help

Pacific Islands Families Study: The Association of Infant Health Risk Indicators and Acculturation of Pacific Island Mothers Living in New Zealand

Jim Borrows1 , Maynard Williams1 , Philip Schluter2 , Janis Paterson3 , and S. Langitoto Helu4

Journal of Cross-Cultural Psychology 42(5) 699–724 © The Author(s) 2011 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/0022022110362750

Abstract The Pacific Islands Families study follows a cohort of 1,398 Pacific infants born in Auckland, New Zealand. This article examines associations between maternal acculturation, measured by an abbreviated version of the General Ethnicity Questionnaire, and selected infant and maternal health risk indicators. Findings reveal that those with strong alignment to Pacific culture had significantly better infant and maternal risk factor outcomes than those with weak cultural alignment. In terms of Berry’s classical acculturation model, separators had the best infant and maternal outcomes; integrators had reasonable infant and maternal outcomes, while assimilators and marginalisors appeared to have the poorest infant and maternal outcomes. These findings suggest that retaining strong cultural links for Pacific immigrants is likely to have positive health benefits. Keywords acculturation, infant health risk, Pacific health, culture and health Introduction and Background People of Pacific ethnicities resident in New Zealand are overrepresented in many adverse social and health statistics. Pacific peoples generally fare worse than the New Zealand population as a whole in statistics relating to health, unemployment, housing, crime, income, education, and nutrition (Bathgate, Donnell, & Mitikulena, 1994; Cook, Didham, & Khawaja, 1999). Despite the growth and employment opportunities in New Zealand, Pacific people are more likely to be living in poor circumstances with restricted access to higher education, home ownership, and access to functional amenities such as automobiles and telephones. Such statistics have significant consequences for Pacific families given that socioeconomic disadvantage has been consistently linked with negative health outcomes (Chen, 2004; Power, 2002). Specifically, the raison d’etre for the Pacific Island Families (PIF) Study, the health of Pacific families, and especially their infants continues to be an issue of major concern for New Zealanders. The total neonatal death rate for Pacific infants at 4.7 per 1,000 live births is twice that of the rate for New Zealanders of European ancestry but still less than the 5.0 of the indigenous Maori population (New Zealand Health Information Service, 2006). Similarly, Pacific infants have high rates of hospitalization, particularly for respiratory illnesses (Ministry of Health & Ministry of Pacific Island Affairs, 2004), and present at hospital with higher severity of illness than other New Zealand children (Grant et al., 2001). These negative infant statistics are somewhat perplexing, especially in a country where primary health care services are available at low cost (free for pre-schoolers) and emergency and hospital care services, including birthing services, are provided free of charge. Also, New Zealand (Abel, Park, Tipene-Leach, Finau, & Lennan, 2001) and Pacific ethnographies (Lukere & Jolly, 2002) show that neonatal and infant care practices are not directly contradictory to accepted Western infant care practices. In Pacific Island settings, themselves changed by 200 years of Western contact, the family is perceived as central in providing traditional protocols for support and advice to ensure infant well-being. Explanation for the current Pacific child health circumstances is likely driven by multiple variables including the immigration process itself. Previous research from the PIF study demonstrated that acculturative orientation had a persistent association with aspects of health status and behaviour for cohort participants (e.g., Abbott & Williams, 2006; Low et al., 2005; Paterson, Feehan, Butler, Williams, & Cowley-Malcolm, 2007), hence the emphasis in this article on testing the association between maternal acculturation and infant and maternal health risk factors. Culture, Health, and Acculturation The interrelationship between culture and health, including associated psychological processes, has been a recurrent theme in the social science literature over much of the last century (Helman, 2000; Sam, 2006a; Stroebe & Stroebe, 1995; U.S. Department of Health and Human Services, 2001). There is now acceptance in the medical and health professional domains that culture should be acknowledged as an important determinant of health status (Corin, 1994; Snowden, 2005; Spector, 2002; U.S. Department of Health and Human Services, 2001) and that concepts derived from anthropologic and cross-cultural research may provide an alternative framework for identifying health issues that require resolution (Kleinman, Eisenberg, & Good, 1978; Savage, 2000). In particular, there is some agreement that many people from minority cultures may not have faith in, or necessarily benefit from, the medical interventions that are being