POLITICS, SOCIOECONOMC STATUS, AND INDIGENOUS

IDENTITY IN LATIN AMERICA: THE BOLIVIAN CASE

A paper to be presented at the 2006 meeting of the Latin American Studies Association,
San J uan, Puerto Rico.






Ral Madrid
Dept. of Government, Univ. of Texas at Austin
Austin, TX 78712
Tel: (512) 232-7246
e-mail: [email protected]
2







Abstract: This paper uses survey data to explore why some speakers of indigenous
languages (or people of indigenous ancestry) in Bolivia identify as indigenous, while
others do not. It examines three main theoretical approaches to understanding ethnic
identification: the assimilation/enclosure model, which would expect that indigenous
identification would be inversely related to socio-economic status and residential and
linguistic integration; the ethnic competition model, which would make the opposite
predictions; and a more explicitly political approach, which would link indigenous
identification to the political socialization of individuals. The study finds some support
for the assimilation/enclosure model and the politicization hypothesis, but it finds no
support for the ethnic competition approach to understanding ethnic identification.
3

Scholars traditionally treated ethnicity as singular, deeply-rooted, and fixed (Geertz
1973; Horowitz 1985; Van Evera 2001). That is, they assumed, implicitly or explicitly,
that individuals have only one ethnic identity and that this identity does not change over
time. In recent decades, however, this approach to understanding ethnicity, known as
primordialism, has been largely supplanted by a different approach, which argues that
ethnic or racial identity is socially constructed. Constructivists maintain that individuals
have multiple ethnic identities and that these identities often change over time (Chandra
2001; de la Cadena 2000; Nobles 2000; Portes 1984). Individuals will, for example,
assume different identities in different social, political, or economic contexts.

The fluid nature of racial and ethnic identity in Latin America clearly provides
support for the constructivist model. Numerous studies have documented how
conceptions of race and identity have shifted over time in the region (Nobles 2000; Telles
2004; Degregori et al 1998; Yashar 2005). Since the colonial era, for example, Latin
American countries have suppressed and devalued indigenous cultures and peoples.
1

During the mid-20
th
century, many Latin American governments sought to reshape their
indigenous populations around their class identities, organizing indigenous people into
peasant associations that were closely linked to the state (Yashar 2005; Degregori et al
1998). The last few decades, however, have witnessed the reemergence of indigenous
identities (Alb 1991; Van Cott 2004; Yashar 2005). Powerful indigenous movements
have emerged that have sought to restore indigenous traditions and identities. These

1
The term mestizo typically refers to people of mixed European and indigenous descent, but at times it
includes people who have other racial or ethnic mixtures as well.
Indians people can erase their Indian surname (e.g. Masmas instead of Mamani, etc.), you can change-
clothes, language, but Indian cannot change their bodily aspect, their body as semantic maker.
4
movements have put pressure on Latin American governments to recognize indigenous
traditions, grant autonomy to indigenous areas, and create state institutions and programs
to cater to the needs of the indigenous population.

Although various studies have shown us how conceptions of race and ethnicity in
Latin America have changed over time, they have not typically explored the determinants
of ethnic identity at the individual level. We still know relatively little about why some
people identify as indigenous, while others with relatively similar ascriptive
characteristics do not. These questions are important not only because they will help us
understand how ethnicity and race is constructed in the region, but also because they will
contribute to broader theories of the determinants of race and ethnicity. Latin America
has long been viewed as an outlier because of the relatively low political salience of race
and ethnicity in the region. Deepening our understanding of the determinants of ethnic
identification in the region will help us build better theories about when and why race or
ethnicity matters.

This paper focuses on the construction of indigenous identities in Bolivia. Bolivia
represents an interesting case because the number of people who self-identify as
indigenous in the country is considerably below what one would predict. There are a
number of reasons to expect that a large percentage of the Bolivian population would
self-identify as indigenous. First, a large percentage of the Bolivian population has
indigenous phenotypes. Indeed, outside estimates have suggested that 60 to 80 percent of
the population is indigenous, although these sources typically do not reveal how
Indians who
speak
Spanish and
are
westernized
5
indigenous status is measured or spell out the methodology used to come up with such
estimates (Deruyttere 1999; Hopenhayn and Bello 2001; Yashar 2005). Second, a large
proportion of the Bolivian population speaks indigenous languages. According to the
2001 census, 49.5 percent of the Bolivian population above 5 years of age reported that
they were raised in a home where an indigenous language was spoken, often in addition
to Spanish (INE 2003: 142). This represents a decline from 55.6 percent of the population
in 1992 and 65.6 percent in 1976, but still constitutes a larger proportion of the
population than in any other Latin American country. Third and finally, Bolivia has an
increasingly powerful indigenous movement, which has sought to promote indigenous
identities. Bolivias indigenous movement has succeeded in forcing the resignation of
two presidents in the last few years and it also recently helped elect an indigenous person,
Evo Morales, as president for the first time.

Although the prevalence of indigenous languages and phenotypes as well as the
strength of the indigenous movement in Bolivia would suggest that the country has a
large indigenous presence, recent surveys have found that only a small percentage of the
population self-identifies as indigenous. Indeed, surveys by the Ministry of Human
Development (Rojas Ortuste and Verdesoto Custode 1997), the United Nations
Development Program (PNUD 2004), and the Latin American Public Opinion Project
(Seligson 1998, 2000 and 2002) found that less than 16 percent of the population self-
identifies as indigenous. These surveys all found that approximately 60 percent of the
Bolivian population self-identifies as mestizo and between 20 and 25 percent of the
population self-identifies as white. Moreover, even among the population that speaks
6
indigenous languages, a relatively small percentage self-identifies as indigenous.
According to the 2002 LAPOP survey, for example, only 20 percent of the population
that speaks indigenous languages self-identifies as indigenous.

