Curran and Rivero Fuentes UNIDAD V

8/2/2019 Curran and Rivero Fuentes UNIDAD V

1/20

Engendering Migrant Networks: The Case of Mexican Migration

Author(s): Sara R. Curran and Estela Rivero-FuentesSource: Demography, Vol. 40, No. 2 (May, 2003), pp. 289-307Published by: Population Association of AmericaStable URL: http://www.jstor.org/stable/3180802

Accessed: 15/01/2009 18:14

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at

http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless

you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you

may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained athttp://www.jstor.org/action/showPublisher?publisherCode=paa.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed

page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the

scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that

promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

Population Association of America is collaborating with JSTOR to digitize, preserve and extend access to

Demography.

http://www.jstor.org
http://www.jstor.org/stable/3180802?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/action/showPublisher?publisherCode=paahttp://www.jstor.org/action/showPublisher?publisherCode=paahttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/3180802?origin=JSTOR-pdf


2/20

ENGENDERINGMIGRANTNETWORKS:THECASE OFMEXICANMIGRATION*

SARA R. CURRAN AND ESTELARIVERO-FUENTES

Thisarticle compares the impact offamily migrantand destination-specific networkson inter-national and internal migration. Weind thatmigrantnetworksare more importantfor internationalmoves than or internal moves and that emale networks are more importantthanmale networks ormoves within Mexico. For moves to the United States, male migrant networksare more importantfor prospective male migrants thanfor female migrants, andfemale migrant networks lower theodds of male migration, butsignificantly increasefemale migration. We uggest that distinguishingthe gender composition and destination content of migrantnetworksdeepens our understandingofhow cumulativecausation affects patterns of Mexican migration.

Two major developments in the study of migration in the past 20 years have been therecognition of the importance of gender and of social networks in the migratory and as-similation process. Until recently, these two fields have remained mostly separate (Curranand Saguy 2001; Grieco and Boyd 1990). Despite the differences that have been demon-strated in the motivations, risks, and norms that govern and promote the movement andassimilation processes for men and women, it has been a common assumption in quanti-tative research on social networks that networks act in the same ways and have the sameeffect on men and women's migration (Pessar 1999b). In addition, this research has oftenassumed that male and female former migrants offer the same kind of help to prospectivemovers and have the same relationship with members of their communities of origin(Curran and Saguy 2001). By examining the gender composition and destination contentof family migrant networks, we deepen interpretations of how cumulative causation andmigrant networks influence patterns of migration.We extend prior research on Mexican migration by comparing the effects of familymigrant networks on men's and women's migration to the United States and within Mexico.We evaluate the impact of migrant networks on the independent migration of young adultswithin Mexico and to the United States because their patterns of migration differ in eachcase. Men have dominated the independent migration of young adults to the United States(Delauney 1995; Donato 1994), whereas women's movement abroad faces strict patriar-chal restrictions (Donato 1994; Hondagneu-Sotelo 1994; Lindstrom 1997). In contrast,women have had a larger and long-standing presence in internal economic migration sincethe 1940s (Arizpe 1975, 1985; Mufioz, de Oliveira, and Ster 1977; Szasz 1999).

*SaraR. Curran,Princeton University, Departmentof Sociology, 153 Wallace Hall, Princeton,NJ 08544;E-mail: [email protected] Rivero-Fuentes, Princeton University, Office of Population Research,229 Wallace Hall, Princeton, NJ 08544; E-mail: [email protected] are grateful for continued institu-tional supportfrom the Office of Population Research, PrincetonUniversity, Center GrantP30 HD 32030, andfellowship supportfor Rivero-Fuentes fromthe Mexican government. We aregrateful to GermanRodriguez forexcellent statistical advice and to Melanie Adams for editorial assistance. Comments from Cecilia Menjivar,Josh Goldstein, MarciaCaldas de Castro,Ann Morning, MarionCarter,andthe anonymous reviewers improvedthe quality of this research. The authors are listed alphabetically, reflecting complete collaboration on thisproject. Earlyversions of this article were presentedby Rivero-Fuentes at the 2001 annualmeeting of the Popu-lation Association of America, the 2001 annual meeting of the American Sociological Association, and theProgramon Urbanizationand Migrationat the University of Pennsylvania. All errors of omission or calculationare the responsibility of the authors.Demography,Volume40-Number2, May2003: 289-307 289


3/20

Demography,Volume40-Number2, May2003BACKGROUND AND MOTIVATIONSocial Networks and MigrationSocial networks in relation to migration are commonly understood as the links betweenresidents in a community of origin and individuals who are living in another place orwho previously migrated, regardless of their current residence (Hugo 1991; Massey1990; Massey et al. 1987). These ties to migrantshave been incorporated nto most theo-ries that are used to explain international migration, including world systems theory(Portes 1978), dual labor markettheory (Piore 1979), social capital (Massey 1990), andnew economics models (Stark 1991; Stark and Levhari 1982; Taylor 1986). The conceptof networks for explaining migration has strong theoretical and intuitive purchase, aswell as empirical support.Social networks facilitate migration in several ways. Contact with migrants showsindividuals thatthey may be better off in a place otherthan their currentresidence (Hugo1991). Migrantnetworks reduce travel costs by providing information on safe and cheaproutes or smugglers and reduce emotional costs. They lessen "assimilation shock" if im-migrantsarrivein an environmentwhere others speak their language (Choldin 1973) andwhere living among other foreigners can easily prevent deportation(Massey 1990). Mi-grantnetworks also increase the expected benefits of migrationwhen contact with previ-ous migrantshelps individuals find jobs. Finally, migrantnetworks help to reduce livingexpenses and provide financial assistance on arrival.Quantitativestudies have demonstrated he importanceof social networksas a mecha-nism that facilitates migration, even after an individual's previous migratoryexperience,individual and household socioeconomic characteristics,and characteristicsof the com-munity of origin are controlled. For Mexican migrationto the United States, Massey andGarciaEspafia (1987) showed that the likelihood of men moving increases significantlywhen at least one member of their household previously migratedto the United States.Massey and Espinoza (1997) also found that kinship networks play an importantrole inincreasing the odds of first and subsequentmigrationsfor heads of households.The importanceof migrantnetworks has received less attentionin studies of internalmigrationin Mexico. To ourknowledge, only two qualitativestudies and one quantitativestudy have examined the role of social networks in promotinginternalmigrationand as-sisting migrants in the assimilation process. In the early 1970s, Arizpe (1975) observedtwo communities in central Mexico and found that migrants' accounts of higher wagesandvast opportunitiesfor employmentmotivatedmany other individuals to migrateevenwhen the informationprovided was faulty. Lomnitz (1977) described the emergence ofrural immigrantneighborhoods in Mexico City during the 1960s and 1970s, where ex-changes of supportamong neighboringmigrantsprovided a social safety net.A decade later, Taylor (1986) found that in two rural communities in Mexico, thepresence of immediatekin of the householdheadin anotherMexican state was only weaklyassociated with the odds of internal migration. Instead, migrant networks to the UnitedStatesmore stronglyaffected migrationto the United States.Taylorarguedthat the weakereffects of migrantnetworkson internalmigrationwere the result of a well-established linkto domestic labor markets, saturatingeach community with informationabout the costs,benefits, and opportunities of moving to urban destinations in Mexico. Unfortunately,other than Taylor's relatively small-scale study of the differential effect of networks forvarious destinations, no other study has comparedinternal and internationalmigration.Engendering Social NetworksThe literatureon gender and migration has shown that the processes, motivations, andsocial norms governing men's and women's movements and the way in which male andfemale migrants settle in the receiving society are different (Curranand Saguy 2001;

