Family Functioning of Portuguese Families with a Clinically Depressed Family Member

There have been a limited number of studies investigating cross-cultural similarities and differences in family functioning. Previous research that has been conducted on the family functioning of families with a depressed member has indicated unhealthy family functioning as compared to families without a depressed member (Keitner, Miller, Epstein, Bishop, & Fruzzetti, 1987; Miller, Kabacoff, Keitner, Epstein, & Bishop, 1986; Sawyer, Sarris, Baghurst, Cross, & Kalucy, 1988). The purpose of this study was to: (1) investigate whether Portuguese families with a depressed member had unhealthy family functioning as measured by the Family Assessment Device (FAD) (Epstein, Baldwin, & Bishop, 1983); and (2) examine whether Portuguese families were comparable in this regard to families from two other cultural groups. Specifically, this study compared Portuguese family functioning with previous findings of American' s and Hungarian's family functioning. The FAD, which is based upon the McMaster Model of Family Functioning (MMFF) (Epstein, Bishop, & Levin, 1978), was used to assess the family functioning of individuals from the Portuguese, American, and Hungarian cultural groups on six specific dimensions and one general scale. The dimensions are Problem Solving, Roles, Behavior Control, Communication, Affective Responsiveness, and Affective Involvement. The General Functioning scale provides an overall assessment of functioning. The individuals in this study met the criteria for Major Depression as outlined in the DSM-111-R (American Psychiatric Association, 1987). These individuals were either seeking outpatient treatment or were inpatients at a local psychiatric hospital.


Acknowledgements
I would like to begin by thanking Robert Kohn, M.D., at the Cultural Psychiatry Program at Butler Hospital. This masters thesis was made possible due to his help and data. I would also like to thank him for taking the time to help me through some of the more difficult aspects of this thesis.
I would like to thank the members of my masters committee for all their support, patience and time in helping me reach this goal. Thanks to Larry Grebstein, Ph.D., for agreeing to take over as my major professor and helping me reach this goal.
Thanks to Jerome Adams, Ph.D., for being patient throughout this long journey.  Acknowledgements _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ iv         There is a paradigm shift taking place in psychology where the need to infuse multicultural issues into curricula and services has assumed major importance (Sue, Bingham, Porche-Burke, & Vasquez, 1999). Indeed, after nearly a half-century of studying White, middle-class males, the field of psychology is recognizing the existence of non-mainstream cultures and the lack of relevant theory to guide their work with these groups. For the field of psychology, this recognition has translated into a necessity of rethinking culture and a willingness to revisit existing beliefs. For clinicians, this recognition has necessitated self-education about the cultural context of the individuals and families with whom we work. With clinical work, the emphasis is on the need for culture-appropriate interventions. Clinicians will need new intervention strategies and more knowledge to appropriately prepare and address the needs of diverse individuals and families.
With issues of diversity and multiculturalism gaining more and more emphasis within this past decade (Hall, 1997;Landrine, 1992;Sue et al., 1999), it is essential that clinicians identify those factors that will assist their work with diverse populations. It is necessary to conduct research that will provide the information needed to support the clinician's training, experience, consultation, and supervision.
To not do so would border on violating the "Ethical Principles of Psychologists and Code of Conduct," (American Psychological Association, 1992, p.1600) which states under the general standard Principle 1.08 Human Differences: Where differences of age, gender, race, ethnicity, national origin, religion, sexual orientation, disability, language, or socioeconomic status significantly affect psychologist's work concerning particular individuals or groups, psychologists obtain the training, experience, consultation, or supervision necessary to ensure the competence of their services, or they make appropriate referrals.
Despite the increased emphasis on issues of multiculturalism in psychology (Hall, 1997;Landrine, 1992;Sue et al., 1999), one area that has received little crosscultural attention is the area of family functioning. Views or perspectives on families are based upon a cultural perspective and may influence services that clinicians provide to families. This may be problematic when differences in ethnicity, race, and class exist between the clinician and a family (Sue, Zane, & Young, 1994). Therefore, when considering the delivery of clinical services, the influence of culture on an individual and familial level should be taken into account. Indeed, Nichols and Schwartz (1998) suggest that the clinician "should consider and respect the unique subculture a family is from... . Perhaps the best advice is to be curious, stay open, but ask questions" (p.103). The behaviors of families may best be understood in their cultural context. Therefore, it is necessary to identify similarities and/or differences among various cultures in the context of a particular behavior or concept. Researchers could measure the similarities and differences between cultures through outlining the different norms and values, definitions of family, and through cultural ideals of healthy family functioning.
Historically, the study of the individual in the context of family and the individual in the context of culture have been treated as two separate topics (Szapocznik & Kurtines, 1993) despite the interconnectedness of family and culture. Sz.apocznik and Kurtines' ( 1993) work with Hispanic families revealed that family and culture cannot be considered separately. Culture exists through individuals, families, and communities in all aspects of life (e.g., family, work, class, ethnicity). Szapocznik and Kurtines' (1993) work led to intervention strategies that were effective in working with Hispanic families and could be generalized to other cultural groups. Their study highlights the potential benefits of studying family functioning across cultures.
This study takes a first step in providing information regarding the family functioning of Portuguese families living in the United States who have a depressed member. One purpose ofthis study is to determine whether or not this group is similar and/or different as compared to findings with two other cultural groups (i.e., Americans and Hungarians) whose families also have a depressed member. This information may then provide an increased understanding of the family functioning of Portuguese families when a family member is ill. Like the work of Szapocznik and Kurtines (1993), the findings from this study may someday lead to the development of new techniques and interventions for clinicians who work with these diverse families.
A clinician without the proper knowledge of the family functioning of different cultural groups may be at risk of performing a disservice to a potential client from one of those groups. Not only does the clinician need to seek out that information but that information must also exist. The concept of family functioning has been applied and researched in areas such as the links between the McMaster Model of Family Functioning (MMFF) (Epstein, Bishop, & Levin, 1978) with developmental and clinical parameters in child epidemiological studies (Maziade, Bernier, Thivierge, & Cote, 1987), the strain of moving (Munton & Reynolds, 1995), and in relation to 3 depression (Keitner, Fodor, Ryan, Miller, Bishop, & Epstein, 1991;Keitner, Miller, Epstein, Bishop, & Fruzzetti, 1987;Keitner, Ryan, Miller, Kohn, Bishop, & Epstein, 1995). One area where little knowledge exists is the relationship between depression and family functioning across different cultural groups.
Findings from research that has been conducted on family functioning indicate that families with a depressed member have unhealthy family functioning (Keitner et al., 1987;Miller, Kabacoff, Keitner, Epstein, & Bishop, 1986;Sawyer, Sarris, Baghurst, Cross, & Kalucy, 1988). Previous findings indicate that American and Hungarian families with a depressed member have more unhealthy family functioning as compared to families of the same culture who do not have a depressed member (Keitner et al., 1991;Miller et al., 1986). The purpose of this study is to determine whether the Portuguese living in Southeastern New England have similarly affected family functioning when they have a depressed family member. The findings regarding depression and family functioning from this study should indicate whether Portuguese families with a depressed member exhibit unhealthy family functioning in a manner that is consistent with the previous findings in two different cultural groups.

Definition of Family and Descriptions of the Three Cultural Groups
It is necessary to clarify what is meant when the terms Portuguese, Hungarian, and American family is used in this discussion. Because they are used so often, it is important to clarify what is actually meant when the reference is made to these families. When the term family is used during a discussion, people might think of Portuguese families as both nuclear and extended in a closed structure system (Moitoza, 1982), American families as the nuclear family unit (Kaslow, Celano, & Dreelin, 1995), and Hungarian families in terms of nuclear and extended families (Smith, 1996). What follows is a more detailed description of each cultural group and the reasons they were selected for this study.
Portuguese families.
The Portuguese have been chosen because of their presence in the Rhode Island and Southeastern Massachusetts area of the United States and the lack of knowledge with respect to family functioning and depression for this population.
Little research has been conducted on this "invisible minority" (Moitoza, 1982, p. 415). This may be due to many factors including the existence of few professionals working with the Portuguese in a research setting and their closed family system.
One of the difficulties in acquiring empirically validated population data for the Portuguese is due to a lack of data collection by government agencies. When the U.S. government collects census data by ethnicity, Portuguese is not included as an option because it is not officially recognized as a specific ethnic or minority group (these individuals are still counted but under a different category). Nonetheless, there are many individuals of Portuguese heritage living in the Southeastern region of New England and it has been estimated that there are roughly 1 million individuals of Portuguese descent living in the United States (Kohn, 1990). Although the Portuguese account for a large percentage of the population in certain regions of the United States, they have been all but ignored by researchers. One assumption often made is that the Portuguese population is like the Spanish-speaking population and therefore knowledge regarding the Spanish-speaking population can be generalized to the 5 Portuguese. This assumption is simply not true. The Portuguese are a separate cultural group that deserves its own body of knowledge for clinical intervention.
Clinicians of Portuguese heritage have the advantages of same language and shared culture, however, the problem is that there are few such clinicians. For those clinicians who are not of Portuguese heritage and find themselves working with this population, any body of knowledge could make invaluable contributions to their work.
Portuguese families are closed and patriarchal (Moitoza, 1982). Portuguese families are closed in that outsiders are viewed as potentially threatening to family cohesion. Clinical treatment is usually sought when "forced by police, local schools, district courts, or their family physicians" (as cited in Kohn, 1990, pp. 1-2). In this patriarchal family, the father's word is not to be questioned but followed (Moitoza, 1982). Most often the mother will manage the children and make the decisions unless an issue is deemed so important that the father is required to make the ultimate decision. In either situation, problems are solved by a parent and his or her decisions are not questioned.
Communication often travels in one direction. There may be more direct communication between mothers and their children and to a lesser degree between mother to father, followed by children with father. Overall, there is a lack of direct and honest communication within the Portuguese family. This is out of respect for authority figures in general, and the father in the patriarchal family in particular (Araujo, 1996). Wishes, desires, and thoughts are not verbalized when they may threaten the honor of the father and so secrecy prevails in Portuguese families (Araujo,l996). There is strict behavior control to retain the family's honor within the 6 community. However, in an attempt to control the behavior of their children, Portuguese parents may attempt to become overinvolved in the daily activities of their children. Individual importance is not valued as much as a family's presentation to the larger community. The focus is more on not shaming one' s family as children's behavior reflects on the parents and one's family as a whole (Moitoza, 1982). The mother is the manager of "both the intrafamilial and family-community interface" (as cited in Kohn, 1990, p. 1 (Araujo, 1996). Males in the Portuguese culture are "expected to keep their feelings to themselves otherwise suffer a diminution of power and respect" (as cited in Kohn, 1990, p. 1 ). Emotions may be expressed at transitional periods in family development such as the death of a loved one. Anger is an emotion that males are free to express but only when there is a culturally-sanctioned reason.

