ETHNICITY AND PSYCHOLOGICAL DISTRESS AMONG LATINO ADULTS: SOCIOECONOMIC STATUS, FAMILISM, AND GENERATIONAL STATUS

Transcription

1 ETHNICITY AND PSYCHOLOGICAL DISTRESS AMONG LATINO ADULTS: SOCIOECONOMIC STATUS, FAMILISM, AND GENERATIONAL STATUS A Dissertation submitted to Kent State University in partial fulfillment of the requirements for the degree of Doctor of Philosophy by Yanmei Xu August 2010

2 Dissertation written by Yanmei Xu B.A. Sichuan Teachers College, 1994 M.A. Kent State University, 2002 Ph.D. Kent State University, 2010 Approved by, Chair, Doctoral Dissertation Committee Dr. Susan Roxburgh, Members, Doctoral Dissertation Committee Dr. C. André Christie-Mizell Dr. Mark Tausig Dr. T. John Akamatsu Accepted by, Chair, Department of Sociology Dr. Richard T. Serpe, Dean, College of Arts and Sciences John R.D. Stalvey ii

3 TABLE OF CONTENTS LIST OF FIGURES...vii LIST OF TABLES...viii AKNOWLEDGEMENTS...x I. INTRODUCTION...1 Ethnicity and Mental Health...2 Epidemiological Catchment Area Study (ECA, )...3 Hispanic Health and Nutrition Examination Survey (HHANES, )...3 National Co-morbidity Study (NCS, )...4 National Latino and Asian American Study (NLAAS, )...4 Gaps in Studies on Latinos' Mental Health...6 Organization of the Dissertation...10 II. ETHNIC DIVERSITIES AMONG LATINOS IN THE US...12 Cubans...14 Mexicans...16 Puerto Ricans...17 III. SOCIAL LOCATIONS, CULTURE, AND MENTNAL HEALTH...21 Socioeconomic Status and Mental Health...22 Threshold Effect...23 iii

4 Measures of SES...24 Familism and Mental Health...28 Generational Status and Mental Health...33 Theoretical Model and Research Questions...36 IV. DATA AND METHODS...40 Data...40 Measures...41 Dependent Variable...41 Independent Variables...42 Control Variables...46 Data Analysis...47 V. SES AND PSYCHOLOGICAL DISTRESS...50 Descriptive Results...50 Ethnic Variations in SES and Familism...50 Generational Distribution of SES, Familism, and Psychological Distress...55 Multivariate Results...58 The Association between SES and Psychological Distress...58 Differential Impacts of SES across Three Ethnic Subgroups...62 Threshold Effects of SES Measures...63 Summary...66 iv

5 VI. FAMILISM, GENERATIONAL STATUS, AND PSYCHOLOGICAL DISTRESS...70 Familism and Psychological Distress...70 Direct Impacts of Familism on Psychological Distress...70 Comparison of the Magnitude of the Impact of Familism...74 SES, Familism, and Psychological Distress...74 Generational Status, SES, Familism, and Psychological Distress...79 The Association of Generational Status and Psychological Distress...79 Moderating Effects of Generational Status...80 Summary...80 VII. DISCUSSION...83 Ethnic and Generational Variations...83 Socioeconomic Status...84 Familism...84 Psychological Distress...86 SES and Psychological Distress...86 Education...87 Household Income...88 Wealth...89 Familism and Psychological Distress...93 Marriage and Household Size...93 Family Cohesion...95 v

6 Family Support...96 SES, Familism, and Psychological Distress...97 Generational Status and Psychological Distress...98 Limitations Conclusion and Future Research REFERENCES APPENDIX vi

7 LIST OF FIGURES Figure 1. Theoretical Model...36 Figure 2. Predicted Psychological Distress As A Function of Education and Family Support for Cubans...76 Figure 3. Predicted Psychological Distress As A Function of Household Income and Family Support for Cubans...77 vii

8 LIST OF TABLES Table 1. Demographic and Socioeconomic Characteristics of Latinos...13 Table 2. Ethnic Differences in Control Variables (Weighted)...51 Table 3. Ethnic Differences in SES and Familism (Weighted)...52 Table 4. Ethnic Differences in Psychological Distress (Weighted)...54 Table 5. Generational Differences in Independent and Dependent Variables (Weighted)...56 Table 6. Psychological Distress Regressed on SES for Cubans...59 Table 7. Psychological Distress Regressed on SES for Mexicans...60 Table 8. Psychological Distress Regressed on SES for Puerto Ricans...61 Table 9. Psychological Distress Regressed on Education and Income Dummy Variables...64 Table 10. Summaries of Findings on Psychological Distress Regressed on SES...67 Table 11. Psychological Distress Regressed on Familism Measures for Cubans...71 Table 12. Psychological Distress Regressed on Familism Measures for Mexicans...72 Table 13. Psychological Distress Regressed on Familism Measures for Puerto Ricans...73 Table 14. Psychological Distress Regressed on SES, Family Cohesion, and Family Support for Cubans...75 Table 15. Psychological Distress Regressed on Generational Status...79 Table 16. Summaries on Psychological Distress Regressed on Familism...80 viii

9 Table 17. Inter-correlations for Cubans Table 18. Inter-correlations for Mexicans Table 19. Inter-correlations for Puerto Ricans ix