290


4/20

Engendering MigrantNetworks 291Hondagneu-Sotelo 1994; Lindstrom 1997; Pedraza 1991; Pessar 1999a, 1999b; Zlotnik1993). Hence, the importanceof networks may vary for men's and women's migration,andthe potential for help is not the same accordingto the genderof the previous migrants.The effect of social networks on the probability of migration may be different formen andwomen because the costs, risks, and benefits of migrationdiffer by gender.Priorresearch on Central American migrationdemonstratedthat women are perceived to, anddo, face greater risks when migrating to the United States (Cerruti and Massey 2001;Hondagneu-Sotelo 1994; Lindstrom 1997; Mahler 1999; Menjivar 2000). As a conse-quence, women benefit more than do men from traveling accompaniedor from informa-tion about safe routes.In addition, men and women have different sources of support in communities oforigin and destination.Lindstrom(1997) found that in ruralMexican communities, men'smigration to the United States relies on the experience of a wide array of friends anddistantkin, as well as close family members,whereas women's migrationdepends exclu-sively on ties to close family members.Because of prevailingideas aboutwomen's vulner-ability and norms of family honor,women can move only if there is a close relative withwhom to travel or with whom to live (Lindstrom 1997). In a related vein, network re-sources within the family are not similarly available to men and women. Mexican men'smigrationis commonly encouraged,and men are generally able to get other family mem-bersto assist them in theirattemptsto migrate.Assistance forwomen's migrationis depen-dent on the kind of move women plan.Whenwomen migrateas partof a family move, theyare able to capitalize on their family networks, independent of the previous migrant'sgender.Whenwomen attemptto migrateon theirown, eitheras independentmigrantsortoreunitewith theirhusbandswho have migratedbefore them, male members of their fami-lies often oppose theirmove. Inthese situations,women mustrely on thehelp of "women'snetworks"composed of female family membersand friends (Hondagneu-Sotelo 1994).Likewise, the demonstrative effect of social networks can be gendered. When themigrationof women is less frequentthan men's, the experience or presence of previousfemale migrants may teach women that they can migrate as well. Mahler (1999) foundthat in El Salvador, the presence of female viajeras (transnationalmigrants travelingbetween El Salvador and the United States) shows young children (especially girls) inthe community of origin that migration is a viable strategy for women's social mobility.Similarly, Cerrutiand Massey (2001) showed that a mother's prior migration increasesthe likelihood that both sons and daughters migrate, but it is more influential on adaughter'smove, and a father's prior migration is significant for a son's migration, butnot for a daughter's.

When there is a genderdivision in the labormarket of the receiving society, migrantsmay benefit more frommigrantsof the same genderwho provide themwith morerelevantinformation or contacts. This is the case for Salvadorans (Menjivar 2000), Mexicans(Hondagneu-Sotelo 1994), andGuatemalans(Hagan 1998) migratingto the United States.Among these migrants, the majorityof women work as domestic laborers,whereas menwork primarilyas unskilled laborers(farm, restaurant,or constructionworkers). Gender-specific migrationis promotedwhen, as Hagan(1998) noted, employersaskmen orwomenforreferences on new workersandkin arecalled fromGuatemalato migrate.Even if thereis little gender segregation among migrantsin the destination labormarket,there may bedistinctly gender-segregatedsocial spheres among migrants,so male and female migrantsdo not have the same resources to help othermigrants.Menjivar(2000) foundthatCentralAmericanfemale migrantsaremore likely to have extensive social networksthanaretheirmale counterparts.These women's networks reach outside immigrantenclaves to takeadvantageof the social services requiredto make ends meet in the United States. Havingaccess to female migrantnetworks may be importantfor prospective female migrantstoovercome not only the economic, but also the social barriers o migration.

Engendering MigrantNetworks 291


5/20

292 Dmgah,Vlm 0Nme ,My20The migrationprocess is dynamic, as aregenderrelations. For example, Kanaiaupuni(2000) found that the gender composition of village migrant networks has diminishingeffects on men's and women's propensities to migrate with time. We also suggest thatgender composition effects will vary, depending on the history and context of men andwomen's migration to a particulardestination. In our study, we evaluated the effect ofsocial networks on men's and women's decisions to migrateto the United States orwithinMexico. Contraryto Cerruti and Massey (2001), we measured the effect of the priormi-grationof any household member.We extend Kanaiaupuni's(2000) analysis by exploringthe effect of the gender composition of the migrants closest to the individuals (thosewithin their immediate families) instead of the village migrantnetworks.Unlike previousstudies of migration, our analysis focused on the migration of single young adults for avariety of methodological concerns addressedlater in the Data and Methods section.

The Case of Mexican MigrationPatternsof migrationare distinctly gendered in the Mexican context. That is, young menhave predominated n the migrationstreamsto the United States, butyoung single womenhave had an importantpresence in migrant streams within Mexico. Decisions to move,especially to the United States, are often proscribedby gendersocialization, expectations,and concerns aboutrisks (Hondagneu-Sotelo 1994; Kanaiaupuni2000; Lindstrom 1997;Szasz 1999). Because of a cultureof "domesticity" Beneriaand Roldan 1987; Kanaiaupuni2000; Lagarde1993), women areperceived to face greaterrisks andaresubjectedto greatercontrol over theirmovements, especially when they move on their own. As a result,womenface differentbarriers o U.S. migrationthanmen do andmustrely on differentsources ofsupportfor getting to the United States (Hondagneu-Sotelo 1994; Lindstrom1997).Young single women have had an important presence in migratory movementswithin Mexico since the 1940s. Although they did not constitute the majorityof internalmigrants until 1980, they dominated the migration flows to metropolitan areas, wherethey worked as domestics and in the service sector (Arias 1995; Garcia, Muiioz, and deOliveira 1979; Szasz 1999). Since 1980, the destinations of women's migration withinMexico have diversified, expanding to Export Processing Zones and to regionalagroindustries, where women work in manual activities and the service sector (Arias1995; Fernandez-Kelly 1983; Szasz 1999). Overall, women's independent, internal mi-gration is associated with moves of long durationand steady employment. The internalmigrationof young men differs from that of young women in various aspects, especiallyin the case of ruralmigrants. Men tend to move to destinations closer to their place oforigin; their migration is mostly seasonal, not permanent;and they take more flexiblejobs, like street vendors, construction workers, or jornaleros (daily wage laborers) inagriculture (Arizpe 1975, 1985). This difference in the migration experience of youngmen and young women implies differences in the ability of migrants to provide re-sources to potential or new migrants and the quality of those resources. The long-termexperience and steady employment of migrant women within Mexico may mean thatthey have a broaderarrayof resources to offer potential migrants than do male internalmigrants. It may also be the case that the informationand resources offered by womeninternalmigrantsto futuremigrantsis less gender specific than are those offered by theirmale counterparts.In contrastto internalmigrationpatterns,young men dominatedthe migrantflows tothe United States throughthe 1980s (Cornelius 1990; Donato 1994). While women's mainmotive for migrating to the United States has been family reunification, men have mi-gratedfor economic reasons (Donato 1994; Hondagneu-Sotelo1994). Men's andwomen'smigration to the United States tends to last longer than does their migration withinMexico. Mexican women, however, participate ess in the American labormarketthan doMexican men (Delauney 1995) and work in very different labor markets. Kossoudji and

292 Demography,Volume 40-Number2, May2003


6/20

Engendering MigrantNetworks 293Ramsey (1984) found that in 1980, 57% of Mexican women, but only 1% of Mexicanmen, working in the United States laboredin privatehouseholds.Given these patternsof migrationand priorresearchon gender, social networks, andmigration,we propose several different kinds of gender-differentiatingeffects in the mi-grationprocess. We have three sets of hypotheses. The first set of hypotheses comparesdifferences across the destinations:

1. Womenwill be less likely thanmen to migrate,but this difference will be greaterfor internationalmigrationthan for internalmigration.2. Social networks will have a positive effect on internationaland internalmigra-tion, but the effect of social networks will be larger in the case of internationalmigration.3. For internationalmigration, the effect of social networks will be positive andmore importantfor women thanfor men, but the effect of social networks will beno different for men and women who are considering internalmigration.

