An Examination of the Policies and Actors that Latin American Voters Associate with Ideological Labels

Transcription

1 An Examination of the Policies and Actors that Latin American Voters Associate with Ideological Labels Kevin Edward Lucas University of Minnesota / SUNY Geneseo Paper prepared for the Annual Meeting of the Midwest Political Science Association, Chicago, IL, April 3-6, Preliminary draft, please do not cite without permission. Direct all correspondence to: luca0058@umn.edu

2 ABSTRACT: Though crucial to understanding Latin America s Left Turn, few scholars have sought to uncover the meanings that Latin American voters attribute to the terms Left and Right. Following Zechmeister (2006-a; 2006-b), I employ Q-method to examine voters understandings of these ideological labels in two countries, Costa Rica and El Salvador. The results of this examination demonstrate that voters in both countries exhibit shared understandings of the policy preferences that are attached to the terms Left and Right. However, whereas voters in El Salvador (where programmatic party-voter linkages are surprisingly strong) also exhibit a shared understanding of which domestic and international political actors are associated with each of these ideological labels, these associations are far less central to voters understandings of the terms Left and Right in Costa Rica (where programmatic party-voter linkages are surprisingly weak). This finding suggests that variation in the strength of programmatic party-voter linkages reflects differences not in the degree to which voters understand what the terms Left and Right imply, but rather in the degree to which they can identify which political actors represent the Left and which represent the Right.

3 Over the course of the past decade, an impressive amount of scholarship has been devoted to examining the causes and implications of Latin America s left turn, the unprecedented series of leftist victories in presidential elections that has spread across Latin America since the beginning of the twenty-first century. The lion s share of this scholarly output has focused on three particular questions. First, what explains the timing of this surge to the left if the extremely high levels of socioeconomic inequality that are prevalent throughout most of the region provides the Latin American Left with a natural base of support for its redistributive agenda, why did the left turn not occur two decades earlier, when a wave of democratic transitions swept across the region? Second, should the left turn be interpreted as a signal of an ideological shift to the left amongst the region s voters, or do these election results simply represent politics as usual in the form of the periodic alternation in power that is an empirical characteristic of presidential democracies? And finally, is it even accurate to refer to the Latin American Left, or should we instead refer to multiple Latin American Lefts? In spite of this demonstrated interest in Latin America s left turn, relatively few scholars have sought to uncover the meanings that Latin American voters attribute to the terms Left and Right. Both in terms of determining whether or not the left turn represents an ideological shift to the left and in terms of understanding cross-country variation within the Latin American Left, this is an unfortunate omission. As Lupu (2009) notes, although explanations of Latin America s left turn ultimately rest on the behaviors of individual voters, few scholars have employed individual-level data in their efforts to explain this wave of leftist presidential victories and to interpret its broader significance. Rather, most studies have relied either on aggregate countrylevel variables, including various retrospective indicators of government performance, or on more purely impressionistic evidence to explain voting behavior in the region (Lupu 2009: 5). 1

4 Early studies that utilized survey data to examine the determinants of voting behavior in Latin America (e.g. Seligson 2007; Arnold and Samuels 2009) examined aggregate public opinion, either at the country level or at the even broader regional level. While Lupu (2009) and Kitschelt et al. (2010) took an important step forward in terms of employing survey data to provide some insight into the question of who votes for the Left in Latin America an approach I have adopted in other work which utilizes survey data to examine the relationships between policy preferences, ideological self-identification, and voting behavior in the region (Lucas 2014-a; 2014-b; 2012) there are clear limitations to the extent to which available survey data can inform us about the meanings that Latin American voters ascribe to the terms Left and Right. One of these limitations concerns the somewhat limited number and relative inconsistency of the policy-related questions that have been included in the two main cross-country public opinion surveys conducted in the region, the Latinobarómetro and the Americas Barometer. More importantly, while the examination of survey data provides important insight into the frequency and/or consistency with which a particular policy preference or symbolic attachment is associated with either the Left or the Right, it does not necessarily allow us to determine which preferences or symbols are considered to be integral components of the meanings that voters assign to these terms. Determining which policy preferences and/or symbolic attachments are central to the voting public s understandings of the ideological labels Left and Right requires a method that allows respondents to not only identify the policy preferences and symbols that they associate with these terms but also to rank their relative salience. It is due to this second limitation of the use of survey data that, following Zechmeister (2006-a; 2006-b), I employ Q-method a scientific approach to the study of human subjectivity and intersubjectivity that counts as one of its main strengths the ability to provide an accurate 2