10 AKNOWLEDGMENTS My deep gratitude goes to Steve and Sharon Bortner, our daughter's guardians. Without their help, I would not have been able to finish my dissertation. They were always there to take care of Hannah when I was busy, when I was sick, and when she missed them. In spite of their tight schedules in this past summer, they were constantly willing to make adjustments to watch her so that I could meet deadlines. I lost count of how many nights Hannah slept over at their house and how many times she traveled with them. They have given her a second home in the US. Words cannot fully express my gratefulness to them. God blesses me with their friendship and love. I would like to thank Dr. Susan Roxburgh for being willing to direct my dissertation and guiding me throughout my graduate study. She is a mentor as well as a friend. She has shown a full understanding of my situation and supported my career decisions. She gave me excellent guidance on the dissertation. She was willing to meet to talk about my work over the weekend to fit in my schedule. She literally coordinated my defense schedule among committee members for me. She was joking that she was a control freak, but I know I was lucky to have her as my committee chair. I look forward to collaborating with her again in the future. I would like to thank Drs. André Christie-Mizell, Mark Tausig, and T. John Akamatsu, and Carol Maier for being always available for consultation and giving me instructive comments on my dissertation. Their suggestions made my dissertation better. x

11 Many friends offered their help throughout my writing. My big thanks go to Dr. Lu Zou, Mingming Lu, Dan Liao, Liwei Li, Dr. Jianru Shi, and Ying Yue. Their help and support made it feasible for me to finish my dissertation on time. And their love for our children is a gift to us. I want to thank my colleagues at Saginaw Valley State University for their full support when I was not able to fulfill my teaching responsibilities due to illness. I am looking forward to going back and having a fresh start. I am grateful to my parents and my mother-in-law for their encouragement and their help with child caring. They show me how to be a good parent with their love, patience, and understanding. A conversation with my parents often throws light into a seemingly dead-end road in our life. Junfeng, my husband, deserves my respect and appreciation for his love, patience, and support. He has lived through every moment of this dissertation. During my threemonth bed rest, he took care of both my daughter and me on top of his full-time job. His perseverance and flexibility in our difficult times always put me right back on track when frustration tended to overcome hope. Lastly, I would like to dedicate my dissertation to my two lovely children, Hannah and Henry. They made many adjustments so that mommy could finish this piece of work. They motivate me to keep going and enjoy life. They ARE the joy of my life. xi

12 CHAPTER ONE INTRODUCTION Latinos are changing the demographic make-up of the United States. According to the US Census, Hispanics represent 14.8% of the total US population, replacing African Americans as the largest racial/ethnic minority group (US Census Bureau 2006). Hispanics are projected to constitute 15.5% of the total US population by 2010 and up to 25%, in 2050 (US Census Bureau 2006). This fast growing population suggests the importance of examining the health of Hispanics, because race and ethnicity shape the pattern of socioeconomic achievement, family composition, and health status in the US. The Hispanic population is a very heterogeneous group composed of subgroups with different immigration histories and settlement patterns. Therefore, it is also important to examine the ways that ethnicity-related factors produce subgroup patterns of health outcomes. The main purpose of this dissertation is to examine the mental health impact of two such factors: socioeconomic status and familism. I will also investigate whether generational status moderates the relationship between SES and psychological distress and between familism and psychological distress within each ethnic group. The US Census Bureau defines Hispanic as a person of Mexican, Puerto Rican, Cuban, Central or South American or other Spanish culture or origin, regardless of race (US Census Bureau 2010). This includes all Spanish-speaking persons from all regions 1

13 2 of Latin America and from the Iberian Peninsula. In this study, however, I limit the use of Hispanic to refer to people from Caribbean and Latin American countries. Therefore, Hispanics is interchangeable with Latino in my dissertation. Ethnicity and Mental Health Mental health is the absence of psychiatric illness such as depression, anxiety, and alcohol abuse, or acute psychological distress (Aneshensel & Phelan 1999). As defined by the World Health Organization, the definition of mental health also includes life satisfaction and ability to realize one s own potential (WHO 2007). Sociological findings indicate that the development and maintenance of healthy mental states are influenced by such social locations as socioeconomic status and race/ethnicity (Brown et al 1999, George 1999, Rosenfiel 1999, Yu & Williams 1999). A high socioeconomic status usually measured as education and income is associated with a low risk of psychiatric disorders or psychological distress (Ettner & Grzywacz 2003, Gavin et al 2010, Williams & Collins 1995, Yu & Williams 1999). Racial and ethnic minorities in the US are overrepresented in the impoverished population. They face economic hardship and obstacles to health services. Given their circumstances, it would be expected that they have negative mental health outcomes. However, findings on the mental health of racial and ethnic minorities are far from consistent. Findings from the National Comorbidity Survey and the National Survey of American Life suggest that African Americans have lower rates of mental illness than do non-hispanic whites, despite their low social standings and poor physical health status (Sue & Chu 2003). Findings on Latinos' mental