American families.
The American families were chosen for a number of reasons. One reason is that recent research involving this group exists and these families reside in the same geographic area as the Portuguese families. Since American values serve as the basis for the MMFF and the Family Assessment Device (FAD), the American families also serve as a reference group. Descriptions of the American families in relation to the MMFF dimensions were derived from the research ofKeitner, Ryan, et al. (1990) and Keitner et al. (1991) who used the FAD to measure differences in family functioning among a community sample. According to this research, American families feel comfortable with their family duties and the responsibilities assigned to them. They are also able to set limits and understand the consequences for rule breaking but do not feel decisive when solving family problems. American families felt that the environment was not open to showing emotion and that some members had difficulty responding emotionally but that family members showed love for one another (Keitner, Ryan, et al., 1990).
''North Americans considered themselves to be less open with each other about their feelings, communicating more indirectly, tending to be autonomous and self-centered and distancing themselves from one another" (Keitner et al., 1991, p. 257 Hungarian families. One reason that the Hungarian families were chosen is that recent research involving this group exists. The Hungarian sample was chosen in past studies because of its geographic location between East and West Europe and because they have a different culture and sociopolitical system (Keitner et al., 1991). Hungary is at the crossroads of Eastern and Western civilization and there exists a mixture of both Eastern and Western values (Smith, 1996).
Descriptions of the Hungarian families in relation to the MMFF dimensions were derived from the research ofKeitner, Ryan, et al. (1990) and Keitner et al. (1991) who used the FAD to measure differences in family functioning among a community sample. Hungarian families were described as feeling better about handling problems or dealing with emotional disputes and taking action to remedy the situation.
Hungarian families are "open, direct, and supportive of one another. They felt that they could work together better as a family unit and communicate and solve problems together" (Keitner et al., 1991, p. 258). Smith (1996)  showing anger, pain, or expressing conflict openly, which may shame a family (Smith, 1996). Hungarian family units work well together but at times family members become overinvolved in each other's affairs to the point that there is a lack of privacy (Keitner, Ryan, et al., 1990;Keitner et al., 1991). Hungarian families were dissatisfied with the lack of clearly defined rules and expectations in regards to each 9 family member's behavior (Keitner et al., 1991). There is a perceived lack of discipline and consequences when rules are broken (Keitner, Ryan, et al., 1990). The Hungarian family system is patriarchal and follows gender divided roles. This may differ from the ideal of healthy roles as defined by Western society. Although Smith (1996)  The McMaster Model of Family Functioning (MMFF) (Epstein, Bishop, & Levin, 1978) is based on empirical research with non-clinical families and systems theory and uses concepts from communication theory, learning theory, and the transaction approach (Byles, Byrne, Boyle, & Offord, 1988;Epstein et al., 1978).
There are five aspects of systems theory that underlie the MMFF. These are: (1) Parts of the family are related to each other; (2) one part of the family cannot be understood in isolation from the rest of the system; (3) family functioning is more than just the sum of the parts; (4) a family's structure and organization are important in determining the behaviour of family members; and (5) transactional patterns of the family system are involved in shaping the behaviour of family members (Epstein et al., 1978, p. 20-21).
The MMFF views a family as an interactional system whose structure, organization, and transactional patterns determine and shape its members' behavior (Byles et al., 1988). Indeed, within this interactional system there are explicit and implicit rules within a family that control family members' behavior (Epstein & Bishop, 1973). These rules relate to the dimensions of the MMFF that conceptualize a family as a system within systems. Part of the system consists of the immediate family members and any dyads within a family (e.g., individual, marital dyad and children). This system then interacts and relates to other systems such as extended family, the school system, the community, work, and religion. Viewed from a systems perspective, a family cannot be understood if just one dimension were isolated and studied (Epstein & Bishop, 1973). Although informative, it would not create a clear picture of how a family functions. To understand a family, one must look at multiple · ns of interaction and how that family operates as a whole, not how one dimeDSIO relates to another or one person's characteristics. What is important is how a person family's rules and its members' behavior govern how everyone in that family will interact (Epstein & Bishop, 1973). For example, it is how a family determines the rules for who will fill what roles, how that will be communicated, how family members will relate to each other in those roles, and how the rules are established so that behaviors are controlled to fit in with the rules.
The instrument used in this study is the Family Assessment Device (FAD) (Epstein, Baldwin, & Bishop, 1983) which is based upon the conceptual model of the MMFF (Epstein, Bishop, & Levin, 1978). The MMFF is the only model of family functioning that was developed on the basis of empirical research with non-clinical families (Epstein, Bishop, & Levin, 1978). The MMFF considers the dimensions of Problem Solving, Communication, Roles, Affective Responsiveness, Affective Involvement, and Behavior Control in conceptualizing and defining a family's functioning (Epstein, Bishop, & Levin, 1978). Therefore, as it pertains to this study, family functioning is defined as how well or poorly a family-unit communicates, defines their roles, solves problems, controls each other's behavior, and responds and is involved affectionately and emotionally in the lives of the other members.

Cross-Cultural Applicability of the FAD
One difficulty in studying differences in family functioning across cultures lies in identifying an instrument that has cross-cultural validity. The MMFF was "d 1 · eve oped man Anglo-American cultural context and is based on a Western Judeo-Christian value system" (Epstein & Bishop as cited in Morris, 1990, p. 105). As a I th FAD measures family functioning in a manner that is consistent with those resu t, e Fo r example, Communication is defined as healthiest when it is clear and values. direct and least effective when it is masked and indirect (Epstein, Bishop, & Levin, 1978 ). For the Behavior Control dimension, a flexible strategy is healthier than a rigid or chaotic strategy. The FAD has been shown to have good psychometric properties (Stevenson-Hinde & Akister, 1995). The extent to which it is a valid measure of family functioning of different cultural groups, however, is unknown.
This study involves three different cultural groups. The American cultural group shares the same value system within which the FAD was created. However, some values in the Portuguese and Hungarian samples run counter to those upon which the FAD is based suggesting the possibility of cultural bias. Indeed, the creators of the FAD caution users of this model to be cognizant of the effect of values "on the judgment and evaluation of behaviour" (Epstein et al., 1978, p. 20).
Research looking at the cross-cultural applicability of the FAD has been conducted. Morris (1990) performed a study comparing Japanese-Americans with Hawaiian-Americans living in Hawaii. She postulated that certain aspects of Hawaiian and Japanese culture that are valued run counter to the North American values inherent in the definition of the dimensions of the MMFF and FAD. In particular, Morris notes that Japanese-American cultural norms run counter to all dimensions except for the Roles dimension. Morris (1990) hypothesized that a Japanese-American family considered to have healthy functioning, according to Japanese-American cultural norms, would be assessed by the FAD as unhealthy functioning on all dimensions except the Roles F or example, Japanese-American families are described as having dimension.
h ·tan· an styles ofleadership whereby the father made family decisions "aut on unilaterally" (Kitano as cited in Morris, 1990, p. 107). This style of problem solving would be considered as unhealthy by the MMFF and would be measured as such by the FAD. Results for the Japanese-Americans indicated healthy functioning on the Problem Solving, Communication, Roles, and Affective Involvement dimensions with unhealthy functioning for Affective Responsiveness and Behavior Control. Morris (1990)  Involvement, and General Functioning dimensions. The scores for the other dimensions were not found to be significant but were still rated as healthy. Morris (1990) believed that since the FAD was based upon a western Judeo-Christian value system, cultures that were most unlike the American culture would be rated by the FAD as having unhealthy family functioning. However, both cultural groups did better than was hypothesized prompting Morris to question whether these effects were a strength of the FAD or a result of the sample. These samples may have 15 Ii tn . g in America for a sufficient enough period of time to allow for the process been v of acculturation and adaptation to have taken place. How did acculturation to American society contribute to the results and was it possible that the FAD was crossculturally valid?
It is important to assess the applicability of the FAD when using it to conduct research on groups whose values may not adhere to those of the Western world.
These questions raise issues about the validity of such an instrument and its potential cultural bias when used in a cross-cultural context. The cross-cultural applicability of the FAD to the Portuguese population is not the purpose of this study. There have been studies conducted to determine whether the FAD translates to the Portuguese (Kohn, 1990) and it is because of these :findings that this study can use the FAD to answer a cross-cultural research question. Although the FAD has successfully been translated into Hungarian, Spanish, Portuguese, and French (Keitner et al., 1991;Keitner, Ryan, Fodor, Miller, Epstein, & Bishop, 1990), the authors of the FAD cautioned test users to be aware of the potential effects of cultural variability on the Affective Responsiveness of families and stated that "consideration obviously must be given to the issue of cultural differences within the confines of our definition" (Epstein et al., 1978, p. 25).
Comments made by the authors of the instrument would suggest proceeding with caution in ipterpreting any results. This implies that researchers must be knowledgeable of a group's culture whenever they are interpreting the results of a cross-cultural study where the administered instrument is based in one value system and the instrument's respondents belong to a cultural group with a different value 16 t Clinicians and researchers must also be aware of their values and those of the sys em. group they seek to assess while acknowledging the possible validity of cultural differences in family functioning that may appear (Epstein et al., 1978). Findings from such research should not be considered invalid simply because the value system of the MMFF and a family may be different. These value systems may relate to a group's culture and how it perceives illness.