The 2001 census did report that a much larger percentage of the population self-
identifies as indigenous--62 percent of the population above 15 years of age--but the
question used by the Bolivian census to ascertain indigenous self-identification has been
widely criticized on several grounds. First, the question did not specifically ask if people
self-identify as indigenous, but rather if they identified as Quechua, Aymara, Guaran,
Chiquitano, Mojeo, another native group, or none of these groups. It may be that people
are more willing to identify themselves as belonging to certain indigenous linguistic
communities, such as Quechua or Aymara, than as indigenous per se. Second, the
question did not provide the option of identifying as mestizo, or mixed, and numerous
surveys have shown that the majority of Bolivians will identify as mestizo if given this
option. Third, the nature of the census question, which obliged people who did not
identify with one of the indigenous categories listed to choose the option none, may
also have encouraged people to choose one of the indigenous categories.

The large difference between the number of people willing to self-identify as
indigenous in various surveys and outside estimates of the size of the indigenous
population suggests that many people who are perceived as indigenous may not self-
identify as indigenous or at least may not do so consistently.
2
Many of the people who

2
Studies of race in Brazil have also noted significant differences in racial self-perception and the way in
which interviewers classify people by race (Telles 2002).
people
believe
that they
are....
contradiction between how people are perceived and how people perceive themselves
SPLIT IDENTITY
7
self-identify as mestizo, for example, have indigenous phenotypes and may be mostly of
indigenous ancestry, which leads them to be perceived as indigenous.
3
Some of the
people who self-identify as mestizo also speak indigenous languages or wear indigenous
dress, which may also lead them to be identified by others as indigenous. In the 2002
LAPOP survey, for example, 37 percent of the population that self-identified as mestizo
was raised in a home where an indigenous language was spoken and seven percent of the
population that identified as mestizo used indigenous dress.
4
Recent anthropological
studies have proposed a new ethnic category of indigenous mestizos, because, they argue,
that many people identify as both mestizo and indigenous (de la Cadena 2000; UNDP
2004).

What leads people to choose one identity over another? Who is likely to identify
with a subordinate ethnic or racial category, and why might they do so? More
specifically, what factors have led some people of indigenous descent in Bolivia to
identify as indigenous, while other people have not?

This paper uses survey data to explore why some people in Bolivia identify as
indigenous, while others do not. It examines three main theoretical approaches to
understanding ethnic identification: the assimilation/enclosure model; the ethnic
competition model; and a more explicitly political approach. As we shall see, the study
finds some support for the assimilation/enclosure model and the politicization hypothesis,

3
Some studies have shown that mestizos in Bolivia have substantial genetic similarities (HLA alleles) to
isolated Amerindian tribes, moreso, in fact than the Peruvian mestizos that were examined (Cervantes 2003;
de Pablo et al 2000).
4
By contrast, 69 percent of the respondents who self-identified as indigenous in this survey spoke
indigenous languages and 23 percent of them wore indigenous dress.
8
but it finds no support for the ethnic competition approach to understanding ethnic
identification. The paper is organized as follows. The first section of this paper explores
some existing theories of ethnic identification and generates some hypotheses from them.
The second section lays out the methodology used to explore these hypotheses. The third
section presents the findings of the paper, and the conclusion discusses some of the larger
implications of these findings.

The Assimilation/Enclosure Model

Existing theories of race and ethnicity offer a number of potential explanations for
ethnic identification. Various studies of ethnicity in the United States, for example, have
linked ethnic identification to socioeconomic status and integration. The assimilation
model, which was developed by Robert Park (1950) and other members of the Chicago
School, argued that ethnic identification gradually waned among members of immigrant
groups as they climbed the socio-economic ladder. According to this approach, rising
levels of income and education would facilitate assimilation, and thus weaken ethnic
identification. Assimilation theorists and other scholars have argued that mastery of the
English language and residential integration also speeds up the assimilation process,
thereby weakening ethnic ties. Another approach known as the enclosure model
emphasized many of the same variables as the assimilation model, but did not view
assimilation (and loss of ethnic identity) among immigrant groups as inevitable (Massey
and Denton 1992; Eschbach and Gmez 1998). Rather, they argued that segregation,
9
isolation, and lack of socio-economic mobility might lead members of immigrant groups
maintain their identities indefinitely as they failed to assimilate to the mainstream.

Indigenous people in Latin America do not represent an immigrant group;
nevertheless they face considerable pressures to assimilate or to adopt a non-indigenous
identity, given the social, economic, and political dominance enjoyed by the non-
indigenous population in Latin America and the high levels of discrimination against
indigenous people. Following the assimilation/enclosure model, we might expect people
of indigenous descent who are wealthier and more educated to be less likely to identify as
indigenous. Indeed, the possibility of assimilation is much greater for wealthier and more
educated people. Wealthier and better-educated people of indigenous descent have a
greater chance than do poorer, less-educated people of being accepted as non-indigenous
and they may find it easier to adopt non-indigenous customs and lifestyles. Wealthier and
better-educated people may also face greater incentives to shed their indigenous
identities, given that they are likely to have more contact with non-indigenous people.
Professional and residential advancement, for example, may depend on adopting a non-
indigenous identity.

