the middle and lower classes. One contributing cause of this type of politics is the economic isolation of the more traditional rural areas from the urban areas which tends to foster the identification of urban politics as the accepted national politics of the country (McDonald, 1969a). Thus one finds again a situation where political sectionalism functions as a response to the uneven levels of economic development and inte- gration within a country (Gil, 1966). Urbanization and the benefits derived from it, i.e., 1 However, education and communication, foster political cosmopolitanism. political cosmopolitanism does not imply political acceptance. Cosmopolitanism in the urban area contributes to the development of new ideologies and reform programs. Once reforms are accepted by the people 1The reader is referred to Everett M. Rogers' book on the Diffusion gf_1nnovations (Glencoe: Free Press, 1962) for a discussion of cosmopolitanism. He defines it as the "degree to which an indi- vidual's orientation is external to a particular social system," p. 17. 33 in the most cosmopolitan area, normally the largest city, these ideas or reforms diffuse to other areas. It is in the rural areas that support for the new reforms or ideologies may have a detrimental effect upon the level of national governmental acceptance and participation. That is, the spatial spread of the ideology of reform may only serve to widen the already existing social and economic differences within a developing country. Political participation cannot be divorced from the social, demographic, and economic elements which together shape the overall cultural patterns of the country. Occupation, race, social class, sex and education are a few of these factors which may influence the rates of participation. If the political milieu is intertwined with the economic and social environment and if there are easily recognizeable dif- ferences in the distribution of the selected indices of this environ- ment, then an analysis of political participation in Guatemala will reflect these differences. Moreover, if a country like Guatemala possesses political parties which are ideologically oriented to a specific class or geographic area, then analysis of participation and rates by party should effect a similar socio—economic bias. This may be one reason why new political parties are often formed in the Latin urban area. If the urban areas in Guatemala are located at the nodes of the information flow, then political participation and parti- cipation for the youngest political party (PR) will be highest in the urban socio-economic setting. 34 CHAPTER III PERSPECTIVE ON THE MAJOR GUATEMALAN POLITICAL PARTIES AND THE 1966 PRESIDENTIAL ELECTION The indices of participation consistently employed throughout this analysis are derived from the 1966 presidential election. One such index is the ratio of total votes cast to the total population of the municipio.1 Therefore a discussion of the election laws and party alignment and composition aids in the understanding and the recognition of the political elements which may directly influence participation. In addition, a discussion of the major issues and party platforms con- tributes to the explanation of the differential participation rates. Like many other Latin countries, Guatemala has undergone a high degree of political instability. An indentification of all the elements contributing to the instability is difficult. However, one such element is the political and economic primacy of the capital city. Two other factors are governmental corruption and tax inequity. Finally, there are vested commercial and landed interests that often exert some control over the elected officials who inspire political extremism among the masses (Johnson, 1967). 1A municipio is the minor civil division in Guatemala, comparable to a county or a large township in the United States. All together there are some 327 of these divisions but the present research pertains only to 323. Four municipios are excluded due to a deficiency of consistent data for them. 35 Confronted by an historical tendency of instability and mass political deprivation, political parties in Guatemala arose. Party proliferation is considered a favorite political pastime particularly around election time (Cehelskey, 1966). However, the constitution limits the number of parties permitted to participate in any national election by requiring each one to have 50,000 'bona fide' members, twenty percent who must be literate. In the 1966 election, the reform oriented and anti-military Christian Democrats were excluded from the election because they failed to meet these requirements (Johnson, 1967). Every literate male and female eighteen years and over is required to vote and for the illiterate, it is an option (Cehelskey, l966). Historically, women in Guatemala voted less than men. One reason for this behavior rests with the cultural values which frowns upon political activity on the part of women in Guatemala (Johnson, 1967). The total number of eligible voters is approximately 46.5% of the population or some 1,950,000 people. This national figure is less than the number of voters permitted to vote in the local elections. The difference is equal to the number of foreigners who are not permitted to vote in the national election. The main issues of the 1966 election were political and military in nature and only partly economic. The foremost issue was the restora- tion of a constitutional government which had been absent for the past three years. Since the 1963 coup overthrew the constitutionally elected president, Ydigoras Fuentens, a military government was conducted under the rule of Colonel Enrique Peralta Azurdia. The second issue of the 36 campaign was the reduction of terrorism and the protection of the civilian population (Cehelskey, 1966). The overall role of the military in the government was an issue only with the more conservative parties. National investment incentives and regulations of foreign companies were less important (Cehelskey, 1966). These salient issues of the 1966 election did not significantly affect voter turnout (Cehelskey, 1966). One explanation for this may rest with the mandated participation. The major determinant of turn- out is related to the amount of freedom with which the election is conducted (Sloan, 1970). It was predicted prior to the election that the freer the election, the better the chances the liberals had of winning (Eder, 1966). Since the liberals did win, this election can be viewed as a relatively free election. In response to the various issues three legally recognized parties presented their platforms. They were Partido Institutional Democratico (PID); the Movimento gg_Liberacion Nacional (MLN); and the Partido Revolucionario (PR). The PID was the incumbent party and it represented the political center of the ideological continuum. It stressed a rebirth of economic nationalism so that private investment would be encouraged (Cehelskey, 1966). For this they won the support of the urban upper classes. In order to win the support of the rural electorate, composed of Indians and the landed aristocracy, the PID emphasized governmental decentralization, road building and land resettlement programs. Urban terrorism, a major issue with the other parties, was downplayed by this party since it was politically respon- sible (Sloan, 1968; Johnson, 1967). 37 The political left of the 1966 election was occupied by the PR composed of reform-minded politicians, the party directed its campaign at the urban areas. The platform stressed the reduction of the military, literacy campaigns, tax reform, consumer and labor cooperatives and a welfare program. Many of the PR's labor reforms and social welfare ideas belonged originally to the Christian Democrats who failed to gain legal recognition. Thus the supporters of the only socialistic party permitted in the election represented a fusion of businessmen, professionals, laborers and students (Cehelskey, 1966). As evident from Table 6, the coalition politics of the PR eventually won the election. Party officials were elated at the support given the party during its first real political battle. They were also optimistic about future growth due to the support given the party in areas of urban migration (Johnson, 1967) (Table 7). The political right of the election was occupied by the MLN. Their major campaign issue was the reduction of terrorism and the elimination of communist insurgency in the rural areas. The MLN drew most of its support from the military officers, clerical personnel and large land owners. Areally the MLN received tacit support in both rural and urban departments but failed to gain a majority in either (Johnson, 1967). It was expected that urbanization would divide the Guatemala electorate into groups favoring or opposing the dominance of the military in governmental affairs (Johnson, 1967). From Table 7 it is 38 TABLE 6 PERCENT OF DEPARTMENT VOTE GIVEN TO THE THREE MAJOR PARTIES DEPARTMENT PR (%) PID (%) MLN (7) Guatemala 62058 55.1 20667 18.3 29997 26.6 E1 Progresso 2161 34.4 2057 32.7 2158 32.8 Sacatepéquez 4012 42.7 2710 28.9 2676 28.4 Chimaltenango 4927 34.0 5560 38.3 3995 27.5 Escuintla 14050 61.8 4206 18.5 4476 19.6 Santa Rosa 6416 41.0 5409 34.5 3819 24.4 561615 3766 35.5 5054 47.7 1770 16.7 Totonicapan 2207 18.3 7074 58.3 2881 23.4 Quezaltenango 12927 42.3 12029 39.5 5558 18.2 Suchitepéquez 13251 57.1 5795 25.1 4131 17.9 Retalhuleu 6109 53.6 3714 32.6 1584 13.9 San Marcos 17487 52.4 10393 31.1 5453 16.3 Huehuetenango 7228 29.7 10268 42.2 6793 27.9 Quiche, 2288 18.3 6674 53.2 3561 28.4 Baja Verapaz 2830 31.2 4074 44.9 2178 24.0 Alta Varapaz 4790 27.9 8418 4911 3912 22.9 Peten 1574 41.6 1547 21.0 659 17.4 Izabal 4266 44.7 3183 33.3 2102 22.1 Zacapa 4136 43.3 3309 34.7 2101 22.1 Chiquimula 5633 37.3 3866 25.6 5627 37.2 Jalapa 3085 41.4 2834 38.0 1536 20.6 Jutiapa 5698 29.5 7443 38.6 6136 31.9 ‘ SOURCE: Johnson, 1967. ‘— 39 TABLE 7 RANK ORDERS OF DEPARTMENT FOR URBAN POPULATION, POPULATION CHANGE AND PARTY VOTE DEPARTMENT PR PID MLN Percent Percent Urban Population Change Guatemala 3 22 8 l 3 El Progresso 15 15 2 9 15 Sacatepéquez 8 18 4. 