The second set of hypotheses proposes that the migratory experiences of previousmigrantsare qualitatively different if the previous migrantsare men rather than women,and vice versa:4. Given the extensive and well-developed migrant experiences of women in theinternalmigrantstreamsof Mexico, female migrantnetworks will have a larger,positive effect thanwill male migrantnetworks on internalmigration.5. Given men's largerandmore diverse presence in the United States, male migrantnetworks will have a larger facilitating effect than female migrantnetworks oninternationalmigration.

The third set of hypotheses proposes that the gender composition of migrant net-works interactswith the gender of the migrantto have a different impact on the propen-sity to move.6. For both internal and internationalmigration, migrant networks that are com-posed of men will have a larger, positive effect on men's migration than onwomen's.

7. Similarly,migrantnetworksthatarecomposed of women will have a larger,posi-tive effect on women's migrationthanon men's.DATAAND METHODSDataThe dataused in this projectcame fromthe 1999 Mexican MigrationProject(MMP).' Thisdataset containsinformationabout 52 communitiesin Mexico thatwere surveyedbetween1982 and 1997. Each communitywas sampledonly once, andfor most of these communi-ties, therewas a complementarysample taken in the United States of households fromtheoriginatingcommunities. We did not use the complementarysample in the United States.The MMP sample was drawnfrom the 13 states that are the source of most migrantsto the United States. The survey was not designed to represent an area of high internal

1. The data set and informationabout the MMP survey is available at http://www.pop.upenn.mexmig,anda description appears in Massey, Goldring, and Durand(1994).



7/20

294 Demography, Volume40-Number2, May2003

migration,but 9 of the states that were included had among the highest rates of internalemigration during the late 1970s and 1980s, and 3 had medium to low rates of internalemigration (Corona 1993; Partida Bush 1993). Even in a region of high migration, therate of adults' emigrationto the United States (the ratio of adults aged 16 and older whomigrated in the year before the survey to all adults aged 16 and older) varied greatlyacross communities, from almost 2%to 66%. The communities sampledalso offer a widevariationin the rate of adult internalemigration(from almost 5%to 41%).Withineach community,households were randomly sampled.Withineach household,the investigators obtained basic social and demographic information at the time of theinterview for the household head, the spouse and children of the household head (even ifthey lived outside the household), and all other individuals who were considered house-hold members (including aunts, cousins, and friends who lived in the household). Migra-tion data about the first and last move were collected about any household memberwhohad some migration experience.Definition of the Study PopulationOnly individualswho were listed as "childrenof the household head,"had never married,and were aged 17-25 at the time of the survey are included in the study. Restricting theanalysis to never-marriedchildren of the household head means that we had a sample ofmen and women who had similar status in the household and whose migration can bedefined as independent or economic. Most of the study migrantsworked at the destina-tion. Of the young adult men who migrated within Mexico, 86% worked, and of thosewho migrated to the United States, 87% worked. The young adult women were slightlyless likely to work, but they worked at the same rate in both destinations (84%).The age and maritalrestriction capturedthose individuals who were most at risk ofmigrationwho still lived in theirparents' households and would be influenced by familymigrantnetworks.2Marriedchildrenwho lived in separatehouseholds would be exposedto a different set of migratorynetworks. The criteriafor establishing the lower marginatage 17 reflects our empirical observationthat the percentage of never-marriedsons whomigrated during the two years before the survey increased from about 5% at age 16 toabout 10%at age 17 for migrationto the United States, and from 3% to 5% for internalmigration. Women's migration startedat somewhat younger ages, especially in the caseof internalmigration,but therewas also an increase at age 17, from 2%to 3%for internalmigration. Furthermore,economic motives for migrationbecome more importantat age17. The percentage of individuals who worked duringtheir last migrationwas lower forthose aged 16 (69.39% for men and 65.22% for women) than for those aged 17 (79.21%for men and 70.97% for women).Estimating the Odds of MigratingOne limitation of the data is that a population at risk of migratingat any time before thesurvey cannot be reproducedbecause informationabout individuals comes from house-holds that were present during the survey, not from households that had disappearedbefore the survey (especially as a result of migration within Mexico, since there is acomplementarysample for those households that moved to the United States, but not forthose that moved within Mexico). The missing individuals affect the sampling propor-tions for migrantsand nonmigrants.If we were to estimate a multinomial regression that compared migrating withinMexico, migrating to the United States, or not migrating at all, the differences in thesampling proportionswould affect the estimation of the P coefficients. This effect doesnot occur when a logistic model is estimated.When the sampling criteriaare independent

2. The survey does not include informationabout migrantnetworks beyond the immediate family.

294 Demography, Volume 40-Number2, May2003


8/20

Engendering MigrantNetworks 295of the covariates included in the analysis, the probabilityof being in the sample factorsout anddoes not affect the estimation of the Ps. This is the reason why the coefficients ina logistic regression are similar,regardlessof whetherthe dataare sampled prospectivelyorretrospectively.However, the interceptin the retrospectivescheme is differentanddoesnot representthe value of the true log-odds (McCullagh and Nelder 1989). We took ad-vantage of this characteristicof the logistic models, and treatedthe data as if they camefrom two matched case-control samples. Doing so allowed us to estimate the association(odds ratios) between the covariates and the odds of migration.However, it did not allowus to estimate the migrationrate.We constructed two separate samples. In one sample, cases were individuals whomigratedat least once across state boundaries within Mexico duringthe two years beforethe survey, and controls were those who did not migrate duringthat period; in the othersample, cases were individualswho migratedto the United States at least once duringthetwo years before the survey, and controls were those who did not migrate. In each in-stance, every migrantwas matched with one or more individuals in the reference or con-trol population (nonmigrants), according to two criteria: age at the time of the survey(specified in single years) and community of origin. Age is directly linked to the prob-ability of migration,andthe communityof origin summarizesseveral othervariables thatrelate to the propensity to migrate, including year of the survey, rural or urban origin,wage and employment opportunitiesin the place of origin, and the prevalence of internaland internationalmigration in the community.Among the study population, only 15 outof 7,513 (0.20%) migrated both within Mexico and to the United States duringthe twoyears before the survey. These cases were eliminated from the sample.To make comparisons across the two samples, we had to guaranteethat they wereindependent. Therefore, we randomlydivided the nonmigrantswho matched the age andcommunityof both internalandinternationalmigrantsinto two groups. Consequently,thenonmigrantsin the sample for internalmigrationwere different from the nonmigrantsinthe sample for internationalmigration.The sample for internal migration was composed of 259 individuals who migratedwithin Mexico duringthe two years before the survey and 1,086 individualswho did not.The sample for internationalmigration is composed of 837 migrants and 3,722 nonmi-grants.These migratorystatuses were used as the dependentvariables in the analysis.CovariatesFour sets of measureswere centralto the study: gender, family migrantnetworks, gendercomposition of family migrant networks, and interactionsbetween gender and family mi-grantnetworks. Genderwas measureddichotomously with male as the omitted category.The measure of a family migrantnetwork is destination specific. We measured fam-ily migrantnetworks within Mexico by coding whether any family member (other thanthe observed individual)-including mothers, fathers, brothers, sisters, sons, anddaughters-migrated within Mexico at least two years before the survey.3Eligible priormigrants included sons and daughtersof the household head who no longer lived in thehousehold. A similar coding scheme was used for family migrantnetworks to the UnitedStates.Men's andwomen's migrationmay be differently influenced by access to andthe useof networks outside the immediate family. We could not measure the size or extent ofthese networks. However, previous research highlighted the importance of family net-works over other networks in men's migration (Massey and Garcia Espania1987) and

3. In work not shown here, we also estimated the effects of a count, ratherthan a dummy variable, andfound similar results. We chose the dummy variablebecause it simplifies the interpretationof the results, espe-cially when we test interactioneffects.