offered by the host society (MacLachlan, 1997). Also recognized is the importance of the interrelationship between migration and health, including seminal New Zealand/Pacific migration studies (Stanhope & Prior, 1976), early international studies (Carballo, Divino, & Zeric, 1998; Ostbye, Welby, Prior, Salmond, & Stokes, 1989), and more recent studies aimed at explaining the link between migration and health (Sam, 2006a). That is, the realization that the well-being of a migrant group is determined by interlinking factors that relate to the society of origin, the migration itself, and the society of resettlement. All three sets of factors need to be considered if one seeks to reduce or merely to understand the level of health disorder in any immigrant group. Despite the recognition of the importance of Borrows et al. 701 culture and migration in determining health status and the explanatory acculturation/health hypotheses that this has generated (Carballo et al., 1998; Sam, 2006a), there have been few empirical attempts to link health with both migration and culture in relation to other demographic, social, and psychological factors operating in given communities in New Zealand or international studies (Snowden, 2005). However, it is now clear that migration at an individual level is a significant life event for individuals impacting on subsequent health behaviour and outcomes. Closely related to culture and migration is the concept of acculturation—that is, “culture change that is initiated by the conjunction of two or more autonomous culture systems” (Social Science Research Council, 1954, as cited in Berry, Poortinga, Segall, & Dasen, 2002, p. 350). The social psychology literature is replete with alternative models of the acculturative process, most of which are multidimensional, involving numerous topics and factors (Stanley, 2003). These multidimensional topics range from those at the personal level, such as personality qualities and psychological adjustment (Ward & Leon, 2004), language retention and community socialization, and external acculturation drivers such as migration experience, micro- and macro-societal policies, and regional setting (Persky & Birman, 2005). Outside of these models, but still incorporating multidimensionality, are the two most common models of acculturation theory: unidirectional and bidirectional models of acculturation. Berry restated Redfield and colleagues’ hypothesis that acculturative adaptations lead to culture changes in either or both of the migrating and host society groups. He went on further to note that it is not inevitable that intergroup contact proceeds uniformly through sequential to ultimate assimilation as there are many other ways of going about it or indeed is potentially bidirectional and reciprocal (Berry, 2006). Such insights generated by this bidirectional model challenges the ethnic melting-pot assumptions and promotes exploration and resolution of political sensitivities among ethnicities (Flannery, Reise, & Jiajuan, 2001). These observations by Berry, Sam, and others, which hint at multiple individual and group acculturation strategies, have been complemented more recently by Boski, who calls for the development of a theoretical model of integration, a key concept in the psychology of acculturation, in which five meanings for this concept identified in the existing literature are positioned as in-depth directed layers of the bicultural psyche (Boski, 2008). That is, the subtleties in the acculturation process at the group and individual level deserve further and more detailed examination. There are many studies that have examined acculturation strategies in nondominant groups. In most studies, preference for integration is expressed over other acculturation strategies, although notable exceptions with Turks both in Germany and in Canada, and in Hispanic immigrant women in the United States, have been cited (Ataca & Berry, 2002; Berry, 2006; Jones, Bond, Gardner, & Hernandez, 2002). All these recent contributions that counter the assimilation and melting-pot models could be seen as underpinning Pacific community perspectives on cultural maintenance within New Zealand society. In New Zealand, there is widespread official government dogma and minority community perception that cultural maintenance is important to health outcomes and that culturally specific information for minority groups on which to base optimal policy and services is necessary. The untested assumption is that such an approach will lead to improved health and social outcomes for Pacific peoples. An alternative “popular hypothesis” in New Zealand would more likely support international perspectives and studies cited above that would expect more positive health outcomes for those effectively embedded in mainstream culture than for those embedded in Pacific culture or those marginalized from both cultures. This dominant cultural and official “cultural maintenance” viewpoint is politically persuasive in New Zealand and as a result became the focus of refutation or support in terms of our working hypothesis outlined as the second aim for this study