Definition of Culture
So why is it interesting or worthwhile to look at the family functioning of Portuguese families in relation to Hungarian and American families? Some researchers argue that due to processes such as increasing globalization, conceptualizing and researching cultural groups as different and separate entities is no longer viable and that alternatives are needed (Hermans & Kempen, 1998). Despite the arguments against studies that investigate cultural dichotomies, there is still a belief in the conceptualization of cultures as having similar and different elements as evidenced by the vast amount of literature in this area.
One definition by the NIMH Culture and Diagnosis Group states: "Culture refers to meanings, values, and behavioral norms that are learned and transmitted in the dominant society and within its social groups. Culture powerfully influences cognitions, feeling, and self-concept, as well as the diagnostic process and treatment decisions" (cited in Lu et al., 1995, p. 481 ). In 1945 Linton defined culture as "being a shared, learned behavior transmitted from one generation to another, having both external and internal components" (cited in Lu et al., 1995, p.481 ). In this study, l l ·s something that is socially transmitted across generations and is defined as a cu ture learned and shared pattern of beliefs, attitudes, norms, roles, emotions, sense of self, t ms and behaviors which influences or shapes the individual's constructural cus 0 ' aruza . tion of said elements (Kaslow et al., 1995;Lu et al., 1995;Triandis, 1995). org Culture is also characteristic of a group that is linked through a shared language, a particular time-period, and people who "live within fairly definite boundaries where they interact socially among themselves" (Berry, 1976, p. 9).
This definition is comprised of many of the points inherent in other definitions.
This definition also explains how culture is kept within a family and transmitted across the generations. This is not to suggest that culture is in some way stagnant and remains exactly the same from one generation to the next but that the amount of change may not deviate far from the current norm. We would not expect a family to suddenly behave in a manner that is inconsistent with its culture's values, beliefs, and customs unless some extraordinary event has occurred. For example, we would expect a family's culture to change if that family immigrated to a new country with a different value system. In this situation the old methods of interaction would still be This is to say that if culture as it is defined in this study were not transmitted groups.
then we could not look at cultural differences across ethnic-cultural lines (one could still look at class differences and the such if one desired).

Culture vs. Ethnicity
Culture is related to but not exclusive to ethnicity. Oftentimes these terms are used interchangeably (Betancourt & Lopez, 1993;Phinney, 1996). McGoldrick and Giordano (1996) state that ethnicity, "refers to a common ancestry through which individuals have evolved shared values and customs (p. 1 )" while Betancourt and Lopez (1993) state that ethnicity refers to groups of individuals who share a common nationality, culture, and/or language. In both instances, ethnicity is marked by something that is shared by a group of people and can be used as a marker to differentiate one group from another (e.g. Portuguese-American, Hungarian, African-American). Ethnicity refers to an individual's sense of belonging to a group of people with common ancestry and history, and with whom the individual shares similar cultural and social beliefs as well as values and customs such as language and religion (Lu et al., 1995;Phinney, 1996). When someone states her or his ethnicity, an implicit message is conveyed about this person' s values, beliefs, and customs (i.e., that persons culture). Depending upon someone' s familiarity with this ethnic group, this implicit message may also carry information about gender roles, methods of communication and even how this person or family interacts with other systems in the social world (i.e., family, friends, community).
There is a culture associated with a given ethnicity but there is also, for example, a business culture that describes the workplace, and an academic culture, and 19 It e of social class status. In each case, culture dictates how individuals within acu ur these systems should interact with one another and other systems. These last three examples could not be called ethnicity. The difference that separates these examples from ethnicity is the lack of shared ancestry. The point is that culture exists in many forms and is not limited solely to ethnicity. This study looks at differences involving three groups that are ethnically different (Portuguese, Hungarian, and American) but also have varying levels of socioeconomic status, have different nationalities, and have different family structures. A distinction is made to clarify what is meant by culture in this study especially since seemingly ethnic labels are used to describe the three groups. My interest is in each culture's value system (that is passed on to successive generations) and its method of family interaction.

Cultural Values and Norms
Individualism and Collectivism.
Portuguese, Hungarian, and American cultural values can be organized around the elements of collectivism and individualism (Triandis, 1995). Individualism and collectivism are not on a continuum and in fact are uncorrelated tendencies (Triandis, 1995). People exhibit both individualistic and collectivist behaviors and it is believed that a balance between the two is optimal (Kagit<;ibasi, 1997; Triandis, 1995).
However, within a given culture, there are overall tendencies towards the values of one or the other. Because of this we can say that one culture tends towards individualism while another tends toward collectivism (Triandis, 1995). Since these tendencies are embedded in our culture, they are passed on from one generation to the next. 20 According to Triandis (1995), collectivism consists of four elements: (1) A social pattern consisting of closely linked and interdependent individuals who not only see themselves as part of one or more collectives (e.g., family, co-workers, tribe, nation) but also emphasize their connectedness to other members of these collectives by sharing common beliefs and a willingness to cooperate; (2) behavior that is primarilY motivated by the social norms of, and duties imposed by collectives rather than by individual pleasure or advantage; (3) prioritizing the views, needs, and goals of collectives over personal views, needs, and goals; and ( 4) the maintenance of established relationships that may not be in one's best interest. Triandis (1995) defines individualism as consisting of four elements: (1) Social patterns that consist of loosely linked individuals who view and define themselves as independent of collectives (e.g., family, co-workers, tribe, nation); (2) behavior primarily motivated by personal preferences, attitudes, personal needs, rights, and personal goals which supercede the goals of the group and may necessitate separating from the group to attain these goals; (3) hedonism where one seeks out and does what is enjoyable to oneself which may not involve relationships; and (4) weighing the advantages and disadvantages of the personal contracts (relationships) they have established with others, and a willingness to disassociate from these others if there is tension, conflict, or a difference in personal goals.
The Portuguese culture tends towards collectivism while the American culture tends towards individualism (Triandis, 1995). As for the Hungarian culture, their cultural orientation is not completely known, but one aspect of their family organization suggests collectivism. Hungarian families can consist of a nuclear family 21 h l .b (Smith 1996). The sib consists of relatives whether by blood or not who andt es ' "d aid to the nuclear family and are morally obligated to help and offer support provi e (Smith, 1996). This extended family suggests collectivism (Triandis, 1995). Hungary sits at the crossroads between the East and the West and it may be the case that Hungarians would fall somewhere between the Portuguese culture and the American culture. However, without any research, deciding whether Hungarians are collectivists versus individualists would be purely speculative.
The elements of individualism and collectivism lend themselves to cultural values that dictate family member's roles; family rules and structure; identifying and solving problems; what, when, and how much affect and emotion may be expressed; and, family communication styles (Kaslow et al., 1995). In this way, individualism and collectivism relate to the dimensions of family functioning in this study. For example, collectivist cultures value conformity and performing the will of the whole, while individualistic cultures place more value on learning to be independent (Triandis, 1995). These values are important in behavior control where the collectivist will attempt to modify the behavior of its members to conform to what is perceived as acceptable behavior by the larger cultural group rather than allowing behavior that could be potentially damaging to a family's standing in the community. This strict upbringing is seen as "love" by the parents and the fulfillment of their duty to make the child a useful member of society (Triandis, 1995). In individualistic cultures behavior control involves more of encouragement to be an individual and to rely less on others. Youth in a collectivist culture tend to perceive parental supervision as love while teenagers from an individualist culture would see the same parental supervision 22 . volvement and intrusive (Triandis, 1995). Behavior control on the parts of asovenn t IS • sometimes received negatively by individualistic youth and is seen as too paren s strict or controlling. "In general, collectivist child rearing is warmer and also more t lling than individualistic child rearing" (Triandis, 1995, p. 65). con ro

Acculturation and adaptation.
Acculturation occurs when one cultural group comes into contact with a new, more dominant cultural group (Berry & Sam, 1997). The non-dominant cultural group changes to adapt to the "new contexts that impinge on them as a result of migration, colonization, or other forms of intercultural encounters" (Berry & Sam, 1997, p. 293).
This cultural change in the group is called acculturation.
Portuguese families in this study began the process of acculturation the day they emigrated from their country of origin. From the day that these families arrived on U.S. soil, their collectivist values have come into contact with individualistic values. Some of the behaviors based on collectivist values may be in conflict with those of the dominant individualistic values in the U.S. As a result ofthis conflict, the Portuguese families would have begun a process that has been termed "behavioral shifts" (Berry & Sam, 1997, p. 298). This term represents the learning of new behaviors that are more appropriate in the new cultural context while possibly unlearning past behaviors that conflict with the new, dominant context. These behaviors can range from changes in food preferences to changing social norms and gender roles.
The level of new behaviors learned and unlearned is directly related to individual factors that include, for example, demographic variables such as age and 23 . as well as the families' acculturation strategy (e.g., integration, education, assimilation, separation, marginalization) and how much they value keeping aspects of their culture unchanged (Berry & Sam, 1997). If, for example, a family's acculturation strategy were separation, then they would attempt to maintain their cultural values while trying to protect against any influence from the dominant culture resulting in a low level of behavioral shift. There are also group-level factors that include, for example, the dominant culture's attitudes and acculturation strategy for new groups, and whether an ethnocultural group exists within the U.S. where the non-dominant migrants will ultimately settle (Berry & Sam, 1997). Settling in an area that is inhabited by a similar ethnocultural group should ease the stress related to the acculturation process while helping new immigrants shift their behaviors to cope better with the new cultural context. This is true for the Portuguese sample in this study that settled in the Southeastern New England region of the U.S. A Portuguese ethnocultural group exists in this area and may act as somewhat of a buffer protecting aspects of the Portuguese culture. The fact that the Portuguese families in this study settled in this region might mean that they have maintained a large percentage of their original cultural values and have experienced little behavioral shift.