Following the assimilation and enclosure models, we might also expect people of
indigenous descent who speak Spanish and who live in mixed or non-indigenous areas to
be less likely to identify as indigenous. People who speak Spanish and live in non-
indigenous areas are more likely to be accepted as non-indigenous, which makes the
adoption of a non-indigenous identity more feasible. Spanish-speakers and people who
parents do not want
their children speak
an indigenous
language
10
live in mixed or non-indigenous neighborhoods may also have greater incentives to adopt
non-indigenous identities since they typically have more contact with non-indigenous
people. They, too, may adopt non-indigenous identities in order to avoid discrimination
or to advance their careers, education or residential situation. People of indigenous
descent who speak Spanish and live in mixed or non-indigenous areas may also feel less
tied to their indigenous identities. By contrast, people who do not speak Spanish and live
in predominantly indigenous areas may feel that they have little possibility or interest in
identifying as non-indigenous or in being identified as such.

The Ethnic Competition Model

Another prominent approach in the race and ethnicity literature in the United States,
the ethnic competition model, would have very different expectations. The ethnic
competition approach has argued that socio-economic progress and contact with other
ethnic groups strengthens ethnic identities rather than weakening them (Portes 1984).
According to this model, as people rise up the socio-economic ladder, master the English
language or move into integrated neighborhoods, they increasingly compete for jobs and
other resources with members of ethnic groups, and they also become increasingly aware
of discrimination. Competition with members of other ethnic groups and greater
awareness of discrimination against his or her own ethnic group, in turn, tend to
strengthen an individuals ethnic identification.

11
An explanation of the determinants of indigenous identity derived from the ethnic
competition approach would focus on many of the same variables as the assimilation
model but would predict the opposite effects. It would expect, for example, that
wealthier and more educated people of indigenous ancestry, especially those people who
speak Spanish and live in integrated areas, would be more inclined to identify as
indigenous because they would be more likely to have competed with members of other
ethnic groups for resources, such as housing, jobs, and school grades and admissions.
According to the ethnic competition model, wealthier and more educated people, as well
as people who speak Spanish and live integrated areas, will also be more likely to be
aware of discrimination against indigenous people, which would make them even more
likely to identify as indigenous.

A political model of indigenous identity

The two dominant approaches in the literature on ethnic identitythe
assimilation/enclosure model and the ethnic competition modelthus focus on economic
and social factors, largely ignoring any potential political determinants of ethnic
identification. A very different approach, however, would suggest that identification with
a particular race or ethnicity is a political act in addition to being a social or economic
one. Race and ethnicity tends to become more politicized in some contexts and periods
than others, and some ideologies may seek to promote racial or ethnic consciousness,
while others may try to suppress it. Political organizations often play an important role in
the politicization of race and ethnicity as well by drawing attention to racial or ethnic
12
discrimination and inequalities. Participation in these organizations can thus lead
individuals to become more aware of this discrimination and inequalities and heighten an
individuals identification with a subordinate ethnic or racial group. Participation in
certain political organizations may also foster a sense of racial or ethnic solidarity,
particularly if those organizations are dominated by members of that group. Thus,
individuals who have grown up in a certain political context, who have participated in
certain organizations, and who have adopted or been exposed to certain ideologies may
be more likely to identify with a particular ethnic or racial group.

The existing literature on Bolivia would suggest that political movements have
played an important role in shaping ethnic identification in the country. The salience of
indigenous identity in Bolivia has varied considerably over time depending in part upon
state policies (Ticona, Rojas and Alb 1995; Patzi Paco 1998; Yashar 2005). In the wake
of the 1952 Bolivian revolution, for example, the Bolivian government sought to suppress
ethnic identification, organizing indigenous people around their class, rather than their
ethnic, identities. During the 1970s, however, an independent indigenous movement
emerged among the Aymara-speaking communities in the highlands of Bolivia and
gradually took control of the peasant unions in the area. This movement, which became
known as the Katarista movement, sought to increase ethnic consciousness, restore
indigenous traditions, and eliminate economic and social discrimination against the
indigenous population. The Katarista movement never gained many adherents outside of
Aymara-speaking areas of the Bolivian highlands, but it did manage to gain control of the
peasant unions in these areas and it raised awareness of indigenous issues nationwide
13
(Hurtado 1986; Rivera Cusicanqui 1986). Moreover, in the early 1990s, a prominent
Katarista leader, Victor Hugo Crdenas, was elected Vice President, and, under his
leadership, some of the traditional demands of the indigenous movement, such as
bilingual education and greater local autonomy, were adopted, although these reforms
failed to satisfy many indigenous leaders.

In the 1990s, an important new indigenous-based movement led by Evo Morales
emerged from the unions of Quechua-speaking coca growers in the Chapare region, and
gradually gained control of the Bolivian peasant union confederation, the Confederacin
Sindical nica de Trabajadores Campesinos de Bolivia (CSUTCB). Morales movement,
like the Kataristas, subsequently formed a political party and entered the electoral arena,
where it achieved considerably greater political success than had the Kataristas. Morales
and his party, the Movimiento al Socialismo (MAS), finished second in the 2002 general
elections, and in 2005 Morales was elected president with a stunning 53 percent of the
vote. Unlike the Katarista parties, Morales and the MAS have drawn support from both
the Aymara and the Quechua populations, as well as from the smaller lowlands Indian
communities. Other important indigenous movements and organizations have also
emerged in recent years. The Movimiento Indgena Pachakuti, a radical indigenous party
led by the Aymara leader Felipe Quispe, emerged to compete in the 2002 presidential
elections. It earned a respectable 6 percent of the national vote in these elections,
although its support was largely confined to Aymara-speaking communities of La Paz.
An indigenous organization designed to represent the lowland Indians, the Confederacin
de Indgenas del Oriente y Amzonia de Bolivia (CIDOB), has also expanded its
14
influence in the last decade, although it has not competed in elections under its own
banner. The various indigenous organizations have carried out repeated marches and
demonstrations to make their political demands known. Indeed, protests led by some of
these indigenous organizations led to the resignation of presidents Gonzalo Snchez de
Lozada in 2003 and Carlos Mesa in 2005.