14 3 health are also mixed. In the following section, I review findings regarding Latino s mental health from four major health surveys in the US. Epidemiological Catchment Area Study (ECA, ) The purpose of the ECA study was to collect data on the prevalence and incidence of mental illness and on the use of mental health services. Data were collected at five Community Mental Health Center catchment areas: New Haven, Connecticut, Baltimore, Maryland, St. Louis, Missouri, Durham, North Carolina, and Los Angeles, California. The ECA sampled Latinos the majority of whom were Mexicans in Los Angeles, CA. Diagnoses were based on the Diagnostic and Statistical Manual of Mental Disorder, third version (DSM-III). Findings indicate that compared with whites, Latinos had lower prevalence of drug abuse and major depressive episodes (see reviews by Agbayani- Siewert et al 1999, Sue & Chu 2003, Williams & Harris-Reid 1999). For other psychiatric disorders, Latinos and non-hispanic whites had similar rates. Findings also suggest that Mexicans born in Mexico had lower rates of depression than their counterparts born in the US (Burnam et al 1987). Hispanic Health and Nutrition Examination Survey (HHANES, ) The HHANES aimed to collect data on nutritional practices and health of Latinos in the US. The HHANES sampled Puerto Ricans in New York City, New Jersey, and Connecticut, Cubans from Dade County, Florida, and Mexicans from Texas, Colorado, New Mexico, Arizona, and California. The age distribution of the samples was from six months to 74 years. Measures of depression in this study provided data on feelings of

15 4 depression, how depression affected everyday life, and help sought during depression. DIS items and Epidemiological Studies Depression Scale (CES-D) items were included to assess mental health of samples of age The HHANES data revealed that Puerto Ricans had higher scores on psychological distress and depression than Cubans and Mexicans (see reviews by Vega & Rumbaut 1991, Williams & Harris-Reid 1999). Mexicans were found to have lower scores on the CES-D scale than non-hispanic whites in previous studies (Moscicki et al 1989). National Co-morbidity Study (NCS, ) The NCS was an epidemiologic investigation designed to study the prevalence and correlates of DSM III-R disorders and patterns and correlates of service utilization. Since the survey was conducted in English, more acculturated Latinos were sampled in the NCS. Therefore, related results may not be applicable to less acculturated Latino immigrants. Latinos in the NCS had the same or higher prevalence of mental disorders than non-hispanic Whites (Sue & Chu 2003). Mexicans alone were found to have lower rates of any lifetime disorders than non-hispanic whites (Ortega et al 2000). Similar to the ECA findings, US-born Mexicans had higher rates than Mexico-born Mexicans (Williams & Harris-Reid 1999). National Latino and Asian American Study (NLAAS, ) The National Latino and Asian American Study is a nationally representative community household survey that estimated the prevalence of mental disorders and rates of mental health service utilization by Latinos and Asian Americans in the United States.

16 5 The NLAAS is the first study that has nationally represented samples of Asian and Latino Americans. The data I use for this dissertation come from the NLAAS. Mental disorders were measured using the DSM-IV and Kessler s K-10 scale. The K-10 scale was designed by Ronald C. Kessler and his colleagues (2002) to measure non-specific distress in the past 30 days. Findings of recent descriptive studies have indicated that Puerto Ricans have the highest overall prevalence rate of psychiatric disorders among the Latino ethnic groups (Alegria et al 2007a, Alegria et al 2007c). The NLAAS data also replicate previous findings on the difference between Latinos who were born in the US and those who were born in home countries, but there is no difference between immigrant and US born Puerto Ricans (Alegria et al 2006). Alegaria and colleagues (2008) made a Latino- White comparison by combining the NCS-R and NLAAS data. They found lower prevalence rates of most psychiatric disorders among aggregate Latinos than among Whites. Three conclusions can be drawn from these findings. First, findings are inconsistent on the white-latino comparison. Findings from the NLAAS indicate that Latinos have lower prevalence rates of psychiatric disorders than non-hispanic Whites whereas the NCS found the opposite. The ECA found that Latinos and Whites have similar rates. One factor that contributes to the inconsistence is that all the abovementioned studies except the NLAAS used regional samples. This limitation constrains the generalizability of findings. To address this issue, I will use the data from the NLAAS for my analyses. Second, findings from the four studies suggest Latino subgroup variations in the prevalence of psychiatric disorders. Although the findings on

17 6 Latino-White comparison are inconsistent, Mexicans were consistently found to have a lower prevalence rate than non-hispanic whites. Moreover, within-group comparisons indicate that Puerto Ricans fare worse than Cubans and Mexicans in terms of psychological well-being. Third, acculturation is positively associated with the risk of psychiatric disorders. Native-born Latinos tended to have lower rates than US-born Latinos, but this finding did not generalize to Puerto Ricans. Given the current findings on subgroup variations in the prevalence of psychiatric disorders, it is important to disaggregate Latinos and examine mental health in each subgroup. The identification of group differences in the association between mental health and its ethnicity-related correlates can offer us a better understanding of ethnic patterns of psychological disorders. Studies in this direction are still lacking. Gaps in Studies on Latinos' Mental Health Although more recent attention has been given to studies of Latino health as the Latino population increases at an unprecedented rate, gaps exist in the current literature. First, many studies mask important heterogeneity among Latinos by focusing on one ethnic group, Mexican Americans (e.g., Burnam et al 1987, Earle 1998, Escobar et al 2000, Finch et al 2000, Franzini & Fernandez-Esquer 2004) or regarding Latinos as a monolithic group (e.g., Finch & Vega 2003, Franzini et al 2002, Gavin et al 2010, Guarnaccia et al 1991, Kim & McKenry 1998). Only a few recent studies examine variations across subgroups (Alegria et al 2006, Alegria et al 2007a, Alegria et al 2007b, Alegria et al 2007c, Rubio et al 2004). These studies find that Puerto Ricans have the highest overall prevalence rate among the Latino ethnic groups. Generational status is a