2 20 Chimaltenango 16 10 7 4 18 Escuintla 1 21 l6 l3 1 Santa Rosa 12 13 9 15 12 Solola/ 14 4 20 6 21 Totonicapan 21. 1 ll 19 14 Quezaltenango 9 8 l7 5 7 Suchitepequez 2 20 18 7 6 Retalhuleu 4 16 22 ll 4 San Marcos 5 17 21 21 ll Huehuetenango 18 6 6 18 10 Quiche/ 21. 2 4. 20 13 Baja Verapaz l7 5 10 17 8 Alta Verapaz 20 3 12 22 17 Pete/n 10 7 19 3 5 Izabal 6 l4 l3. l2 2 Zacapa 7 12 13. 8 l6 Chiquimula l3 19 l 14 19 Jalapa 11 ll 15 10 22 Jutiapa l9 9 3 16 9 40 TABLE 7 (cont'd) SOURCE: calculated by the author. 41 clear that PR and the ranks of urbanized departments are close in agreement. Likewise the ranks of MLN and PID votes and urbanization approach disagreement. Similar results are obtained in a subsequent chapter dealing with the party participation at the municipio level. The 1966 election in Guatemala substantiated the existance of party ideological cleavages. Similar cleavages were recognized in Colombia between the conservatives, liberals and leftists on one side and the rightistsand fascists on the other (Johnson, 1965). According to electoral analysts, a pre-election compromise was established between the PID and PR which would have eliminated the MLN from entering the governmental machinery (Sloan, 1968). As evident from Table 8, this compromise covered a wide ideological gap. Perhaps a better compromise could have been established between the MLN and PID. Undoubtedly such a coalition is ideologically and statistically sound. As evident from the results, the PR constituency was relatively independent of the MLN and PID constituency. These last two parties robbed each other of votes. From Table 8 it is empirically verified that the PR relied more upon the urban areas than the other parties. In addition this party was dependent upon large changes in population. Those areas in Guatemala which have undergone the most pronounced population changes are also the most urban. 42 TABLE 8 SPEARMEN'S RANK ORDER COEFFICIENT FOR PARTIES, URBANIZATION AND POPULATION CHANGE PR PID MLN URBAN POP. CHANGE PR 1.00 PID - .51 1.00 MLN - .57 + .62 1.00 URBAN + .63 - .21 - .25 1.00 POP. CHANGE + .75 + .09 - .05 + .67 1.00 SOURCE: calculated by the author. 43 CHAPTER IV RESEARCH DESIGN STATEMENT OF HYPOTHESES From the previously cited literature and analysis, it is concluded that urban areas in Latin America stimulate economic viability, produce cultural invigoration and act as destination points in the migra- tion flow. In addition the urban environment tends to realign the politi- cal activities of the migrant (Horowitz, 1967). Since the urban areas within Guatemala perform these functions, then they will also very likely stimulate political participation. Distance affects the diffusion of political information and as distance from urban areas increases, the intensity of the information flow tends to decrease (McDonald, l969a). Due to their location the urban areas often act as the centers of the spatial spread of information. Description of Data The variables were chosen for this study on the basis of their level of aggregation, overall relevance to the socio-economic environment, and availability. A total of fourteen variables were collected for 323 municipios (Table 9). Excluded from the list of possible variables were many distinctly collinear variables, i.e., percent urban and percent rural. When this occurred only one variable was chosen. The fourteen variables were then statistically analyzed within a factor analytical model. 44 TABLE 9 DESCRIPTION OF DATA VARIABLE DESCRIPTION YEAR SOURCE COLLECTED 1 LITERACY RATIO 1950 l 2 PERCENT URBAN 1964 3 3 PERCENT MALE 1964 3 4 PRESENCE OF ABSENCE OF INDUSTRY 1964 2 5 RATIO OF INDUSTRIAL WORKERS TO 1964 2 URBAN MALES 6 PRESENCE 0R ABSENCE 0R BUDGET 1957 4 SURPLUS ' 7 VALUE ADDED BY MANUFACTURING 1964 2 8 PERCAPITA MUNICIPIO TAX RECEIPTS 1957 4 9 PERCAPITA MUNICIPIO TAX 1957 4 EXPENDITURES 10 PERCAPITA MUNICIPIO TAX EXPENDITURES ON PUBLIC WORKS 1957 4 11 WEIGHTED HIGHWAY DISTANCE1 1966 6 12 PERCENT POPULATION INCREASE 1950-1964 1964 2 13 PERCENT INDIAN OF THE TOTAL POPULATION 1950 1,3 14 RADIO STATION COVERAGE 1969 5 l5 VOTES CAST IN THE 1966 ELECTION 1966 7 (TOTAL AND BY PARTY) 45 TABLE 9 (cont'd) SOURCES: ./ 1. Ministerio de Economia (1950). Sexto Censo g§_Poblacion, Guatemala City: Ministerio de Economia. / 2. Ministerio de Economia (1964). Trimestre Estadistico. Abril, Mayo and Junio. III Censo Industrial. Guatemala City: Ministerio de Economia. 3. Ministerio de Economia (1964).Pob1acioh Total d§_la_Ré ublica por _epartmento, Sex y_areas, Urbana y_Rural, Conforme a Censo Effectivado de 18 a1 26 de Abril de 1964. Ministerio de Economia. 4. Direccion General de Estadistica (1957). Boletjn Mensual. Numero l_, Guatemala C. A. Direccion General de Estad1stica. 5. World Radio and Television Handbook (1969). Copenhagen: Lund Johansen. 6. Robert N. Thomas (1968), "Internal Migration to Guatemala City," Ph.D. Dissertation, Department of Geography, Penn State University. 7. Kenneth Johnson (1967), Analysis of the 1966 Presidential Election in Guatemala, March, 1966. Elect10n Analys15 Series, Washington: Institute for the Comparative Study of Political Systems, Operation and Policy Research, Inc. 1' The weighting is one mile of hard surface road equals two miles of dirt road. 2' The variable was derived by first locating the munici io in which each of the 55 medium range radio stations are located. Then concentrics circles were drawn, the radius proportional to the kiowatts of the emitting station. It was conservatively estimated that a 10 kilowatt station would reach 100 miles in every direction. The entire country has only six stations of the 10 kilowatt category. Most of the stations emit .5 to 2 kilowatt of power. A basic assumption is a flat to rolling topography. Once the circles were drawn, the maximum number of radio stations that could be received in each municipio was calculated. 46 Basic Outline of Factor Analysis Factor analysis begins with a matrix of intercorrelation values and attempts to find the smallest set sufficient to explain the observed set of correlations (Gullahorn, 1966). The assumptions of the model are not binding and are almost tautological. The assumptions are: that a group of variables does have a set of common factors; that the summed component variance equals total variance; that the scores of the individual factors can be more economically represented in terms of reference factors (Fruchter, 1954; Russet, 1965). 1 2 3 Factor analysis has many varied applications. Some of these applications are: reducing data for economical description; delineating patterns; discovering primary structure by reducing complexity and increasing homogenity of units; classifying variables into categories; reducing and compressing the number of observations on the basis of independent scalar quantities; testing for the existence of hypothesized relations; transforming data to meet requirements for other techniques; permitting the mapping of social terrain (Rummel, 1967; Dingham, 1969). AAA AA AAA CO \I 05 m «b (A) N —" VVV vv VVV Within geography factor analysis has reached widespread use and has gained descriptive importance. Perhaps the acceptance of the technique can be attributed to the pioneer work of Berry (1960, 1961, 1962, 1965, 1968); King (1966); Smith (1965), and Carey (1966) and others. Two recent uses of the factor analytical model using both principal axes and common analysis in voting behavior are Cox (l968c) and Brunn et a1. (1969, 1970a). In the factor analytical models there are two major approaches used that produce different results. The first is principal component analysis (PCA), and the other is common factor analysis (CFA). In 47 principal component analysis the objective is to determine a center of gravity solution on the basic structure in a 'm' dimensional space in which 'm' variables load on 'r' dimensions (m > r) (Harvey, 1970; Rummel, 1967; Brunn et a1“ 1970a). Variance is at a maximum with the first extracted factor and subsequent factors partition the residual variance. In common factor analysis the number of factors K is determined a priori and it is always less than m dimensions. In component analysis the factors are always statistically independent (orthogonal) whereas in common factor analysis the factors may be orthogonal or dependent. One of the major differences between the two approaches lies in the diagonal values of the correlation matrix, the starting point for the model. In PCA the diagonals are equal to 1 (one) while in CFA the diagonals are less than 1 (one). Thus an error term is assumed in the latter. The factor loading is the measure of the importance of one variable with all of the others in the extracted factor. In PCA and CFA the factor loading equals ZJ] = VJR°fR where fR equals the value of the variable and VJR is the characteristic root. However, in CFA the M (2J1)2 i=1 ... p & p < 1n is less than one while in PCA the z )2 (2.11 1 1 II ll (‘1'!) I 1 ... m is equal