9/20

296 Dmgah,Vlm 0Nme ,My20showed thatfamily networks are one of the most importantdeterminants n the migrationof both men and women (Cerrutiand Massey 2001). Our measure of family migrantnet-works is similar to Massey and Garcia Espafia's (1987) "household network ties,"Kanaiaupuni's(2000) "migrantchildren networks," and Lindstrom's (1997) network of"close family members."We also measured the gender composition of the migrantnetwork. Female migrantnetwork within Mexico measures whether any woman in the family had migratedwithinMexico two or more years before the survey. Similarly, male migrant network withinMexico measures whether any male memberof the family had migratedinternallytwo ormore years before the survey. We constructed similar measures for migrantnetworks tothe United States. This distinction of male and female networks is the best measure of thegender composition of the networks, given the sample size available. Disaggregating thenetworkmeasuresfurther nto female-only,male-only, andmale-and-femalenetworkswasnot possible because of the few cases in some alternatives.For instance, only 3% of thesample for internationalmigrationhad a female-only network.Besides controlling for age and community of origin throughour matching process,we also controlled for previous migratory experience, family migrant networks to thealternativedestination,4educational level, and family size. Priormigration experience isdestination specific and accounts for the informationandpersonal contacts an individualgains that may otherwise be obtained through his or her family's migrant networks(Massey and Garcia Espafia 1987). Previous migrationexperience within Mexico or theUnited States measures whether the individual made a first trip to the destination at leasttwo years before the survey.Educationallevel was measuredcategorically, dividing years of schooling into thosewith 0-6 years, 7-9 years, 10-12 years, and 13 or more years. This codification capturesthe nonlinear relationship between the propensity to migrate and education observed inpriorresearchon Mexican migration.Migrationis higher amongthose with some second-ary education, but lower among those with primaryandmore than a secondaryeducation(Jasso and Rosenzweig 1990).Employment before migration may also affect the probability of men and womenmigrating (e.g., see Kanaiaupuni2000). Unfortunately,we could not control for the indi-vidual condition of employment at the moment at which the risk of migration began inour sample. However, by controlling for community,we indirectly controlled for the rateof male and female employment in the place of origin, among othercommunity factors.Family size was measured by summing the numberof currenthousehold membersand the number of sons and daughters of the household head who did not live in thehousehold. Previous research demonstratedthat family size increases the probability ofmigration. Part of this effect may result from larger families having larger social net-works and thereforegreateraccess to migrationpossibilities (Massey et al. 1987; Masseyand GarciaEspafia 1987; Starkand Taylor 1991).The means and proportionsof all variables in the analyses are presented in Table 1.About 19%in the internal-migrationsample and about 18%in the international-migrationsample had migratedwithin the past two years. For the two samples, more than 59% ofthe individuals had a family migrant network to the respective destination, primarily amale network in both samples. With regard to education, the level of education wasslightly higherin the internal-migration amplethanin the international-migrationample.The two samples had slightly more men than women, the mean age was about 20 years,and the average family size was aboutnine members.

4. Previous studies have shown thatmigrantnetworks to the United States may have a negative effect onthe odds of internal migration, whereas internal migrant networks may have a negative effect on the odds ofinternationalmigration (Lindstromand Lauster2001).

296 Demography, Volume 40-Number 2, May 2003


10/20

Engenderin -- 2

Table 1. Description of Dependent and IndependentVariablesProportionn Subsample Proportionn SubsampleVariable forInternalMigration forInternational igrationDependentVariables

Migrated ithinMexico .193Migratedo theUnitedStates .184

Independent ariablesSexMen .533 .579Women .467 .421

Familymigrant etworkwithinMexico .594 .550Familymigrant etworko theUnitedStates .557 .681Femalemigrant etworkwithinMexico .250Malemigrant etworkwithinMexico .524Femalemigrant etworko theUnitedStates .250Malemigrant etworko theUnitedStates .653PreviousmigratoryxperiencewithinMexico .098Previousmigratoryxperienceothe UnitedStates .200Educationalevel0-6 years .306 .402

7-9 years .326 .29310-12 years .233 .20813or moreyears .135 .091

Meanage(standardeviation) 19.8(2.376) 20.2 (2.397)Mean amily ize(standardeviation) 9.3 (3.140) 9.6 (3.167)Number f communities 43 51Sample ize 1,345 4,559

ModelsWe include the notationonly for our first model (Eq. (1)). Ourlogistic-regression estima-tion takes into account the fact that the matching mechanism yields serially correlatedobservations of individuals of the same age andfrom the same community.We fit a fixed-effects model within these two variables. The other models we estimate build on thebaseline model in an additive way, and the text explains each of them. To test our firsthypothesis, thatwomen's propensityto migrateis lower thanmen's but significantly moreso for internationalmigration,we examine the coefficient for womanwithin each equationestimated on the separate destination samples (P,) and then compare Pinternal > 1international

( ( J(dest) (agecommunit =P(not _ move) a commuty = agemmnity + [ woman + [l (controlsi). (1)(n~~~~~~~~~~ Og,otmmovey



11/20

298 Demography, Volume40-Number2, May2003In our second model, we test whether family migrantnetworks have a positive effect onthe decision of young adults to migrate internationallyand internally,but that the effectof networks is largeron internationalmigrationthanon internalmigration.We add a mea-sure of the family migrant network to the baseline model and estimate a coefficient, I2.According to our hypothesis, P2international > P2internal > 0.Our third model tests whether the general effect of family migrant networks is thesame formen andwomen who areconsideringinternalmigration,butnot the same for menand women who are considering internationalmigration. It is done by adding an interac-tion term for woman with family migrant network (woman x family migrant networkdes)and estimating a coefficient 3 We examine whether 3 is equal to zero in the model forinternal migration and significantly greater than zero in the model for internationalmigration.In the fourthmodel, we test whether networks that are composed of female migrantshave a larger effect than do networks that are composed of male migrants on the likeli-hood that young adults will migrateby adding a measure offemale migrantnetworkdestoour baseline model and estimating a coefficient, 4. In the same model, we test whethernetworks that are composed of male migrantshave a larger,positive effect on migrationthan do networks that are composed of female migrants by adding a measure of malemigrant networkd,,,,nd estimating a coefficient, 6. We evaluate supportfor our hypoth-eses by examining whether P4internal > P6internal > 0 and whether 36international> P4international 0.We use a one-tailed Waldtest to comparethe coefficients.5Ourfifth model builds on the fourth, testing whether female networks have a larger,positive effect in facilitating the migrationof young adultwomen than do male networksand whether male networks have a larger,positive effect in facilitating the migration ofyoung men. We add interaction effects of women x female migrant networkdest estimat-ing 5) and women x male migrant networkdest estimating P7).We expect to find P5> 0and 0 > 57 for both internaland internationalmigration.The results of our estimations are presented in Tables 2 and 3. For ease of presenta-tion, the log-odds coefficients are transformedinto odds ratios (exp(P)). The general in-terpretationis that odds ratios between 0 and 1 indicate decreased odds of migration,given a certain condition, and odds ratios greaterthan 1 increase the odds of migration.We evaluate supportfor our hypotheses on the basis of one-tailed t tests of significancefor the log-odds coefficients.RESULTSResults for Internal MigrationThe coefficients for educational level, migratoryexperience, and household size do notvary amongthe five models. Educationallevel is not significant, implying that the odds ofinternalmigrationdo not varywith years of schooling. Family size is also not significantlyassociated (at the .10 level) with the odds of migration. Previous migratory experiencedoes contributesignificantly to explaining internalmigration.In the five models, the oddsof migrating within Mexico are almost double for individuals who had prior migrationexperiences than for those who had never migrated.Contraryto the results of LindstromandLauster(2001), family migrantnetworksto the United States do not have a significanteffect on the odds of migrationwithin Mexico once one controls for age, gender, educa-tional level, previous migratoryexperience, family size, and communityof origin.There is some evidence that women have slightly lower odds of migrating than domen for the age group and region considered. The odds of migration are 17%lower for

5. The Wald statistic in this case is W= (3, - 6)2/var(4 - 136),and is distributedas a chi-square with 1degree of freedom.