5 reflection of the broad spectrum of discourses that exist within a larger population from a relatively small sample of participants (Doody et al. 2009: 1130) to examine voters understandings of the ideological labels Left and Right in contemporary Latin America. I do so by examining the results of Q-method studies completed in two Central American countries, El Salvador and Costa Rica, which vary considerably in terms of the degree to which voters leftright self-identification predicts their voting behavior. 1 The results of this examination demonstrate that voters in both of these countries exhibit shared understandings of the policy preferences that are attached to the terms Left and Right. However, whereas voters in El Salvador where programmatic party-voter linkages are surprisingly strong also exhibit a shared understanding of which domestic and international political actors are associated with each of these ideological labels, these associations are far less central to voters understandings of the terms Left and Right in Costa Rica where programmatic party-voter linkages are surprisingly weak. This finding suggests that programmatic party-voter linkages are weak in Costa Rica not because voters do not understand what the terms Left and Right imply, but rather because they cannot identify which political actors are associated with each label. More broadly, it also suggests that variation in the strength of programmatic party-voter linkages in the region may reflect differences not in the degree to which voters understand what the terms Left and Right imply, but rather in the degree to which they can identify which political actors represent the Left and which represent the Right. 1 This research was conducted during the final two months of 2011 (El Salvador) and the first two months of 2012 (Costa Rica). 3

6 II. Case Selection and Method While my decision to use Q-method to examine the meanings that Latin American voters ascribe to the terms Left and Right was influenced by Zechmeister s (2006-a; 2006-b) use of Q- method to examine voters understanding of these ideological labels in Mexico and Argentina, my decision to conduct this examination in El Salvador and Costa Rica stems from the results of research described elsewhere (Lucas 2012; 2014-a; 2014-b) in which I employed data from two series of public opinion surveys (the Latinobarómetro surveys and the Americas Barometer surveys) to measure (1) the extent to which survey respondents policy preferences predict their left-right self-placements and (2) the extent to which survey respondents left-right self-placements predict their voting behavior. Contrary to the prediction that these two relationships would be strongest in those countries that had experienced a relatively high level of economic development already before World War II, longer periods of democratic competition after 1945, and the implementation of comparatively broad policies of social protection for at least the urban population in that time period (Kitschelt et al. 2010: 8-9), this research finds that left-right self-identification is a better predictor of voting behavior in presidential elections in El Salvador than in any other Latin America country, 2 and that the 2 Following Mainwaring and Torcal (2005) and using data the Americas Barometer surveys, I ran logistic regressions on each of the three pairs formed by the three leading vote-getters in each of the 32 presidential elections held in Latin American during the years , using survey respondents left-right self-identification (as measured on a 10-point scale) as the sole independent variable and their reported vote choice as the dependent variable, and then created a summary measure of the strength of programmatic party-voter linkages by calculating the average of the Nagelkerke R 2 values from each of these three logistic regressions, entering.00 instead of the reported Nagelkerke R 2 value if the logistic regression coefficient was not significant at the.95 confidence level. The Salvadoran presidential elections of 2004 and 2009 achieve the two highest scores on this measure (.260 and.277); the scores for the Costa Rican presidential elections of 2002 and 2006 (.020 and.013) are lower than both the mean (.065) and median (.037) values (Lucas 2014-a). 4

7 relationship between policy preferences and left-right self-placement is stronger in El Salvador than in all but two other countries in the region. 3 While the surprising meaningfulness and relevance of ideological labels in El Salvador makes that country an interesting venue for a more thorough examination of the meanings that Latin Americans ascribe to the terms Right and Left, Costa Rica serves as an interesting counterexample because the relationships between policy preferences, left-right self-identification, and voting behavior are much weaker than expected. With the longest history of continual electoral democracy in all of Latin America and with Central America s highest level of socioeconomic development and its most extensive social welfare program, Costa Rica should have developed a party system characterized by strong programmatic party-voter linkages. However, both my own analysis of survey data and the analysis conducted by Kitschelt et al. (2010) find that party-voter linkages in Costa Rica are weakly grounded in issue positions. Therefore, following the same logic that Zechmeister (2006-b: 155) used in choosing Mexico (a country where politicians assign both symbolic and policy content to left-right labels and, to a fair extent, use ideological labels in their political rhetoric ) and Argentina (where the political elite hold much weaker substantive conceptions of the left-right semantics and infrequently use the terms in political discussion ) as her two case studies, I selected Costa Rica as my second case. Now, what is Q-method? First introduced by William Stephenson in 1935 as an instrument designed to study individuals subjective beliefs and attitudes, Q-method differs from the more familiar R-method approach to the analysis of public opinion data in that it aims to 3 Using data from the Latinobarómetro surveys, I constructed a summary measure of the extent to which survey respondents preferences on a range of economic, political, and social issues regional economic integration, privatization, the economic role of the state, income inequality, support for democracy, violent crime, support for the police and the armed forces, the legalization of abortion, gay rights, and the appropriate role for women predict their self-placement on an 11-point Left-Right scale. El Salvador (0.77) ranks third on this measure which is reported in terms of standard deviations above or below the regional mean behind Chile (2.30) and Uruguay (1.88); Costa Rica (-0.38) ranks eleventh on this measure (Lucas 2014-b) 5