Given the relatively recent emergence of powerful indigenous movements in
Bolivia and the rising political salience of ethnicity in the country, we might expect
relatively young individuals to be more likely to self-identify as indigenous. Younger
people will have come of age at a time of increased indigenous mobilization and ethnic
pride. They therefore might be more likely to adopt an indigenous identity as a political
statement than older people who tended to be politicized around class for much of their
lives. Given the rising political salience of ethnicity, we would also expect the percentage
of people who self-identify as indigenous to be greater in recent surveys than in older
ones.

Owing to the long tradition of indigenous organizing in Aymara areas of Bolivia,
we might also expect people of Aymara heritage to be more likely to identify as
indigenous than people from other backgrounds. Indigenous organizations and parties
have traditionally been much stronger in Aymara-speaking areas than elsewhere in
Bolivia. As a result, we would expect the Aymara population to be more politicized than
other indigenous groups and Aymara individuals to have greater ethnic pride and
consciousness. It is unclear whether there would be differences in the level of ethnic self-
15
identification among other indigenous communities, however. Given the emergence of a
strong lowlands indigenous organization, it is possible that members of lowlands
indigenous communities might have higher rates of politicization and ethnic
consciousness, however.

Much indigenous organizing has focused around the peasant unions or associations
that were created after the Bolivian revolution. These peasant organizations have played a
central role in formulating and presenting the demands of indigenous communities and in
promoting ethnic consciousness. People who participate in peasant unions or associations
should therefore be more likely to identify as indigenous as well. We might also expect
people who participate in other types of unions, such as the miners and teachers unions,
to have greater ethnic consciousness as well. Many of these unions have a tradition of
radicalism and intense political activity, which may promote ethnic as well as class
consciousness. In Bolivia, men have traditionally participated in these types of political
organizations much more frequently than women, so we might also expect men to be
more ethnically politicized and more likely to self-identify as indigenous than women.
5


Finally, we might expect people who espouse left-wing views to be more likely to
identify as indigenous than people on the center or right ideologically. Left-wing leaders
and parties have done much to promote indigenous demands in Bolivia and to raise ethnic
consciousness. The traditional left-wing parties carried out a great deal of organizing in
indigenous areas in Bolivia in the 1970s and 1980s, which helped politicize some sectors

5
By contrast, a more culturally-driven theory of indigenous identity might expect women to be more likely
to self-identify as indigenous than men, given that indigenous women often maintain certain cultural
traditions, such as traditional dress, more widely than do men.
16
of the indigenous population and may have indirectly promoted ethnic consciousness
(Ticona, Rojas and Alb 1995; Van Cott 2005). Left-wing parties have also gone further
than other parties in embracing traditional indigenous demands and recruiting indigenous
candidates. Indigenous parties, meanwhile, have adopted traditional left-wing positions as
well as indigenous demands. The MAS, for example, has vigorously opposed neoliberal
reforms and U.S. intervention in Bolivia at the same time that it has pushed for greater
autonomy and land rights for indigenous peoples and trumpeted the achievements of
indigenous civilizations. The Katarista parties and, especially the MAS and the MIP,
have also drawn heavily on the human and material resources of the Bolivian left (Zegada
Claure 2002; Romero Ballivin 2002; Van Cott 2005). As a result, indigenous language
speakers who hold left wing views should tend to have greater sense of ethnic
consciousness and pride and be more likely to self-identify as indigenous.

Measurement

To explore these hypotheses, I use data from three surveys carried out in Bolivia in
1998, 2000, and 2002 under the direction of Mitchell Seligson as part of the Latin
American Public Opinion Project.
6
The surveys, which were financed by the U.S. Agency
for International Development, have several advantages over most Latin American
surveys for the purposes of this study. First, they used national samples, rather than
strictly urban samples, which is particularly important in studying indigenous people who
are disproportionately located in rural areas. Second, unlike most Latin American polls,
these surveys included a question about ethnic identity. Third, the surveys were

6
For more details on the surveys, see Seligson 1999, 2001, and 2003.
17
conducted in Quechua and Aymara as well as in Spanish, which meant that they did not
exclude those indigenous people who do not speak Spanish.
7


The survey question about ethnic identity was phrased as follows: Do you consider
yourself a person of the white race, mestizo, indigenous or black?
8
The interviewers
also accepted two other categories: cholo, which is a term commonly used to describe a
person of indigenous or mixed descent who lives in urban areas; and other. People who
responded that they were originario, or original people, were included in the indigenous
category. Approximately 62.8 percent of the surveys respondents self-identified as
mestizo, including many people who reported that they had spoken an indigenous
language at home in their childhood. The second most common response was white,
which accounted for 23.9 percent of the surveys respondents including some people who
reported that they had spoken an indigenous language at home in their childhood. Only
10.0 percent of the surveys respondents self-identified as indigenous. The least common
responses were cholo and black, which accounted for 2.3 percent and 1.0 percent of the
sample respectively.

