Mexicans and Colombians at Home and Abroad: A Comparative Study of Political Engagement

Although a large number of democracies have extended political rights to expatriates, relatively little is known about the depths of transnational political engagement. How attentive are expatriates to politics in the country of origin? When expatriates judge leaders “back home,” are their evaluations based on the same ideological considerations as those of citizens in the country of origin? Drawing from original surveys conducted during presidential elections in Mexico (2006) and Colombia (2010), in which both emigrants and citizens within the country were sampled, this study addresses these questions. The results indicate that for each nationality group, living abroad is not associated with a drop in political attentiveness, and time abroad does not in and of itself depress attention to politics from a distance. Moreover, emigrants and individuals in the country of origin do not vary in the extent to which ideological preferences are used to judge presidents, which is a key marker of political sophistication. These results suggest that in the context of Mexican and Colombian politics, living abroad does not markedly diminish the potential for effective democratic engagement.


Introduction
Some thirty-eight million individuals of Latin American origin reside today outside of their native country, with the majority of emigrants settling in the United States in search of better employment opportunities (United Nations Department of Economic and Social Affairs 2017). Immigration scholars have long noted that the process of migration can disrupt one's personal life in many ways (see, e.g., Portes and Rumbaut 2006). It can be difficult for an emigrant to maintain close ties to family and friends left behind; new roles and identities in the receiving country may displace earlier cultural customs and traditions; and migrants must often accept employment positions after settling abroad that they had not previously held. This article considers whether emigration leads as well to disengagement from politics in the country of origin.
Do expatriate citizens living outside of the national territorial boundaries become less attentive to and less cognitively invested in public affairs and campaigns in that country? Or does political engagement persist even after emigration?
Such questions have become increasingly important in recent decades, as many of the major migrantsending countries of Latin America have extended political rights to expatriates, including the right to vote either in person or via absentee ballot. In some cases, this expansion of the franchise came about in reaction to emigrant lobbying. In others, legislators extended voting rights to expatriates in an effort to reinforce symbolic ties to their country of birth, a move that could in turn lead to increased remittances and other forms of economic investment. Transitions to democracy in ex-authoritarian regimes have also favored the inclusion of the politically displaced for either democratic or legitimacy-building purposes or for both. Strategic political considerations have likewise shaped voting reform initiatives. Domestic partisan leaders are much more open to enfranchising expatriates when they anticipate that these future voters will turn out from a distance to support them (Bauböck 2005;Caramani and Grotz 2015;Collyer 2014a;Escobar 2007Escobar , 2017Lafleur 2011Lafleur , 2013McCann, Cornelius, and Leal 2009).
Whatever the political dynamics, it is incumbent on comparative researchers investigating the quality of political engagement and representation in democracies to examine public opinion across the full range of the "public"-including newly incorporated emigrants. Yet to date, strikingly little scrutiny has been given to the political attitudes of citizens who live outside the territorial boundaries of a democratic nation and how these attitudes compare to those of individuals within the country. Turnout rates for expatriates who are eligible to take part in sending-country elections tend to be quite low (IFE 2006;Lafleur and Calderón Chelius 2011). This may be a reflection of apathy and disengagement; emigration could breed a sense of withdrawal or detachment from the political system. Before jumping to this interpretation, however, it is important to trace the orientations of expatriates more directly and systematically through survey interviewing. After all, many administrative or logistical factors, such as registration requirements or the mode of balloting, may depress transnational turnout rates regardless of an expatriate's level of motivation to take part from the distance.
Drawing from original surveys conducted during the presidential elections of 2006 in Mexico and 2010 in Colombia-investigations in which both expatriates living in the US and citizens within each country were randomly sampled and interviewed at the same point in time-we model the difference that residence outside of the country makes. No previous research on transnational political engagement has drawn from such large-N comparative surveys that were fielded in the heat of a major national campaign. The findings from these studies indicate that in both the Mexican and Colombian contexts, emigrants are not appreciably less attentive to public affairs or less politically competent vis-à-vis compatriots living in the sending country. In the following section, a wider theoretical backdrop is provided for these results. Then, in the third part, the empirical models themselves are detailed.

Theoretical Backdrop: Assessing Political Engagement
The degree to which citizens in a mass public are engaged in political affairs bears directly on the caliber of representation. The rich scholarly literature on public opinion in contemporary democracies offers a variety of perspectives on what it means to be "politically engaged." Two traits among others stand out as essential: a desire to follow politics, and a capacity to render political judgments based on overarching ideological postures or values. For this reason, no academic survey of public opinion would be complete without items gauging whether an individual pays attention to politics and whether concepts such as, say, "being on the ideological left," "being on the right," "being a liberal," or "being a conservative" are salient and meaningful.