8 discover the universe of perspectives that can be found within a population rather than to measure the frequency of occurrence of certain perspectives in that population (the goal of the frequentist approach). 4 Lying at the heart of Q-method is the Q-sort, a process through which Q- participants rank-order a set of Q-statements (expressions of perspectives on the topic of interest to the researcher, drawn from the concourse of possible perspectives) into a quasi-normal distribution according to some explicit rule, as instructed by the researcher. Factor analysis is then used to identify patterns in the placement of the Q-statements in different Q-participants Q- sorts; the resultant factors, which reflect the deeper organizing principles present within the population, are referred to as social perspectives. Compared to the frequentist approach which defines the survey respondent as the subject, the population as the set of possible respondents, and the survey question as the variable Q-method inverts the subjects and the variables. The Q-sort generated by each Q-participant is the variable and the Q-statements are the subjects, while the population consists of the concourse of all possible Q-statements that contain expressions of the perspectives that exist on the topic at hand. One of the advantages of Q-method is the ability to produce statistically significant results with a relatively small number of participants. Webler et al. (2009: 9-10) provide two useful rules of thumb for determining the appropriate number of Q-participants and Q-statements needed to produce a reliable Q-study. Since Q-studies typically uncover 2-5 social perspectives, and since 4-6 individual Q-participants are generally sufficient to define a perspective, the number of participants can be anywhere between eight and thirty. As for the number of Q- statements to include, a Q-statement to Q-participant ratio of 3:1 is often used; ratios below 2:1 are deemed inappropriate. 4 This frequentist approach to the study of individual beliefs and attitudes is dubbed the R-method due to its reliance upon statistical tests used to produce Pearson product-moment coefficients (referred to as r statistics). 6

9 In her study of voters understanding of the terms Left and Right in Mexico and Argentina, Zechmeister used a total of sixty-two Q-statements; thirty-six of these are policyrelated statements, twenty-one are symbolic items (references to certain domestic political actors, political parties, and societal groups), and the remaining five are described as valence items. While I followed Zechmeister s example by including policy-based, symbolic, and valence items that might be attached to the ideological labels Left and Right in my concourse of Q-statements, I also made some noteworthy changes designed to facilitate comparisons between the results of my Q-method research and the results of other work that uses elite survey data and/or public opinion survey data to examine the social perspectives on ideology that exist in Latin America. In the Q-studies I conducted in El Salvador and Costa Rica, I used a total of sixty-seven Q-statements: thirty-nine statements representing seven categories of policy-based items (economic inequality, the economic role of the state, domestic taxation, free trade, support for democracy, law and order issues, and social issues), twenty statements that refer to four categories of symbolic referents (domestic political parties and politicians, groups in society, international political actors, and personal characteristics), six valence items referring to government priorities, and two country-specific statements. 5 Twenty-eight of these statements are either the same as or very similar to the statements used by Zechmeister; the main innovation in my concourse of Q-statements is the inclusion of the symbolic references to four international political actors (the United States, George W. Bush, Hugo Chávez, and Fidel Castro), an inclusion motivated by my finding (Lucas 2012; 2014-b) of a strong relationship between 5 The full list of Q-statements, along with English-language translations, is reported in Appendix A. 7

10 attitudes towards these four actors and Left-Right self-placement amongst Latinobarómetro survey respondents. 6 A second difference between my study and Zechmeister s concerns the instructions given to the Q-participants as they conducted their Q-sort. Whereas Zechmeister randomly assigned her Q-participants to sort the Q-statements either for the term Left or the term Right, I instructed Q-participants to simultaneously sort for both terms. 7 This was done in order to reinforce the idea that these two terms constitute the endpoints of a one-dimensional continuum, just as they are used on public opinion surveys when respondents are asked to place themselves and/or prominent domestic and international political actors on a Left-Rights scale. The third and perhaps most substantial difference between my study and Zechmeister s concerns the recruitment of Q-participants. Ideally, Q-participants are selected not at random, but rather, they are chosen because they have different and well-formed opinions on the matter at hand (Webler et al. 2009: 9). The parallel here is with the selection of survey questions in frequentist research; just as the researcher, when designing his/her survey instrument, does not choose questions at random, but rather selects questions that he/she expects will elicit interesting and meaningful insights, the researcher conducting a Q-study selects Q-participants because he/she believes that these individuals will reveal interesting and meaningful perspectives that are expected to represent the entire range of perspectives that could be found in the population. Since it seems reasonable to assume that people who are actively involved in partisan politics are more 6 Elsewhere (Lucas 2014-b), I utilize data from the Latinobarómetro surveys to examine the bivariate relationship between left-right self-placement and a total of seventy-six variables fifty-six variables that measure policy preferences regarding (1) economic nationalism, (2) the economic role of the State, (3) taxation and public spending, (4) economic inequality, (5) labor unions, (6) support for democracy, and (7) questions regarding the rule of law and tradeoffs between order and liberty, fourteen demographic indicators, and six variables that measure respondents attitudes towards the United States and selected foreign leaders. Attitudes towards George W. Bush, Fidel Castro, and Hugo Chávez are the three best predictors of left-right self-placement in the region; attitudes towards the United States serve as the fifth-best predictor. 7 Appendix B provides a detailed description of the manner in which these Q-sorts were completed. 8