18 7 significant correlate of psychiatric disorders for Mexicans but not for Puerto Ricans. These findings on subgroup differences suggest the importance of disaggregating the generic category Latino into specific national-origin groups. Latino ethnic subgroups differ markedly on a variety of dimensions including their settlement patterns and socioeconomic circumstances. Latinos come from different countries and regions including, Cuba, Mexico, Puerto Rico, and other Central and South American countries. They arrived in this country with various cultural backgrounds; they are in a wide range of occupations and earn different incomes; they are assimilated in different ways into the host society and attached to their traditional values to varying degrees. The heterogeneity of this population warrants a consideration of ethnic variations in terms of health and its correlates. Knowledge generated from studies in this direction can help us better interpret ethnic group patterns of mental health. The second reason why it is important to investigate group patterns of health correlates such as SES among Latinos is that although the strong positive relationship between SES and mental and physical health has been confirmed among non-hispanic Whites and African Americans (see reviews by Williams & Collins 1995, Williams et al 1992, Yu & Williams 1999), there is so far no consensus on such an association among Latinos. Some researchers find that SES has no effect on mental health among Mexicans (Breslau et al 2006, Mirowsky & Ross 1980). Others find an inverse relationship between socioeconomic status and psychological distress for Latinos in aggregate (Cuellar & Roberts 1997, Franzini & Fernandez-Esquer 2004, Guarnaccia et al 1991). Findings regarding one group (i.e., Mexican Americans) may not apply to other ethnic

19 8 subgroups. Similarly, findings from research, in which Latinos are lumped together, do not tell us anything about between-group variations. Given the fact that socioeconomic attainment varies significantly among Latinos subgroups, it is crucial to examine whether the positive association of SES and mental health hold for all Latino subgroups. Research on the group variation in the association between SES and mental health becomes especially important in light of the "Hispanic Health Paradox". The Hispanic Health Paradox refers to the fact that Latinos enjoy relatively good physical health in spite of their low socioeconomic status in the US (Franzini et al 2002, Markides & Coreil 1986, Morales et al 2002, Scribner 1996). On average, Latinos have lower educational attainment and income than the general population (US Census Bureau 2006). They also face serious financial and structural barriers to health services. On the basis of increasing evidence suggesting that individuals with high SES have better health status, one would expect Latinos to have relatively poor health outcomes. However, the health of Latinos is favorable relative to other racial/ethnic groups, especially on the indicators of life expectancy, adult mortality, and infant mortality (Morales et al 2002). The Hispanic Health Paradox suggests that the confirmed association between SES and health may not hold among Latinos. In addition, one important finding related to the Hispanic Health Paradox is that Mexicans who have the lowest SES among Latinos have the lowest ageadjusted mortality rate, whereas Puerto Ricans have a much greater mortality rate, followed by Cubans who are financially more secure than other Latinos (Morales et al 2002, Rosenwaike 1987). This finding further suggests that Latino subgroups may be different in terms of the relationship between SES and health. Despite these findings

20 9 regarding Latino physical health, limited studies have been done on the group variation in the association between SES and mental health among Latinos. Researchers have proposed explanations for the weak association between SES and health among Latinos. The ethnic-culture explanation refers to protection of traditional values against the negative impact of socioeconomic disadvantages (Franzini & Fernandez-Esquer 2004, Franzini et al 2002, Mirowsky & Ross 1980). Familism is an intrinsic value of Latino culture, which has been frequently invoked to interpret the finding of the Hispanic Health Paradox (Franzini et al 2002, Guarnaccia & Rodriguez 1996, Mirowsky & Ross 1980, Mirowsky & Ross 1984, Rubio et al 2004). Familism can be beneficial to psychological well-being by offering Latinos a sense of security and promoting self-esteem (Mirowsky & Ross 1980). Cultural resources embedded in familism could also provide a cushion for absorbing economic hardship and consequently moderate the negative impact of SES on mental health. People who value family cohesion may be less likely to worry about low SES than those who value socioeconomic achievement. For example, immigrants may feel good about sending money home and increasing the standard of living of the family members back home although they are poor by the standards of the host country (Suarez-Orozco & Suarez-Orozco 1995). Being able to contribute to their families, financially in this case, boosts their family pride and their level of satisfaction. Consequently, their mental health may be affected less by their current socioeconomic circumstances. The negative impact of low SES may thus be lessened for Latino immigrants who have a strong sense of familism.

21 10 A final important issue about within-group differences in mental health concerns the fact that the importance of SES and familism may change over generations. According to the acculturation hypothesis, more acculturated immigrants may lose protections of traditional values and thus endure negative health outcomes compared to less acculturated individuals (Franzini et al 2002). However, evidence is mixed regarding the association of acculturation and mental health. Some studies find positive associations and others report negative relationships (Lara et al 2005, Rogler et al 1991). Generational status is an important indicator of acculturation. Later generations are more likely than the first generation to be English speakers, identify with American mainstream values and compare themselves to American Whites in terms of socioeconomic status (Guarnaccia et al 2007, Loue 1998, Molina & Aguirre-Molina 1994, Zhou 1997). Based on the acculturation hypothesis, generational status could moderate the mental health impact of SES and familism because the importance of SES increases and the significance of familism decreases over the generations. Latergeneration immigrants may be at greater risk for internalizing negative messages associated with discrimination and low SES and at the same time lose protection of traditional vultures. Empirical studies on the moderating effect of generational status are limited. Organization of the Dissertation In Chapter Two, I review the information on immigration histories and current socioeconomic circumstances of three major Latino groups in the US: Cubans, Mexicans, and Puerto Ricans. Chapter Three introduces a theoretical framework that is useful for