presented below. Based on all these considerations, we applied Berry’s acculturation model to the relationships between acculturation and health, in this case operationalised as poor outcomes for maternal and 702 Journal of Cross-Cultural Psychology 42(5) infant health risk factors. Thus, in the context of understanding the process and outcomes of acculturation strategies adopted by Pacific families, this study had two principal aims: namely, to (a) investigate the association between mother and infant health variables that might act as infant risk indicators and adaptation to living in New Zealand and (b) test the New Zealand view that strong cultural alignment to the original Pacific culture is associated with significantly better outcomes in terms of maternal and infant health risk factors and that weak cultural alignment is associated with significantly poorer outcomes in terms of maternal and infant health risk factors. For reasons outlined in the Method section, an abbreviated version of the General Ethnicity Questionnaire (GEQ; Tsai, Ying, & Lee, 2000) acculturation measurement instrument was employed. As a result, a secondary aim was to establish the validity and reliability of the modified instrument. Migration and Pacific People in Contemporary New Zealand Society To give a context to this study, it is necessary to describe the place played in New Zealand’s migration history by people of the Pacific Islands (as distinct from indigenous Maori descent) and their place in contemporary society. Polynesian settlement of the Pacific was completed around 1200-1300 AD when Te Ika o Maui (the mythical fish of Maui), the North Island of New Zealand, was the last Pacific archipelago to be discovered and settled by the ancient Polynesians (Prickett, 2001). These Polynesian ancestors became the New Zealand indigenous Maori. Major European settlement, and subsequent colonization, commenced from the late 18th century. Polynesian post-Maori contacts in the 18th and 19th centuries were limited, and at the 1945 New Zealand Census of Population and Dwellings, only about 2,000 people were recorded as being of Pacific origin. A second great wave of Polynesian migration took place in the relatively short period between the 1950s and 1980s, when Pacific peoples arrived from the islands of Samoa, Tonga, Cook Islands, Niue, Fiji, and the Tokelaus. This modern Polynesian migration was based principally on opportunity provided by largely economic imperatives in New Zealand (Macpherson, Spoonley, & Anae, 2001) or economic sustainability of small island groups such as the Tokelaus (Prior, Welby, Ostbye, Salmond, & Stokes, 1987; Salmond, Joseph, Prior, Stanley, & Wessen, 1985), supplemented more recently by matters relating to renewing or continuing links of kinship and family. Currently, Pacific peoples are a very significant and growing proportion of New Zealand’s population. More than 6% (231,801 people) in New Zealand were of Pacific ethnicity at the time of the 2001 Census (Statistics New Zealand—Te Tari Tatau, 2002a), and Pacific people are projected to make up more than 8% of the population by 2021 (Statistics New Zealand—Te Tari Tatau, 2005). The biggest concentration of Pacific people is in Auckland, New Zealand’s largest metropolitan area. Sixty percent of people of Pacific ethnicity were born in New Zealand; of those born overseas, 40% had arrived in New Zealand by 1981 and 30% between 1981 and 1990 (Statistics New Zealand—Te Tari Tatau, 2002a). This latest migration of Pacific people influences the nature of both New Zealand and the home island societies. For example, in the islands, it is significant in terms of reducing the overall population and in providing economic support to home communities by way of individual and family remittances to relatives. Table 1 illustrates the large proportion of Pacific people residing in New Zealand in relation to their respective home island populations. Since the migration wave of the late 20th century, Pacific people have actively participated in the New Zealand economy and society. In economic terms, Pacific people have relatively high labour force participation rates, particularly in the manufacturing sector. This sector has declined since the mid-1980s as a proportion of total employment but has been offset with Pacific people employment participation in the growing consumer service industries (such as hotels, restaurants, and retail) and the employment of younger people in more skilled technical and professional Borrows et al. 703 occupations (Statistics New Zealand—Te Tari Tatau, 2002b). However, people of Pacific ethnicities remain underrepresented in managerial and professional occupations yet overrepresented in trades and elementary occupations. Overall current labour force participation rates for people of Pacific ethnicities are at 62.9%, lower than the national rate of 68.5%, and unemployment rates are at 6.9%, higher than the national rate of 3.7% (Department of Labour—Te Tari Mahi, 2007). Maori rates for 2007 in labour force participation and unemployment are 67.6% and 7.6%, respectively. In terms of demography, Pacific people