The surveys also asked interviewees: What language have you spoken at home
since you were little? Five alternatives were provided--Spanish, Quechua, Aymara,
other native, or other foreign languagebut more than one alternative was accepted. In
the full sample, 59.2 percent of the respondents reported that they only spoke Spanish in
the home when they were little, 34.8 percent reported that they spoke an indigenous

7
The number of interviews conducted in the indigenous languages was quite small, however, in large part
because the vast majority of Bolivians who speak indigenous languages also speak Spanish.
8
In Spanish, the question was: Usted se considera una persona de raza blanca, mestiza, indgena o negra?
18
language and Spanish, and 6.0 percent reported that they only spoke an indigenous
language. Of the respondents that spoke an indigenous language, 63.8 percent spoke
Quechua and 37.7 percent spoke Aymara, and only 3.4 percent spoke some other
indigenous language.
9


The total sample size of the (national) surveys was 9,000 (approximately 3,000
for each survey), but I restricted the sample to those people who might have a cultural
reason to identify as indigenous, namely those people who had spoken an indigenous
language at home in their childhood. I subsequently carried out analyses on a somewhat
larger sample of people who acknowledge having indigenous ancestry since these people
too might have some reason to identify as indigenous. The latter category included
people who self-identify as mestizo, cholo, or indigenous. Only 75 percent of the national
samples acknowledge having some indigenous ancestry and 41 percent of the national
samples reported speaking an indigenous language. Restricting the sample to these
categories of people, of course, means that the samples are not nationally representative.

To carry out the analyses, I created a dummy variable which is scored as 1 if the
respondent identifies as indigenous and 0 if he or she identifies with some other ethnic
category. To test the assimilation and ethnic competition hypotheses, I included variables
measuring the socio-economic status of the respondents in the analyses, specifically their
monthly household income (in income ranges) and their level of education (last year of
study completed). To measure linguistic assimilation, I included a dichotomous variable
measuring whether the respondents only spoke an indigenous language at home when

9
These figures do not sum to 100 percent because some people spoke more than one indigenous language.
19
they were little. Unfortunately, the survey did not include a question about the degree of
residential integration, but it did include a question about the level of urbanization. In
Bolivia, cities tend to be more ethnically mixed than rural areas, so I used the
urbanization rate as a rough proxy for the degree of residential integration. To test the
politicization hypotheses, I included variables measuring whether they participated often,
occasionally, almost never, or never in labor unions and in trade associations--the latter
category specifically included peasant organizations as well as associations of
professionals, merchants, and producers. I also included an ordinal variable measuring
the respondents ideology (on a ten point scale from left to right), as well as variables
measuring the age and gender of the respondents. Finally, I included dummy variables for
people who spoke Aymara at home when they were little as well as for people who spoke
other native languagesthat is, lowlands indigenous languages. Quechua speakers
were thus the base category. Table 1 lists the variables included in the analysis and their
predicted effects according to the different theoretical models examined

Data analyses

Table 2 presents summary data on the percent of the respondents with different
demographic, socio-economic, and political characteristics that self-identified as
indigenous in the 2002 survey. The first column presents data for those people who
reported speaking indigenous languages at home since they were little, whereas the
second column presents data for those people who report having some indigenous
ancestry (self-identified mestizos, cholos, and indigenous people). The table shows that,
20
as assimilation theory would predict, people of lower socio-economic status tend to self-
identify as indigenous more frequently than do people of higher socio-economic status.
This is true for indigenous languages speakers as well as for people who report having
some indigenous ancestry. For example, 25.4 percent of indigenous language speakers
who have completed less than 6 years of school self-identified as indigenous in the
survey, as opposed to only 9.3 percent of people who had completed 13 or more years of
school. Similarly, 27.6 percent of indigenous language speakers who earned less than 250
Bolivianos per month self-identified as indigenous, compared with only 9.0 percent of the
respondents who earned more than 1,000 Bolivianos monthly. Speakers of indigenous
languages (or people with indigenous ancestry) who live in very rural areas or who are
monolingual also tend to self-identify as indigenous more frequently than do indigenous
language speakers (or people with indigenous ancestry) who live in cities or who are
multilingual. Whereas these findings provide preliminary support for
assimilation/enclosure theory, they contradict the expectations of ethnic competition
theory.

Table 2 also provides some preliminary support for a theory of indigenous
identification focusing on political factors. As the table shows, speakers of indigenous
languages (or people with indigenous ancestry) who participate in trade associations,
including peasant organizations, frequently, or even occasionally, tend to self-identify as
indigenous at much greater rates than do indigenous language speakers (or people with
indigenous ancestry) who never or almost never attend such meetings. The frequency of
participation in union meetings, however, does not have a linear relationship with
21
indigenous self-identification for indigenous language speakers, although it does for
people with some indigenous ancestry. Speakers of indigenous language (or people with
indigenous ancestry) who are male and who have left-wing views also self-identify as
indigenous more frequently than do indigenous language speakers (or people with
indigenous ancestry) who are female or who have centrist or right wing views. Contrary
to the expectations of the politicization model, however, speakers of indigenous language
who are relatively young (under 40) are not more likely to self-identify as indigenous.
10

Finally, speakers of Aymara also tend to self-identify as indigenous much more
frequently than do Quechua speakers (or the monolingual Spanish-speaking population),
which may be a result of the long tradition of indigenous activism in Aymara
communities. Surprisingly, however, speakers of lowlands indigenous languages tend to
identify as indigenous much more frequently than either the Quechua or Aymara
population. This may be partly a result of the growing levels of political organizing and
mobilization among the indigenous population in the Amazon. However, it is also
possible that a high percentage of speakers of lowlands indigenous languages identify as
indigenous in large part because these communities have traditionally been relatively
isolated and only weakly assimilated into the rest of the Bolivian population. As a result,

10
Nevertheless, the LAPOP surveys do indicate that the proportion of people who self-identify as
indigenous in Bolivia is increasing. A nationally representative survey carried out by LAPOP in 2004
found that 16 percent of the population self-identified as indigenous, as opposed to only 11 percent in 2002,
9 percent in 2000 and 10 percent in 1998 (Seligson, Morales and Blum 2004: 39). The difference between
the percentage of people who self-identified as indigenous in 2004 and the percentage of people who self-
identified as indigenous in earlier surveys was statistically significant. These data refer to the number of
people who answered indigenous or native when they were asked whether they identified with the white,
mestizo, indigenous or black race, but other questions yield similar results. The 2004 LAPOP survey, for
example, also asked a question about ethnic self-identification that was identical to that which appeared on
the 2001 census. It found that the percentage of people who identified as belonging to some indigenous
group (i.e., Quechua, Aymara, Guarani, etc.) represented 74 percent of the population in 2004, as opposed
to only 62 percent in the 2001 census.
22
this finding may support an assimilation/enclosure model of indigenous identity, rather
than a politicization approach.