to one (Harvey, 1970; King, 1969). The square summation of row factor loading 2 (ZJ)2 is the portion of variance explained. In PCA this value must equal the value of the diagonal cells (rii) of the correlation matrix which is equal to one while in CFA the value is less than one. Therefore, PCA explains all the variance 48 of a characteristic while in CFA a portion of variance is attributed to random noise. Since the researcher attributes a portion of explanation to unknown factors in CFA, he must have a yardstick with which to choose the most meaningful column vectors. The rule of thumb adopted in this study and which is supported by others, such as King and Rummel is to choose only those vectors with a latent root greater than unity. By doing this, the researcher is then compressing the variance of his data. A latent root in column one which is twice the size of an adjacent root in a second column tells the researcher that the first column has accounted for twice the variance of the second column. Dividing the la- tent root by the number of rows yields the percent of explained variance in the first column. The rating of each observation on the particular factor is called the factor score. Each column vector has a column vector of scores, where [Y1] = [Ut] ' [x]. In order to facilitate column by column comparison, the component scores are standardized. The new formula then becomes [Yi'] = [ A - t] - [Yi]' It is noteworthy to mention that standardization does not imply normalization. In summary PCA is most valuable for a basic assessment of structure. However, if the structure is too vague, the researcher may want to use CFA which is specifically designed for an assessment of the clustering of variables around extracted factors. Normally the factors are rotated through the center of the cluster either main- taining orthogonality or not. As a general rule, if the researcher 49 hypothesizes a priori the existance of certain factors then the re— searcher should use common factor analysis with either correlated (oblique) or uncorrelated (orthogonal) factors. However, if the re- searcher merely wants to reduce dimensional variation, for example data transformations or scaling, then the principal components analysis is best (Kendall, 1957). In geography there has been a paucity of research based on principal components analysis. In many studies utilizing factor analytical models a principal components approach is combined with a rotation producing essentially a common factor analytical model (King, 1969). Most of the previously mentionned geographic applications of factor analysis have used common factor analysis. Basic Outline of Regression Coefficient Analysis The following section of the research design proposes and a statistical technique capable of answering the research problem proposed earlier. Regression coefficient analysis is needed to evaluate the influence that each of the segments of the socio-economic has upon the participation rates. Analysis of correlation coefficients could possibly be employed within the research design. However, this type of analysis has many inherent weaknesses, some of which are: a) The correlation coefficient is designed for locating important variables not in the assessing of change of one unit with another; b) The percent of explained variation in a correlation problem 50 measures the degree of scatter around the least squares line. Conse- quently, a high percent of variation represents a low degree of scatter. Voting involves numerous intricacies the complete identification of which is close to impossible; c) The correlation coefficient is ideal for the measurement of the nature of a relationship. That is, does a particular variable vary directly or indirectly with other variables and what is the goodness of fit of the relationship? Correlation coefficients do not measure the form of the relationship. The research problem of this paper requires a research design stressing the form of the relationship between the socio-economic cleavages and voter participation. It is assumed, based upon related research in political science, that a relationship does exist between the dependent and the independent variables; d) The correlation coefficient is susceptible to confounding influences. The size of the correlation coefficient depends not only on the variation in the dependent variable, but also on other con- founding variables affecting the variation in the dependent variable (Blalock, 1964). The regression coefficient is unaffected by con- founding variables as long as the researcher assumes that any mani- pulations'ofdata occurred solely with the independent variables. A manipulation is any overweighting of the sample, selection of certain cases, grouping of certain cases of shifting of areal units (Blalock, 1964). In the present study, the universe was studied with no a priori weighting in the collection or formulation of the research design. Also the independent variables were not grouped on the basis of aggregate units; e) The correlation coefficient is based upon a set of events occurring at place (i,j). However, the events at (i,j) may be related to the events at (i+m, j+n) which in turn leads to a correlation of each observation with its neighbor (Tobler, 1966; Cliff and 0rd, 1969). It is assumed that many of the variables chosen for this study represent events which are autocorrelated. However, it is beyond the scope of the present study to prove or disprove the above assumption. Therefore, the inherent weaknesses of the correlation coefficient have eliminated its use in the present study. Substituted for it is the regression coefficient which overcomes most of these weaknesses. The regression coefficient b represents the units change in yx Y associated with a unit change in X (Yeates, 1968). In reference to 51 this problem byx represents the amount of change in political partici- pation associated with a unit change in one of the socio-economic dimensions. Often the regression coefficient is symbolized with a period to the right of the variables. These variables to the right of the decimal are held constant or controlled (King, 1969). In order to compare the values of the various slopes of the regression equation, a standardization of units is necessary. The formula for this is: Bj = bj (Sj/Si) where bjis the regression coefficient and Si and Si are the standard deviations of the independent and dependent variables (King, 1969). Two assumptions which must be met prior to a valid interpretation of regression coefficients are: a) means of zero, and b) standard deviations of one The following section describes the normalization of the data, a process necessary in order to meet these assumptions. ASSESSMENT OF DATA NORMALITY The determination of the normality of the data is accomplished through the use of the two-sided Kolmogorov—Smirnov test.1 Since this test is noneparametric, the interval factor scores are changes to orginal variables. The variables are then classified into one of four— teen classes which represent oneehalf standard deviation units above and below the mean. For each class the cumulative frequency is determined. 1For most tests, two-sided levels of significances represents the probability that the samples were drawn from identical populations. 52 The Kolmogorov-Smirnov test is based upon the acceptance or rejection of a null hvpothesis.2 If the actual difference between the test distribution and the actual distribution is less than the expected, then the null hypothesis is accepted. If the actual difference is greater, then the null hypothesis is rejected. The null hypothesis states that the two independent samples have been drawn from identical populations, one of which is normal. Therefore, the cumulative frequency distributions for the sample are equal (Blalock, 1960). To determine at what point to reject the null hypothesis the chi-square test is used. The chi-square values for the sample distribu- tions are calculated from the following formula: 2(N1N2) N1 + N2 X2 = (4a) Where N1 and N2 equal the total number in each rank and 'd' is the largest difference between the actual data and the normal data. The last value is obtained from the fourteen categories. The above chi—square value is compared with the chivsquare value based on N1 and N2 degrees of freedom. If the sample value is less than the expected, then the hypothesis is accepted. Likewise, if the sample value is greater then the expected, then the hypothesis is rejected. Table 10 lists the variable, the actual difference and the expected difference at a certain level of confidence. The fourth 2The Kolmogorov-Smirnov test was incorporated into a computer program written by Theodore Miller, Geography Department, University of Iowa. Norm was adapted for use on the CDC 6500 by Robert Wittick of the Computer Institute for Social Science Research. 53 TABLE 10 NORMALITY 0F VARIABLES EMPLOYED IN STUDY VARIABLE ACTUAL EXPECTED DIFFERENCE TRANSFORMATION DIFFERENCE AT A CONFIDENCE REQUIRED LEVEL Per Capita Vote .355 .068-10% log10 Percent Vote (PID) .052 .068-10% ----- Percent Vote (MLN) .081 .091- 1% ----- Percent Vote (PR) .109 .091- 1% ----- Factor 1 .176 .068-10% log1O Factor 2 .155 .068-10% log10 Factor 3 .057 .063-15% ----- Factor 4 .156 .068-10% exponential Factor 5 .047 .063-15% ----- SOURCE: calculated by author. 54 column contains a transformation which produces, in some cases a smaller actual difference. Repeated tests at less restrictive levels of con- fidence were conducted to see if the non-normalized variables were indeed normal. Since these variables remained non-normalized, it is assumed that the transformation indicated in Table 10 represents the closest approximation of normality attainable within the present research design. 