22 11 examining Latino s health. I review the literature on the links between SES and mental health. I also review the literature on familism, generational status and their relationships to mental health. Chapter Four describes the data I use, measures, and the analytical methods. Chapter Five describes the results that addresses the group distribution of all measures and examines relationships between three SES indicators and psychological distress across three Latino subgroups. I compare the magnitude of health effects of SES between subgroups. I also analyze the threshold effect of SES for each subgroup. In Chapter Six, I examine direct and moderating effects of familism and generational status on mental health across three groups. Chapter Seven concludes the dissertation by summarizing findings and discussing major limitations.

23 CHAPTER TWO ETHNIC DIVERSITIES AMONG LATINOS IN THE US Latinos immigration to the US can be traced back to the mid-19 th century and it reached its peak in the second half of the 20 th century. Over 40 million Latinos live in the US and the population is growing by more than 1.5 million annually (Tienda & Mitchell 2006a). Mexicans constitute the largest subgroup (64% of the whole population), followed by Puerto Ricans (9.6%) and Cubans (3.5%) (US Census Bureau 2004a). From over 20 countries in Latin and South America and the Caribbean, Latinos share a language and can trace their cultural origins to both Europe and their native land. While they are likely to reside in certain states of the US, including California, Texas, Florida, and New York, increasingly Latinos are dispersed across the country. In Table 1, I present recent US census data on socio-demographics for Latinos, subgroups, and the general population. These data show that Latinos as a whole do share certain characteristics which differentiate them from the general population. 12

24 13 Table 1. Demographic and Socioeconomic Characteristics of Latinos Total Population Latinos Cuban Mexican Puerto Rican Median Age Married 50.2% 47.4% 50.8% 49.4% 37.7% Female headed household Living with grandchild(ren) Language other than English Less than high school diploma (age 25+) Bachelor s degree (age 25+) Employed in civilian labor force Median family income 12.5% 18.5% 13.0% 16.9% 25.4% 3.5% 7.0% 4.9% 7.7% 5.8% 19.7% 78% 84.4% 78.7% 68.0% 15.5% 39.4% 24.6% 45.8% 27.4% 17.4% 8.7% 16.0% 6.2% 10.7% 60.3% 52.3% 57.3% 63.3% 51.4% $61,173 $42,074 $52,113 $40,419 $42,300 Families in poverty 9.5% 18.5% 10.2% 20% 21.1% House ownership 67.2% 49.9% 61.0% 51.2% 40.3% (Source: 2007 American Community Survey 1-Year Estimates, US Census Bureau) As shown in Table 1, Latinos in the US are much younger, are less likely to be married and more likely to live in a female-headed household. Compared with the general population, they are more likely to live with their grandchildren in an extended household and speak a language other than English at home. Latinos' socioeconomic status is much lower than that of the total population. Relative to the total population, Latinos are less likely to graduate from high school, receive a bachelor's degree, and be

25 14 employed in the civilian labor force. They have a much lower median family income and a higher poverty rate. They are also less likely to own a house. Table 1 also shows subgroup variations in terms of socio-demographics. Cubans are older than Mexicans and Puerto Ricans. Mexicans have lower education and income but are more likely to live with grandchildren than Cubans and Puerto Ricans. The rate of female-headed households is much higher among Puerto Ricans than Cubans and Mexicans. These data highlight the fact that the Latino population is diversified. Since this paper focuses on Cubans, Mexicans, and Puerto Ricans, I describe their immigration histories and compare their socioeconomic statuses in the following sections. This information will be helpful for our understanding of group variations in health impact of SES and familism. Cubans According to the US Census 2006 American Community Survey, about 1.5 million Cubans are currently living in the US (US Census Bureau 2007). Cubans migrated to the United States to seek refuge in large numbers after Fidel Castro came to power in The first wave of Cuban immigrants, who came to the US between 1959 and 1962, were mostly well-educated professionals and businessmen. Many of them settled in Miami, Florida. The second wave of Cubans came via airlifts between 1967 and This wave was primarily composed of family members of those who had earlier sought exile in the US. Since men aged between 14 and 27 were prohibited from leaving Cuba, this cohort was mainly composed of the elderly and women (Perez 2004). The age make-up of these cohorts accounts for the fact that Cubans tend to be older than

26 15 other Latino groups. Immigration was temporarily suspended between 1974 and After 1980, the door to the US was open again. Because of the US government s very positive attitude toward Cuban immigrants, Cuban refugees received unprecedented financial aid from the US federal government in terms of relocation, business financial aid, health care, job training and other welfare subsidies in the 1960s and 1970s (Grenier & Perez 2003). Partially as a result of this assistance, Cubans have turned Miami into an international trade center. Their major role in transforming Miami economy is a testimony to their economic achievement in the US. However, their current socioeconomic status needs to be examined from two perspectives. Based on the census data in Table 1, compared with the total population, Cubans are less educated, and have lower family income and employment rates, and a higher poverty rate, and are less likely to own a house. In spite of these disadvantages, they fare much better on socioeconomic indicators than Mexicans, Puerto Ricans, and Latinos as a whole. Cuban ethnic enclaves in the South, particularly in Florida and the threegeneration living arrangement help to maintain native language and cultural practice (Molina & Aguirre-Molina 1994, Stepick & Stepick 2002). Recent immigrants are likely to settle in successful and self-sustained enclaves established by their predecessors because of job availability and easy transition in an ethnic community (Stepick & Stepick 2002). Compared to other Latinos, Cubans have a higher rate (over 84%) of individuals who speak a language other than English at home (US Census Bureau 2004a, US Census Bureau 2007). Cubans' strong family commitment is reflected in their family work ethic.