To test the hypotheses more rigorously and examine the independent effects of
each variable, I carried out logistic regression analyses on both the sample of indigenous
language speakers (model 1) and the sample of people with at least some indigenous
ancestry (model 2). The results, which are presented in Table 3, provide further support
for an assimiliation/enclosure model of indigenous self-identification, but not for an
ethnic competition model. The coefficients of income, education and the level of
urbanization are negative in both models, indicating that wealthier, better educated,
urban-dwelling indigenous languages speakers (or people with indigenous ancestry) are
less likely to identify as indigenous than are poorer, less-educated and rural dwelling
indigenous language speakers (or people with indigenous ancestry). All three
coefficients are highly statistically significant in the sample that includes all people who
report having some indigenous ancestry, but the income variable falls below conventional
levels of statistical significance in the model that is restricted to speakers of indigenous
languages. The substantive effects of education are particularly strong. A simulation
based on Model 1 suggests that there is a 23 percent probability that a speaker of
indigenous languages in Bolivia will identify as indigenous if he or she has no education,
as compared to only a 7 percent probability if he or she has a university education,
holding all other variables at their means.

23
Both models also suggest that linguistic isolation makes self-identification as
indigenous more likely, as assimilation/enclosure theory (but not the ethnic competition
theory) would expect. Indeed, the dummy variable for monolingual indigenous language
speakers has a positive sign and is highly statistically significant in both models,
indicating that people who speak only indigenous languages are more likely to identify as
indigenous than people who also speak Spanish or who only speak Spanish. The
substantive effect of this variable on indigenous self-identification is also quite large. A
simulation based on Model 1 suggests there is a 24 percent probability that someone who
only speaks an indigenous language will self-identify as indigenous, whereas there is
only a 13 percent chance that someone who speaks Spanish in addition to an indigenous
language will do so.

The logistic regression results also provide some support for a politicization
model of indigenous identification. As the results show, participation in meetings of trade
associations, including peasant organizations, has a positive and statistically significant
effect on indigenous identification in both models. Ideology, meanwhile, has a negative
and statistically significant coefficient in both models, indicating that indigenous
language speakers (or people with indigenous ancestry) with left wing views are more
likely to self-identify as indigenous than are indigenous language speakers (or people of
indigenous ancestry) with centrist or right wing views. Participation in labor unions does
not have a statistically significant effect on the likelihood of self-identifying as
indigenous in either model, however. Moreover, ohe substantive effect of participation in
trade associations or holding left wing views on the likelihood of self-identifying as
24
indigenous is somewhat weaker than the effect of education or linguistic isolation. For
example, indigenous language speakers who participate frequently in meetings of trade
associations have a 17 percent probability of self-identifying as indigenous, as opposed to
less than 12 percent for those indigenous language speakers who never participate in
these meetings, holding all other variables at their means.

Several other variables that may be related to politicization levels--age, gender,
and linguistic group--also proved to have a significant effect on the likelihood of self-
identifying as indigenous. The positive sign for the male variable indicates that men are
more likely to self-identify as indigenous than are women, and the negative sign on the
coefficient of the age variable suggests that the young are more likely to self-identify as
indigenous than are older people. As discussed earlier, men may be more likely to self-
identify as indigenous because in Bolivia they typically participate at higher levels in
political parties, peasant associations, and other political organizations that have raised
ethnic consciousness. Young people, meanwhile, may be more likely to self-identify as
indigenous because they have come of age during a period of increased ethnic
consciousness and politicization.
11
The substantive effects of these variables are not
particularly strong. According to a simulation based on Model 1, male indigenous
language speakers have a 15 percent probability of self-identifying as indigenous,
whereas female indigenous language speakers have a 10 percent probability of doing so,
when all other variables are held at their means.


11
The coefficient of the variable measuring year that the survey was taken is also positive, indicating that
more people self-identified as indigenous in recent surveys. However, it is only statistically significant in
the sample that includes all people who acknowledge having some indigenous ancestry.
25
The most powerful effects are generated by the dummy variables for linguistic
group, both of which are highly statistically significant and in the expected direction in
both samples. A simulation based on model 1 shows there is a 24 percent probability that
a speaker of Aymara will self-identify as indigenous, as opposed to only a 9 percent
probability for speakers of other indigenous languages when all other variables are held
at their means. As noted above, this may be in part a reflection of the traditionally higher
levels of indigenous organizing among Aymara communities than among the Quechua-
speaking population. The substantive effect of speaking a lowlands indigenous language
is even greater. According to Model 1, speakers of lowlands indigenous languages have a
51 percent probability of self-identifying as indigenous, whereas speakers of other
indigenous languages have a 12 percent probability on average. As noted previously,
however, it is not clear whether speakers of lowlands indigenous languages are more
likely to self-identify as indigenous because they are more politicized or because they are
less assimilated.