11 likely to have well-formed opinions regarding the meanings they attach to ideological labels than people who are not politically inclined, and since it is also reasonable to predict that different political parties supporters might differ in their understandings of these terms, known party sympathizers may form the ideal set of Q-participants for a study of the policies and actors that are associated with the terms Left and Right. Therefore, whereas Zechmeister recruited her Q-participants at random from the student population of elite private universities in Mexico City and Buenos Aires, I recruited most of my Q-participants by visiting national offices (and, in the case of El Salvador, regional and local offices) of the four main political parties in each country the FMLN, ARENA, PCN, and PDC in El Salvador; the PLN, PAC, PML, and PUSC in Costa Rica to identify party activists, lowlevel party officials, and other known party sympathizers who were willing to complete a Q-sort and a follow-up interview. In both countries, additional Q-participants were recruited amongst university students to represent the population of voters who are not closely affiliated with any political party. 8 III. Findings: El Salvador Both in El Salvador and in Costa Rica, the factor analysis performed on the data provided by the Q-sorts identified two distinct and statistically significant social perspectives regarding the meanings attributed to the terms Left and Right. Forty of the forty-four Q-sorts were identified 8 In both countries, the distribution of partisan Q-participants was designed to roughly approximate the relative electoral strength of the four largest political parties based on their vote percentage in previous election cycles. In El Salvador, this translated to seven Q-participants each from ARENA and the FMLN and two each from the PCN and the PDC. In Costa Rica, there were eight Q-participants from the PLN, six from the PAC, and two each from the PUSC and PML. In Costa Rica, four unaffiliated Q-participants were recruited amongst students of the Universidad de Costa Rica and the Universidad Estatal a Distancia (Cartago campus). In El Salvador, three unaffiliated Q- participants and one activist from a minor party (the CD) were recruited amongst students of the Universidad de El Salvador and the Universidad Don Bosco. 9

12 as defining sorts on one of these two factors. 9 As for the other four Q-sorts, one (completed by a non-partisan Q-participant from Costa Rica) was identified as a negative defining sort (i.e. her Q-sort loaded heavily on the first factor, yet she associated the terms that others had identified as being related to the Left with the Right, and vice versa), while the remaining three (one completed by a PLN activist, one by an ARENA activist, and one by a PCN activist) did not load heavily on any factor. The first factor that emerged from the Q-sorts conducted in El Salvador identifies eleven Q-statements that are closely associated 10 with the Right and fifteen Q-statements closely associated with the Left; these statements are reported in Tables 1 and An examination of these twenty-six highly salient Q-statements provides meaningful insights into the use of the ideological labels Left and Right in El Salvador. The eleven statements most closely associated with the Right can all be placed in one of two categories: five are references to domestic political actors or groups in society, while the other six express agreement with neoliberal economic beliefs and/or policies. Similarly, fourteen of the fifteen statements most closely associated with the Left can also be placed into two categories: eight statements refer to international or domestic political actors or to groups in society, while another six express support for expanding the state s role in managing the national economy. In every single instance including the remaining Q-statement that is closely associated with the Left (support for reinstating the colón) 12 the 9 The PQMethod software (Schmolck 2012) used to perform factor analysis on Q-sort data identifies a maximum of seven factors from any set of Q-sorts. For each of the two Q-studies I conducted, only the first two factors had an Eigenvalue greater than The Q-statements that I describe as closely associated with the Right are those with a Z-score greater than 1.00; those described as closely associated with the Right are those with a Z-score less than The four tables included in Appendix C report the Z-score for each Q-statement on each of the two significant factors discovered in each country. 12 Although this policy does not map onto the traditional Left-Right divide regarding the appropriate economic role of the state, the identification of support for reinstating the colón as being closely associated with the Left is consistent with the placement of the other twelve economic policy-related Q-statements that appear in Tables 1 and 2 inasmuch as the FMLN opposed dollarization when it was implemented in 2001 (by an ARENA president, 10

13 placement of these specific economic beliefs and policy preferences corresponds to the public policy stances offered by El Salvador s two main political parties, ARENA and the FMLN. Similarly, the symbolic identifications associated with the Left and the Right on this factor are also correct ARENA, that party s founder (Roberto d Aubuisson), former ARENA president Alfredo Cristiani, businessmen, and the rich are all associated with the Right; Hugo Chávez, Fidel Castro, communists, the FMLN, long-time FMLN leader Schafik Handal, workers, the poor, and the people are all associated with the Left. Table 1. Q-Statements Associated with the Right, El Salvador, Factor 1. No. Statement Z-Score 9 Privatizations have benefited the national economy Identify oneself with the rich Identify oneself with Roberto d Aubuisson Capitalism is the best economic system for this country Identify oneself with ARENA Economic production should be left in the hands of the private sector Identify oneself with Alfredo Cristiani Identify oneself with businessmen The free trade agreement with the United States will benefit the national economy It is more important to increase economic production than to reduce inequality Economic growth will result in poverty reduction Francisco Flores) and made reinstating the colón one of the central components of its campaign in the 2003 legislative election and the 2004 presidential election. 11