27 16 Many Cubans own businesses, in which three generations in a household work together (Grenier & Perez 2003). This economic cooperation within the family has contributed to Cubans' economic success. Extended households help to pass on traditional values. Later generations of Cubans are able to speak and understand Spanish and have close extended-family ties in the process of being assimilated into mainstream America (Stepick & Stepick 2002). Mexicans The immigration of Mexicans started in the 19 th century when the United States acquired a large section of Mexico in 1848 (Bean & Tienda 1987). A large-scale immigration took place after 1986 and reached its peak in the 1990s (Zinn & Pok 2004). Mexican immigrants came to seek economic relief in the US. They are concentrated in the West and South, particularly in Arizona, Texas, Colorado, New Mexico, and California. Unlike previous Mexican immigrants who came to work in farms, most new arrivals are urban workers in metropolitan areas. Mexicans' immigration is characterized by a chain movement. This refers to a pattern whereby one family is usually followed by their extended-families relatives. Other extended families from the same areas then follow their friends or relatives to the US. Chain migration results in the creation of immigrant neighborhoods where many relatives and friends live in close proximity to one another. They share information on housing and job opportunities and help more recently arrived immigrants to settle down (Garcia 2002). Mexicans are the youngest group among Latinos because of the young age of immigrants and high fertility rate. Mexicans socioeconomic status lies at the low end of

28 17 the distribution among Latinos. Despite their high level of participation in the labor force, Mexicans receive less family income than Cubans and Puerto Ricans partly because Mexicans are concentrated in low-wage occupations in the service and construction industries (US Census Bureau 2007). About 46% of Mexicans aged 25 or above do not have a high school diploma and only 6% have earned a bachelor s degree. However, they are more likely to own a home than Puerto Ricans. The impoverishment experienced by Mexicans has not weakened their commitment to core cultural values (Tienda & Mitchell 2006b). Mexican immigrant families are more likely to include extended family members, usually grandparents. Families are the primary source of help and support. For example, a Mexican family is willing to raise a niece or a nephew until the rest of the family settles in the US (Garcia 2002). Mexican families also serve as repositories of Mexican cultural heritage. Over 78 percent of Mexicans speak a language other than English at home (US Census Bureau 2004a, US Census Bureau 2007). Thanks to extended family networks, second generations are able to frequently visit relatives and exchange goods and services among family members (Velez-Ibanez 1996). Puerto Ricans Puerto Ricans started immigrating to the United States in the early 20 th century (Malina & Aguirre-Molina 1994). The early immigrants predominantly settled in New York City and were hired as seasonal contract laborers. The great majority of Puerto Ricans came after the island achieved Commonwealth status in 1952 (Perez y Gonzalez 2000). The years from 1946 to 1964 witnessed an unprecedented number of Puerto

29 18 Ricans immigrating to the US. By 1973, 40% of Puerto Rico s population was living in the United States (Perez y Gonzalez 2000). They migrated to work in the manufacturing and garment industries in the US (Perez y Gonzalez 2000). With the decline of these industries in the 1980s, many Puerto Ricans became unemployed and lived in poverty (Molina & Aguirre-Molina 1994). Unlike other Latinos, Puerto Ricans are born with US citizenship, which allows them to move freely back and forth between the island and the continent. They may feel less family dislocation than other groups of immigrants (Alegria et al 2007c). They are likely to settle in the states of New York, New Jersey, Connecticut, and Massachusetts (US Census Bureau 2004a). Puerto Ricans levels of education and income fall between those of Cubans and Mexicans. However, they are less likely to be employed than Cubans and Mexicans (US Census Bureau 2004b). This might be due to a sharp decline in jobs in manufacturing industries in the 1980s, which employed 60 percent of Puerto Ricans (Malina & Aguirre- Molina 1994). They are also less likely to own a home than the other two groups. Today, Puerto Ricans have the highest poverty rate among Latinos, partly due to an increasing number of female-headed households (US Census Bureau 2004a). Puerto Ricans are highly acculturated and urbanized. Puerto Ricans on the island have a strong tradition in extended families and commitment to families (Carrasquillo 2004). However, their families in the US are changing and becoming unstable, though extended family structures continue to play an important role in providing informal support (Carrasquillo 2004, Vega 1990a). Their marriage rate (37.7%) is much lower than the general population and the other two ethnic groups. The rise in divorce and non-