With many of these findings, it is difficult to be certain if there is a causal
relationship between these variables and, if so, in what direction the causality runs. We
cannot be sure, for example, whether participating in trade associations or holding leftist
views leads people to self-identify as indigenous or whether self-identifying as
indigenous causes people to participate in trade associations and hold leftist views.
Similarly, it is not clear whether having little education and living in a rural area makes
some one more likely to self-identify as indigenous or if those people who self-identify as
indigenous are less likely to gain an advanced education or move to urban areas. It is also
26
quite possible that other variables, such as skin color or experiences with discrimination,
may shape indigenous identification and socio-economic status or the degree of
politicization. The ideal way to examine these thorny issues of causality would be to
track the same people over time to examine how changes in their socio-economic status,
political views or participation affected their ethnic identities. Unfortunately, there are no
panel data available that examines ethnic identification among the citizens of these
countries over a long period of time, so the issue of causality must therefore remain
unresolved. Nevertheless, the present study at least sheds light on the correlates, if not the
determinants, of indigenous identification.

Conclusion:

This study has found support for both the assimilation model as well as a more
political approach to understanding indigenous identity in Bolivia, but not for the ethnic
competition model. As the assimilation model would predict, socio-economic
advancement and linguistic integration reduce the likelihood that indigenous language
speakers (and people of indigenous ancestry) will self-identify as indigenous. However,
the probability that an indigenous language speaker will self-identify as indigenous also
increases with participation in meetings of trade associations, such as peasant
organizations, as a politicization model would expect. Indigenous self-identification is
also higher among Aymara speakers, who have a long tradition of indigenous organizing
and politicization, as well as among speakers of lowlands indigenous languages.

27
These findings have both practical and theoretical implications. On a practical
level, they suggest that indigenous identities in the region will be subject to two
countervailing trends in the future. On the one hand, the rapid decline in the number of
monolingual indigenous speakers, increased urbanization, and progress in the socio-
economic standing of most people of indigenous ancestry in Latin America may
gradually reduce the proportion of people who self-identify as indigenous. On the other
hand, the rising power of indigenous movements and the growing politicization of ethnic
identity may increase the number of people who self-identify as indigenous. Which trend
proves to be more powerful may determine the future of indigenous identities in the
region.

On a theoretical level, the findings of this study suggest that existing theories of
ethnic identity need to take politics more seriously. This study found that indigenous
identification in Bolivia is shaped not only by socio-economic or demographic factors,
but also by politics. Political experiences may shape ethnic identities by heightening
peoples awareness of discrimination, instilling a greater sense of ethnic solidarity, and
exposing people to new ideologies of race and ethnicity. As these findings make clear,
ethnic self-identification may be a social or economic act, but it is also profoundly a
political one.

28
References

Alb, Xavier. 1991. El Retorno del Indio. Revista Andina 2 (December): 299-345;
Cervantes, J orge. 2003. Genetic Differences Between Mestizo Populations of Bolivia
and Peru. Bulletin dInstitut Francais des Etudes Andines. 32, 1: 213-222.
Chandra, Kanchan. 2001. Cumulative Findings in the Study of Ethnic Politics. APSA-
CP (winter), 12, 1: 7-11.
de la Cadena, Marisol. 2000. Indigenous Mestizos: The Politics of Race and Culture in
Cuzco, Peru, 1919-1991. Durham, N.C.: Duke University Press.
Degregori, Carlos Ivn et al. 1998. La Construccin de la Nacin y la Representacin
Ciudadana en Mxico, Guatemala, Per, Ecuador, y Bolivia. Guatemala:
FLACSO.
De Pablo, R. et al. 2000. HLA Class I and Class II Allele Distribution in the Peruvian
Population. Tissue Antigens. 56, 6: 507-514.
Deruyttere, Anne. 1999. Nativos en nmeros. Revista BID Amrica. (Sept.-Oct.): 15.
Eschbach, Karl, and Christina Gmez. 1998. Choosing Hispanic Identity: Ethnic
Identity Switching among Respondents to High School and Beyond. Social
Science Quarterly. 79, 1 (March): 74-90.
Geertz, Clifford. 1973. The Interpretation of Cultures. (New York: Basic Books) Ch. 10,
pp. 255-279.
Harris, David R., and J eremiah J oseph Sim. 2002. Who is Multiracial? Assessing the
Complexity of Lived Race. American Sociological Review 67 (August): 614-
627.
29
Hill, Mark E. 2002. Race of the Interviewer and Perception of Skin Color: Evidence
from the Multi-City Study of Urban Inequality. American Sociological Review
67 (Feb.): 99-108.
Hopenhayn, Martn, and Alvaro Bello. 2001. Discriminacin tnico-racial y xenophobia
en Amrica Latina y el Caribe. Serie Polticas Sociales 47 (May).
Horowitz, Donald L. 1985. Ethnic Groups in Conflict. Berkeley: Univ. of California
Press.
Hurtado, J avier. 1986. El Katarismo. La Paz: Hisbol.
Instituto Nacional de Estadstica (Bolivia). 2003. Bolivia: Caractersticas
Sociodemogrficas de la Poblacin. La Paz: INE.
Massey, Douglas S., and Nancy A. Denton. 1992. Racial Identity and the Spatial
Assimilation of Mexicans in the United States. Social Science Research 21: 235-
260.
Nobles, Melissa. 2000. Shades of Citizenship: Race and the Census in Modern Politics.
Stanford, CA: Stanford University Press.
Park, Robert. 1950. Race and Culture. Glencoe, IL: The Free Press.
Portes, Alejandro. 1984. The Rise of Ethnicity: Determinants of Ethnic Perceptions
among Cuban Exiles in Miami. American Sociological Review 49: 383-397.
Rivera Cusicanqui, Silvia. 1986. Oprimidos pero no Vencidos: Luchas del Campesinado
Aymara y Qhechwa de Bolivia, 1900-1980. Geneva: Naciones Unidas.
Rojas Ortuste, Gonzalo, and Luis Verdesoto Custode. 1997. La Participacin Popular
como reforma de la poltica: Evidencias de una cultura democrtica boliviana.
La Paz: Ministerio de Desarrollo Humano.
30
Romero Ballivin, Salvador. 2003. Geografa Electoral de Bolivia: Tercera Edicin
Actualizada. La Paz: Fundemos.
Programa de Naciones Unidas para el Desarrollo (PNUD). 2004. Interculturalismo y
Globalizacin: Informe Nacional de Desarrollo Humano 2004. La Paz: PNUD.
Seligson, Mitchell A. 1998. La Cultura Poltica de la Democracia en Bolivia: 1998. La
Paz: U.S. Agency for Internacional Development.
--------. 2000. La Cultura Poltica de la Democracia en Bolivia: 2000. La Paz: U.S.
Agency for Internacional Development.
--------. 2003. Auditoria de la Democracia: Bolivia, 2002. La Paz: U.S. Agency for
Internacional Development.
Seligson, Mitchell A., Daniel Moreno Morales, and Vivian Schwarz Blum. 2004.
Democracy Audit: Bolivia 2004 Report. La Paz: U.S. Agency for Internacional
Development.
Telles, Edward E. 2002. Racial Ambiguity among the Brazilian Population. Ethnic and
Racial Studies. 25, 3 (May): 415-441.
--------. 2004. Race in Another America: The Significance of Skin Color in Brazil.
Princeton, NJ : Princeton University Press.
Van Cott, Donna L. 2005. From Movements to Parties in Latin America: The Evolution
of Ethnic Politics. New York: Cambridge University Press.
Van Evera, Stephen. 2001. Primordialism Lives! APSA-CP (winter), 12, 1: 20-22.
Yashar, Deborah J . 2005. Contesting Citizenship: Indigenous Movements and the
Postliberal Challenge in Latin America. New York: Cambridge University Press.
31
TABLE 1
PREDICTED EFFECTS ON THE LIKELIHOOD OF SELF-
IDENTIFYING AS INDIGENOUS