55 CHAPTER V ANALYSIS OF PARTICIPATION IN GUATEMALA DELINEATION OF ASPECTS OF THE SOCIO-ECONOMIC ENVIRONMENT The rotated factor analysis model for this study consists of five orthogonally related dimensions with roots greater than one. The variables loading on the five dimensions have communality values ranging from .43 to .84 (Table 11). The low communality values for some of the variables indicates that these variables load on factors with latent roots less than one. Once the model is constructed, the next most important step is the labeling of each dimension. Although the identification of the factors is not mandatory, it does aid the reader in the subsequent interpretation of the results. The enumeration of each factor in this study is based upon the following criteria: (1) The specific variables loading highest on the factor; (2) The sign of the factor loading; and (3) The spatial distribution of the factor scores. As evidenced from Table 11 the most important factor, number one, had an urban orientation. The variables loading high on this factor are Value Added, Tax Receipts and Tax Expenditures on municipio services. Higher factor scores for this dimension are found in the more urban municipios (Figure 4).1 Analysis of the factor scores indicates that 1The computer program SYMAP (1969) was used to construct Figures 4-8. 56 Figure 4: FACTOR l The Development Dimension MAP LEGEND N Represents the Lower One-Third of the Distribution Dot Represents the Middle One-Third of the Distribution P Represents the Upper One-Third of the Distribution H Represents Scores which are Greater than +.99 L Represents Scores which are Less than -.99 SOURCE: calculated by the author. W W... l'I'III'III..I."I'I'.' mm IIII . . . . . . - _ . . . . . '. ‘_.__ ”P” 6.0-. - -' WMD.” ”I ...........-.-.-.-.. In. ... .' lll.‘-..II'III.|I‘C'.'I.I. n—ru . ..nI-n.I-nn;Ifl I -..".b”'p_.-""":I-uc-nuu in.-. m ”II... ' .IIIIIOIIIIIOIJIIII Inna-Inn“ nun... ...... I. ~‘v'IIIIIUII --. ”O a I . A —.<——— unoc-I.I..IIIIO.II II... ' ‘ -.-n.1:n"flI.AI-IOWW‘ . _ . . a..I.-oIII.I-II.' III-III ..._. .-I'¢I.ool NI- ...-...... .......-.... .I'OIIIICI.D'O.I". huh-hfllIM‘ -. .-.-u.-.uo..I--o-.III m- 'V'I" —. ’I’I..IIIOOIIIIQ.'I'II.II on.-....--.-.-..ou.-... IIIIIIIIQIIIIIQ'I' “VI'UVVVU .I.-.-II--vn..I--.I. An- ”0 0.. M ... .1 . n...n.-o-n-unnIo-I _ m m m . - -.1I'.I1I ~ 11‘ 111 .. .:--n....':u. .uon.nvn.‘.rnn .. I '. ' ' ' I ......u. _ --. . . . I I ...”... ,. - . ' I Iv A...|'.. ‘ l . . _ . _ . .u....;....u nun-"Innerv'sluo ... - ....u -. .- - - ...........'Io "I” --- . . -~- - . .u . l['-' u--- .. V7.7 . ,,..,..--. o . . ' - - .I'I'u ‘ ' ‘ an...c...o-----u... ..III---.- . . I ' .. -I-\.1.I..-. . nun- . ---- ' ......”nun-............-.-..n...-.-.. . m m ..no-u llIII . . ll...’..".'. .--.0 --.--.u:-.-I--II-o- . . ' 'II I-.....:-.-I..II-.I--.-.-..-.--no-...o'-o- " " - . . . .. '. “...-nu." ...-.0... I I‘I. ... n- "7'1 I-vn.I|II-:I-I"'va-Ocvvvvnmv\ IIIIIIIII-.I.I.I I. ‘ I II. l.ll....l.ll"'..'0 l‘l ‘ nihu'mon mn-nununnur. ~ . .I I u ...-.....-.-.-..-.o.- on-IouoI--.-.u.-.o.o u. . _ .. 2“ - ww—r‘nv‘n‘h . .. . ..... "Hun-”h“. . " ' ' ’ . .. mu .-.-. ........ nu. .. . ' H I . rthvr- ._ 7 ”In". ' .... .-n. n. ..........-.........u.............. CW“. ou....oc...n..I---Ioa....-.vI-o.- M n...”* n-oIn.-I.I.o.o.n .....o... a 77" I I I v "1‘1" 5...... Cw .I-Icl.i . 1' —nnu--q .-....'.......:...TI.'..'..l...- M m ' * ,.... u ... ' ' .. ‘. .......n......u..-ou- m on— — .- nu.....nIIuI-n..--........ "‘|‘|‘W.‘ “- Io-IIIIII.I.I.I.I.I.I.I ‘—-—‘ - 4- ‘ I -IIIIIII 'U..'...'I." I'I'IIIII' ' .... ”91"." an . w VIIV.IIII.I.I‘-.I.o.'.01ltl I'OI I .u.-.- D.I.O'I.I ......IIOII- .I.I.I.I I.I.I...-.-- MI ...-I. ...-II. I........I...-:-.n:. ~ ....n...-..- I.-.II-lo-n-n.-.I.o.c.I.I.I....... . ' O. . I. v“ I ' ...-......” ...... . .0 .-.-.-I III Inl-II'.IIIIIIIII oI.-.-.u.o.. . INIICIOOO'IOIW’OOIU40:..IIIIII ...II.....I'...-.......1.I.l'.. I 5 IIII.I.I-I.I.I.I.I..-..-.-u - n...l.u. on. .I..Iu-.I.I.IIIII II.I.I.I.I.I.I. ...-- 1" IIIOIUIOUIIII..I.I'..III' II-I.I.I.I.I. .I.--o.-................ “I.‘..".""' II. 'I'I'.IIQ ..- - .- ‘vvv‘W ' m“ lin- ...... - ...-...... , * In-va—‘NI-w II ...—IIII- III.I.., - ' I .II .I- ....n 'ICOUIOI' .I.I I. mnm I ...oo--.II.-.n I. .fi HWL'I -“wv-Wmvrfi. ’V'I' v w- n "...-....- ”...-"... . . AL .— vvv v —v III-I. WVW- v v nun-nu- ...~ “It I ..‘ I ... . 'I'yl . I "UI'IIIII'IIIOII. ...-'uI-Inou-o I n- a .. I . """" I .. ”I. . I ”nun-Inn.” w. -. .... ~-*”—‘ng * ' ° ‘ ""' " ' ...”... human- uu... ..." I.o.l-0-'-I-I.. Vim-III" lgvnfn‘. I “...—0....” . Mm" “In I... -.-....... n * ‘fi . I “ ~IIII— «runny-"M . - .... n . .I II. I .-.~-.--9..I.I I.II¢II~¢II- ‘ - W H‘ I." I m - '. . ‘ ...”... fl . N- nvmrn *gw- _ - .... . III-Iva... — . ... “ “tun .- W 7.7.... ...-5....- o.-.I OIIIIIO II ....I.-.no-.III..I.III. "‘ .‘ I.-.I.-.III.-.II.I--I° '0'.- "v'vw I n. ..I- an“ Man- ” .- . n "... “I.“ “F. *3 .... w "IIIN‘I. “5W3.“ sun-una- Nun-“um“ .. till..— - u r -- «wfiv-OWI-un-IQQA. 00- MW 58 TABLE 11 FACTOR LOADINGS AND COMMUNALITIES VARIABLE FACTOR COMMUNALTIES 1 2 3 4 5 1%) Literacy Ratio + - .81 - - - 78.3 Percent Urban + + .51 + - - .42 64.0 Percent Male - + .56 + + + 53.0 Presence/Absence + + - .90 + - 84.1 of Industry Budget Surplus + .42 + - .47 - - 43.1 Industrial Workers + + - .88 + - 82.7 Value Added by + .81 - - + - 70.8 Industry Tax Receipts + .68 + .45 - - + 78.6 Tax Expenditures + .80 + - - + 86.5 Tax EXpenditures + .81 - + + + 70.9 on Public Works Weighted Highway + - + - + .84 75.8 Distance Percent Population + + - + .88 + 79.9 Increase (1950-1964) Percent Indian/ + - .62 + - - 51.7 Total Population Radio Station + + - + .70 - .54 81.1 Coverage Proportion of 20.2 14.7 14.6 11.3 10.3 71.4 Variance 59 TABLE 11 (cont'd) Eigenvalues 3.74 1.72 1.36 2.02 1.14 + II loading less than or equal to .40 loading less than or equal to - .38 SOURCE: calculated by author, 60 the department capitals are generally all above one standard deviation and positive. Guatemala had the highest score, a 14.8, followed by Quezaltenango, Mazantenango, Antigua, Puerto Barrios, 501015, Totonicapgn and Escuintla. This factor representative of economic integration and urbanization is labeled the Development factor. The second factor, similar in areal extent to factor two, yet more important in overall explanation, includes the variables percent male, percent urban, percent literate and inversely related to all of these, fifteen percent Indian. Factor scores are negative and less than one for most large urban departments. Factor scores greater than plus one are found in the departments of Retalhulue, Suchitépequez, Escuintla and along the Pacific Piedmont paralleling the Pacific Coast Highway. (Figure l and Figure 5). High positive scores are also found in the department of Petén and Izabal. The high positive scores in this part of Guatemala are associated with recent agricultural settlement schemes where male participation is high. Therefore, this factor is called the Agricultural Settlement dimension. The highest factor loadings on factor three are the nominal variables industry, industrial workers, urban males and budget surplus. All of these variables are directly related and indirectly related in the factor. High positive scores are found along the northern and northwest sections of the Central Highlands and in the southeastern section of the country. Areas with low negative scores are predominantly urban centers, for instance, Retalhuleu, Mazatenango, Quezaltenango, Puerto Barrios, Zacapa and Guatemala City. Other extensive areas of 61 Figure 5: FACTOR 2 The Agricultural Settlement Dimension MAP LEGEND N Represents the Lower One-Third of the Distribution Dot Represents the Middle One-Third of the Distribution P Represents the Upper One—Third of the Distribution SOURCE: calculated by the author. 63 low scores occurs along the Pacific coast (Figure 6). Since this factor is composed of variables implying economic advancement and agricultural endeavors and since the scores are most significant in selected urban and rural areas, it is labeled the Industrial-Commercial Farm index. The fourth dimension, labeled the Migration factor, is composed of the distance variable radio station coverage and population growth, both of which vary directly within the factor (Table 11). High positive factor scores are found in the department of Guatemala, Retalhuleu, San Marcos, Escuintla, Suchitepgquez, and Zacapa (Figure 7). According to Table 12 these departments have undergone a rapid popu- lation growth within the past twenty or so years. Negative scores are principally found in the central highlands around 801013 and Totonicapgn and in the major northwest portions of Huehuetenango and Quiche: Loading highest on factor five are the distance variables. They include weighted highway distance from Guatemala City; potential radio station coverage and percent urban. The two distance variables are directly related to each other and indirectly related to the per- cent urban. The lowest factor scores for the factor are located in the more urban areas especially those adjacent to Guatemala City and Quezaltenango (Figure 8). High factor are numerous in the northern departments of Petén, Alta Verapaz, Quiché and Huehuetenango. These departments contain a strong majority of the Indian population who live in remote municipios and daily operate outside the political and economic sphere of the large urban cetners. In addition the rugged topography 64 Figure 6: FACTOR 3: The Migration Dimension MAP LEGEND N Represents the Lower One-Third of the Distribution Dot Represents the Middle One-Third of the Distribution P Represents the Upper One—Third of the Distribution SOURCE: calculated by the author. . III‘IODII 9" ...