Conceptions of Illness
Cultural groups have varying value systems; therefore, standards regarding healthy or unhealthy family functioning will vary accordingly. What one cultural group considers healthy family functioning could be perceived as unhealthy by another. Culture influences concepts of illness and the expression of symptoms (Lu et 24 I 1995 ). Cultural norms will dictate whether certain behaviors are judged as normal al., d fun ctional and so, may differ across cultures (Kaslow et al., 1995). The manner or ys in which individuals express, experience, and cope with feelings of distress has been termed "idioms of distress" (Nichter as cited in Lu et al., 1995, p. 489). These idioms of distress may be expressed in a somatic manner. Recent research has found that somatization is evident in all cultural groups and societies (Kirmayer & Young, 1998).
In the Portuguese culture for example, "nerves" is what a woman has when she is experiencing or beset by feelings of sadness, anxiety and worry and behaves in a manner that is consistent with these symptoms (there is no parallel experience for Portuguese men). However, Portuguese people would not describe her behavior as depressed. Instead, they would attribute her behavior to the "nerves" and consequently the woman would not be inclined to seek any help because "nerves" is seen as normal or accepted behavior in Portuguese women (Moitoza, 1982). From a Western point of view, this behavior in a person is seen as unhealthy behavior requiring help. This is evident in Kohn's (1990) work translating the English version of the FAD into Portuguese. Of the Portuguese depressed patients involved, 58% of the sample did not identify a person with major depression as having mental illness but identified paranoid schizophrenia, schizophrenia, and alcoholism as mental illnesses (Kohn, 1990). Of those who did not identify the depressed individual as having mental illness, 36% did not believe that there was anything wrong with the individual. Depression in the Portuguese culture was felt to be a "normal state" (Kohn, 1990, p. 8). 25 According to a Western viewpoint, these attitudes may lead to Portuguese . 'd ls or families not identifying depression when it exists. These individuals or jnd1vi ua fiunilies may not seek help because there is no perceived problem, which in turn could exacerbate any problems associated with the illness experienced by an individual and 8 family. However, from a Portuguese perspective, it could be the case that since depression is seen as a normal state, then these families may have inherited mechanisms (passed down through the generations) to cope with these "nerves" in a way that limits serious impact on family functioning. Alternatively, they may perceive depression as something that comes and goes and that a family simply needs to readjust and accommodate all its members during that time. This perspective suggests that depression as an illness that affects family functioning would not be real because it is a 'normal state' that does not require any differing response. In fact, depression would not be considered an illness at all. If depression is not seen as an illness, then how could it possibly adversely impact a family's functioning?
These factors suggest that for the Portuguese, depression is not seen as a mental illness. Depression is part of life much like when someone in a family catches a cold or gets sick; it is just dealt with. Even if the Portuguese perspective explained what was occurring, it does not mean that someone having depression does not adversely impact her or his family. It could be the case that Portuguese families are unaware of depression because they have not been educated about mental illness and how to identify it and its effects. Even if the social construction of mental illness is different between these two groups, it does not necessarily mean that the functioning of a family will not change if some of its members began to show the effects of 26 This change in behavior will impact a family's interaction patterns and depression. modes of operating. Depression and its impact on families may exist whether or not a particular culture identifies it as such.

Family Functioning and Illness
The implicit and explicit rules that govern a family's behavior are usually known to all its members and become threatened when a family member is ill. When someone is depressed, whether or not it is called or perceived as an illness, that person's ability to function at a pre-illness level changes, and this change will impact a fiunily's functioning. Rose (1998) has posited that families go through a process of •'family transformations" (p. 364) before coming to accept the illness. During that time, a family may experience or be forced to restructure and reorganize tasks, develop new coping skills, decide upon new role definitions, and deal with the loss of or added responsibility (Papadopoulos, 1995). For example, the ill member may no longer be able to fulfill his or her family role forcing others to carry the responsibility.
Other functions that a family fulfills that may be impacted by an illness include meeting the basic needs for safety, security, and survival; meeting emotional and educational needs; coping with crises that may arise other than the current illness; and advocating for other family members when necessary (Marsh & Dickens, 1997).
Another factor that contributes to the impact of illness on a family' s functioning is due to changes in the mental health system where there has been a shift m mental health care from inpatient care to outpatient care (Backer & Richardson, l989;Mitchell & Reaghard, 1996), termed deinstitutionalization (Backer & Richardson, 1989;Lefley, 1989). In the past decade the mental health field has 27 d a shift from inpatient care to outpatient care. This deinstitutionalization of witnesse ta l health system (Backer & Richardson, 1989) has been accompanied by the men many issues such as concerns over quality of care, shifting of care from institutions to fiunilies with the accompanying new family burdens, and basic questions regarding recovery.
With the increased emphasis on outpatient care, more of the responsibility for caring for the mentally ill person has shifted from formal care networks to families (Backer & Richardson, 1989;Rose, 1998). Family members provide approximately two-thirds of the care given to mentally ill persons (Backer & Richardson, 1989).
Although many mentally ill persons live with their families and their family members act as caregivers, families are often not prepared to care for a mentally ill member. It can create enormous strain on relationships, tax or exhaust a family's resources, and move a well functioning family into dysfunction (i.e., unhealthy family functioning).
What may further compound the stress that accompanies having an ill member is the duration of the illness. The illness is not always temporary and oftentimes fiunilies are uncertain of whether and when the illness will remit. In the meantime, fiunilies are dealing with certain burdens as they cope, manage, and come to terms with the illness of their family member and what that means for them and the rest of their family. Consequently, feelings of neglect, frustration, resentment, and poor communication can arise (Papadopoulos, 1995).
To date, research comparing inpatient and outpatient services has been inconclusive in determining which is better. There have been some studies that report no differences in inpatient/outpatient treatment (Cornwall & Blood, 1998; Van den 28 Ernrn elkamp Kraaykamp, & Griez, 1988) and yet other studies indicate that Hout, ' d Orn e circumstances an inpatient treatment setting is more effective than an un er s outpatient setting (Lerner, Gelkopf, Oyffe, & Sigal, 1995;Longabaugh, 1996;). Both settings are paths to recovery; the difference may lie in the manner in which patients improve vis a vis the philosophy of treatment or services they may receive.

Family Functioning and Depression
Depression has been defined in general terms as a mental illness ''that affects the physical, psychological, and social functioning of an individual" (Culbertson, 1997, p. 29 ; that's close to 12.4 million women and 6.4 million men in the United States. Major depressive disorder affects 6.5% of women (6.7 million) and 3.3% of men (3.2 million) each year and is the leading cause of disability in the United States (National Health, 2000). Consider that these statistics are based within a

Institute of Mental
given year and it is safe to assume that at some point, over a lifetime, a family is likely to experience the effects of depression on one of its family members. What has not been detennined is whether depression leads to unhealthy family functioning or if it is unhealthy family functioning which precipitates depression ).
This dilemma is further complicated by the fact that there is evidence supporting both 29 . s . In general, there appears to be support for the persPCct1ve b ·d"ty of unhealthy family functioning (dysfunction) and depression. comor 1 1  analyzed literature on the relationship between ·on and family functioning. Results from various studies indicated that depress1 fiunilies with depressed members reported more dysfunction, marital maladjustment, and demonstrated impaired family functioning with specific problem areas in Communication, Roles, Affective Involvement, and Problem Solving (Keitner et al. , 1987). Also, 40% of adults living with a depressed family member were distressed to the point of meeting criteria for therapeutic intervention. Families with a depressed fiunily member also experienced substantial difficulties in many areas of family life and more difficulties than families with members diagnosed as schizophrenic, bipolar, with rheumatoid arthritis, or cardiac disease.
Depression is a mental illness that can negatively impact a family that is unprepared to cope and manage it. "Be it onset, maintenance, or change of illness and health behaviours, the family through support, modeling and communication can contribute to, or hinder the successful treatment or coping with an illness" (Papadopoulos, 1995, p. 32). Those families that are not prepared or may not have the resources appear to be at risk for family dysfunction and high levels of stress that are so severe as to create the need for an intervention. Fortunately, research indicates that depressed family members can recover from their depressive episode more rapidly if his or her family's overall functioning improves during the course of the illness . Improving a family ' s overall functioning serves to protect a fiunily from experiencing a need for an intervention, concomitantly speeding the 30 time of the ill member. It serves to mitigate a negative cycle where recovery unhealthy family functioning, which occurs in part as a result of depression of a family ber increases the duration of a family member's depression, which in turn mem , continues to strain a family to the point of dysfunction and the need for an intervention. Thus, professionals can learn from the families that are able to cope and manage the illness in order to inform their work with other families. Consultation that empowers and educates families to manage and cope with the illness more effectively may result in the alleviation of family stress and possible future dysfunction and give fiunilies a sense of mastery or competence in dealing with the illness.