14 Table 2. Q-Statements Associated with the Left, El Salvador, Factor 1. No. Statement Z-Score 4 Combating poverty and inequality should be one of the government s top priorities Identify oneself with the FMLN Some public services are too important to be left in the hands of the private sector Identify oneself with the poor Identify oneself with Schafik Handal The income distribution in this country is unjust Believe that... the colón should be reinstated as the national currency Identify oneself with the communists Identify oneself with Hugo Chávez The State should... guarantee that people have access to basic foodstuffs Identify oneself with Fidel Castro The State should spend more on education, health care, and other social services Identify oneself with this country's people State is responsible for reducing the differences between the rich and the poor Identify oneself with workers Just as interesting as the Q-statements that do emerge as highly salient components of the meanings attributed to the terms Left and Right on this factor are those statements that do not appear to be closely associated with these ideological labels. None of the ten Q-statements that refer to law and order issues, none of the four Q-statements that refer to social issues, and none of the six Q-statements that address popular support for democracy are highly salient to this social perspective on the meanings of the terms Left and Right. Quite plainly, this factor is defined by two considerations: policy preferences regarding the economic role of the state and symbolic attachments. Whereas this examination of the Q-statements that emerge as salient to this first social perspective on ideological labels in El Salvador does not uncover any surprises in terms of the policy preferences, political actors, and group identities associated with the terms Left and Right, the same cannot be said of the second factor that emerges from the Q-sorts conducted in El Salvador. Thirteen Q-statements are closely associated with the Right while ten are closely 12

15 associated with the Left on this factor; these statements are reported in Tables 3 and 4. If the first factor can be labeled as a Symbolic / Economic Role of the State perspective on the meaning of ideological labels, this second factor can perhaps best be described as a Nationalist Right / Symbolic Left perspective. As shown in Table 4, six of the seven statements that are most closely associated with the Left along this factor consist of symbolic references to domestic and international political actors. While the three policy-related variables that are closely associated with the Left along this factor opposition to dollarization, support for maintaining legal protections for persons accused of breaking the law, and support for the legalization of abortion are all consistent with policy statements made by the FMLN, it is noteworthy that none of these refer to the economic role of the state, the main policy dimension that differentiates the Right and the Left along the first social perspective. Table 3. Q-Statements Associated with the Right, El Salvador, Factor 2. No. Statement Z-Score 44 Identify oneself with this country's people Democracy is always preferable to any other form of government Identify oneself with ARENA The State should be responsible for the wellbeing of all citizens Identify oneself with workers Public security should be one of the government s top priorities Environmental protection should be one of the government s top priorities Be patriotic Be honest The best way to combat public insecurity is to create more jobs and opportunities Combating corruption should be one of the government s top priorities Protecting human rights should be one of the government s top priorities This country should try to attract more international investment

16 Table 4. Q-Statements Associated with the Left, El Salvador, Factor 2. No. Statement Z-Score 31 Identify oneself with the FMLN Identify oneself with the communists Identify oneself with Hugo Chávez Identify oneself with Fidel Castro Identify oneself with Schafik Handal Believe that... the colón should be reinstated as the national currency Identify oneself with Mauricio Funes Be corrupt The government should always obey the law Support the legalization of abortion The picture of the Right that emerges from this second social perspective on ideological labels in El Salvador defies any simple explanation; in addition to four nationalist/populist statements (identification with workers and with the people and the positive characteristics of honesty and patriotism), all four non-economic valence items, support for democracy, and identification with ARENA, three policy-related statements are strongly associated with the Right on this factor. However, while one of these policy preferences (the encouragement of foreign investment) is consistent with the policies implemented by ARENA presidents, the other two (the belief that the State should be responsible for the wellbeing of its citizens and the belief that the best way to combat public insecurity is to create more jobs and more opportunities for young people) are not only not policy preferences traditionally associated with the Right in Latin America, but they also stand in stark contrast to the wave of privatizations and series of iron fist programs designed to combat the country s street gang epidemic (Mano Dura and Super Mano Dura) that were introduced by ARENA presidents. The identity of the Q-participants whose Q-sorts are identified as defining sorts on each of these two factors is suggestive. As shown in Table 5, which indicates the number of Q- participants affiliated with each party whose Q-sorts are identified as defining sorts on each of 14