30 19 martial childbearing among Puerto Ricans signals what some scholars term a family decline (Gonzalez 2000). As indicated in Table 1, over a quarter of Puerto Rican families are headed by single women. The rate is the highest among Latinos and double the rate of the general population. Puerto Ricans suffer economic hardships and discrimination experienced by African Americans because a significant number of Puerto Ricans are phenotypically black (Malina & Aguirre-Molina 1994, Suarez-Orozco & Suarez-Orozco 1995). The above description indicates that the Latino population not only shares certain characteristics but also differs significantly in terms of demographic and socioeconomic characteristics. These differences reinforce the point that Latinos are not one homogeneous group. Moreover, these differences could affect group patterns of the health effect of SES and familism. Many Cubans were businessmen and professionals before they came to the US and are much more economically successful than Mexicans and Puerto Ricans. Thanks to settlement in enclaves, they are able to maintain many traditional values including familism. Their economic success combined with a strong commitment to family make it possible that both high SES and familism could yield health benefits for Cubans. Mexicans, though socio-economically disadvantaged, retain the value of familism (Garcia 2002, Zinn & Pok 2004). With a large number of recent immigrants, Mexicans may feel a sense of security with their well-maintained cultural values (Suarez-Orozco & Suarez-Orozco 1995). Their low SES may have relatively low impact on their health. We might expect a strong impact of SES on the mental health of Puerto Ricans who are likely to experience discrimination and economic hardship but

31 20 may receive less protection from traditional value of familism because of a family decline. A systematic study of the health impact of ethnicity-related factors including SES and familism will contribute to a deeper understanding of the etiology of psychological well-being among Latino ethnic groups.

32 CHAPTER THREE SOCIAL LOCATIONS, CULTURE, AND MENTAL HEALTH According to the fundamental cause model (Link & Phelan 1995), social conditions are fundamental causes of disease because they influence multiple health outcomes in spite of changing disease-related risk factors. Social conditions refer to factors that involve a person s relationship to other people and range from life events to SES to social network (Link & Phelan 1995:81). Social conditions define how power, property, and prestige influence the quality and quantity of resources an individual can deploy to promote health. Phelan and her colleagues (2004) emphasize that socioeconomic status is a fundamental cause of diseases. SES influences multiple health outcomes such as mortality and depression (Phelan et al 2004, Williams & Collins 1995). The importance of SES in health studies makes it necessary to examine the association between SES and psychological distress among Latinos, whose mental health is understudied. Moreover, race/ethnicity and SES are highly correlated in the US. Non- Hispanic whites are better educated, earn more, and possess more assets than African Americans and Latinos (US Census Bureau 2003). Among Latinos, Cubans are better educated and earn more than Mexicans and Puerto Ricans. Given the link between ethnicity and SES, it is important to explore and assess the extent to which the association of SES and mental health varies by ethnicity. 21

33 22 Socioeconomic Status and Mental Health Socioeconomic status refers to actual resources individuals own and their relative position in social hierarchies (Krieger et al 1997). Resources are material and social resources, including income, education credentials, and wealth. The impact of socioeconomic resources on morbidity and mortality endures despite changes in risk factors and intervening mechanisms. Higher socioeconomic status is strongly associated with lower mortality (House 2002, Mirowsky & Ross 2000, Phelan et al 2004) and lower rates of physical illness (Ferrer & Palmer 2004, Williams et al 1997). The socioeconomic gradients have also been observed for psychiatric morbidity. Lower SES is associated with a higher risk for psychiatric disorders (Bruce et al 1991, House 2002, Kessler & Cleary 1980, Marmot 2004, Mirowsky 1999, Syme & Berkman 1976, Williams et al 1997, Yu & Williams 1999). Although the inverse relationship between SES and mental health has been confirmed among non-hispanic whites, evidence is inconsistent regarding the association between SES and mental health among Latinos. Some researchers find that SES has no effect on depression or psychological distress (Breslau et al 2006, Gavin et al 2010, Golding & Karno 1988, Mirowsky & Ross 1980, Moscicki et al 1989) whereas others find a significant association between the two (Caetano 1987, Cuellar & Roberts 1997, Franzini & Fernandez-Esquer 2004, Guarnaccia et al 1991, Roberts & Roberts 1982, Ross et al 1983, Swenson et al 2000, Vega et al 1987). To find out whether the inverse relationship between SES and mental health is replicable for each ethnic subgroup, I will examine the association of SES and psychological distress among

34 23 Cubans, Mexicans, and Puerto Ricans respectively. I will also compare the magnitude of the association across groups to examine the differential impact of each SES measure. Threshold Effect While some studies find a stepwise progression of risk in the relationship between SES and health status, other studies find that the SES gradient in health is characterized by a threshold that predicts a weakening of the association between SES and health (Krieger et al 1997, Marmot 2004, Morales et al 2002, Williams & Collins 1995). This type of association between SES and health is called a threshold effect. Much of health impact occurs below a certain threshold of socioeconomic status. Additional effects become modest once a certain level of SES is reached. Previous studies found evidence for a threshold effect among Whites and African Americans (Ferrer & Palmer 2004, Mechanic 1978, Williams & Collins 1995). The National Co-morbidity Study found that the associations between SES and anxiety and affective disorders were characterized by a threshold effect (Yu & Williams 1999). There were large reductions in the risk of psychiatric disorders associated with income at low levels, but smaller declines in risks linked to additional income at higher levels of SES. The effect become non-significant above $70,000 (Yu & Williams 1999). Studies on the threshold effect of SES are still limited. Many questions need to be answered. First, few studies have been done to test the threshold effect among Latinos. The findings of two studies indicate a threshold health effect of household income among Latinos as an aggregate group (Alegria et al 2007c, Moscicki et al 1989). That is, income gains are associated with much larger changes in health status among the poor Latinos as