Studies conducted in both the established and developing democracies around the world generally find that most citizens are only moderately or modestly interested in public affairs (see, e.g., Almond and Verba 1963;Converse 1964;Dalton 2008;Delli Carpini and Keeter 1996;Domínguez and McCann 1996;Helms et al. 2016;Prior 2010). This is understandable given the potentially high costs of becoming engaged in politics, in terms of both time and money, and it is not necessarily detrimental to democratic performance provided a sufficiently large minority within the public follows politics closely (Downs 1957). A critically important question is how attention to politics varies across subgroups within an electorate. In many countries, political interest correlates with education level, age, gender, family income, and other indicators of socioeconomic status; the higher the SES, the greater the tendency to follow public affairs. Correlations such as these have raised many red flags among researchers in that they signal potential inequities in political representation (Schlozman, Verba, and Brady 2012).
Does living outside the country similarly depress attention to politics? It is entirely possible that expatriates grow detached from public affairs after settling abroad. Traditional models of immigrant assimilation would lead to this expectation (e.g., Gordon 1964). However, much research on childhood socialization demonstrates that general orientations toward politics take hold at a young age, well before one is eligible to vote, and persist in large part throughout the life cycle (Prior 2010;Sears 1983). People who are exposed to politics as children and encouraged to follow the news likely carry these habits with them into adulthood. This could include emigrants, who in many cases depart from their native country after their formative childhood years. Residing abroad thus may not have an effect on interest in sending-country politics, at least once conventional SES background factors linked to political engagement are taken into account. In fact, some studies have found that pre-migration electoral experience is a good prediction of participation among expatriates in receiving countries (Waldinger, Soehl, and Lim 2012). It is less clear, however, given a lack of prior research in this area whether early childhood socialization habits continue to orient migrants toward politics in the country of birth, though this is theoretically plausible.
Yet nonresident citizens may be less attentive to specific electoral contests compared to co-nationals within the country. In Mexico, Colombia, and other countries with large extraterritorial populations, political parties and campaign organizations do not engage in much overt outreach to expatriates. 1 For the most part, the accoutrements of modern campaign politics-open-air rallies, candidate appearances, voter mobilization drives-are missing in emigrant communities. Of course, migrants have informal opportunities to stay informed about candidates and campaigns in their country of birth. Interpersonal networks have been one of those channels of information. Many emigrants, except for refugees who must flee the country and not look back, send funds "home," and large numbers are in contact at least occasionally with family and friends there (DeSipio et al. 2003). Such exchanges could conceivably transmit a considerable amount of information about sending-country politics (cf. Ahn, Huckfeldt, and Ryan 2014;Cramer Walsh 2004;Finkel and Smith 2011;Parker, Parker, and McCann 2008). Recently, satellite television and ethnic media outlets as well as social networks have provided permanent and instantaneous sources of information of home-country news. While access to these resources is by no means uniform among expatriates, it has clearly reduced the information gap between domestic and external constituencies (Lafleur 2013). It might still be overly burdensome for some expatriates to pay attention to specific details of electoral campaigning; but for others, technology has made this task easier. In spite of the reduction of the gap and the fact that it might affect some people more than others, access to information from abroad and to discussions concerning this information is a challenge to expatriates that domestic voters do not have to face.
Continuing with this line of reasoning, the roots of political evaluations for emigrants may differ substantially from those of co-nationals living within the sending country. Political competition within democracies typically takes place along an ideological spectrum-the left versus the right, or liberals versus conservatives. Among political scientists, one widely accepted traditional marker of political sophistication within a mass public is how well one understands the ideological stakes of policy debates and evaluates public officials or party platforms based on ideological criteria (Converse 1964). This can be a demanding standard for citizens to meet, given that ideological labels refer to complex abstract beliefs about how politics work or should work. Political sophisticates meet this standard. Those citizens who opt to live outside of the national borders could become less politically sophisticated in this sense, in that they might be less exposed to the ideological signals and cues disseminated by elites in the sending country. Much prior research conducted within conventional electoral settings suggests that such stimuli can foster or reinforce mass-level ideological sophistication (e.g., Nie, Verba, and Petrocik 1976;Zaller 1992).
The analyses below probe patterns of political interest and ideological sophistication among Mexican and Colombian expatriates in the United States. The decision to focus on these groups stems from several considerations. First, concentrating on only a single nationality group would clearly limit our ability to extrapolate from any findings. As discussed in greater detail in the following section, the logistical challenges in gathering comparable data on Latin American citizens at "home" and "abroad" are formidable. Doing this in two national contexts is doubly formidable but is nonetheless warranted in order to strive for broader generalizations.