17 the two social perspectives that emerged from this Q-study (with the political parties ordered according to their presumed placement on the Left-Right scale), fourteen of the twenty-two Q- sorts are identified as defining sorts on the first factor, including those of all seven FMLN activists and the sole CD activist. On the other hand, while only one ARENA activist s Q-sort is identified as a defining sort on this first factor the factor that combines symbolic referents and policy preferences regarding the economic role of the state in ways that are consistent with the policy statements and symbolic referents associated with El Salvador s two main political parties five of the six Q-participants whose Q-sorts are identified as defining sorts on the second factor the factor that includes relatively little (and somewhat inconsistent) policy content are affiliated with ARENA, a pattern that explains why honesty is associated with the Right, and corruption with the Left, on this factor. Table 5. Defining Sorts by Party Affiliation, El Salvador Party Affiliation Factor 1 Factor 2 None FMLN CD PDC PCN ARENA No party affiliation Total Eigenvalue Explained Variance 35% 17% Of course, further research would be necessary to determine whether this pattern whereby persons affiliated with the FMLN express an understanding of the terms Left and Right that is grounded in economic policy preferences in a manner that is entirely consistent with the public discourse authored by the country s two main political parties while persons affiliated with ARENA exhibit an essentially non-programmatic understanding of these ideological labels 15

18 extends beyond the small sample of party activists who participated in this Q-study. The historical development of El Salvador s two main political parties provides some reason to predict that this pattern might indeed be found in a national sample of Salvadoran voters. ARENA, founded in 1981, had made few efforts to promote its ideological commitments prior to suffering a surprise setback in the 2003 legislative elections, an electoral defeat that prompted the party to inaugurate a permanent political training center in San Salvador and to embark upon an ambitious ideological outreach program in preparation for the 2004 presidential election. On the other hand, the various social movements and later guerrilla organizations that were formed during the 1960s and 1970s and that eventually merged to create the FMLN had since their birth relied on promoting their ideological distinctiveness to win popular support. As a prototypical example of an externally mobilized party that emerged from outside the ruling circles of power, the FMLN initially push[ed] for programmatic competition because programs [were] all they [had] to offer (Bornschier 2009: 8). Although the grassroots organizing efforts that characterized the opposition to El Salvador s military regime were scaled back to a considerable degree during the country s civil war and also during the period immediately following the war s conclusion when the FMLN underwent a difficult transition from guerrilla army to political party, they were never completely abandoned. Beginning in 2002, when the revolutionary Left (the ortodoxos), led by long-time Communist Party leader Schafik Hándal, gained complete control of the FMLN, the party s efforts to broadcast its ideological commitments were revitalized, and the FMLN opened a number of training centers designed to provide ideological instruction to party leaders, candidates, and rank-and-file party members (Gutiérrez Castro 2010). 16

19 If the FMLN has been more successful than ARENA in terms of making ideological education the starting point through which sympathizers are incorporate into the party, as it almost certainly has, 13 then it should perhaps come as no surprise that FMLN officials, activists, and sympathizers exhibit a more ideologically coherent understanding of the terms Left and Right than do their ARENA counterparts. IV. Findings: Costa Rica The first factor that emerges from the Q-sorts conducted in Costa Rica identifies thirteen Q-statements that are closely associated with the Right and thirteen Q-statements closely associated with the Left; these statements are reported in Tables 6 and 7. Quite clearly, economic policy preferences associated with support for or opposition to neoliberal economic reforms distinguish between the Right and the Left on this first factor; the twenty-three policy items that load heavily on this factor are all related to questions of the economic role of the state, economic inequality, domestic taxation, and free trade, and they are all located at the expected end of the ideological spectrum. Similarly, the three symbolic associations that are highly salient components of the meanings attributed to the terms Left and Right on this factor identification with businessmen (Right), with workers (Left), and with communists (Left) are also consistent with a purely economic policy-based conception of political ideology. 13 According to Orlando Cocar Romano, the manager of ARENA s political training center (the Instituto de Formación Política Mayor Roberto d Aubuisson), approximately 60,000 ARENA sympathizers received ideological instruction during the twelve months between the 2003 legislative election and the 2004 presidential election, and an additional 40,000 people had been trained as of November Nonetheless, during an interview (conducted November 3, 2011, in San Salvador), Cocar expressed admiration at the breadth and success of the ideological education efforts undertaken by the FMLN. 17