12/20

Engendering MigrantNetworks 299Table2. Logistic RegressionEstimatesof InternalMigration (Odds Ratios Presented)MigrantVersusNonmigrant Parameter Model 1 Model 2 Model 3 Model 4 Model 5(Man)Woman(Has No FamilyMigrantNetwork)FamilyMigrantNetworkWithin MexicoWoman x FamilyMigrantNetwork(Has No FemaleMigrantNetwork)Has FemaleMigrantNetworkWithin MexicoWoman x Female

MigrantNetwork(Has No Male MigrantNetwork)Has Male MigrantNetworkWithin MexicoWoman x Male MigrantNetwork(Has No FamilyMigrantNetwork to the United States)FamilyMigrantNetwork

to the United States(Has No MigratoryExperience)Has MigratoryExperienceWithin Mexico(0-6 Yearsof Education)7-9 Yearsof Education10-12 Yearsof Education13 or MoreYearsof EducationFamilySizeLog-LikelihoodLR Chi-SquareN

exp(P1) 0.830 0.830 0.766 0.833 0.666

exp(p2)

exp(P3)

exp(P4)exp(pS)

exp(Pf)exp(]37)

1.456* 1.380- 1.112

1.807* 1.841*.~-.-.~-. - . ..972

~- ~- ~- 1.202 0.997.~-.~.~ -.~- ~ 1.494

exp(p8) 1.095 1.098 1.098 1.110 1.116

exp(o9) 2.072* 1.875* 1.880* 1.668* 1.701*

exp(pl0i) 0.821 0.819 0.818exp(P,o2) 1.095 0.928 0.923exp(P,03) 0.847 0.841 0.836exp(P,l) 0.989 0.983 0.983

-424.2214.34*1,345

-421.6719.64**1,345

-421.6019.79**1,345

0.8100.8730.7980.978

-417.0628.87**1,345

0.8050.8710.7990.980

-416.2130.56**1,345

*Aone-tailedest ndicateshat hecoefficient ehaves spredictedtp < .05.**Aone-tailedest ndicateshat hecoefficient ehavesspredictedtp < .01.

women thanfor men in threeof the models, 24%lower in the thirdmodel, and 33% lowerin the fifth model. However, this difference approaches significance only in the fifthmodel, with a probabilityof .10 (for a one-tailed test).The results in the second model show that family migrantnetworks within Mexicoincrease the odds of internalmigrationby 46%. The significance of family migrantnet-works disappearsin the thirdmodel, which adds an interactiontermwith gender.Never-theless, the odds ratio of family migrantnetworks is similar (exp([2) = 1.380) to that in


13/20

300 Demography,Volume40-Number2, May2003Model 2, while the odds-ratio coefficient for the interactionterm (exp(,3)) is close to 1.Thus, the presence of internalmigrantnetworks increases the odds of internalmigrationfor both men and women.

In our fourthmodel, having female migrantnetworks has a largerpositive influenceon migrationthan does having male migrantnetworks. The effect of female migrantnet-works is positive and significant, whereas that for male migrantnetworks is not differentfrom zero. With regard to the difference between the coefficients, the effect of femalemigrantnetworks is significantly largerthan the effect of male networks (Table2, Model4: log(1.807) > log(1.202), X2= 2.52, p < .056). The odds of internalmigration of indi-viduals with female migrants in their families are 1.8 times the odds of those withoutfemale migrants (significant at the .01 probability level). Because the model controls formale migrantnetworks, the interpretationof the effect of female migrantnetworks shouldbe understoodto be the effect over and above that of having a male migrantin the family.In our fifth model, female migrant networks positively and significantly affect themigrationof both men andwomen andareequally important or women and men. Femalemigrantnetworks continue to increase the odds of migration for men (4P >0, p < .004),but the effect is not statistically different for women (testing the significance of the coef-ficient for the interaction,P5,in Model 5). Male migrantnetworks, on the otherhand, donot affect the odds of internal migration of either men or women. The effect of malemigrant networks on men's migration (P6) and on women's migration (p6 X P7) is notsignificantly different from zero.One explanationfor the positive and significant effect of having female migrantnet-works on the internalmigrationof both men andwomen is that a female migrant'sexperi-ence is qualitatively different from a male's. Because women's migrationwithin Mexicois more permanentand female migrant'semployment is more stable than male migrants',this finding may not be surprising.Permanence in a place of destination may yield di-verse access to resources thatwould be useful to men or women, yielding greaterbenefitsfrom female migrantnetworksthan from male migrantnetworks.Results for International MigrationThe effect of education on the odds of internationalmigration is not linear, but also notsignificant. The direction of effects is similar to that of previous studies (see Jasso andRosenzweig 1990), and we suspect that the lack of significance is due to our studypopulation's young age. Family size has a positive but insignificant effect. Previous mi-gratory experience has an importantand significant effect on the odds of internationalmigration.The odds of migratingto the United States were at least 70% higher for thosewho had previously migrated internationallythan for those who had not. The odds ofinternationalmigrationarethe same for individualswith family internalmigrantnetworksthan for those without such networks.Women from the same community who are of a similar age, educational level, andmigratoryexperience arenoticeably less likely thanare men to embarkon an independentmigrationto the United States duringtheiryoung adultyears. The odds of women migrat-ing internationallyare about 80% lower than the odds of men migrating internationallyfor all five models.The odds ratio of family migrant networks in Model 2 (exp(P2)) confirms that themigration experience of other family members has a positive and significant effect (at aprobability level of .01) on young adults' migration to the United States. The odds ofmigration for those with migrantnetworks are 2.4 times the odds of migration for thosewho have no migrant networks. We do not find support for our hypotheses that familymigrantnetworks have a differential effect on men's and women's migration. The oddsratio of the interactionterm (exp(33) in Model 3) is more than 1, but is not significant,which implies that the increase in the odds of internationalmigrationassociated with the



14/20

Engendering MigrantNetworks 301Table3. Logistic RegressionEstimatesof InternationalMigration (Odds Ratios Presented)MigrantVersusNonmigrant Parameter Model 1 Model 2 Model 3 Model 4 Model 5(Man)Woman(Has No FamilyMigrantNetworks)FamilyMigrantNetwork tothe United StatesWoman x FamilyMigrantNetwork(Has No FemaleMigrantNetwork)

exp(Pl) 0.222** 0.212** 0.180** 0.213** 0.185**

exp(,2)

exp(P3)

2.408** 2.333**- 1.209

Has FemaleMigrantNetworkto the United StatesWoman x Female

MigrantNetwork(Has No Male MigrantNetwork)Has Male MigrantNetworkto the United StatesWoman x Male MigrantNetwork(Has No FamilyMigrantNetworkWithin Mexico)Has a FamilyMigrantNetwork