It is reasonable to gather data on Mexicans and Colombians in particular, given that they make up a substantial share of the foreign-born Latino population in the United States. Some thirteen million Mexicans, or approximately one-tenth of the Mexican-born population, live outside of the country, nearly all in the US. Mexicans form by far the largest immigrant group in the United States, with approximately 27 percent of the foreign-born population of the country being of Mexican origin. Historically, Mexicans have tended to settle in California, Texas, and other states in the American Southwest. These destinations remain popular, but in recent decades, many of the small cities and towns in Midwestern and Southeastern states have become attractive, for example, Indiana, Iowa, Kentucky, Georgia, and the Carolinas (López 2015b).
Colombian migration to the US has been somewhat more limited. Approximately 3.4 million Colombians (8 percent of the Colombian-born population) live outside of Colombia. After Venezuela, the United States is the principal destination country, with approximately 650,000 Colombians now residing there (López 2015a). Compared to Mexican migration, Colombian settlement in the United States is relatively recent, having started as a trickle in the interwar period, increasing in the mid-twentieth century as Colombia faced a civil war (La Violencia). Colombian migration surged in the 1980s, as a response to an economic crisis, and then again in the 1990s and early 2000s as a major political crisis deepened. Substantial Colombian communities have become established in two major metropolitan areas, Miami and New York. Like emigrants from other South American nations, emigrants from Colombia are characterized by higher levels of education and socioeconomic resources-generally the ones able to afford the more distant and therefore more costly migration-than the typical Mexican or Central American migrant. While a significant proportion of Colombian emigrants have been forced out by the violence that has afflicted the country for many years, most of these forced migrants have gone to Europe and other Latin American countries, which offered better policies for refugees than the United States (Silva and Massey 2015).
It is further fitting to focus attention on these two nationality groups since Mexicans and Colombians residing abroad retain in principle many of the political rights and the same civic standing as co-nationals who live within the territorial boundaries of their respective country. In 1996, the Mexican national legislature passed a constitutional reform to allow Mexican citizens living north of the border to be dual nationals and participate in presidential elections in that country. Legislation to implement absentee balloting, however, was blocked that year by partisan infighting. Almost a decade later following the historic fall of the longruling Institutional Revolutionary Party (PRI), the logjam in the Mexican Congress was broken, and expatriates were permitted to vote via postal ballots in advance of that country's 2006 presidential elections. Analysts predicted that hundreds of thousands, if not millions, of emigrants would vote. In the end, however, only 33,131 ballots from abroad arrived, much less than 1 percent of the Mexican-born adult population that resided in the United States (McCann 2012). The high costs of transnational balloting through registered mail was said to have played a major part in depressing participation rates that year, and the expatriate turnout rate did not change markedly in the next round of presidential elections in 2012. Nevertheless, an important threshold was crossed as residence was eliminated as a condition for the exercise of the franchise. A more recent law (Ley General de Instituciones y Procedimientos Electorales 23 de mayo de 2014) allows Mexicans residing abroad to elect legislators to the upper house.
In the Colombian context, the national government that followed the deposed military regime in 1961 enacted a law that permitted Colombians with permanent residence outside of the country to vote in the presidential elections the following year. Decades later, a new constitution that took effect in 1991 allowed dual citizenship and the participation of expatriates in congressional elections, a right that was first exercised in 1998 for Senate elections and in 2002 for a newly created special district in the House of Representatives. Unlike in the Mexican case, the extension of electoral rights to Colombian expatriates was not delayed due to partisan opposition. The implementation of external voting rights was quite gradual, however, due to the fact that expatriates were only one of many minority groups in the Colombian context seeking to become incorporated into the franchise (Escobar 2007).