-Jun WWW-Hunt.» IVOUCIOIIO'C‘U" 'n OTCPIrIrtI - {Inn'- ...IO‘III ” u ...-v. uuo'. ~ Try-BINJE‘u-uunu 'I.:::l. . :::11:::::"" "I In .. ......U. ”...-nun. m .. ... a m .....u-IIII .IIg-I'I. I I. l I «III-In"loo-huIII-Ih- In I - u _ 66 Figure 7: FACTOR 4: The Migration Dimension MAP LEGEND N Represents the Lower One-Third of the Distribution Dot Represents the Middle One-Third of the Distribution P Represents the Upper One-Third of the Distribution H Represents Scores which are Greater than +1.49 L Represents Scores which are Less than a -l.19 SOURCE: calculated by the author. - _ “ — _ ~ — * ~ - — — _ — - - — - _ - w m m m m m m m mm. " __ ‘b‘ hnt“: r: _ _‘ SLLI LI \I II ILL .WD““"' 0“! l\~“‘.~‘ G‘ "M I‘ ‘.‘--' :0 0" - \~§~‘V 0:; . . ~ ...c I . . . . W mum-um“. I. , . W * ~“”"‘ “—0..." v m m ‘m‘ .... “M u ..I.-.. . . . . ...-u...-.o . . g u... - - _ “ “ I I .IIII-II-III-II...I.IIIIII 1111111111 -I. II-IIII-.IIIIIIII-.I.I...-......I.I..... I..-........, ' ' T -l _ :2 : n..........u-IHuuu”...-......“ ‘ o O - n . w .... .I-0 I In ' IIIIIlIIQOOOIO-OIIOIIIIICCIltnullIIOIIIOIIIO — ...... fl . ... . . m “ m . m -u ....u-.....I-......I-u..-uu-.-u.u..- W W II-II 1 11 ' “I ' m : _ , fl ' ... . . m u m “MW ' ‘ ‘ . W “MW-“‘ILMI“ m . .. m WWML m IIIOIIQIno.0:l0lIODI.IIIIOIOII*Ilnl. *- “Bl .. . . ' fl rrnvrmr ................. ....................I... "'23. ““W" WW“ ....II-I "Wu...“ m I. .-I. ...I... I.‘ ......III ......I... m .... _ m I "“'"'" "u *- ... nu... .... m ... I m- """' ""'°"" "" . """"" "'"t . --.-.-.uu-....--. m .... m 00...... I..IIIO. In. IIIquI-III-IIIII IIIIII .I-...--II.-.I.I.. . m . II.I.I. . . m IIIIII IIIIOII IIIII'IIIIIII .I-III-I. " ' W IIIIIIII III I I... - IIIII.-II ' ..I-I-III-II. -- ' ' ’ “ uII-IIIIIIIIIIIIII . u I. . I.I'.I "I. .. ........, - I I.I IIIIIIII III-I IIII'II'IQOIIIII-l .- "IIIIII-I-III I .... III-IIIIIn-IIIIII. ....II-.I - ' I...I.-. .-.o- 'II'I..IIIIIOII|II ....n-I-I-III-IIIIII n-Iouo- :::::...:::...::I ‘ I... .: .... . "I: II-IIIIII..III OOIII'O . I -.I.n.a.I-Ino.-II.I Inc-In- III-.III. . . o-I I... ‘- .o-III In. ...-IOIIOIIIIIII I IIII.I.I.O' mun -. ..- It. “COO-u“ ..IIIIHIIII ...I. * III- ..uu -I--I.-...IIIII I 68 Figure 8: FACTOR 5: The Rural Isolation Dimension MAP LEGEND N Represents the Lower One-Third of the Distribution Dot Represents the Middle One—Third of the Distribution P Represents the Upper One—Third of the Distribution H Represents Scores which are Greater than +1.99 L Represents Scores which are Less than -1.99 SOURCE: calculated by the author. ..—. I- . .~-_‘.~- ... ,. - I —--.-- mom "* mw- --~ ~- --- — ~— ...-.0" -‘ '- '1- -“ “0-- W ...-.4... ~ -‘ ‘ "I”..I- 9'”! "I. ImI. "M‘Ou ... '9’! II- IIIII .‘O‘ :I'I'? I I‘” III I ' FWD... II D DI~ ”O .....I-II-I I-I.- IUIDIIQIt'IIIIIIIQII- .- I I. "*mr‘v "n m " II. 03-... ”...-I III-IIIIIIIIII l . .. .. ...... Lu...“ I . I x . ' Q n. A III. . l .....l. oInIo IIII III - ~ I. III nu ul- ” -.I‘ . w 0 I In D 1 v II ' u“ - ' “5 . . 70 TABLE 12 SOCIAL STATISTICS AND FACTOR SCORES FOR DEPARTMENTS AND CABECERA DEPARTMENT SOCIAL STATISTICS CABECERA POPULATION PERCENT OF FACTOR SCORES INCREASE POPULATION IN URBAN 1 2 3 4 5 (1950-1964) AREAS Guatemala 85.3 78.3 14.21 -1.75 .29 2.01 -1.61 E1 Progresso 39.4 27.2 .02 .74 -2.02 - .47 - .61 Sacatepequez 33.8 72.8 2.63 1.56 -1.50 - .34 - .84 Chimaltenango 34.8 37.2 31 - .05 -2 52 - .57 -l.17 Escuintla 118.0 24.6 1 47 .72 -2 46 .06 - .46 Santa Rosa 41.5 20.6 - .01 .52 -2.19 - .26 - .52 So1o15' 31.2 33.6 1 35 -4.68 -2 82 -2.57 -3.21 Totonicapan 40.5 13.9 - 11 - .55 -2 05 .03 .15 Quezaltenanto 46.0 35.8 3 17 .67 -2 08 - .69 -1.02 Suchitepéquez 49.7 31.4 2.75 .22 -1.75 - .07 .40 Retalhuleu 83.7 26.6 .40 1.66 - .61 - .92 .49 San Marcos 42.8 13.3 - .31 .89 -2.06 - .41 .48 Huehuetenango 43.4 15.9 .30 - .82 -1.59 - .68 .20 Quiche] 41.6 13.4 - .03 -l.O6 .54 - .04 .87 Baja Verapaz 44.2 15.8 1 91 -2.46 -1 51 .67 .05 Alta Varapaz 36.9 11.7 .28 -l.20 .54 - .26 2.13 Peten 68.2 52.0 - .24 .41 .82 - .51 2.56 71 TABLE 12 (cont'd) DEPARTMENT SOCIAL STATISTICS CABECERA POPULATION PERCENT OF FACTOR SCORES INCREASE POPULATION IN URBAN I 2 3 4 5 ( 1950-1964) AREAS Izabal 107.8 20.9 - .52 .56 .34 .62 .78 Zacapa 38.0 28.0 - .11 -1.02 .36 .49 - .33 Chiquimula 34.0 21.2 - .31 - .24 -1 10 .13 -2.76 Jalapa 30.0 26.3 .13 - .23 -1.85 - .15 .15 Jutiapa 43.2 18.5 2.9 .95 -1.66 - .80 - .04 SOURCES: l. Ministerio de Economia (1950). Secto Censo dg_PoblaciOh. Guatemala C.A.: Ministerio de Economia. 2. Ministerio de Economia (1964). P blacion Total de la Republica por Departmento, Sexo y_areas, Urbana y_Rural, Conforme a1 Censo Effectivado g§_l§'al_2§_Abril' de 1964. Guatemala, C.A.: Ministerio de EconomTa. 3. calculated by author 72 and the poor roads in the northern and northwest portions of Guatemala increases the cost and the time of movement to the large urban areas. For these reasons, this factor is called the Rural Isolation dimension. IDENTIFICATION OF THE DYNAMIC CORRELATES OF POLITICAL PARTICIPATION The aim of the next three sections is to identify those socio- economic dimensions which are most important in affecting a change in political participation. The analysis is conducted on two levels: first total participation and second, party participation. The generalizations derived from both are easily comparable. Analysis of Overall Participation The five dimensions listed in Table 11 and the necessary trans- formation of the variables are statistically analyzed within a linear regression model. The dependent variable is log10 votes cast per one- thousand people.1 By grouping the dependent variable it is hoped that the impact of exogenous influences is reduced and that the angle of the slope subsequently increases (Blalock, 1964). The regression coefficients, both standardized and the non- standardized are presented in Table 14. The top portion of the table 1When per capita vote was used as the dependent variable, the slope of the regression line wss almost horizontal. Based upon prior research dealing with electoral geography, it is assumed that the slope should have a value greater than zero. Undoubtedly confounding influences are inhibiting the recognition of the relationship which in turn reduces the magnitude of X1 and X0 in the following formula: [2X0X1-(X111 . (2X0)/NI x [(21112— (2x012/N1 0' [1: x02 - (2 x072 / NJ”2 x [2 x02 - (z x012- (1: x012/ N1 As evident from the coefficient formula, once the dependent variable is aggregated into larger units, the numerator increases in value. The empirical proof for this change is found in Table 13. 1/2 B 1/2 73 TABLE 13 SUM AND SUM OF SQUARES OF THE REGRESSION MODEL VARIABLE SUM SUM OF EQUARES (X) (X ) BEFORE AGGREGATION OF THE DEPENDENT VARIABLE Factor One 4.2 2l9.83 (Urban Development) Factor Two 3.7 229.83 (Migration) Factor Three .0004 326.5l (Agricultural Settlement) Factor Four .0006 322.l2 (Industrialization and Commercial Farm) Factor Five .OOOl 324.7l (Rural Isolation) Per Capita Vote_ 42.4 23.6 (dependent variable) AFTER AGGREGATION OF THE DEPENDENT VARIABLE Votes Per lOOO people 65l.4 l324.l SOURCE: calculated by author. 74 TABLE 14 REGRESSION COEFFICIENTS FOR OVERALL PARTICIPATION VARIABLE COEFFICIENT STANDARD STANDARDIZED DEVIATION COEFFICIENT BEFORE DELETION Votes/1000 ----- .1847 ----- Factor One .0376 .3038 .0634 (Urban Development) Factor Two -.0212 .6125 -.O704 (Migration) Factor Three .0540 1.0069 .2961 (Agricultural Settlement) Factor Four .0084 1.3952 .0636 (Industrial-Commercial Farm) Factor Five .0018 1.0042 .0103 (Rural Isolation) AFTER DELETION AT THE .10 LEVEL OF SIGNIFICANCE Factor Two .0307 .6125 -.lO64 Factor Three .0539 1.0069 .3139 SOURCE: calculated by author. 75 contains all of the coefficients while the bottom portion contains only those coefficients which are significant at the .10 level of significance. This level represents the same level attained consistently with the data transformations. A regression coefficient such as BOl.2345 assumes that variables two through five are held constant and that the mean value of the dependent variable does not change (Blalock, 1964). The value B02.3 is different from the preceding in that this value assumes that only variable three is held constant while variables one, four and five are eliminated. Thus in the latter situation a clearer picture of the slope of the relationship is established. Is is evident that for every unit change in the factor labeled agricultural settlement a subsequent change in the magnitude of approximately thirty percent occurs in the overall voter participation rates (Table 14). The most significant foci of increased voter activity are found not only in the urban areas but also along their periphery, particularly in the Northeast. Therefore, where the agricultural schemes are dominant, participation in political affairs in Guatemala tends to increase. Also, the regression of the factor dimensions implies that political participation is higher among the more literate. Furthermore, it is readily established that participation decreases as distance to najor urban centers increases and at the same time, participation tends to decrease in areas experiencing rapid population growth (Table 14). As a result migration to major urban centers in the Central Highlands and in the Northeast has a