20 Table 6. Q-Statements Associated with the Right, Costa Rica, Factor 1. No. Statement Z-Score 22 Economic production should be left in the hands of the private sector Capitalism is the best economic system for this country Economic growth will result in poverty reduction Favor eliminating taxes on imports and exports in order to encourage international trade It is more important to increase economic production than to reduce inequality Economic growth should be one of the government s top priorities Taxes should be kept to a minimum Privatizations have benefited the national economy Each person/family should be responsible for its own wellbeing The best way to combat public insecurity is to provide more resources to the police Identify oneself with businessmen The free trade agreement with the United States will benefit the national economy This country should try to attract more international investment Table 7. Q-Statements Associated with the Left, Costa Rica, Factor 1. No. Statement Z-Score 1 State is responsible for reducing the differences between the rich and the poor Some public services are too important to be left in the hands of the private sector The State should spend more on education, health care, and other social services The income distribution in this country is unjust Combating poverty and inequality should be one of the government s top priorities The State should be responsible for the wellbeing of all citizens Identify oneself with workers The State should... guarantee that people have access to basic foodstuffs Identify oneself with the communists This country's economic problems are a consequence of... international markets Environmental protection should be one of the government s top priorities Protecting human rights should be one of the government s top priorities It is the State s responsibility to provide jobs A comparison of this first, Economic Role of the State factor that emerges from the Q- study conducted in Costa Rica to the first factor uncovered from the Q-study conducted in El Salvador identifies interesting similarities and differences. In both countries, economic policy preferences are the only policy items that are highly salient to the dominant social perspective on the meanings associated with the terms Left and Right neither support for democracy, preferences on law and order issues, nor attitudes on social issues play an important role in 18

21 distinguishing the Left from the Right along the dominant social perspective identified in these two Q-studies. Where the first factor identified in these two countries differs quite dramatically is in terms of the symbolic referents that are associated with ideological labels. In El Salvador, five of the six Q-statements expressing identification with domestic political actors and two of the four Q-statements that express identification with international political actors load heavily onto the first factor; in Costa Rica, none of these ten symbolic references to political actors is highly salient. I examine the implications of, and possible explanation for, this difference in the final section. Nine Q-statements are closely associated with the Right and nine are closely associated with the Left on the second social perspective on the meaning of ideological labels in Costa Rica; these statements are reported in Tables 8 and 9. As in El Salvador, the second factor that emerges from the Costa Rican Q-study defies any simple explanation. The Q-statements closely associated with the Right along this factor represent a rather confused mix of economic and noneconomic valence items, policy preferences, and symbolic associations with social groups, while the picture of the Left that emerges on this factor is to some degree steeped in symbolic associations with international political actors. A number of the relatively few policy items that are highly salient on this factor appear to be placed incorrectly; for instance, the belief that the state should be responsible for the wellbeing of all citizens is associated with the Right, while the belief that economic production should be left in the hands of the private sector is associated with the Left. Perhaps the best way to summarize this social perspective would be to label it Democratic Right / Communist Left, in reference to the Q-statement that is most strongly associated with each pole of the ideological spectrum on this factor. 19

22 Table 8. Q-Statements Associated with the Right, Costa Rica, Factor 2. No. Statement Z-Score 41 Democracy is always preferable to any other form of government Combating poverty and inequality should be one of the government s top priorities Protecting human rights should be one of the government s top priorities Combating corruption should be one of the government s top priorities The best way to combat public insecurity is to create more jobs and opportunities Public security should be one of the government s top priorities The State should be responsible for the wellbeing of all citizens Identify oneself with the poor Identify oneself with this country's people Table 9. Q-Statements Associated with the Left, Costa Rica, Factor 2. No. Statement Z-Score 27 Identify oneself with the communists Identify oneself with Fidel Castro In certain circumstances, it may be better to have a non-democratic government Identify oneself with Hugo Chávez It is better to live in a society where order is maintained Be corrupt Support the legalization of abortion Government should not be allowed to censor the media Economic production should be left in the hands of the private sector The identity of the Q-participants whose Q-sorts are identified as defining sorts on each of these two factors is again illustrative. As reported in Table 10, the Q-sorts of all six Q- participants affiliated with the PAC and of seven of the eight Q-participants affiliated with the PLN are identified as defining sorts on the first factor. The group of six Q-participants whose Q- sorts are identified as defining sorts on the second factor include both of the Q-participants affiliated with the PUSC and both of the Q-participants affiliated with the PML. Again, as in El Salvador, persons affiliated with parties that are located towards the left end of the political spectrum exhibit a policy-based understanding of the terms Left and Right while those who are affiliated with right-of-center parties demonstrate a much more idiosyncratic understanding of these ideological labels. This finding, which again would need to be confirmed by further 20

23 research, is rather surprising, given the non-polarized, largely non-ideological nature of Costa Rica s main political parties. Table 10. Defining Sorts by Party Affiliation, Costa Rica Party Affiliation Factor 1 Factor 2 None PAC PLN PUSC PML No party affiliation 1 2 1* Total * Eigenvalue Explained Variance 48% 13% * The Q-sort of one nonpartisan Q-participant is identified as a defining sort on factor 1, but with a negative factor loading. V. Conclusions In this paper, I have employed the results of Q-method studies conducted in El Salvador and Costa Rica, two countries that vary considerably in terms of the extent to which left-right self-placement predicts voting behavior, in order to examine the manner in which political actors in contemporary Latin America understand the ideological labels Left and Right. Three main findings merit emphasis. First, these studies suggest that, when focusing on the relative salience of different policy areas to the social perspectives on the meaning of the terms Left and Right, economic policy preferences are paramount in both countries. In both countries, a number of economic policy preferences are strongly associated with the meanings that are attached to these two ideological labels. Moreover, at least along the dominant factor in each country, statements related to economic policy preferences appear in the correct place policy preferences in line with the 21