In addition to the demographic and political similarities that make these cases comparable, in both Colombia and Mexico the establishment and expansion of political rights and the enactment of dual citizenship were accompanied by the development of state programs and offices for emigrants who, in the contemporary context of global and regional markets, acquired a new status in the eyes of their homeland political elites (Escobar 2007;Bermudez 2016;Guarnizo 1998;Lafleur 2011Lafleur , 2013. Colombia and Mexico are thus prime examples of a more general worldwide trend whereby states seek to redefine and reinforce their relationships with expatriates (Bauböck 2005;Lafleur 2013;Collyer 2014aCollyer , 2014b. Notwithstanding these similarities, there are important differences between the two cases that make this comparison instructive. First, in Mexico the extension of absentee voting rights to expatriates occurred much later than in Colombia and came about largely in response to lobbying efforts on the part of Mexican emigrants (Lafleur 2011(Lafleur , 2013. Furthermore, the federal structure of the Mexican political system, which is absent in Colombia, has allowed various states to develop their own particular laws and customs regarding expatriate participation in elections. 2 Emigrant voters in the Colombian context must register in a consulate or embassy, and voting during an election takes place primarily in these consulates and embassies; in neighborhoods with dense Colombian populations, additional voting stations are established in other public locations. Mexico in contrast adopted a postal ballot, a much costlier mode of participation for emigrants McCann 2015, 2014;Lafleur and Calderón Chelius 2011;Leal, Lee, and McCann 2012). In addition, Colombian expatriates enjoy not only the right to vote (i.e., "active rights") but also the right to be elected (i.e., "passive" or candidacy rights), although not in local elections; no such "passive" rights are extended to their Mexican counterparts (Escobar 2015;Palop-García and Pedroza 2017). Finally, the mode of representation differs in both cases. In Mexico, expatriates are incorporated as former residents of territorial entities, and their votes are therefore added into the general pool of votes of all the residents of those territorial entities; this serves to dilute votes from abroad (Palop-García and Pedroza 2017). In contrast, Colombian expatriates enjoy special representation in the Lower House, where one legislative seat is reserved for emigrants. 3 As a consequence, Colombian political parties may have a somewhat greater incentive to reach out to emigrants (cf. Østergaard-Nielsen and Ciornei 2019). However, this difference in the mode of representation between the two countries should not be overstated, given that there are no specific advantages granted to Colombian expatriates in the Upper House, which operates as a single national district.
The empirical analysis below addresses whether the above differences between the Mexican and Colombian cases make a difference in practice with respect to transnational political engagement.

Research Design and Findings
Public opinion polling in Mexico and Colombia is not as extensive an enterprise among scholars, policymakers, and journalists as it is in the United States and other Western democracies. In both of these nations, however, it is now common during elections for citizens to be surveyed extensively about their level of engagement in the campaigns, their judgments about the candidates, and their beliefs about the direction of the country. Many of these surveys are publicly available for researchers at archives such as the Roper Center for Public Opinion Research and the Latin American Public Opinion Archive. Yet to our knowledge, no surveys of civic attitudes and political involvement in either Mexico or Colombia have included emigrants who have a formal right to participate in elections from abroad. 4 This gap in sampling coverage is understandable given the daunting administrative challenge of selecting respondents who are living outside of the country in any number of locations, and conducting interviews in sync with those administered within the national boundaries. We partially close this surveying gap by drawing from our own more specialized studies of Mexican and Colombian expatriates living in selected locations in the US.
In the Mexican context, 753 emigrants were interviewed in June of 2006 by telephone in three sampling sites, Dallas (N = 350), San Diego (N = 126), and north-central Indiana (N = 277), including Indianapolis but excluding the Chicago region. Respondents were recruited randomly through records obtained from a wellestablished marketing research firm specializing in the Latino community. 5 The three sampling areas were 2 Such measures in some cases complement state-level efforts to promote emigrant investment in the local economy (cf. Waddell 2015). 3 See Escobar (2015). A second seat obtained for the district of Colombians abroad in 2013 (Acto Legislativo 01 del 15 de Julio de 2013) and held for only one term (2014-2018) was stripped in 2015 (Acto Legislativo 2 de 2015). 4 To broaden this point, we are not aware of an election-year survey of political attitudes and participation in any of the major migrant-sending countries around the world, except our own, that has sought to include expatriates in the sampling frame. 