Assessing Family Functioning in Families With an Ill Member
Demographic differences on the FAD.
Some studies have investigated the relationships between participant sociodemographic information and the FAD and family functioning in general. Wenniger, Hageman, and Arrindell (1993) in their study of the Dutch version of the FAD found that the Affective Involvement and Behavior Control Scales were influenced by educational level. Ninety-four percent of the Portuguese sample went to school in their country of origin while it is expected that the majority of American and Hungarian respondents attended school in their respective countries of origin. It is difficult to assess the influence of educational level on family functioning ratings when respondents attended school in different countries with different educational systems and standards. This study was also missing some data on years of education across the three samples. For these reasons, the educational levels of the three samples were not compared. 31 Maziade et al. (1987) assessed the potential confound of SES and did not find an association between SES and family functioning but cautioned that they had a low sample size.  tested the relationships of SES variables with the FAD dimensions and did not find any clinically meaningful correlations between the two.
More recent research looking at the effects of SES and marital status on family functioning suggests that two-parent families and higher SES families are less likely to experience unhealthy family functioning Hayden et al., 1998).
Unmarried mothers reported less healthy family functioning than their married counterparts Hayden et al., 1998). No differences in the father's FAD ratings existed although few of the unwed fathers participated Hayden et al., 1998). There were no differences in the self-report ratings oflow vs. high SES families on family functioning . Eighty-eight percent of the Portuguese and 93% of the American sample respectively were married with the remainder of respondents divorced, separated, or widowed. All Hungarian respondents were married. The lack of a varied distribution of marital status minimizes the need to determine if there is any influence of marital status on family functioning ratings.
There is evidence to support a relationship between individual ratings of family functioning and age. Sawyer et al. (1988) found that older adolescents rate their family's functioning as less healthy than do younger adolescents. This may be of llnportance depending upon the number and age of any adolescents taking the FAD as it may skew results. Sawyer et al. (1988) also found that in both samples, adolescents . tly reported unhealthier functioning than their parents did. For this study, constSten individuals under the age of 18 were not included in the sample.
In developing the Portuguese FAD, there was a problem with the validity of the Affective Responsiveness scale. Kohn (1990) suggests that the problem is one of differing conceptualizations of affective responsiveness between Portuguese and American samples. Kohn (1990) did not alter the scale and suggested that it may serve to illuminate immigrants' adaptation and acculturation towards developing an American conceptualization of healthy Affective Responsiveness.

Comparisons of clinical and non-clinical samples.
The MMFF was based upon clinical and non-clinical samples and does discriminate between healthy and unhealthy family functioning. The developers of the FAD established the initial psychometric properties of the instrument on a sample that included large proportions of families with one member in a clinical setting and family members from a non-clinical setting (Epstein, Baldwin, & Bishop, 1983). Results from a study conducted by Miller et al. (1986) indicated that families with a depressed member scored as more pathological than the non-clinical families on all of the FAD's seven scales. Overall, there was a significant difference between clinical and nonclinical families' FAD ratings across disorders of major depression, alcohol dependence, schizophrenia, adjustment disorder, and bipolar mania. The most severe cases of unhealthy functioning were evidenced in the families of depressed patients. Keitner et al. (1991) compared family functioning of Hungarian families with a mentally ill member to families without a mentally ill member. All family members over the age of 12 were asked to complete the FAD. Results indicated that Hungarian faIIlilies with a depressed member reported worse family functioning than those without a depressed member. Significant differences were found for the Communication and General Functioning dimensions. Roles was the only dimension where the control group reported unhealthy family functioning. Sawyer et al. (1988) compared the FAD ratings of family members of a clinical sample to family members of a non-clinical sample in South Australia. The clinical sample consistently rated their families as lower on functioning than the nonclinical sample. Similar findings came from studies where family members of fumilies with a depressed member reported significantly worse family functioning than families without a depressed member (Keitner et al., 1987;Miller et al., 1986).
There is evidence that family functioning improves as depression remits but not to the point where the families experience less dysfunction than non-clinical families (Keitner et al., 1987). Positive changes in overall family functioning were associated with faster recovery (Keitner et al., 1987; and, patients that view their families as having healthy functioning during the acute phase of major depression have a significantly greater chance for recovery (Keitner et al., 1995). In tum, depressed patients were three times as likely to relapse in families where members exhibited high levels of "expressed emotion" and 59% of patients whose spouses exhibited high levels of expressed emotion relapsed .
Expressed emotion is "critical, hostile, or emotionally overinvolved attitudes expressed by key relatives in reference to the depressed patient" , p. 1130). 34 One critic questioned the feasibility of assessing effective family functioning in families with an ill member. Roy (1990) felt that families with an ill member may function well or as well as possible under their current circumstances but would be assessed by the FAD as functioning poorly. Roy (1990) stated that a family with an ill member must function at a level that is on par with a healthy family in order to be assessed as functioning well by the FAD. A family with a depressed member may be functioning at its optimal level but this in and of itself should not be interpreted as an indication of healthy family functioning. This is not to understate the importance of identifying the level at which a family with an ill member is functioning. Other measures could be created that are sensitive to the scenario presented by Roy. The result of such measures would mean that these families could receive clinical services and interventions tailored to their level of need and current situation.

Cross-cultural comparisons.
The FAD, which is based on Western values, has been used in studies to measure the family functioning of different cultural groups with and without a depressed member. Keitner, Ryan, et al., (1990)

Implications for the current study.
The implication of the previously mentioned studies on clinical versus nonclinical samples in relation to the current study is to acknowledge that depressed patients and their families will report more dysfunction within their family than nonclinical families. The FAD adequately discriminates between healthy functioning fiunilies and families with unhealthy functioning. Even if a family with a depressed 36 be were functioning at the highest level in its situation, that family would still niem r report unhealthy functioning in relation to a healthier functioning family.

Findings from the cross-cultural research study suggest that depressed families
can score as healthy on some of the FAD dimensions. These findings relate to the current study in that the family functioning scores of the Portuguese clinical sample may be better (i.e., healthier) than what would be found in a community sample of Portuguese families in Southeastern New England. Thus, it would be an error to generalize results to the same cultural group in the community. A study comparing a Portuguese non-clinical with a clinical sample, while considering the effects of time on cultural aspects such as acculturation and generational differences, would be necessary to draw conclusions regarding differences in family functioning.
Knowing that the Hungarian and American cultures had their differences in family functioning (Keitner, Ryan, et al., 1990) and assuming that the FAD is not culturally biased, this study sought to examine whether the family functioning of a Portuguese family from Southeastern New England with a depressed member might resemble in any way the family functioning of an individualist (American) or possibly collectivist (Hungarian) cultural group.

Predictions
This study makes three predictions involving Portuguese family functioning.
Due to a lack of research on the Portuguese, predictions will be guided by literature that describes the Portuguese culture and by information from personal communications with clinicians who work with this population. Two specific predictions are made for Portuguese family functioning. The final prediction is made 37 h of the three cultural groups: Portuguese, Hungarian, and American. There is for eac . s research to guide the Hungarian and American predictions (Keitner, Ryan et preVIOU al, 1990).
The creators of the FAD developed cut-off scores to rate a family as healthy or unhealthy on a specific dimension (Miller, Epstein, Bishop, & Keitner, 1985). Since any response to a FAD question ranges from I (strongly agree) to 4 (strongly disagree), the average score for a specific dimension will fall within that range. In order to obtain the mean cut-off scores, the mean scale scores for each FAD dimension for each individual was first computed. The individual scale scores were then aggregated and a mean cut-off score for each dimension was computed. In general, a mean FAD score of2.0 or higher suggests unhealthy family functioning while a mean ofless than 2.0 indicates healthier family functioning. The cut-off score that corresponds with a specific dimension are listed within each prediction.

Prediction I.
The FAD will assess Portuguese families as having healthy functioning on the dimensions of: (a) Problem Solving; (b) Roles; and (c) Behavior Control. This corresponds to a mean FAD score that is lower than the cut-off scores of 2.2, 2.3, and 1.9 respectively for each dimension (Miller et al., 1985). Kohn' s (1990) work with the Portuguese version of the FAD indicated that the depressed patients were rated as unhealthy on every dimension with the exception of Problem Solving.

Prediction 2.
The FAD will assess the Portuguese family's functioning as unhealthy on the dimensions of: (a) Communication; (b) Affective Responsiveness; and (c) Affective 38 l me nt This corresponds to a mean FAD score that is higher than the cut-off 1nvo ve · f 2 2 2.2 and 2.1 respectively for each dimension (Miller et al., 1985). seores o · ' ' Prediction 3.
The General Functioning scale measures overall health-dysfunction of family functioning. Since these families are dealing with a major illness, it is hypothesized that all three groups will have overall unhealthy family functioning. This corresponds to a mean FAD score that is higher than the cut-off score of2.0 (Miller et al., 1985).
Wrthin the range of scores, it is predicted that the Hungarian sample will have the healthiest rating of the three groups followed by the American sample. The Portuguese will have the worst family functioning.

Participants
The present study involved analyzing data that was previously collected at the for Major Depression. Demographic information for the three cultural groups is presented in Table 1. Of the 73 respondents, 68 were currently married, 1 was single and had never been married, and 4 were widowed, divorced or separated. The average number of children was 2.2. All participants were inpatients at a local psychiatric hospital.
There was also a sample of American outpatients for which there were no individual data but for which there were a total n, SDs, and mean ratings on the FAD 40 ·ons This data was used in the preliminary analyses discussed later in the mmenst .
Results section.