Within Mexico(Has No Migratory Experience)Has MigratoryExperienceto the United States(0-6 Yearsof Education)7-9 Yearsof Education10 - 12 Yearsof Education13 or More Yearsof EducationFamilySizeLog-LikelihoodLR Chi-SquareN

exp(P4)

exp(P5)

exp(P6)exp(p7)

- ~- 1.049 0.7155.387*

~- ~- - ~ 2.210** 2.520**0.524

exp(p8) 1.074 1.029 1.029 1.031 1.045

exp(p9) 2.008** 1.731** 1.734** 1.711* 1.670*

exp(P10i)exp(P,02)exp(,103)

1.2540.7920.663

1.2430.7600.670

1.2410.7600.670

1.2410.7550.666

exp(Po,) 1.041 1.023 1.023 1.023-1,328.97

384.15**4,559

-1,302.05438.00**4,559

-1,301.86438.37**4,559

-1,304.70432.70**4,559

1.2200.7540.6751.027

-1,278.18485.72**4,559

*Aone-tailedest ndicateshat hecoefficient ehaves spredictedtp < .05.**Aone-tailedest ndicateshat hecoefficient ehavesspredictedtp < .01.

presence of previous migrants in the family is the same for young adult men and youngadultwomen.In Model 4, male migrantnetworksaremore important hanfemale migrantnetworksfor predicting internationalmigration. The coefficient for male migrantnetworks is sig-nificantly largerthanthe coefficient for female migrantnetworks(i.e., comparingwhether6> P4 in Model 4: log(2.210) > log(1.049), x2 = 18.90, p < .000). Individualswith male


15/20

302 Demography, Volume 40-Number2, May2003migrant networks are 2.2 times more likely to migrate than are those without such net-works. The presence of female migrantsdoes not significantly increase the odds of migra-tion. These findings mirror hose for internalmigrationin which female migrantnetworkswere more important han male migrantnetworks. In this case, the longer history of men'smigrationto the United States and the greaterarrayof job opportunitiesfor men decreaserisks and increase benefits to a potentialmigrant.In the fifth model, we test whether female migrantnetworks are more importantforwomen and male migrantnetworks are more importantfor men. We find thatthe interac-tion effect of woman andfemale migrantnetwork is significantly greaterthan 1 at a prob-ability level of .001 (exp(p5)) and indicates that the effect of female migrantnetworks isgreaterfor women than for men. Similarly,the interactioneffect of woman and male mi-grant network is less than 1 and significant at a probability level of .05, indicating thatmale migrantnetworks aremore importantfor men than for women.Multiplyingthese interaction effects by the networkeffects yields the effect of femaleand male networks on the odds of women's internationalmigration(exp(54)x exp(55)andexp(P6) x exp(f7), respectively). A comparison of these odds ratios with the effects ofnetworks on the odds of men's migration (exp([4) and exp(P6)) reveals that the oddsof migrationamong young adultmen with male migrantnetworks are2.5 times the odds ofmigration of those with no male migrantnetwork. The presence of male migrants in thefamily, however, does not have a significant effect on the odds of migrationof young adultwomen (exp(p4)x exp(P5)= 1.326, p > .10). Female migrantnetworks, on the otherhand,are associated with an increase, by 3.8 times, in the odds of young women migrating.Butfemale migrantnetworks diminish the odds of men's migrationby 30%comparedto menwithout female networks.Comparing Internal and International MigrationThe difference between men's and women's propensityto migrateis significantly greaterfor internationalmigrationthan for internalmigration.Womenhave much lower odds ofmigratinginternationallythando men. For internalmigration,women have lower odds ofmigratingbut the difference is not statistically significant. In a comparisonof P, in Model1 of Table 2 and Table 3, we find that Ilinternal > Plinternational(log(0.831) > log(0.222),t = 7.148). In addition, family migrant networks play a larger role for migration to theUnited States than for migrationwithin Mexico. Contraryto Taylor(1986), we find thatthe presence of migrants in the family has a positive, significant effect on the odds ofinternalmigration.This effect, however, is less importantthanthe effect of U.S. migrantnetworks on migrationto the United States.The coefficient for internalmigrantnetworks,

2internal, is smaller than the coefficient for international migrant networks, 2intemrational,(log(1.456) < log(2.408), t =-2.422).When migrant networks are divided into male and female networks, furtherdiffer-ences between internaland internationalmigration are revealed. Male migrantnetworksfacilitate internationalmigration and female migrantnetworks facilitate internal migra-tion. The magnitudeof the effect of male migrantnetworks on internationalmigration isno different from the effect of female migrantnetworks on internalmigration(log(2.210)> log(1.807), t= 0.945).Furthermore,Model 5 in Tables 2 and 3 shows that male and female migrant net-works differentially influence the internationalmigrationof men and women, butnot theirinternalmigration. Male and female migrantnetworks and their interactionwith genderare all significant at a probability level of .05 for internationalmigration. In the case ofinternalmigration, only the coefficient of female migrantnetworks is significant, but itseffect is the same for men andwomen.These results reflect four observations about the migration process. First, internalmigration is less risky and difficult than internationalmigration and hence individuals



16/20

Engendering MigrantNetworks 303need less informationand resourcesprovidedby migrantnetworksto migrate.These find-ings affirm those of Taylor's (1986) smaller-scale study. Second, the barriersto interna-tional migration are much greater for women than for men, but are little different forinternalmigration.Third,female migrantnetworkswithin Mexico are more useful to bothmen and women than are male migrantnetworks. Fourth,male and female migrantnet-works to the United States reflect differentarraysof resources, andpotential migrantmenand women have different access to, or make different use of, the information and re-sources available throughthese gender-differentiatednetworks.CONCLUSIONSThe results from our study contributeto our understandingof the relationship betweensocial networks and migration, as well as gender and migration. We presented evidencefrom the literatureson both topics to show how migrant networks influence migrationthroughvariousgenderrelations. We compared young adult men's and women's propensi-ties to move, dependingon the destination. Wecomparedthe differentialimpactof migrantnetworks on young adult men's andwomen's decisions to move to particulardestinations.We evaluated how the gender composition of migrantnetworks influences migration todifferent destinations. In addition, we compared the effect of gender-differentiatedmi-grantnetworks on the migrationof young adultmen and women across destinations. Ouranalysis of the independentmigrationof young Mexicans to the United States andwithinMexico confirmedmany of our hypotheses.Our most importantfinding is that the effect of being linked to a former migrantwithin the household varies by the gender of the previous migrantfor both internationaland internalmigration.Being female andhaving access to a priorfemale migrantnetworkis importantfor facilitating internationalmigration;similarly,being male and having ac-cess to a male migrantnetworkis important or facilitating internationalmigration.How-ever, female migrantnetworks within Mexico are equally helpful to men andwomen andmale internalmigrantnetworkshave no influence on internalmigration.Although our study and the dataavailable to us did not give us informationabout thecontent of migrant networks within households, beyond their gender composition, weknow from prior ethnographicresearch that men and women live the migrationprocessdifferently.We observed that men andwomen face different barriersto migration,whichrequire the mustering of different kinds of resources to migrate. Upon their arrival in aplace of destination, their sources of social supportand access to jobs and opportunitiesmay differ, depending on how gender-segregatedthe labormarketor migrantenclave orhow well-established the migrantnetwork. It is not surprisingthat female migrant net-works aremoreuseful for facilitatingmigrationwithin Mexico because a significant num-ber of women have been participatingin the migration streams for more than 40 years,and these women have obtained steady and diverse employment, yielding a broad arrayof resourcesfor male andfemale migrantsalike (Arias 1995;Arizpe 1975; Lomnitz 1977).As other studies have shown for Mexico-U.S. migration,the barriers o female inter-national migration are much higher than they are for male internationalmigration.Eventhoughmen have been migratingto the United States for many years, they are not equallyhelpful to potential male and female migrants.This difference is partially due to genderdifferences in places of origin and destination.Womenhave to garer more resources andinformationeither to convince their families that they can migrate or to resist their fami-lies' influence and migrate anyway. The resources that priormale migrantsprovide maybe unavailable to women or insufficient to overcome these barriers.Priorfemale migrantsmay have important demonstrative effects, as was shown in prior ethnographic work(Mahler 1999), as well as different resources and informationaboutdestinations that canhelp to overcome the barriers o migration(Hondagneu-Sotelo 1994; Menjivar2000). Eth-nographicresearch has also shown that male migrants can be more isolated in places of