5 Funding for these surveys was provided by the Carnegie Corporation of New York, the Public Policy Institute at the University of Texas, the College of Liberal Arts and Purdue University, and the University of California, San Diego. Wayne A. Cornelius and David L. Leal collaborated in the development of the survey design. Nearly all interviews were conducted in Spanish and on average lasted just over twenty minutes. Interviewing Services of America (Van Nuys, California), a firm with a long track record of academic survey research on Latinos, administered the interviews. Since no ready-to-use listings of immigrants are available in the three regional sites, random samples of "Mexican heritage" households were obtained from Geoscape International (Miami, Florida). Up to fifteen attempts were made to reach a respondent. Because the telephone records contained both US-born Mexican Americans and immigrants in unknown proportions, and many lines were out of service, there is no straightforward way to calculate a rate of response. If disconnected telephone lines, calls that were never answered, busy signals, and individuals who asked to be contacted again before interviewers could determine whether they fit the study profile are counted as "non-responses" (per the RR1 calculation of the American Association for Public Opinion Research), the estimated response rate is a rather low 11 percent. However, if the response rate is defined as the ratio of completed interviews to attempted interviews of subjects known to fit the chosen to maximize variation in settlement areas and demographic profiles. Dallas and San Diego are major traditional destinations for migrants, with a current combined Mexican-born population at the time of the survey of over one million (Batalova 2008 . While the number of Mexicans now living in Indiana is much smaller than in California or Texas, this rapid expansion of immigrant communities was unprecedented in the state. If patterns of transnational political engagement vary across settlement sites, this three-region sampling framework should be sufficient to detect any noteworthy differences. These surveys were timed to coincide with the height of presidential campaigning in Mexico. The contest of 2006 was deeply polarizing, with Felipe Calderón, the nominee of the right-of-center National Action Party, winning by the closest of margins in the elections in early July. 6 As this survey of Mexican expatriates was being conducted, an additional 305 randomly selected Mexicans living in Mexico were interviewed as part of the "Mexico 2006 Panel Study" (Domínguez, Lawson, and Moreno 2009). 7 In both the expatriate and in-country surveys, comparable instrumentation appeared on the questionnaires. This allows us to pool the datasets, control for a number of demographic traits that have been linked to civic competence, and gauge whether living away from Mexico leads to "poorer quality" engagement in Mexican politics.
A similar comparative design strategy was implemented to examine the quality of transnational political attitudes in the Colombian context during the 2010 presidential election when Juan Manuel Santos, a candidate chosen by the right-leaning incumbent president Alvaro Uribe, 8 competed against Antanas Mockus, the candidate of the newly created Green Party. In the weeks leading up to the 2010 presidential elections in that country, 413 Colombian expatriates in New York City (N = 155) or Miami, Florida (N = 258) were surveyed. All interviews were conducted in person in Spanish. These surveys were administered at random to Colombians who approached the consulates of these two cities to use any of their services (e.g., the issuing or renewal of Colombian IDs, passports, permits, and certificates). 9 This sample is not necessarily representative of all Colombians residing in the selected cities. Nevertheless, the sampling frame likely covers the bulk of the Colombian population, given the wide array of services that consulates offer. 10 The questionnaire was designed to be self-administered; however, the various teams conducting the surveys, composed mostly of students and professors of the various universities involved in the research, helped the interviewees when they had questions or needed assistance.
As in the Mexican surveys, the samples of Colombian expatriates were paired with surveys from Colombia proper. Shortly before the 2010 presidential elections, 458 Colombians in five cities (Barranquilla, Bogotá, Cali, Medellín, and Pereira) were interviewed using questionnaires that were comparable to those employed in the United States. 11 Here too randomized selection procedures were used to choose respondents. In these study protocol (i.e., RR5 in the AAPOR guide), this figure is dramatically higher at 89 percent. Whatever the method for calculating response rates, it is worth noting that with respect to key background variables such as gender, age, level of education, church attendance, and language use at home, study participants are similar to the Mexican-born respondents in other large-scale surveys peace talks with the FARC. At the time of the election, however, Santos had the support of the incumbent president. 9 The response rate in each city was reasonably high. Seventy percent of all Colombians who were approached for an interview agreed to participate in the study. Respondents were selected using randomization procedures that are commonly employed when conducting exit polls; at each sampling location, the interviewer approached respondents at a pre-specified interval, e.g., every third adult entering the consulate area was targeted (cf. Associated Press 2014). See Escobar, Arana, and McCann (2014) for additional methodological details. The scholars involved in the collection of data were Cathy McIlwaine, Anastasia Bermúdez, Marion Magnon Peñuela, William Mejía, Milena Gómez, Claudia Cujar, Jean-Michel Lafleur, and Sarah Mahler. 10 Although we were not able to target survey participants randomly from lists, as in the Mexican sampling, the procedures for recruiting Colombians at consulates resulted in a broadly representative pool of respondents. The distributions of gender, education, and age groups in the Colombian survey conform to those from census data on the adult Colombian-born population nationwide in the 2010 American Community Survey. These comparisons are available upon request. 11 Given the urban origin of the majority of Colombian migrants in the US, these five cities were identified as the most appropriate sampling locations. five cities, two to four strategic public places that offered the most diverse possible sample were selected for conducting interviews among passersby. 12 With these datasets at hand, political engagement among Mexicans and Colombians is examined by addressing three interrelated questions. To what extent are citizens generally attentive to politics in that country? How interested were respondents specifically in the presidential elections of 2006 or 2010? And did ideological positions shape evaluations of the president-Vicente Fox in Mexico or Álvaro Uribe in Colombia-which could indicate a more sophisticated understanding of politics and governance in that particular national context?