Measures
Description of the FAD. 1983) The FAD was created as a screening instrument "to collect &BIS op, · information on the various dimensions of the family system as a whole, and to collect this information directly from family members" (Epstein, Baldwin, & Bishop, 1983, p. 171 ). While other researchers focus on single dimensions such as communication or role behaviors, the creators of the MMFF felt that a family's complexity could best be assessed or explained using multiple dimensions (Epstein et al., 1978). The FAD consists of seven scales, six of which measure the dimensions of the MMFF and the seventh that is called General Functioning. Participants rate their agreement or disagreement to a statement by selecting an answer on a four-point Likert scale of 1 (Strongly Agree) to 4 (Strongly Disagree).
Problem Solving is defined as a family's ability to solve problems at a level that maintains effective family functioning. The Problem Solving dimension consists of six items (a= .80), one of which is, "After our family tries to solve a problem, we usually discuss whether it worked or not." Epstein et al. (1978) define a family problem as "an issue that threatens the integrity and functional capacity of the family, the solution of which presents difficulty for them" (p. 21-22).
Communication is defined as the verbal exchange of information within a fiuni.ly. This dimension consists of nine items (a= .70), an example of which is, "We are frank with each other." The most effective communication style would be clear and direct communication (Epstein et al., 1978).
Roles are the established patterns of behavior by which individuals handle fiunil.y needs, including whether tasks are assigned appropriately and carried out 42 "bly This dimension consists of 11 items (a= .69), an example of which is, resp<>OSl · "We are g enerally dissatisfied with the family duties assigned to us." Affective Responsiveness is the ability to express appropriate affect in quantity and quality over a range of situations, and as such, focuses on the familial patterns of responding to affective stimuli (Epstein et al., 1978). The Affective Responsiveness dimension consists of six items (a= .73), an example of which is, "We do not show our love for each other." Affective Involvement is the degree to which a family shows appreciation and values each other's activities, concerns, and interests. "The focus is on how much and in what way family members can show an interest and invest themselves in each other" (Epstein et al., 1978, p. 25). This dimension consists of seven items (a= . 78), an example of which is, "Our family shows interest in each other only when they can get something out of it." Behavior Control is the pattern a family adopts for maintenance and handling of standards of behavior "in physically dangerous situations, situations involving the meeting and expressing of psycho biological needs and drives, and situations involving socializing behaviour both inside and outside the family" (Epstein et al., 1978, p.26). This dimension consists of nine items (a= .73), an example of which is, "Anything goes in our family." Finally, the General Functioning scale consists of 12 items (a= .84) and assesses the overall health/pathology of a family (Epstein et al., 1983;Stevenson-Hinde, & Akister, 1995). An example of an item is, "We don' t get along well together." 43 Psychometric properties of the FAD.
Based on psychometric studies of the FAD, the creators of the screening st ·onnaire have concluded that it has good psychometric properties (Stevensonque 1 ffinde & Akister, 1995). What follows is a description of the psychometric studies involving the FAD.
Factor analyses were conducted on the seven scale, 60-item FAD (Kabacoff, Miller, Bishop, Epstein, & Keitner, 1990). The study by Kabacoff et al. ( 1990) was the first factor analytic study of the FAD to be published. Results indicated that the FAD had a "highly similar factor structure in non-clinical, psychiatric, and medical samples" (Kabacoff et al. , 1990, p. 438). Ninety percent of the FAD's items loaded on the factors that had been hypothesized by the MMFF (Kabacoff et al., 1990). "The confinnatory factors accounted for more test variance than a random model and closely approached the maximum possible variance accounted for by a six-factor principle components model" (Kabacoff et al. , 1990, p. 438). The overall variance accounted for was similar to other measures that correspond to the FAD. Overall, the structure of the FAD corresponds with the theoretical structure of the MMFF (Kabacoff et al., 1990).
Internal reliability was established using Cronbach's Alpha for the original 53item FAD (Epstein et al. , 1983) and then later for the 60-item FAD (Kabacoff et al. , 1990). Kabacoff et al. ( 1990)  Alphas for the medical sample ranged from .69 for Roles to .86 for General functioning. Kabacoff et al. (1990) recommend caution in using the Roles scale since it demonstrated "marginal reliability" (p. 438). Miller et al. (1985) assessed the test-retest reliability of the FAD. The testretest reliabilities for the seven scales ranged from .66 for Problem Solving to .76 for Affective Responsiveness.
The original six scales of the FAD were tested for intercorrelation (Epstein et al., 1983). During the preliminary psychometric testing of the FAD, Epstein et al. (1983) found that the scales were highly intercorrelated. Those items responsible for the high intercorrelations were removed and then used to create the General Functioning scale. With these items removed, partial correlations between the scales approaches zero. The remaining six dimensions are "moderately independent" with correlations ranging from .40 to .60 (Epstein et al., 1983, p. 171). The General Functioning scale is highly intercorrelated with the other six scales but when the variance of the General Functioning scale is partialled out, intercorrelations between the remaining six scales is nearly nonexistent (Epstein et al., 1983;Miller et al., 1985).
For this reason the General Functioning scale acts as a composite scale of the six scales. Findings from a study by Byles et al. (1988) lends support to the construct validity of the General Functioning scale and further supports previous findings of reliability. 45 Discriminative, concurrent and predictive validity were also assessed at the tiJile of the FAD's development and later (Epstein et al., 1983;Miller et al., 1985).
Discriminant validity was originally established by clinicians predicting whether a fumily came from the clinical or non-clinical group (Epstein et al., 1983). The predictions were found to be statistically significant. A later test of discriminative validity (Miller et al., 1985) compared ''the FAD scores of a family with an experienced family therapist's clinical ratings of the same family" (p. 350). Results indicated that clinical ratings corresponded with family scores on six of the seven dimensions (Behavior Control was not significant). Perosa and Perosa ( 1990) found conflicting evidence for the psychometric properties of the FAD. For example, they found evidence to support the convergent validity of the Affective Involvement scale but much weaker support for the Problem Solving, and less support for the Behavior Control and Roles scales. They then found that there were problems with discriminant validity for the FAD. Most notably, Perosa and Perosa found that the Behavior Control, Problem Solving, and Roles scales correlated highly with the Affective Involvement scale "and present the most serious problems with discriminant validity" (p. 861).
Concurrent validity was originally assessed through comparison with the Philadelphia Geriatric Morale Scale and the Locke Wallace Marital Satisfaction Scale (Epstein et al., 1983). Later assessment compared the FAD with the Family Unit Inventory (FUI) and FACES II (Miller et al., 1985). Results from the FUI and FAD comparison provide good evidence for the concurrent validity of the FAD (Byles et 46 1988 ;Miller et al., 1985). The results from the FACES II comparison were not as al., cJear.
The FAD was correlated with a social desirability scale (Miller et al., 1985).
Social desirability was assessed through an administration of the FAD with the Marlowe-Crowne Social Desirability Scale (Miller et al., 1985). Their results indicated that "social desirability does not appear to exert a strong influence on FAD scores" (Miller et al., 1985, p. 347).

Beck Depression Inventory and Hamilton Rating Scale for Depression.
The Beck Depression Inventory (BDI) (Beck, 1967)

and the Hamilton Rating
Scale for Depression (HAM-D) were used to measure depression severity. Both the BDI and the HAM-D were originally designed to monitor severity of depression in patients who had already been diagnosed with depression, not as screening tools (Hotopf, Sharp, & Lewis, 1998;Yonkers & Samson, 2000).
The BDI is a 21-item self-administered questionnaire that lends greater weight to cognitive symptoms and "emphasizes the subjective experience of depression" (Fitzgibbon, Cella, & Sweeney, 1988, p. 374). Each item is scored from 0-3 (from absent or mild to severe) and then all scores are summed to create a severity score as follows: 0-9, minimal; 10-16, mild; 17-29, moderate; and 30-63, severe (Yonkers & Samson, 2000). There is support for the internal consistency of the BDI with a Cronbach's Alpha range of .76 to .95. There is also support for high concurrent validity with other depression severity measures such as the HAM-D. Correlation coefficients for BDI and HAM-D depression severity scores in psychiatric patients ranged from .61 to .86 (Yonkers & Samson, 2000). The Portuguese version of the 47 BDI was used with the Portuguese patients and has been found to be reliable (a= .86) nstetn, . Andrade, Filho, Tung, & Artes, 1999). (Gore The HAM-Dis based on an unstructured clinical interview (Hotopf, Sharp, & Lewis, 1998) and "emphasizes physiological (somatic) components" (Fitzgibbon et al., 1988, p. 374). The scale was initially created with 21 items but has since been modified to include only the first 17 items due to the infrequent occurrence of the last 4 items (Yonkers & Samson, 2000). Items are scored on a scale of0-4 or 0-2 and then summed to create severity scores as follows: 0-7, normal; 8-13, mild; 14-18, moderate; 19-22, severe; 23-50, very severe (Yonkers & Samson, 2000).
There is research that suggests that the two scales measure the same underlying construct (Fitzgibbon et al., 1988;Hawley, Gale, Smith, & Sen, 1998;Hotopf et al., 1998). This issue is important since this study assessed depression severity using the BDI with the Portuguese families and the HAM-D with the American families. Hotopf et al' s (1998) study of 49 patients attending an urban general practice found "considerable overlap" (p. 27) between the BDI and the HAM-D (r = .77). Fitzgibbon et al.'s (1988) study also compared the BDI and the HAM-D with a sample of 108 participants including 36 participants from each of the following groups: hospitalized inpatients with a diagnosis of major depression, inpatients on a renal service, and healthy volunteers. Fitzgibbon et al. (1988) report that the correlations between the BDI and the HAM-D were highest for the total heterogeneous group (r = .90). The correlations for the homogeneous samples were lower yet still significant (e.g., for depressed inpatients, r = .52, p < .001). Findings from a study by Hawley, Gale, 48 Smith, and Sen (1998) suggest that the HAM-D and BDI correlate strongly as indices of severity and when used together do not provide any new information.

First Prediction
The first prediction of this study was that the Portuguese families would have healthy functioning on the dimensions of: (a) Problem Solving; (b) Roles; and (c) Behavior Control. This corresponds to a mean FAD score that is lower than 2.2, 2.3, and 1.9 for each dimension respectively (Miller et al., 1985). Table 2 includes the means and standard deviations of the six dimensions for the Portuguese sample. SD "" .31) dimensions.
As illustrated in Table 3, 81 % of the Portuguese sample was rated as healthy on the Problem Solving dimension while 97% was rated as unhealthy on the Behavior Control dimension. Note. an= 67.

Second Prediction
The second prediction was that the Portuguese families would have unhealthy functioning on the dimensions of: (a) Communication; (b) Affective Responsiveness; and (c) Affective Involvement. This corresponds to a mean FAD score that is higher than 2.2, 2.2, and 2.1 respectively for each dimension (Miller et al., 1985).