17/20

304 Demography, Volume 40-Number2, May2003destination and know less about the wider arrayof social services and supportavailableto ensure safety and survival (Menjivar 2000), which are especially importantfor pro-spective female migrants.This difference between male and female migrants'experiencesmay be limited to the CentralAmerican migrantsstudied by Menjivarand not extend toMexican migrantswho have had a longer history of migrationto the United States. Nev-ertheless, our results do not suggest otherwise.The gender division of laborfor Mexican migrantsto the United States explains whymen with female migrantnetworks have significantly lower odds of internationalmigra-tion. Having a female migrant network limits prospective male migrants' opportunitiesbecause female migrantsto the United States are less likely to work and to know aboutjob opportunities. If they do work, female migrants may have access only to jobs thatmen would not accept (e.g., as domestic workers or hotel maids). The opportunitiesthatfemale migrantsoffer are so gender specific thatmen with female migrantnetworks pre-fer not to migrate.Although we cannot show exactly why the gender composition of thenetwork matters,ethnographicwork suggests several reasons for why it should.Otherways in which gendermay influence the content of migrantnetworks and sub-sequent patterns of migration, but that we were not able to study, are the frequency ofmale and female returnmigration,male and female migrants' visits to their communitiesof origin, male and female contact through telephone or remittances, male and femaledifferences in documentedand undocumentedstatus, or differentialrates of failed or suc-cessful migration. Ethnographicevidence from a variety of places has suggested impor-tant gender differences in many of these patterns(Curranand Saguy 2001). For example,in two systematic studies of the Dominican Republic, men and women were found tohave different patternsof remittances, investments in communities of origin, and expec-tations aboutreturnmigration(Georges 1990; GrasmuckandPessar 1991). This may alsobe true for Mexican migrants.Replicating these models using data from other settings may yield different results,given different combinations of restrictions on migration and behavior or the immigrantlabor market. For example, the ethnographicliteratureon Filipino immigrantsto HongKong suggests that under conditions of similarly segmented labor markets, the effect ofgendered networks on men's and women's migration differs from the Mexican case be-cause the barriersto internationalmigration for both men and women are lower, whilethe obstacles to men's employment are higher in the place of destination than they arefor women's employment. Female migrants from the Philippines to Hong Kong have awell-developed, destination-specific migration network that channels them into the do-mestic work sector of the Hong Kong economy. Channeledthroughthese female migrantnetworks, a growing number of Filipino male migrants to Hong Kong are clustered indomestic jobs (as child care providers, drivers, gardeners,and odd-job repairmen)(Con-stable 1997; Trager 1988)-jobs they would not normally have occupied either in theircommunities of origin or other destinations (like the United States or the Middle East).Comparativestudies of variation in the barriersto migration and in the size, structure,composition, and content of social networks should yield greater insights into how andwhy social networks influence migrationprocesses and result in particularmigrationpat-terns and outcomes.Our results not only deepen our understandingof how cumulative causation influ-ences Mexican migrationto the United States but also help to explain an increase in thepresence of women in this flow, especially after the increase in female migrationduringthe late 1980s and early 1990s. Women who migrated to the United States in a family-reunificationstrategyafterthe Immigrationand Reform ControlAct of 1986 will serve inthe near futureas importantsources of supportwithin migrantnetworks that are particu-larly importantfor otherwomen who want to migratefromMexico for economic reasons.We suspect that the rates of female internationalmigration are likely to grow faster than



18/20

Engendering MigrantNetworks 305are those of male internationalmigrationin the near future.The limitations of these data,the retrospective study design, and our logistic regression estimation did not allow us toestimate a migrationrate to test this speculation.Our study investigated young adult, unmarriedmigrants.Although this group is animportantpartof the migrantstreamout of Mexican communities, it is not representativeof Mexican migrants (either internalor international).The dynamics found here may beless relevant to older or marriedmigrants,especially those who migratefor family reuni-fication motives. For example, marriedfemale migrants may be more likely to rely ontheirhusbands'networks, eithertheirhusbands' own migrationexperiences andresourcesor those of their in-laws.

Furthermore,our study limited our measure of migrantnetworks to those within theimmediate family. We controlled for variation in community-level migrantnetworks, butdid not test for its effects. At this date, there are no publicly available data that wouldallow for prospective studies of the influence of extended networks of migration (eitherfamily based or friendshipbased) on individualmigration.However, researchin this areais an importantnext step. Ethnographicstudies have suggested that the weak or strong tieof migrantnetworks in places of origin and destination may differently influence men'sand women's migrationbehaviors.These limitations andourstudy suggest futureavenues of analysis anddatacollection.By demonstratingthe gendered effects of migrantnetworks, we have effectively linkedethnographicresearch, which has shown importantgender differences in the barrierstomigrationin places of origin and migrationexperiences in places of destination,with thequantitativeliteratureon migrantsthathas shown the importanceof social networks. Ourcomparison of internaland internationalmigration shows how gender-differentiatedmi-grantnetworks interactwith gender barriersto migrationto have a differential influenceon young men's and women's independent economic migration. Futureresearch, eitherqualitativeor quantitative,should continue to deepen ourunderstandingof this process bypursuing questions (that cannot be answeredwith these data) about how and why genderrelationsaffect migrationpatternsandhow genderinfluences the size, structure,composi-tion, and content of migrantnetworks.REFERENCESArias,P. 1995. "LaMigraci6nFeminina nDosModelosdeDesarrollo: 940-1970y 1980-1992"[Women'smigrationntwomodelsof development 940-1970and1980-1992].RelacionesDeGenero y Transformaciones Agrarias: Estudios Sobre el Campo Mexicano, edited by M.Gonzalez.MexicoCity:El ColegiodeMexico.Arizpe, L. 1975. Indigenas en la Ciudadde Mexico: El Caso de Las Marias [Indigenous in MexicoCity:Thecaseof "lasmarias"].MexicoCity:Secretaria e EducacionPiblica.. 1985. Campesinado Migracion Peasantry ndmigration].MexicoCity:Secretaria eEducaci6nPublica.Beneria, L. and M. Roldan. 1987. The Crossroads of Class and Gender: Industrial Homework,Subcontracting,and HouseholdDynamics in Mexico City.Chicago:University of Chicago Press.Cerruti,M. andD.S. Massey.2001. "OntheAuspicesof FemaleMigrationFromMexicoto theUnitedStates."Demography8:187-200.Choldin,H.M. 1973."KinshipNetworks n theMigrationProcess." nternationalMigrationRe-view 10:163-75.Constable,N. 1997. Maid to Orderin Hong Kong: Stories ofFilipina Workers. thaca,NY: CornellUniversityPress.Cornelius,W.1990."Impactsf the 1986U.S. ImmigrationawonEmigration romRuralMexi-can Sending Communities."Pp. 227-50 in UndocumentedMigration to the UnitedStates: IRCAandtheExperience f the1980s, editedby F.Bean,B. Edmonston,ndJ.S.Passel.Washing-ton,DC:TheUrban nstitute.