Interest in home-country politics
Turning to the question of general attention to politics, Mexican respondents in all sampling regions indicated their level of interest based on a four-point scale (none, little, some, or a great deal). In the Colombian case, a six-point scale was used (0 = no interest, 5 = much interest). These items were regressed on place of residence and four demographic variables (age, household affluence, education level, and gender). The results from this regression are presented in Table 1. 13 In the top part of this table (first column), there is evidence that living abroad, whether in San Diego, Dallas, or Indiana, is associated with lower levels of overall interest in Mexican politics than in the home country (the baseline reference value). The effects for these regional dummy predictors are all negative, implying a lesser degree of attention. However, the joint F-statistic for this model falls far short of statistical significance. This suggests that on average, Mexicans residing relatively close to the border (Dallas, San Diego) or far away (Indiana) are as attentive on average as Mexicans in Mexico.
The second model in this table demonstrates the impact of sociodemographic variables in shaping levels of political interest, rather than region of residence. More highly educated Mexicans expressed significantly greater interest in public affairs, as did older Mexicans. These strong correlations are in keeping with traditional explanations of political engagement, which emphasize the key impact of schooling and life experience.
Quite comparable findings emerge in the Colombian case (lower part of Table 1). Expatriates were actually more attuned to politics in Colombia, with New York-based Colombians being especially interested. These two dummy predictors are significantly related to political attentiveness (p < .05). The significance of these dummies is much diminished, however, when the set of background demographic variables is introduced. As with the Mexican sample, the age of the respondent and level of education condition interest in politics. Males were also significantly more attuned to Colombian politics, another finding that is in keeping with prior research in many countries.
On the whole, the results in Table 1 demonstrate that Mexican and Colombian expatriates are generally not less interested in sending-country politics than their counterparts in the home country. This interpretation is sustained in the regression findings presented in Table 2. In this instance, the dependent variable is the degree of interest specifically in the presidential election, again measured through a four-point (Mexico) or six-point (Colombia) scale. Among Mexicans, place of residence on its own is significantly associated with attention to the 2006 presidential contest (p < .01); expatriates in all three regional sites were less interested. But when background demographic variables are included (second column), these effects wash out, which implies that distance away from Mexico per se has little to do with interest in Mexican campaigns. As in the previous model in Table 1, the more significant predictors are age and education level.
These patterns can be generalized to the Colombian case. Region appears to be linked to political interest in a simple model without controls. When demographic predictors are factored in, these effects are rendered insignificant. Again, the more consequential determinants are age, education, and gender. Living outside of the country did not affect attention to the presidential race. 12 As with interviewing in the US, prospective study participants were targeted using a pre-specified interval rule to help ensure randomized selection. However, when comparing the demographic breakdown of this sample to census data collected in Colombia in 2005, which was the most recent census in that country, some notable discrepancies were discovered with respect to gender and education level. More males than females agreed to take part in the interview when approached by the survey team (65 percent versus 35 percent), and highly educated Colombians were somewhat better represented in the sample. Massey and Tourangeau (2013) note that nonresponse such as this is a growing concern among survey researchers across all disciplines; absent information about the factors that prompt an individual to decline to be interviewed, it is customary to weight the data based on known population characteristics. In this case, applying weights so that these demographic distributions match those in the 2005 census does not affect the substantive interpretation of the findings presented below. 13 As noted at the bottom of this table and the others, missing values are imputed multiple times via chained equations. This step helps to preserve statistical efficiency, in that all available observations are retained for analysis. The results for "complete case" analysis are available upon request; they do not differ substantially from those presented here. Replication files for the models presented in this table and the others are available as additional online files.