Variable Prediction of
Assimilation
Model
Prediction of
Ethnic
Competition
Model
Prediction of
Politicization
Model
Socioeconomic
Variables


Income

Negative Positive
Education

Negative Positive
Residential/linguistic
integration variables


Urbanization rate

Negative Positive
Speaks only
indigenous languages

Positive Negative
Political Variables


Participation in trade
association meetings
Positive
Participation in union
meetings
Positive
Leftist ideology

Positive
Demographic
variables


Aymara speakers

Positive
Age

Negative
Male

Positive
32
TABLE 2
Characteristics of self-identified indigenous respondents in Bolivian surveys
(% of people with each characteristic that self-identify as indigenous)

Indigenous language
speakers
Respondents with some
indigenous ancestry
Monthly income
0-250 Bolivianos 27.6% 25.2%
250-500 Bs 17.1 13.7
500-1,000 Bs 14.1 9.5
Over 1,000 Bs 9.0 5.0
Education
0-5 years of school 25.4% 21.7%
6-8 years 19.2 15.2
9-12 years 13.8 9.7
13 or more years 9.3 5.6
Residence
Urban>20,000 11.2% 7.1%
Urban/Rural 500-20,000 20.8 13.8
Dispersed rural 22.6 19.2
Linguistic isolation
Spoke indigenous language only 31.5% 35.2
Spoke Spanish 15.6 10.7
Participation in trade groups
Almost never or never 16.9% 11.3%
Occasionally 18.4 13.8
Frequently 22.5 16.6
Participation in union meetings
Almost never or never 17.4% 11.6%
Occasionally 19.4 15.0
Frequently 17.7 16.6
Ideology
Left 19.9 14.0
Center 14.6 10.1
Right 14.7 9.9
Linguistic group
Quechua speakers 12.8% 14.8
Aymara speakers 24.6 26.9
Other native languages 52.8 62.7
Age
Less than 40 17.8% 11.5%
40 and over 17.8 13.4
Gender
Male 19.3% 13.1
Female 16.0 11.2
33
Table 3
Correlates of Indigenous Self-Identification
(Logistic Regression Analyses)

Model 1
Indigenous Language
Speakers
Model 2
All people who report
indigenous ancestry
Constant -117.838
(69.924)
-191.803**
(56.911)
Income level -.099
(.060)
-.182***
(.048)
Years of education -.078***
(.015)
-.084***
(.013)
Urbanization level -.185***
(.0478)
-.185***
(.038)
Only spoke an
indigenous language
.888***
(.152)
.986***
.151)
Participation in union
meetings
.019
(.066)
.041
(.056)
Participation in trade
association meetings
.143*
(.057)
.141**
(.048)
Ideology (10 point
left-right scale)
-.067**
(.026)
-.051*
(.021)
Spoke Aymara 1.055***
(.129)
1.153***
(.114)
Spoke a lowlands
indigenous language
2.059***
(.225)
2.469***
(.240)
Male .419**
(.126)
.317**
(.100)
Age -.011*
(.004)
-.014***
(.004)
Year of survey .059
(.035)
.096**
(.028)
Pseudo R
2

.1218 .1436
N

2,428 5,147
* p<0.05, ** p<0.01, *** p<0.001 (two-tailed tests)
(Standard errors in parentheses)