deterimental 76 effect upon increases of political activity. Perhaps the migrant to these areas lacks concern about the national issues. Some economic considerations identified in the factor analytical model have only a minimal influence upon any increase of political activity, i.e., industrialization, commerical agriculture and their derived benefits apparently have little effect upon political activity. In contrast to these institutions is the relatively recent agricultural resettlement schemes in the North and Northeast. In all probability this type of economic activity and the residual benefits derived encourages more intensive political participation. It is inferred from the empirical evidence that the middle class in Guatemala is a politically active force especially in the urban areas. These people, many who migrated recently to the urban area, have become technologically valuable due to their education, training and urban acculturation (Thomas, 1958; Ningo, 1967). Likewise these people who actively engage in frontier settlements also possess this economic and cultural advantage. It is these groups which have the highest rates of political participation. Analysis of Party Participation The foregoing analysis of participation paints a broad mosaic of the social and economic dynamics of participation. Missing from that analysis are the finer regional details of participation. This section will present some of these. Despite the fact that is is difficult to generalize about 77 political behavior, this study has empirically verified that the socio- economic environment influences political activity. Undoubtedly exceptions to this statement are easily found. However, specific exceptions should not detract from the validity of the generalization. What is needed in the face of numerous exceptions is a redefinition of participation. A broadening of the earlier generalization occurs in this paper by an examination of participation by party using the same factor analytical model. The parties chosen for the study are the PID, PR and the MLN, all of which were defined earlier. These political interest groups possess a spatially diverse constituency, ranging from the Highland Indian to the urban migrant. By analyzing the party participation rates and correlates of the socio-economic environment, it is hoped that this empirical analysis validates this spatial variance. The results of the regression coefficient analysis (Tabie 15) substantiated the earlier regression results. Once again the most salient and consistent correlate of participation is the migration dimension. However, the direction of the relationship varies among the parties. The most important and significant correlates of P10 parti- cipation are the migration and agricultural settlement dimensions. Both of these factors tend to adversely effect the PID dominance in rural and urban areas. In most cases once the non—significant factors are eliminated, the direction of the relationship becomes more definite. This clarification is easily observed from Table 15 particularly with the PID. 78 TABLE 15 REGRESSION COEFFICIENT FOR POLITICAL PARTICIPATION BY PARTY VARIABLE REGRESSION STANDARD STANDARDIZED COEFFICIENT DEVIATION COEFFICIENT PID - DEPENDENT VARIABLE BEFORE DELETION 11.2 Factor 1 3.37 .05 .015 (Urban Development) Factor 2 -l.60 .92 .133 (Agricultural Settlement) Factor 3 .40 2.54 .098 (Industrial-Commercial Farm) Factor 4 - 8.25 .05 .175 (Migration) Factor 5 .70 .89 .057 (Rural Isolation) AFTER DELETION .05 LEVEL Factor 4 -40.6 .05 .186 MLN - DEPENDENT VARIABLE BEFORE DELETION 6.7 Factor 1 -32.2 .05 .231 Factor 2 .0004 .92 .0002 Factor 3 - .17 2.54 .052 Factor 4 10.5 .05 .0785 Factor 5 .42 .89 .056 79 TABLE 15 (cont'd) VARIABLE REGRESSION STANDARD STANDARDIZED COEFFICIENT DEVIATION COEFFICIENT AFTER DELETION .05 Factor 1 - .31.8 .05 - .237 PR - DEPENDENT VARIABLE BEFORE DELETION 12.5 Factor l 27.9 .05 .113 Factor 2 1.6 .92 .121 Factor 3 - .22 2.54 - .048 Factor 4 27.7 .05 .111 Factor 5 - 1.13 .89 - .093 AFTER DELETION .05 Factor 1 21.7 .05 .088 Factor 4 29.7 .05 .119 SOURCE: calculated by author. 80 Relating the results of Table 15 to the specificcnmposition of the dimension, it is evident that the PID rates are negatively associated with a factor containing a negative percent Indian. Therefore, the PID rates are positively related with this variable. Likewise, the agri- cultural settlement dimension is composed of a positive urban variable and it is negatively related to the PID rate. In addition, population growth and urban proximity, two variables composing the migration dimension, are negatively associated with PID support. Participation for the MLN varies negatively with the urban- development dimension. Once again, a negative dependent variable is associated with a factor composed largely of negative loadings. The result is a positive relationship between the dimension and the dependent variable. MLN support is also negatively paired with the industrial commercial farm index and positively related with the migration dimension. These last two dimensions contain many positive variables. The regression coefficient for the PR participation exhibits some similarity in direction with the preceding two parties. However, in this case the direction of the regression line is opposite of the earlier rates. Positively associated with PR participation are the migration and the urban-development dimension, both of which are composed mainly of positive variables. Negatively associated with the only socialistic party are two rural based, social and economic dimension, rural isolation and industrial-commercial farm. An analysis of the dynamic correlated using the same factor 81 analytical model but choosing only those municipios which ranked one standard deviation above the urban or rural mean produced the results in Table 16. The results of this analysis are consistent with the earlier findings at the higher levels of aggregation. In these highly urban areas of Guatemala, the migration and urban development dimensions produce a high positive change in PR and MLN participation while the migration dimension produces a low negative change in the PID rate and the overall participation rates. MLN participation in the very urban or rural areas is less than the rates for the other two parties in these areas. This cursory analysis of party participation not only provides a substantiation but also an expansion of the previously expounded empirical generalizations. PID participation appears to be the strongest in the rural Indian areas which have experienced only minimal levels of population growth. MLN participation is highest in rural areas of limited government activity as well as urban areas experiencing rapid population growth. Some departments of future MLN support appear to be Guatemala, Quezaltenango, Alta Verapaz, Baja Verapaz, and Huehuetenango. The urban migration and the economic and social integration of the city appear to favorable influence increased PR participation. Therefore, the future locations of H2 strength would seem to rest in the most urban departments such as Guatemala, Sacatepeguez, Quezaltenanto, Solola: Isabal and Chimaltenango. In summary, this chapter has not only identified some important socio-economic dimensions of voter participation but also has provided an 82 TABLE 16 REGRESSION COEFFICIENT FOR POLITICAL PARTICIPATION BY PARTY IN SLEECTED AREAS HIGHLY URBAN AREAS N = 48 VARIABLE COEFFICIENT O ERALL PID MLN PR PARTICIPATION Factor 1 .165 .026 - .377 .241 (Urban Development) Factor 2 .051 - .O79 - .050 .145 Agricultural Settlement) Factor 3 .1125 .057 - .051 - .037 (Industrial-Commercial Farm) Factor 4 - .014 - .1648 .248 .410 (Migration) Factor 5 .253 .159 - .053 .1770 (Rural Isolation) HIGHLY RURAL AREAS n = 133 Factor 1 .033 .117 - .245 .024 Factor 2 .223 .001 .158 - .079 Factor 3 - .023 .032 .032 - .042 Factor 4 .142 - .257 - .035 .225 Factor 5 - .158 .092 - .054 - .047 SOURCE: calculated by author. 83 assessment of each dimension's influence upon increased participation. With this information the comparison of the participation correlates is treated in the next chapter. 84 CHAPTER VI CONCLUSIONS AND SUGGESTIONS FOR FUTURE RESEARCH DISCUSSION The logical rationale for the following conclusions comes from a synthesis of political participation and its dynamic correlates on the three levels of aggregation: overall, part and urban. (1) Political participation not only reflects regional, social and economic differences, but it also illustrates the degree of economic integration, government influence and social mobility of that region. In areas of high political participation these social and economic forces are most intense. Evidence for this generalization is derived from the large average coefficients for the urban-related dimensions, that is, factors 1,2 and 3,4 (Table 17). (2) The urban environment disrupts the political stability of the migrant; however, if the political parties stabilize the environment and if the researcher views the stabilization only in the urban areas, then the migrant tends to participate in the political activities. Proof for this supposition comes from the fact that the migration dimension is negatively associated with overall participation but positively associated with certain party rates. Within Guatemala the party which appears to motivate migrant participation is the PR. MLN participation also is consistently positive with the migration dimension. 