Ideological preferences
Did expatriates rely on ideological preferences to the same degree as co-nationals living within the sending country when evaluating the head of state? Mexican and Colombian respondents indicated during interviews whether they considered themselves to be on the left (coded 1), in the ideological middle (2), or on the right (3). If these attitudes are meaningful, as opposed to superficial or ephemeral, they should be linked to assessments of leading political figures, such as the president. Both Vicente Fox and Álvaro Uribe were positioned on the political right within their respective political systems. If living abroad leads expatriates to adopt new political identities and new understandings of what it means to be on the left or right, there would be systematic differences in how ideological positions shape presidential assessments. This possibility is explored through a series of regression models in Table 3. In the Mexican case, evaluations of Vicente Fox were gauged through an eleven-point scale (0 = very bad, 10 = very good). When this item is regressed on left-right positions, a significant effect emerges (b = .52, p < .05); the more right-wing the respondent, the more positive the evaluation of President Fox. In the second model, . See the appendix for item wordings. * p < .10; ** p < .05; *** p < .01. the three regional dummies are added along with the four demographic background predictors. These additional variables contribute little explanatory power, and political ideology remains a strong predictor of presidential evaluations. Furthermore, as shown in the third model, the impact of ideological orientations does not fluctuate across sampling regions; the three interaction terms (ideology × region) are collectively insignificant. The fourth model shows, however, that the impact of political ideology varies across educational groupings. The "ideology × education" interaction term is positive and statistically significant, which implies that more highly educated Mexicans were more prone to evaluate the president in this more sophisticated manner. To show this more parsimoniously, the fifth model drops the statistically insignificant interaction terms. A clearer story line thus becomes apparent in the Mexican sample: living north of the border does not affect abstract ideological judgments of the president.
Turning to the Colombian samples, a similar pattern is evident. In this case, the dependent variable is a dichotomous measure of presidential approval (0 = disapprove of Uribe, 1 = approve), and a logistic regression model is fit. The impact of ideology stands out clearly in the first specification. The regression coefficient of 1.42 is highly significant. This suggests that a one-unit move toward the ideological right leads Note: OLS coefficients and standard errors. In the Mexican sample, the dependent variable is a four-point scale (1 = no interest in the 2006 presidential campaign, 4 = much interest); in the Colombian sample, the dependent variable is a six-point scale (0 = no interest in the Colombian elections, 5 = much interest). The excluded regional dummy category is "living in Mexico/Colombia." N = 1,058 (Mexicans) and 870 (Colombians). To retain all available information, multiple imputation of missing values was employed via chained equations (MICE, with 20 imputations).

Sources: Authors' surveys of Mexican/Colombian immigrants in the US and Colombians in Colombia; Mexico 2006 Panel
Study (http://mexicopanelstudy.mit.edu/). See the appendix for item wordings. * p < .10; ** p < .05; *** p < .01. Table 3: Assessing variations across regions in the impact of "left-right" ideological postures on presidential evaluation.  to an increase in the odds of approving Uribe by over a factor of four (i.e., exp [1.42]), a strikingly large effect.
The "ideology × sampling region" interactions presented in the third model are jointly insignificant, which tells us that the forcefulness of left-right positions does not vary across sampling sites. Moreover, the impact of ideology does not vary based on age, income level, education, or gender (Model 4). Taken together, the findings in Table 3 highlight some noteworthy differences in ideological reasoning between Mexicans and Colombians. For Mexicans, the more highly educated are more apt to evaluate the president based on ideological considerations. Among Colombians, no such distinction is present. But in both cases, expatriates are no less ideologically minded in comparison to compatriots living within the national boundaries. This is yet more telling evidence that distance from the country does not in and of itself depress substantive engagement in public affairs.

Time abroad
The length of time that an expatriate has resided abroad also does not affect transnational political engagement to any noteworthy extent. Mexicans and Colombians who had lived in the US for many years are approximately as attentive to politics and campaigns in their country of origin as relative newcomers to the United States. This is evident in the findings presented in Table 4. Focusing only on the expatriate imputations. * p < .10; ** p < .05; *** p < .01. samples (N = 753 for Mexicans and 413 for Colombians), the item on general interest in sending-country politics was regressed on the set of sociodemographic background variables plus the amount of time living in the US, coded in years. Among Mexican expatriates, the number of years spent abroad is associated with a decline in interest in Mexican politics, but this effect is not strong or significant beyond the .10-level ( Table 4, first column). 14 When interest in the 2006 presidential campaign is modeled (second column), this predictor carries even less weight.
In the Colombian sample, the length of time an expatriate had lived in the United States is similarly negatively related to attention to Colombian politics in general and interest in the presidential contest in particular (b = −.04 in each case). These effects, however, are far from statistically significant. For all practical purposes, time in the United States has no impact on transnational political interest. Much previous research has shown that the longer an immigrant resides in the receiving country, the more engaged he or she becomes in politics within that country (Guarnizo, Portes, and Haller 2003). The findings in Table 4 tell us that these processes of political acclimation do not necessarily wipe away foundational orientations toward politics in one's nation of birth.