Third Prediction
Preliminary analyses.
Prior to conducting any comparisons for the third prediction, there were some methodological issues that were addressed.
The first step was to determine whether there was any relationship between the socio-demographic variables and the FAD scales. Rather than pool the sociodemographic data across the three groups, each comparison was analyzed separately.
One reason to not pool the data was that the same socio-demographic data was not collected across the three groups. Another reason was that the groups differed in respect to their inpatient-outpatient status as well as culture, therefore, it was possible that relationships with the FAD scales could have been influenced differently by these factors. The socio-demographic variables and analyses conducted by cultural group are listed in Table 4.
An ANOV A was conducted with variables such as marital status and family position (i.e., male or female head of household; child of head of household; other relative). Linear regressions were conducted with variabl~s such as age, years of marital status, and number of children. There were no significant differences across these variables. 51 Table 4 . The final analyses conducted were t-tests looking at gender and the FAD scales. There was a significant difference for the Behavior Control dimension in the Portuguese sample, t(43.80) = 2.80,p < 0.008 (two-tailed) 2 , and for Affective

demographic Variables and Analyses Conducted by Group
Responsiveness in the American sample, t(71) = -2.05, p < 0.044 (two-tailed). This suggests that gender relates to at least two of the FAD scales. Because of this relationship, gender was entered into a chi-square test with the variable country (which represents cultural group) to assess its relationship with the independent variable. The 52 results indicated that gender was not related to country, i}(2, N = 177) = 4.41, p < . l l O; therefore, gender was not a confounding variable that would need to be controlled for in future Linear Regressions.
The second methodological issue pertained to the differing inpatient-outpatient status of the samples. It was not the focus of this study to investigate which setting (inpatient vs. outpatient) was more effective in treating depression but whether these samples could be compared. Three sets oft-test comparisons were conducted with the Bonferroni method used to control for multiple comparisons. The results of these ttest comparisons can be found in Table 5.
The first t-test compared the outpatient with the inpatient American FAD scale means. The two groups had significant differences on the Problem Solving, t(269) = 2.61,p < .01 (two-tailed), and Roles dimensions t(269) = 2.38,p < .01 (two-tailed). In both cases, the American outpatient sample had worse family functioning. There was no evidence that the outpatient sample had better family functioning than the inpatient sample. The two samples were either equal or the outpatient sample was worse.
The second set oft-tests first compared the outpatient Portuguese FAD scale means with the American outpatient sample. The Portuguese and American outpatients had significant differences on the Problem Solving, t(264) = 7.09,p < .01 (two-tailed), Behavior Control, t(264) = -9.19,p < .01 (two-tailed), and General Functioning scales t(264) = 2.53, p < .01 (two-tailed). The Portuguese had better functioning on the Problem Solving and General Functioning scales while the Americans had better functioning on the Behavior Control scale.  The results of the t-test comparisons allowed for a cautious assumption that family functioning in inpatient and outpatient depressed samples behaved in a similar enough pattern to permit comparisons across the three groups.
The third issue pertained to the assessment of depression severity in the three groups. The BDI was used with the Portuguese sample and the HAM-D was used with the American sample while there were no depression severity ratings for the Hungarians. A composite depression severity score was created based upon the 55 categorical ratings used with the BDI (1 = minimal, 2= mild, 3 = moderate, 4 = severe) and HAM-D (1 =normal, 2 =mild, 3 =moderate, 4 =severe, 5= very severe). One modification was made to the HAM-D categorical ratings where the severe and very severe categories were grouped into a single severe rating. The minimal and normal categories were treated as similar.
In order to validate the composite depression severity score, each depression measure was entered separately into a linear regression with the FAD scales. There were four steps involved. At each step it was necessary to determine whether depression severity related to family functioning in a consistent manner.
The first step was to take the total score from the BDI and HAM-D and enter them into a linear regression with the FAD scales. The American HAM-D depression severity total scores were significantly related to the Behavior Control scale F(l , 71) = 4.67,p < .03. The Portuguese BDI depression severity total scores were significantly related to all of the FAD scales except for Problem Solving. Table 6 contains the results of these analyses.
The second step was to take the BDI and HAM-D categorical ratings and enter them into a linear regression with the FAD scales. There were no significant differences found. The same was true for the next step, which was to take the modified HAM-D categorical ratings (1 =mild to 4 =severe) and enter them into a linear regression with the FAD scales. There were no significant findings for these linear regressions due to the loss of power from lost variance. The American sample was primarily severely depressed. When the total scores ranging from 0 to 50 were categorized into ratings ranging from 1 to 4, the distribution became truncated. The 56 Portuguese sample had more variance but it too was restricted when it went from total scores to categorical ratings of depression severity. Another important finding was that Portuguese depression severity, as measured by the BDI, was predictive of family functioning. The more severely depressed a family member, the worse that family will be rated on the dimensions of family functioning.
Since depression severity was related to family functioning, it was necessary to determine ifthere were differences in depression severity by cultural group (country).
An ANOV A was conducted and a significant difference was found between country and depression severity F(l, 130) = 74.983, p < 000. The American sample was more depressed than the Portuguese sample. Because depression severity was related to both country (IV) and family functioning (DV), it was a confounding variable that needed to be controlled for in the final analysis to test the third prediction.
Since depression severity was a confounding variable and since the Hungarian sample did not have depression severity ratings, this group could not be included in the final analysis. Without the depression severity ratings, there was no way to control for its effects on family functioning in the Hungarian sample.

Findings.
The third prediction was that the American, Hungarian, and Portuguese samples would have overall unhealthy family functioning on the General Functioning dimension of the FAD. This corresponds to a mean FAD score that is higher than 2.0 58 (Miller et al., 1985). It was predicted that the Portuguese sample would have the worst family functioning and that the Hungarian sample would have the lowest rating of the three groups and thus the healthiest score.
As a result of the preliminary analyses, the Hungarian sample was dropped from this analysis. With the subtraction of the Hungarian sample, the third prediction became that the Portuguese would have worse family functioning than the American sample. In order to measure this prediction depression severity was controlled for as a confounding variable.
Two linear regressions were conducted. One analysis was conducted with country and depression severity by the FAD scales and the other with country by the FAD scales. The reason for having conducted two regressions was to be able to detennine how much of the variance was explained by the cultural group versus depression severity as well as which variable could have been responsible for any significant differences. The results of these analyses can be found in Table 7.
There was no significant difference between groups on the General Functioning Scale, F(I, 139) = .235,p < .629. Both the Portuguese and American groups scored above the mean cut-off (i.e., unhealthy) for General Functioning with means of2. l 7 (.37) and 2.20 (.50) respectively. Cultural group accounted for less than 1% of the variance whereas depression severity accounted for 3.8% of the variance and was predictive of General Functioning, t(l31) = 2.286, p < .05.
The findings indicate that both the American and Portuguese cultural groups scored in the unhealthy range on the General Functioning scale. Also, neither group was healthier than the other on this scale therefore the third prediction was not 59 supported. Depression severity, rather than cultural group, was a better predictor of the overall health/dysfunction of a family as measured by the General Functioning scale. Table 7 Regression Analysis for Cultural Group and Depression Severity on the General

Functioning scale
Variable B Step 1 Country 3.613 Step 2 The predictions were made on the premise that the FAD was cross-culturally applicable and then by looking at the items of the FAD and projecting how the Portuguese would respond. As evidenced by the earlier description of the Portuguese families, some of the values in the Portuguese culture run counter to those within which the FAD was created. In Portuguese families, for example, communication tends to flow in one direction, from the top -down (e.g., from father to mother and from mother to children). Communication tends to be clear and direct from the topdown but indirect from the bottom -up. Communication, as defined by Epstein et al. (1978), is optimal when it is clear and direct. This is usually true of the father or mother speaking to the children but is not the case when children and even mother (at times) communicates with the male head of household in a Portuguese family. As a result, it was believed that certain dimensions, such as Communication, that might be considered healthy within the Portuguese culture would be assessed as unhealthy by an instrument developed in a Western cultural perspective. Recall that Morris (1990) had postulated that certain aspects of Hawaiian and Japanese culture that were valued run counter to the American values inherent in the definition of the dimensions of the MMFF. Specifically, Morris (1990)  The findings were congruent with studies on depression and family functioning between clinical and non-clinical samples. Previous research has shown that families with a depressed member have unhealthy family functioning (Keitner et al., 1987;Miller et al., 1986;Sawyer et al., 1988 Solving dimension (see Epstein, Bishop, & Levin, 1978). As predicted, the Portuguese were rated as having healthy functioning on the Problem Solving dimension.
This finding may be due to the immigrant status of the Portuguese sample.
Leaving one' s country and acculturating to a new society may have forced this group to become problem solvers in order to adapt to a new culture. It may also be a cultural 62 difference in regards to living (and perhaps surviving) in a different sociopolitical and economic environment prior to immigrating. Another possibility could be that there was a selection bias difference between those Portuguese families that immigrated and those that remained in the country of origin. Perhaps those families that choose to immigrate had better problem solving skills to begin with. One question that could be answered with future research is whether a link between immigration and problem solving exists solely for the Portuguese or if it is common among most immigrant groups.
Another explanation may be that depression does not impact the Portuguese family's problem solving when the depressed member is the female head of household. Portuguese families are patriarchal, therefore, the male head of household is the person primarily responsible for making decisions and solving problems. The female head of household will manage the home but even in this instance, the male will make the final decision when necessary. The majority of respondents were female and the healthy rating on problem solving functioning may reflect the idea that the women are somewhat protected from making the final family decisions, resulting in an indirect experience of the pressures associated with problem solving for a family unit.
Following the logic ofthis explanation, it would seem likely that the male respondents would have had low ratings on problem solving. This would have been reflective of the father' s experience ofloss of power within his family. If the male head of household is the depressed member in a Portuguese family he may be unable to fulfill his role as the decision maker and as a result, that family' s ability to solve 63 problems would suffer. Any movement by another family member to fill the role of problem solver could be seen as a move to undermine the authority of the male head of household. This could also serve to worsen the male head of household's depression. Failure to fulfill one's family duties and responsibilities would be realized when other family members began to function as the head of the family. What could make this situation worse is if a family member that fills in for the male head of household does a good job. However possible, this study did not seek to provide any evidence for the existence of gender-specific relationships with the FAD dimensions.
Also of note was that nearly all the Portuguese respondents were rated as unhealthy on the Behavior Control dimension when it had been predicted that they would have scored as healthy. Most of the Portuguese respondents had children who were being raised in the United States, which provided a basis for the following explanation. It was believed that the Portuguese families had a rigid behavior control style (see Epstein et al., 1978) in part due to their patriarchal family system. The style offamily behavior management and socialization used by Portuguese families would be clear to its members and should have been consistent enough that the respondents answering the survey would have felt that their family's behavior was being managed properly. The rigid style is neither the best style nor is it the worst (Epstein et al., 1978). The children of these Portuguese immigrants were growing up in an individualistic culture which would value the flexible behavior control style (purported to be the best of the four styles) but which would rebel against a rigid style.
It is possible that children who have been living in the United States and exposed to depressed, that they would be rated as unhealthy. It was also believed that the Portuguese culture differed the most in terms of its value system from that of the FAD and thus they would have had the worst family functioning.
The Hungarian sample was removed from the final analysis and the prediction was modified to reflect this fact. The third prediction was supported in that both the Portuguese and American samples were rated as unhealthy on the General Functioning Scale. The Portuguese sample was not significantly worse then the American sample.
The results are consistent with past research in the literature. Previous studies have consistently supported the unhealthy family functioning of families with a depressed member (Keitner et al., 1987;Miller et al., 1986;Sawyer et al., 1988).