19/20

306 Demography, Volume 40-Number2, May2003Corona, R. 1993. "Migracion, Permanente Interestatal e Internacional, 1950-1990" [Permanentinter-statemigrationand internationalmigration]. ComercioExterior 43:750-62.Curran,S.R. and A.C. Saguy. 2001. "Migrationand CulturalChange:A Role for Gender and SocialNetworks." Journal of International Women i Studies 2(3):54-77.Delauney, D. 1995. "Mujeres Migrantes: Las Mexicanas de Estados Unidos" [Migrant women:Mexican women in the U.S.]. Estudios Demograficos y Urbanos 56:607-50.Donato, K. 1994. "CurrentTrends and Patterns of Female Migration: Evidence From Mexico."International Migration Review 27:748-68.Fernandez-Kelly, M.P. 1983. "Mexican Border Industrialization,Female LaborForce and Migra-tion." Pp. 205-23 in Women,Men and the International Division of Labor, edited by J. Nashand M.P. Fernmndez-Kelly.Albany, NY: State University of New YorkPress.Garcia, B., H. Mufioz, and 0. de Oliveira. 1979. Migracion, Familia y Fuerza de Trabajoen laCiudad de Mexico [Migration, family and labor force in Mexico City]. Mexico City: CentrodeEstudios Sociologicos, El Colegio de Mexico.Georges, E. 1990. TheMaking of a TransnationalCommunity:Migration, Development and Com-munityChange in the Dominican Republic.New York:ColumbiaUniversity Press.Grasmuck,S. and P. Pessar. 1991. Between Two Islands: Dominican InternationalMigration. Ber-

keley: University of California Press.Grieco, E.M. andM. Boyd. 1990. "WomenandMigration:IncorporatingGenderInto InternationalMigrationTheory."WorkingPaperSeries. College of Social Science, Florida State University.

Hagan, J.M. 1998. "Social Networks, Gender, and ImmigrantIncorporation:Resources and Con-straints."American Sociological Review 63:55-67.Hondagneu-Sotelo, P. 1994. Gendered Transitions:Mexican Experiences of Immigration. Berke-

ley: University of California Press.Hugo, G. 1991. "Village-CommunityTies, Village Norms, and Ethnic and Social Networks:A Re-view of the Evidence From the ThirdWorld."Pp. 186-225 in Migration Decision Making, ed-ited by G. DeJong and R.W. Gardner.New York:PergamonPress.Jasso, G. and M. Rosenzweig. 1990. The New Chosen People: Immigrants in the United States.New York:Russell Sage Foundation.Kanaiaupuni,S.M. 2000. "Reframingthe Migration Question: An Analysis of Men, Women, andGender in Mexico." Social Forces 78:1311-48.Kossoudji, S. and S. Ramsey. 1984. "The LaborMarketExperience of Female Migrants."Interna-tional Migration Review 18:1120-43.Lagarde, M. 1993. Los Cautiverios de las Mujeres:Madresposas, Monjas, Putas, Presas y Locas[Women's Captivities: Mothers-wives, Nuns, Whores, Prisoners, and Crazy]. Mexico City:Universidad Nacional Aut6noma de Mexico.Lindstrom,D.P. 1997. "The Impactof TemporaryU.S. Migration on Fertility of Female Migrants:The Case of TemporaryMigrationin a RuralMexican Township."WorkingPaperSeries. Popu-lation Studies and TrainingCenter,Brown University, Providence, RI.Lindstrom,D.P. and N. Lauster.2001. "Local Economic Opportunityand the Competing Risks ofU.S. and Internal Migration in Zacatecas, Mexico." International Migration Review 35:1232-54.Lomnitz, L.A. 1977. Networks and Marginality: Life in a Mexican Shantytown.New York:Aca-demic Press.Mahler, S. 1999. "EngenderingTransnationalMigration:A Case Study of El Salvadorans."Ameri-can Behavioral Scientist 42:690-719.Massey, D.S. 1990. "Social Structure,Household Strategies, and the CumulativeCausationof Mi-

gration."Population Index 56(1):3-26.Massey, D.S., R. Alarcon, J. Durand,and H. Gonzalez. 1987. Returnto Azatldn: The Social Pro-cess of International Migration From WesternMexico. Berkeley: University of California.Massey, D.S. and K.E. Espinoza. 1997. "What's Driving Mexico-U.S. Migration?A Theoretical,Empirical, and Policy Analysis." AmericanJournal of Sociology 102(4):939-99.



20/20

Engendering MigrantNetworks 307Massey, D.S. and F. GarciaEspafia. 1987. "The Social Process of Migration."Science 237:733-38.Massey, D.S., L. Goldring, and J. Durand. 1994. "Continuities in TransnationalMigration: AnAnalysis of Nineteen Mexican Communities."AmericanJournal of Sociology 99:1492-533.McCullagh, P. and J.A. Nelder. 1989. Generalized Linear Models. London:Chapmanand Hall.Menjivar,C. 2000. Fragmented Ties: Salvadoran ImmigrantNetworks in America. Berkeley: Uni-versity of CaliforniaPress.Mufioz, H., O. de Oliveira, and C. Ster. 1977. Migracion y Desigualdad Social En La CiudadDeMexico [Migrationand Social Inequality in Mexico City]. Mexico City: Universidad NacionalAutonoma de Mexico.PartidaBush, V. 1993. "Niveles y Tendencias de la Migraci6n Internaen Mexico a Partirde lasCifras Censales, 1970-1990" [Levels and trends of internalmigration in Mexico based on cen-sus data, 1970-1990]. Revista Mexicana de Sociologia 1:1955-73.Pedraza,S. 1991. "WomenandMigration:The Social Consequences of Gender."Annual Review ofSociology 17:303-25.Pessar, P. 1999a. "EngenderingMigration Studies: The Case of New Immigrants in the UnitedStates."American Behavioral Scientist 42:577-600.. 1999b. "The Role of Gender,Households, and Social Networks in the MigrationProcess:A Review and Appraisal."Pp. 53-70 in Handbook of International Migration: TheAmericanExperience, edited by C. Hirschman,P.Kasinitz, and J. DeWind. New York:Russell Sage Foun-dation.Piore, M. 1979. Birds of Passage: Migrant Labor in Industrial Societies. Cambridge, England:CambridgeUniversity Press.Portes, A. 1978. "Migrationand Underdevelopment."Politics and Society 8(1):1-48.Stark,0. 1991. TheMigration of Labor. Cambridge,MA: Basil Blackwell.Stark, 0. and M. Levhari. 1982. "OnMigration and Risk in LDCs." Economic Development andCultural Change 31:191-96.Stark,0. and J.E. Taylor. 1991. "Relative Deprivation and InternationalMigration."American So-ciological Review 51:273-86.Szasz, I. 1999. "La Perspectiva de Genero en el Estudio de la Migraci6n Femenina en Mexico"[Gender perspective in the studies of female migration in Mexico]. Pp. 167-205 in Mujer,Genero y Poblacion en Mexico, edited by B. Garcia. Mexico City: El Colegio de Mexico ySociedad Mexicana de Demografia.Taylor,J.E. 1986. "Differential Migration, Networks, Information and Risk." Research in HumanCapital and Development: Migration, Human Capital and Development 4:147-71.Trager,L. 1988. The City Connection: Migration and Family Interdependencein the Philippines.Ann Arbor:University of Michigan Press.Zlotnik, H. 1993. "Womenas MigrantsandWorkers n Developing Countries."InternationalJour-nal of ContemporarySociology 30(1):39-62.

Engendering Migrant Networks 307

Documents

Curran and Rivero Fuentes UNIDAD V