35 24 compared to the wealthy. Little is known about whether such a threshold exists for each ethnic subgroup. Second, little is known about which SES indicator has the threshold effect on mental health. Third, research is needed to identify the threshold after which the weaker health effect of SES occurs. To address these issues, I plan to test the threshold effect of education and household income for Cubans, Mexicans, and Puerto Ricans. My analyses will explain the question of which SES measures have threshold effect on psychological distress in each group and identify the thresholds beyond which weaker effects of SES are observed. Measures of SES One factor contributing to the inconsistent findings on the association between SES and mental health among Latinos is that researchers seldom distinguish the effect of SES for Latino sub-groups. SES may exert an effect on mental health for a particular Latino subgroup but have no impact for another subgroup. Another important reason is that various indicators of SES used in research may differently impact mental health. Evidence suggests that even in a single study, patterns of association vary for different indicators of SES (Williams 1990). I include in my analyses three indicators of SES: education, income, and wealth. These three SES measures capture distinct aspects of the socioeconomic well-being of the individual. Education and income are usually included in research on health; yet, few studies have assessed the health effect of wealth.

36 25 Education Education is one of the commonly used measures of SES in mental health studies. It is easy to measure and applicable to the unemployed. Considerable evidence demonstrates that individual s education level is an important predictor of mental health in the United States (House 2002, Kessler & Cleary 1980, Miech & Shanahan 2000, Williams 1990, Williams & Collins 1995, Williams et al 1992b, Williams et al 1997, Yu & Williams 1999). Regarding the Latino population, findings of some studies indicate a higher risk of mental illness associated with lower education (Moscicki et al 1989, Swenson et al 2000, Vega et al 1987) whereas other evidence points to its different impact for distinctive ethnic groups. Bratter and Eschbach (2005) find that education has less impact on Mexicans than Puerto Ricans. Rubio and colleagues (2004) find that low education elevated the risk of psychological distress among Puerto Ricans but not among Cubans. Their findings indicate the distinct effect of SES across ethnic subgroups. Cubans and Mexicans are more tied to traditional values such as familism than Puerto Ricans who are more acculturated into mainstream America. Individuals who perceive a high level of familism may receive less benefits from SES. Education itself is not a sufficient measure of SES because of its limitations in capturing the inequality in health effect. First, education is generally stable over adult lifespan and thus may not capture how changes in economic well-being in adulthood can affect mental health status (Krieger et al 1997, Williams & Collins 1995). Second, education as an individual measure of SES does not include the economic well-being of other family members (Williams & Collins 1995). Krieger and colleagues (1997)

37 26 stressed that SES needs to be measured at the level of the individual, the household, and the neighborhood. Household economic well-being largely determines how many financial resources are available to individual members and these resources produce health benefits or risks for the individual. Furthermore, inequalities in health associated with income are larger than those associated with education (see review by Williams & Collins 1995). To address these limitations of education, I will include household income in my study. Household Income Household income measures various sources of individual income in a household, including wage earnings, child support, and pensions. It has consistently been found that high income is associated with a low risk of psychiatric disorders (see reviews by Williams & Collins 1995, Williams & Harris-Reid 1999, Yu & Williams 1999). This pattern has been confirmed for the Latino population in aggregate and for a nationalorigin subgroup (Caetano 1987, Golding et al 1990, Moscicki et al 1989, Ross et al 1983, Rubio et al 2004, Swenson et al 2000, Vega et al 1987). I will expand this line of research by investigating subgroup patterns of the association between income and mental health. I will examine whether household income has the same magnitude of effect on mental health for Cubans, Mexicans, and Puerto Ricans. Wealth Wealth is a critical component of financial well-being; yet, researchers have paid little attention to the consequences of this aspect of socioeconomic status. Wealth refers

38 27 to accumulated assets, which people can use in time of emergencies or economic shocks. It is a source of economic security and power for a family (Krieger et al 1997). It is especially important to include wealth as a measure of SES in research on social gradients of health because wealth is more unequally distributed than income in the US (Keister & Moller 2000). Race/ethnicity affects wealth ownership. Non-Hispanic Whites own more assets than African Americans and Latinos (O'Connor et al 2001, Yu & Williams 1999). Among Latinos, about 60% of Cubans own a house whereas only 50% of Mexicans and 37% of Puerto Ricans own a house (US Census Bureau 2004a). The median value of houses owned by Cubans is also much higher than that of houses owned by Puerto Ricans or Mexicans. Excluding wealth from health research is problematic because of the important health implication of this aspect of socioeconomic stratification. Correlations between income and wealth are relatively low (Keister & Moller 2000, Lerman & Miksell 1988). Leman and Miksell (1988) found that the correlation between income and wealth was about 0.5 and it reduced to 0.26 when asset income was removed from total income. Households with comparable incomes can differ greatly in their total net worth. Even within same income level, individuals differ in amount of wealth (Marotta & Garcia 2003, Williams & Harris-Reid 1999). Furthermore, household structures affect wealth accumulation. Increased family size and family dissolution through divorce or separation decrease wealth ownership (Keister & Moller 2000). Among Latinos, Cubans and Mexicans tend to live in extended households and Puerto Ricans have a high rate of female-headed families. These ethnic variations in family structure make wealth especially important for a study of Latino mental health. Although