With respect to political sophistication, it is equally apparent that time spent in the United States does not influence the extent to which the Mexican or Colombian head of state is evaluated using ideological criteria ( Table 5). In the Mexican expatriate sample, regressing evaluations of Vicente Fox on the four background factors, regional dummies, the number of years spent in the US, and an interaction term for education level and political ideology shows a slight statistical interaction between education and left-right ideological preferences. Highly educated Mexicans were slightly more likely to judge Fox based on ideological leanings, which is in keeping with the findings in Table 3 (Model 5). 15 Allowing time in the US to interact with 14 If the relationship between time in the US and attention to Mexican politics is specified as a quadratic fit (i.e., "Time Spent"-squared is included as an additional predictor to take into account a potential curvilinear effect), little is changed in the model. 15 The main effect for education level and the interaction term for education and political ideology both approach statistical significance (p < .20). As in Table 3, these are the only two effects on evaluations of Fox that appear to be potentially consequential. education and left-right ideology has no notable effect on attitudes toward the president. These additional interaction terms are all clearly insignificant. 16 In the Colombian case, education was not found to moderate the effect of political ideology on evaluations of President Uribe; consequently, no such interaction effect between the two predictors is estimated. In the first model, approval or disapproval of Uribe is regressed on four sociodemographic background factors, region of residence in the US, ideological stance, and time spent in the country. Several of these predictors significantly shape judgments of the president, particularly political ideology, as in Table 3 (Models 1-3). Adding an interaction term for left-right ideological position and the number of years in the US contributes little to this model. This implies that expatriates who had resided abroad for years or decades utilized abstract ideological criteria when rating the president to the same degree as more recent emigrants. This is further evidence that living outside of the country of origin does not necessarily result in markedly different patterns of engagement in Colombian politics.

Conclusion
Within any democratic system, the boundaries of the electorate are inherently inexact. Determining who should be counted "in" or "out" of the public is a complex task, with legal theorists, strategically minded policymakers, and electoral bureaucracies all weighing in. These boundaries are not only somewhat fuzzy, but also dynamic. One of the most remarkable trends in democratic representation over the last several decades has been the expansion of many electorates to include sizeable expatriate communities. Such expansions have been quite controversial in many cases. This study considers whether residence outside a country is corrosive for democratic engagement.
Few scholars have previously investigated this matter with a comparative eye. Our focus on two nations in Latin America that have followed distinctive paths to democratization and expatriate inclusion, Mexico and Colombia, permits a preliminary degree of generalization. The empirical findings collectively suggest that 16 To gauge whether the length of time an immigrant has stayed in the US moderates the impact of ideology on evaluations of Fox for the whole sample or primarily for the more highly educated Mexicans, it is important to include three interaction effects: "Time in the US" by left-right ideological position, education, and the composite "Left-Right × Education" multiplicative term (Brambor, Clark, and Golder 2006 Note: Coefficients are ordinary least squares (Mexicans) and logistic regression (Colombians) estimates with standard errors in parentheses. Missing values imputed via chained equations (MICE), with twenty imputations. N = 753 (Mexicans) and 413 (Colombians). In the Mexican sample, evaluations of President Vicente Fox were measured on an eleven-point scale (0 = very bad, 10 = very good). In the Colombian sample, evaluations of President Álvaro Uribe were coded dichotomously (0 = disapprove, 1 = approve). * p < .10; ** p < .05; *** p < .01. emigrants retain a relatively high level of intellectual involvement in sending-country politics long after settling abroad, at least when compared to co-nationals within that country. Merely residing abroad does not diminish enthusiasm for following public affairs. This finding may prove reassuring for policymakers and theorists within migrant-sending countries who advocate a deeper commitment to transnational incorporation.
Yet before closing we should note the need for further work that may put a finer point on the comparisons that are drawn here. By design, this study investigates transnational political engagement as defined most broadly-general interest in public affairs and campaigns, and the capacity to evaluate leading public figures using abstract ideological benchmarks. These are essential facets of democratic involvement, but there are many others that are quite worthy of scrutiny (González-Rábago and Blanco 2016). At this juncture, social scientists have vast opportunities to apply the kinds of systematic social science research techniques that have shed so much light on public opinion and engagement in conventional political settings to expatriate electorates. How well does information about specific domestic policy issues get diffused across borders to emigrants, and what are the primary conduits of communication? Does a sense of civic duty to take part in politics in one's native country-an attitude that has long been recognized as among the most consequential for prompting democratic participation-fade when one lives abroad? How are attachments to political parties and interest groups in the native country affected when one moves abroad? Such questions, among others, must be addressed in fuller detail, with many more comparative cases added to the analysis, before any general claims can be offered concerning political capabilities and aspirations in emerging transnational contexts.