85 TABLE 17 AVERAGE COEFFICIENTS FOR THE CORRELATES OF POLITICAL PARTICIPATION IN GUATEMALA CORRELATES SUM OF TWELVE MEAN STANDARD COEFFICIENTS DEVIATION Factor 1 1.650 .1375 .0982 Factor 2 1.101 .1101 .0820 Factor 3 .646 .0537 .0331 Factor 4 1.919 .1593 .0878 Factor 5 1.199 .1001 .0580 SOURCE: calculated by author. 86 However, based upon its platform, it is assumed that the MLN support in areas of high population growth arises among urban people who may unfavor- ably react to migration, such as the upper class. (3) Location and distance surrogated influence the spatial diffusion of politically relevant information which subsequently pre- pares the area for higher participation. The large urban areas in Guatemala are nodes in the network of this flow. Since the flow is most intense at the nodes, here it is most easily regulated. Likewise as shown by this study, the largest changes in political participation whether overall or party are caused by urban related measures (factors 1, 2 and 3). In addition, the consistent positive coefficient of the newest legally recognized party in Guatemala, the Partido Revolucionario, with the major urban indicators substantiates this generalization. (4) It is predicted that the socialistic oriented parties of Guatemala will find their support increased if the middle and lower classes continue their integration into the modern day economy. This prediction is based upon the assumption that the middle class in Guatemala maintains and increases its present economic and political viability. A measure of this strength is found in the coefficients between factors 1 and 2 and the overall and PR participation. As evidenced from these values and its party philosophy, is is suggested that the PR which presently controls, will continue to maintain control in the future. LIMITATIONS OF THE STUDY One of the major restrictions of the present analysis is the time 87 delay incorporated into the study by the limited availability of data. As evident from Table 8, this covers a period of fourteen years. The variables chosen for this study were deemed the only ones available which also fit the research problem. For this reason the time element was consciously neglected. The decision to vote represents the interaction of not only social and economic factors but also psychological variables. The latter were empirically ignored in this study due to the level of aggregation. The procurement of psychological indices for a given population, other than sample data, is almost impossible to obtain for a country such as Guatemala. Finally, the level of aggregation prevents any elaborations or application of the findings of this research to individual political behavior. An application of this type can only occur if the individual composes the basic unit of data aggregation. SUGGESTIONS FOR FUTURE RESEARCH Voting tends to be the product of group and individual factors, most of which vary regionally. Generalizations or conclusions from empirical works often find themselves localized and mathematically deterministic. For this reason, it is recommended that the research design of future electoral analyses incorporate some probabilistic aspects which will enhance the applicability and the validity of the empirical model. Field research is valuable when data are unavailable. The psychological factors mentioned above can only be procured presently by means of a sample survey. More of these types of data procurement 88 are required in electoral analyses. Future research may also concentrate on the analysis of parti- cipation as it varies between groups of people of similar socio- economic background. For example, this type of research could then be used to expand or reject the conclusions derived from present highly aggregated analysis. The present study of political participation in a developing country besides yielding some elements of voting participation levels in a developing country has raised some questions about the nature of political participation. It is hoped that the heuristic nature of the conclusions will serve not to discourage but foster similar electoral research in other developing countries. 89 BIBLIOGRAPHY Adams, Richard S. (1968), The Problems Lf Political Development in Light Lf Recent Socio- Political History Lf Guatemala. Washington: Brookings Institute. Alexander, Robert. (1965), Latin American Politics and Government. New York: Harper and Row. Alschuler, Lawrence R. (1967), "Political Participation and Urbanization in Mexico," An Unpublished Ph.D. Dissertation, Department of Political Science, Northwestern University. Benney, M. and A. Gray. (1956), How People Vote: A_Study pf_Electoral Behavior jp_Greenwich. New York: Humanities Press. Berelson, BA R. W. N. McPhee and P. F. Lazarsfeld. (1966), Voting, Study_ of Opinion Formation in a Presidential Campaign. Chicago. University of Chicago Press. Berry, B. J. L. (1960), "An Inductive Approach to the Regionalization of Economic Development," in N. Ginsberg (ed.), Geography and Economic Development," Resource Paper No. 62, Department of Geography, University of Chicago. (1961), "A Method for Deriving Multifactor Uniform Regions." (1962), "Sampling, Coding and Storing Flood Plain Data, U.S. Department of Agriculture, Farm Economics Division, Agriculture Handbook No. 237. (1965), "Identification of Declining Regions," in W. D. Wood and R. S. Thomain (eds.), Areas gf_Economic Stress jp_ Canada. Kingston, Ontario: (1968), “Essays on Commodity Flows and the Spatial Structure of the Indian Economy," Resource Paper No. 111, Department of Geography, University of Chicago. Billet, Jean. (1968), "L'Expression Politique in Gresivaudan et son Interpretation Geographique," Revue gg_Geographie Alpine, vol. 46, pp. 97-128. Birdsall, Stephen S. (1969), "Preliminary Analysis of the 1968 Wallace Vote in the Southeast," Southeastern Geographer, vol. 9, pp. 55-66. Blalock, Hubert M. (1960), Social Statistics. New York: McGraw-Hill. 90 (1964), Causal Inferences ip_Non-Experimental Research. Chapel Hill: University of North Carolina. Blankstein, George L. (1959),"Political Groups in Latin America,” American Political Science Review, vol. 53, pp. 106-127. Brunn, Stanley 0., Wayne L. Hoffman and Gerald Romsa. (1969), "The Defeat of the Youngstown School Levy - A Study of Urban Political Geography," Southeastern Geographer, vol. 9, pp. 67—79. Brunn, Stanley 0. and Wayne L. Hoffman. (1970a), "The Spatial Response of Negroes and Whites Toward Open Housing: The Flint Referendum," Annals pf_the Association pf_American Geographers, vol. 60, pp. 18-36. Brunn, Stanley 0., Wayne L. Hoffman and Gerald Romsa, (1970b), l'The Youngstown School Levies - A Geographical Analysis in Voting Behavior," Urban Education, vol. 5, pp. 20-48. Burghardt, A. F. (1964), "The Basis of Support for Political Parties in Burgenland," Annals gf_the Association pf_American Geographers, vol. 54, pp. 372-390. Burnett, B. G. and K. F. Johnson. (1968), Political Forces ip_Latin America: Dimensions gf_the Quest for Stability. California Wadsworth Press. Byron, Robert. (1968), "Politics in a Neighborhood of Guatemala City," Sociology, vol. 2, pp. 185-203. Carey, G. W. (1966), "The Regional Interpretation of Manhattan Popula- tion and Housing Patterns Through Factor Analysis," Ihg_ Geographical Review, vol. 56, pp. 551-569. Cehelskey, M. (ed.) (1966), Election Factbook for Guatemala, 1966. Washington: Institute for Comparative Studies of Pol1tical Systems. Cliff, A. and K. 0rd. (1969), "The Problem of Spatial Autocorrelation," in A. Scott (ed.), Studies jp_Regional Science. London: Pion Limited, pp. 1-27. Converse, Philip E. (1958), I'The Shifting Role of Class in Political Attitudes and Behavior," in E. Maccoby, T. Newcomb and E. Hartley, (eds.), Readings 1p Social Psychology. New York: Holt, Rinehart and Winston. Cornford, James. (1964), "The Adoption of Mass Organization by the British Conservative Party," in E. Allardt and Y. Littunen (eds.). Cleavages, IdeaolOgies and Party Systems. Helsinki: Westermarck Society, pp. 400-425. 91 Cope, O. G. (1966), "The 1964 Presidential Election in Chile: The Politics of Change and Access," Inter-American Economic Affairs, vol. 19, pp. 3-29. . (1968), "The 1965 Congressional Election in Chile: An Analysis," Journal pf_Inter-American Studies, vol. 10, pp. 256-276. Cox, Kevin. (l968a), "Spatial Interactional Model for Political Geography,” East Lakes Geographer, vol. 4, pp. 58-76. Cox, Kevin and G. Demko. (1968b), “Conflict Behavior for a Spacio- Temporal Context," Sociological Focus, vol. 1, pp. 58-67. Cox, Kevin. (l968c), "Suburbia and Voting Behavior in the London Metropolitan Area," Annals pf_the Association pf_American Geographers, vol. 58, pp. 111—127. . (l969a), "The Voting Decision in a Spatial Context," in B. R. Chorley, P. Haggett and D. Stoddart (eds.). Progress ip_Geograph§. International Reviews of Current Literature, vol. , Lon on: Edward Arnold, Ltd. . (1969b), "The Spatial Structure of Information Flow and Partisan Attitudes," in M. Dogan and S. Rokkan (eds.). Quantitative Ecological Analyses jg Social Science. Cambridge: M.I.T. Press, pp. 55—68. (1970a), "The Spatial Evolution pf_National Voting Response Surfaces - Theory and Measurement, Department of Geography Ohio State University. Publication No. 10. (1970b), "Residential Relocation and Political Behavior: Conceptual Model and Empirical Tests," Acta Sociologica, vol. 13, pp. 40-53. . (1971), "The Spatial Component of Urban Voting," Economic Geography, vol. 47, pp. 27-35. Dean, V. K. (1949), "Geographic Aspects of the Newfoundland Referendum,” Annals gf_the Association pf American Geographers, vol. 39, pp. 70-77. Dingham, Harvey. (1969), "The Scientific Method and Reproducibility of Results," Multivariate Behavioral Research, vol. 4, pp. 517-522. Eder, Richard. (1966), “Free Vote Urged on Guatemalans," New York Times, February 23, 1966, pp. 22:1. Festinger, L. (1947), "The Role of Group Belongingness in a Voting Situation," Human Relations, vol. 1, pp. 154-181. 