24 Washington Consensus are uniformly associated with the Right, while preferences for greater state involvement in the economy are consistently associated with the Left. Second, these Q-method studies provide evidence that contradicts the suggestion that the Latin American Left can be reduced to anti-american sentiments (Arnold and Samuels 2009). Although public opinion surveys in the region do find that respondents attitudes towards the United States and selected foreign leaders strongly predict their left-right self-placement, neither identification with the United States nor identification with George W. Bush emerge as highly salient components of either one of the two significant social perspectives that were identified in either El Salvador or Costa Rica. Finally, the finding that domestic political actors are strongly associated with the terms Left and Right in El Salvador but not in Costa Rica suggests that variation in the strength of programmatic party-voter linkages in these two countries reflects differences not in the degree to which voters understand what the terms Left and Right imply, but rather in the degree to which they can identify which political actors are on the Left and which are on the Right. Costa Rica exhibits a lower than expected degree of programmatic political competition not because voters are ignorant of the economic policy alternatives associated with the labels Left and Right, but because they cannot identify the ideological placement of the country s main political parties. Salvadoran voters, for their part, vote correctly (in the sense that individual voter s policy preferences predict left-right self-placement which in turn predicts voting behavior) not because El Salvador enjoys a high level of socioeconomic development (it does not), a long history of electoral democracy (it does not), or extensive social welfare programs (it does not), but because the country s main political parties (particularly the FMLN) have succeeded in educating voters about their ideological preferences. 22

25 Bibliography Alcántara, M. and C. Rivas The Left-Right Dimension in Latin America Party Politics. Paper presented at the 2006 Annual Meeting of the American Political Science Association, Philadelphia. Altman, D. J. P. Luna, R. Piñeiro, and S. Toro Partidos y sistemas de partidos en América Latina: Aproximaciones desde la encuesta a expertos Revista de Ciencia Política, 29(3): Arnold, J. and D. Samuels Public Opinion and Latin America s Left Turn. In S. Levitsky and K. Roberts, eds., Latin America s Left Turn. Cambridge: Cambridge University Press. Bornschier, S Cleavage Politics in Old and New Democracies. Living Reviews in Democracy, 1. Castañeda, J Latin America s Left Turn. Foreign Affairs, 85(3): Cleary, M Explaining the Left s Resurgence. Journal of Democracy, 17(4): Colburn, F Latin America at the End of Politics. Princeton: Princeton University Press. Corrales, J The Backlash against Market Reforms in Latin America in the 2000s. In J. Domínguez and M. Shifter, eds. Constructing Democratic Governance in Latin America, Third Edition. Baltimore: The Johns Hopkins University Press. Doody, D. G., P. Kearney, J. Barry, R., Moles, B. O Regan Evaluation of the Q-method as a method of public participation in the selection of sustainable development indicators. Ecological Indicators, 9: French, J. D Understanding the Politics of Latin America s Plural Lefts (Chávez/Lula): social democracy, populism and the convergence on the path to a post-neoliberal world. Third World Quarterly, 30(2): Gutiérrez Castro, E. P Estudio comparado sobre los procesos de socialización y formación política de los partidos en El Salvador y Nicaragua: casos FMLN y FSLN. M.A. thesis submitted to the Universidad Centroamericana José Simeón Cañas (Antiguo Cuscatlán, El Salvador). Hagopian, F. and S. Mainwaring, eds The Third Wave of Democratization in Latin America: Advances and Setbacks. Cambridge: Cambridge University Press. Kitschelt, H., K. Hawkins, J. Luna, G. Rosas, and E. Zechmeister Latin American Party Systems. Cambridge: Cambridge University Press. Levitsky, S. and K. Roberts, eds Latin America s Left Turn. Cambridge: Cambridge University Press. Lora, E. and M. Olivera The Electoral Consequences of the Washington Consensus. Economía, 5(2): Lucas, K. E a. Programmatic Political Competition in Contemporary Latin America. Unpublished manuscript b. The Meanings and Meaningfulness of Ideological Labels in Contemporary Latin America. Unpublished manuscript Programmatic Political Competition Where We Least Expect It: Party System Development in El Salvador. Paper presented at the 2013 Annual Meeting of the Western Political Science Association, Hollywood Mass Understandings of the Terms Left and Right in Contemporary Latin America. Paper presented at the 2012 Annual Meeting of the Midwest Political Science Association, Chicago. Lupu, N Electoral Bases of Leftist Presidents in Latin America. Paper presented at the 2009 Annual Meeting of the American Political Science Association, Toronto. Madrid. R The Origins of the Two Lefts in Latin America. Paper presented at the 2009 Annual Meeting of the American Political Science Association, Toronto Indigenous Voters and Party System Fragmentation in Latin America. Electoral Studies, 24(4):