Implications
The MMFF is based upon aspects of systems theory including: a family is interconnected and one member's behavior impacts on other members' behavior; a family' s structure and organization as important in determining the behavior of family members; and the assumption that one family member's behavior cannot be understood in isolation from the family context within which it exists (Epstein et al., 1978). The findings of this study were congruent with the aspects of systems theory underlying the MMFF.
Findings from this study support a relationship between depression severity as measured by the BDI and family functioning as measured by the FAD within the Portuguese community. This finding adds to a growing body ofliterature that reports on the existence ofthis relationship within other cultural groups as well as findings from research suggesting that level of depression severity is a better predictor of family functioning than a DSM diagnosis . 66 Other research has provided evidence that family functioning improves as depression remits but not to the point where the families experience less dysfunction than non-clinical families (Keitner et al., 1987), that positive changes in overall family functioning were associated with faster recovery (Keitner et al., 1987;, and that patients who view their families as having healthy functioning during the acute phase of major depression have a significantly greater chance for recovery (Keitner et al., 1995).
Depression severity is predictive of family functioning and the depressed family member's best chance for recovery is to have their family's functioning specifically addressed. Based upon these :findings, a strong case can be made for a combination of individual and family treatment versus individual-centered treatment.
Addressing the individual's depression may help the situation but it will not improve a family's functioning to a point that is on par with non-clinical families. This relates to the interconnectedness of a family in that addressing one member's depression separately from the shift that has occurred in a family system may not be sufficient for effecting change. A depressed family member does not exist in isolation from the rest of his or her family but is part of a whole that has had its resources taxed.
Providing treatment for the depressed family member alone may not be sufficient or as efficient as treating the depression while also addressing the family system. It was mentioned earlier that it is unclear whether depression leads to poor family functioning or poor family functioning leads to depression. To this point, research findings support a comorbid relationship between the two variables . It would seemingly behoove researchers to investigate multi-67 treatment approaches that deal with both the individual and family levels yet research has primarily focused on the efficacy of treatment approaches such as interpersonal therapy versus cognitive behavior therapy and whether or not medication was involved. Ultimately this may not be much of a surprise when considering that the research was primarily conducted within an individualistic framework. Individualistic cultures view individuals as separate from collectives or groups. It would not make as much sense to approach the problem of depression in an individual by addressing their family system. An argument could be made that a combination of individual and family treatment might not be appropriate for all individuals and families. This combination may run counter to an individualistic value system where an individual would be more inclined to perceive a problem as his or her own and wanting to deal with it in that way through individual treatment and not family treatment (and may be urged by family members to do so). In the same scenario, an individual from a collectivist culture may find himself or herself receiving family treatment as a result of their family seeking help as a whole for one family member"s problems. In both instances, positive results may be influenced by matching treatment modality with patient values.
While there are effective methods for the treatment of depression in individuals, the findings of this study suggest that clinicians may be shortchanging the individuals they treat by not addressing the entire family' s functioning and giving them the best chance for recovery and protection against remission. Until the relationship is better understood, clinical interventions that treat the comorbidity of family functioning and depression in a family as a whole should be considered. Future 68 1 research should investigate treatment efficacy between the different approaches (i.e., treating the individual alone, treating a family alone, treating both simultaneously) while considering any relationship or fit issues between the client's cultural value system (individualistic/collectivistic) with the treatment approaches. Based upon the evidence presented herein and in other research, it may be the case that treating both the individual and family simultaneously would result in quicker recovery with lower rates of remission regardless of the cultural value system.
The direct clinical importance for mental health service providers working with the Portuguese population is that assessment of an individual as depressed may act as an indicator for services to be provided not only to the individual but to the entire family. Knowing that the depression severity of a family member reflects upon his or her family system as a whole provides for a more appropriate treatment plan.
The difficulty will still remain in engaging the entire family in the therapeutic process.
This new knowledge has two effects. One, it would provide clinicians with a research-based argument to present to Portuguese immigrants when advising for family therapy while potentially increasing their comfort level by offering an approach that is congruent with a collectivist-based value system. Secondly, it suggests to clinicians that when treating depression, it would be optimal to treat the entire family and use family therapy to build supports that would protect against and mitigate future depressive episodes while possibly decreasing the amount of time for recovery from the current episode. 69

Limitations
There were a number of limitations to this study including the creators of the FAD having stressed caution in interpreting any results with cultural populations outside of North America. This study was also missing key demographic information including yearly income and years of education. Because the data was collected at two separate sites, the collection of demographic information was inconsistent and incomplete.
There were other issues that needed to be addressed prior to running any analyses. One such issue was related to basing the health or pathology of a family's functioning on the reports of one family member, in this case the depressed member.
The FAD is a self-report measure, therefore it is one family member's perception of his or her family's functioning. This limitation notwithstanding,   If the Portuguese sample has experienced little acculturation, then FAD dimensions such as Affective Responsiveness might be rated as unhealthy. This dimension is based on the premise that families show a wide and appropriate range of affection towards one another as situations dictate (see Epstein, Bishop, & Levin, 1978). This may not be true for the Portuguese families where members are concerned with acting in a manner that does not shame their family and the father.
The Western value is based on individual need and attention whereas the Portuguese value is based on maintaining the "face" of one's family as a whole when dealing with the situation. For example, the Portuguese father may not respond to the individual emotional needs of a family member but rather, would take steps to deal with the situation and respond to that family member in a manner that maintains the family's "face". If the Portuguese sample has acculturated to the American culture, then these families might be rated as having healthy family functioning on the Affective Responsiveness dimension. The possibility also exists that this FAD dimension would have been rated as healthy with Portuguese families who had just arrived or were 71 living in Portugal and as unhealthy for Portuguese immigrants who had acculturated and been living in the U.S. for some time. The point is that without having measured the level of acculturation, it is difficult to offer possible explanations for the results and whether the FAD is cross-culturally applicable.
Not knowing the level of acculturation and behavioral shift weakens the explanations that can be offered for discussing the family functioning ratings. Future research should include an acculturation measure to see how it correlates with family functioning as measured by the FAD. Another solution may be to use the FAD and a measure of individualism/collectivism in a cross-national study between Portuguese immigrants in the United States and those still living in Portugal.

Measurement.
Prior to conducting any comparisons, there were some methodological issues that needed to be addressed. One methodological limitation of this study pertained to the differing inpatient and outpatient status of the sample. The Portuguese sample consisted of outpatient respondents while the American and Hungarian samples consisted of inpatient respondents. The Portuguese ·sample was assessed prior to receiving treatment while the Hungarian and American samples were assessed for family functioning during their inpatient stays. To date, research comparing inpatient and outpatient services has been inconclusive in determining which is better. It was also not the focus ofthis study to investigate which setting (inpatient vs. outpatient) was more effective in treating depression but rather whether these samples could be compared.

72
Because this study involved a comparison of group means of outpatient and inpatient samples, it was necessary to find some way to support such a comparison.
The results of the analyses allowed for a cautious assumption that the inpatient and outpatient samples were similar enough to proceed with testing the predictions.
Despite the analyses that were conducted, the different inpatient-outpatient sample status remained a limitation of this study.
Another limitation of the study was that different measures were used to assess level of depression. Respondents from all three groups did meet the DSM-11/-R criteria for depression. There is also research that suggests that the two scales measure the same underlying construct and correlate strongly as indices of severity such that when used together, they do not provide any new information (Fitzgibbon et al., 1988;Hawley, Gale, Smith, & Sen, 1998;Hotopf et al., 1998). The BDI was used with the Portuguese sample and the HAM-D was used with the American sample while there were no depression severity ratings for the Hungarians. The lack of depression severity ratings proved to be critical since it was found that depression severity was a confounding variable that needed to be controlled for. As a result, the Hungarian sample was withdrawn from the final analysis weakening the third prediction.
The final limitation had to do with the use of cut-offs established on U.S. and Canadian samples to determine the health/dysfunction of the Portuguese sample's family functioning. This sample might not be clinically unhealthy in a Portuguese context but within an American context they were. There is no normative data for the Portuguese population but it is also very difficult to collect such data for a group of immigrants who are acculturating to a new society. The argument could be made that 73 such a group would not have a normative representation of health/dysfunction cut-off scores since they are changing and adapting to a new society (i.e., experiencing behavioral shifts). It would also seem impractical to collect such data since it could be obsolete by the time it was completely analyzed and would only represent a small group for an unknown limited amount of time.

Future Directions
The focus of this study was on the family functioning of families who have a member diagnosed with depression according to DSM-111-R critera. Findings from this study support a relationship between depression severity as measured by the BDI and family functioning as measured by the FAD within the Portuguese immigrant community. The direct practical importance for mental health service providers working with Portuguese families is that if they see an individual who is depressed then there may be a clinical indication for services to be provided to the entire family.
Replication studies of the relationship between depression severity and family functioning within the (un)acculturated, (non)clinical Portuguese community in the United States using measures other than and includfug the BDI and FAD are warranted.