92 Fitzgibbion, R. H. (1967), "Measuring Democratic Change in Latin America," Journal pf_Politics, Vol. 29, pp. 129-166. Flinn, W. L. and A. Comacho. (1969), "The Correlates of Voter Participation in a Shantytown Barrio in Bogota, Columbia,” Inter-American Economic Affairs, vol. 22, pp. 47-58. Fruchter, B. (1954), Introduction to Factor Analysis. Princeton, N. J.: Van Nostrand Co. Inc. Germani, Gino. (1970), "Mass Immigration and Modernization in Argentina,” in I. L. Horowitz (ed.). Masses ip_Latin America. New York: Oxford University Press, pp. 319-333. Gil, F. G. (1966), The Political System 9f Chile. Boston: Houghton Mifflin Co. Gillin, J. and K. Silvert. (1956), "Ambiguities in Guatemala," Foreign Affairs, vol. 34, pp. 469-482. Gomez, R. A. (1962), Government and Politics jp_Latin America. New York: Random House. Gullahorn, Jeanne E. (1966), "Multivariate Approaches in Survey Data Processing: Comparison of Factor, Cluster and Guttman Analysis and of Multiple Regression and Correlation Methods," Multivariate Behavioral Research Monograph. Harvey, David. (1969), Explanation jp_Geography. New York: St. Martin's Press. Hoffman, Wayne L. (1966), "A Statistical and Cartographic Analysis of the 1954 and 1964 Urban Renewal Referendums for Columbus Ohio," An Unpublished M.A. Thesis, Department of Geography, Ohio State University. Horowitz, I. L. (1967), “Electoral Politics, Urbanization and Social Development in Latin America, Urban Affairs Quarterly, vol. 2, pp. 3-35. Ingalls, Gerald L. (1970), "A Geographic Analysis of Three Costa Rican Presidential Elections: 1958, 1962 and 1966," An Un- published M.A. Thesis, Department of Geography, University of Florida. James, Preston E. (1969), Latin America (4th edition). New York: Odyssey Press. Johnson, Kenneth. (1967), Analysis gf_the 1966 Presidential Election ip_Guatemala, March, 1966. Election Analysis Series, Washington: Institute for the Comparative Study of Political Systems, Operation and Policy Research, Inc. 93 Kasperson, Roger E. (1965), "Toward a Geography of Urban Politics: Chicago, A Case Study," Economic Geography, vol. 41, pp. 95-107. Kendall, M. G. (1957), A_Course ip_Multivariate Analysis. New York: Hafner Publishing Company. Key, V. O. (1949), Southern Politics ip_State and Nation. New York: Vintage Press. Key, V. 0. and F. Manger, (1959), "Social Determinism and Electoral Decision, The Case of Indiana," in E. Burdick and E. Brodbeck (eds.). American Voting Behavior. New York: Free Press of Glencoe. King, Leslie J. (1966), ”Cross-Sectional Analysis of Canadian Urban Dimensions: 1951 and 1961,” Canadian Geographer, vol. 10, pp. 205-224. (1969). Statistical Analysis ip_Geography. Englewood Cliffs, N. J.: Prentice-Hall, Inc. Kirkpatrick, E. (1964), "The Impact of the Behavioral Approach on Traditional Political Science," in A. Ranney (ed.). Essays pp_the Behavioral Study_pf_Politics. Urbana, Illinois: University of Illinois Press. Lazarsfeld. P., B. Berelson and H. Gaudet. The People's Choice: Hgy_ the Voter Makes Up_Hi§_Mind jp_a PresidentiaT'Campaign. New York: Columbia University Press. Lerner, Daniel. (1958), The Passing pf_Traditional Sociegy, Glencoe: Free Press. Lewis, P. F. (1965), "Impact of Negro Migration on the Electoral Geography of Flint, Michigan: A Cartographic Analysis," Annals gf_the Association pf American Geogrpphers, vol. 53, pp. 1-25. Lipset, Seymour M. (1959), "Some Social Requisites of Democracy: Economic Development and Political Legitimacy," American Political Science Review, vol. 53, pp. 69-105. (1964), "Political Cleavages in Developed and Emerging Politics," In E. Allardt and Y. Littunen (eds.). Cleavages, Ideaologies and Party Systems. Helsinki: Westermarck Society. Mangin, William. (1967), "Poverty and Politics in Cities of Latin America," in W. A. Bloomberg and H. J. Schmandt (eds.) Power, Poverty and Urban Policy. Urban Affairs Annual Review, vol. 2, Sage Publications, pp. 397-432. McCarthy, Harold H. (1960), "McCarthy on McCarthy: The Spatial Distribution of the McCarthy Vote, 1952," An Unpublished Manuscript, Iowa City: Department of Geography, State University of Iowa. 94 McDonald, Ronald, (1967a), "Political Protests and Alienation in Voting - The Case of Columbia," Inter-American Economic Affairs, vol. 21, pp. 3-19. (l967b), "Electoral Systems, Party Representation and Political Change in Latin America," Western Political Quarterly, vol. 20, pp. 694-708. (1969a), "National Urban Voting Behavior: The Politics of Dissent in Latin America," Inter-American Economic Affairs, vol. 23, pp. 3-20. (l969b), "Apportionment and Party Politics in Santiago, Chile," Midwest Journal pf_Political Science, vol. 13, pp. 445-470. . (l969c), "Electoral Behavior and Political Development in El Salvador,” Journal pf_Politics, vol. 21, 00. 397-419. Micklin, Michael. (1969), "Urbanization, Technology and Traditional Values in Guatemala: Some Consequences of a Changing Social Structure," Social Forces, vol. 47, pp. 438-446. Nash, Manning. (1958), "Political Relations in Guatemala," Social Economic Studies, vol. 7, pp. 67-75. Needler, Martin. (1966), I‘Political Development and Military Intervention," American Political Science Review, vol.60, pp. 616-626. Nelson, H. J. (1952), l'The Vernon Area, California: A Study of Political Factor in Urban Geography,“ Annals gf_the Association gj_American Geographers, vol. 42, pp. 177-191. NORM, Computer Institute for Social Science Research. Programmer: Theodore Miller, Department of Geography, University of Iowa. Adapted to CDC 6500 by Robert Wittick, 1970. Pertti, Personen. (1964), “Studies on Finnish Political Behavior,“ in Austin Ranney, Essays pp_the Behavioral Study ngPolitics. Urbana: University of Illinois. Prescott, J. R. V. (1969), "Electoral Studies in Political Geography," in Roger Kasperson and Julian Minghi (eds.), The Structure pf_Political Geography. Chicago: Aldine. 95 Reynolds, David R. (1969), "A Friends and Neighbor Voting Model as a Spatial Interactional Model for Electoral Geography," in K. Cox and R. Golledge (eds.), Behavioral Problems 1p Geography: A_Symposium. Evanston: Northwestern University, Department of Geography, No. 16. Rogers, Everett. (1962). Diffusion gf_Innovations. New York: Free Press. Rokkan, Stein. (1962), IIThe Comparative Study of Political Participa- tion: Notes Toward a Perspective on Current Research," in A. Ranney (ed.), Essays pp the Behavioral Study gj_Politics. Urbana: University of Illinois. Rossi, Peter. (1959), "Four Landmarks in Voting Research,“ in E. Burdick and A. Brodbeck (eds.), American Voting Behavior. New York: Free Press of Glencoe. Rumage, K. W. and M. C. Roberts. (1965), "The Spatial Variations in Urban Left Wing Voting in England and Wales in 1951," Annals pf_the Association pf_American Geographers, vol. 55, pp. 161-178. Rummel, R. J. (1967), "Understanding Factor Analysis," Journal pf Conflict Resolution, vol. 11, pp. 444-480. Russett, Bruce. (1966), "Discovering Voting Groups in the United Nations," The American Political Science Review, vol. 60, pp. 327-340. Simmons, J. W. (1967), "Voting Behavior and Socio-Economic Characteris- tics - The Middlesex East Federal Election, l965,II Canadian Journal pf_Economics and Political Science, vol. 33, pp. 389-400. Sloan, John. (1968), ”The 1966 Presidential Elections in Guatemala - Can a Radical Desiring Fundamental Change Win An Election in Guatemala?" Inter-American Economic Affairs, vol. 22, pp. 15-32. Smith, R. H. T. (1965), "Method and Purpose in Functional Town Classification," Annals pf_the Association pf_American Geographers, vol. 55, pp. 539-548. (1970), "Electoral Frauds and Social Change: The Guatemalan Example," Science and Society, vol. 34, pp. 78-91. Soares, G. and R. L. Hamblin. (1967), "Socio—Economic Variables and Voting for the Radical Left: Chile 1952," American Political Science Review, vol. 61, pp. 1053-1065. Stokes, Donald. (1967), "Parties and Nationalization of Electoral Forces," in W. Chambers and W. Burnham (eds.), The American Party Systems: Stages pf_Political Development. New York: Oxford University Press, pp. 182-202. u! If I. Elli-1.4:; D . ‘. ... he 96 SYMAP, Laboratory for Computer Graphics, Department of City and Regional Planning, Graduate School of Design, Harvard University, Cambridge, Massachusetts, Modified by the Computer Institute for Social Science Research, Technical Report 199; Michigan State University, East Lansing, Michigan, June, 1969. Thomas, Robert N. (1968), "Internal Migration to Guatemala City," Ph.D. Dissertation, Department of Geography, Pennsylvania State University. Thompson, W. E. and J. C. Horton. (1960), l'Political Alienation as a Force in Political Action," Social Forces, vol. 38, pp. 190-195. Tobler, Waldo. (1966), "On Geography and Geometry," Unpublished Manu- script, Department of Geography, University of Michigan. Van de Berghe, Pierre. (1968), "Ethnic Membership and Cultural Change in Guatemala," Social Forces, vol. 46, pp. 514-522. Whetten, Nathan L. (1961), Guatemala, The Land and the People. London: Yale University Press. Wingo, Lowdon, Jr. (1967), "Recent Patterns of Urbanization Among Latin American Countries," Urban Affairs Quarterly, vol. 2, pp. 81-110. Yeates, Maurice. (1968), Ap_1ntroduction to Quantitative Analysis_ip Economic Geography. New York: McGraw Hill.