living in New Zealand have a relatively young age structure and a high fertility rate. While people of Pacific ethnicities currently have a lower life expectancy than the total population, it is higher than that for the indigenous Maori population (Cook et al., 1999). The Pacific population is proportionately more likely than the national population to be in the lower income bands, even after age standardization. Employment and income aside, the degree to which people of Pacific ethnicity participate in New Zealand society, and are hence not marginalized in ethnic group terms, is illustrated in Figure 1, with the number of births resulting from interethnic marriage between three of the major four ethnic groups in New Zealand. Interethnic marriage between the Pacific and Asian ethnic groups is not as common. Geographically, Pacific peoples are principally resident in major urban areas. Eighty-one percent of peoples of Pacific ethnicities reside in the major urban areas, including the Auckland Region (66.0%), Wellington (12.4%), Christchurch (3.6%), and Hamilton (1.9%). No other New Zealand city, town, or district had more than 4,000 residents of Pacific ethnicity (Statistics New Zealand—Te Tari Tatau, 2006a). Choice of residential locations was driven by migration history and economic imperatives mainly to low socioeconomic status neighbourhoods that have persisted along with maintenance of kinship and family ties often irrespective of changes in standard of living. There was no formal overt or covert official state or local determination for spatial distribution or segregation—unlike that experienced in some migration histories elsewhere (Musterd, Breebaart, & Ostendorf, 1998). Consequently, the New Zealand location of Pacific families remains concentrated in relatively deprived mixed-ethnicity urban areas, with the major concentrations in the sprawling central, western, and southern suburbs of greater metropolitan Auckland and in Wellington. At the 2006 New Zealand Census, 14% of the Auckland region’s population was of Pacific descent, compared with European (55%), Asian (18%), and Maori (11%). In terms of the PIF study at recruitment, all participants in the study were resident in the catchment area for Middlemore Hospital, the principal birthing hospital for the Counties Manukau District Health Board (CMDHB). This catchment area is located predominantly in Manukau City, South Auckland. In 2005, just under half the CMDHB population was made up of European and other ethnicities (48%), with significant minorities being Pacific (20%), Maori (17%), and Asian (15%). More than a third (36%) of all Pacific people in New Zealand live in CMDHB (2008). Table 1. Pacific People in New Zealand (New Zealand 2001 Census) and Pacific Islands of Origin (South Pacific Commission 2001 Estimate) New Zealand Island of Origin PIF Cohort Pacific Population Population Population N % N N % Samoan 115,017 48.6 170,900 647 52.9 Tongan 40,716 17.2 99,400 287 23.5 Cook Island Maori 52,569 22.2 19,300 229 18.7 Niuea n 20,148 8.5 5,400 59 4.8 704 Journal of Cross-Cultural Psychology 42(5) The CMDHB area comprises a highly diversified community in a country (New Zealand) that by international standards ranks as a moderate to highly diversified society, ranking equivalent to the United States, ahead of Australia, and behind only Canada and Israel. The authors of a recent international study on immigrant youth claim that the “diversity index” portrays the degree of cultural pluralism present in society and reflects the potential for interethnic and interlinguistic contacts that people experience in a given society (Berry et al., 2006). Pacific peoples live in a positively oriented multicultural society with ample exposure to other cultures, including the majority culture, both in work and play, with a significant degree of intermarriage with people of European and indigenous Maori ancestry (Figure 1). Compared with some migrant communities elsewhere and some rural indigenous communities in New Zealand (Maori) and Australia (Australian Aborigines), people of Pacific ethnicities who arrived in New Zealand as late 20th-century migrants have had relatively high involvement in the New Zealand labour force, have located in multi-ethnic urban (if poorer) areas, and have significant social, sporting, and cultural links with the wider New Zealand society. They provide another cultural dimension alongside indigenous urban Maori, Pakeha (New Zealanders of European ancestry), and people of Asian ethnicities in a rapidly evolving but largely empathetic society that has a moderately positive attitude toward the principles of multiculturalism and integration as preferred acculturation strategies (Sang & Ward, 2006). Method Participants Data were gathered as part of the PIF study, a longitudinal investigation of a cohort of 1,398 infants (22 pairs of twins) born at Middlemore Hospital, CMDHB, South Auckland, New Zealand during the year 2000. Middlemore Hospital was chosen as the recruitment site as it has the largest Figure 1. Pacific Children’s Live Births 2003: Distribution by Ethnicity (Data From Statistics New Zealand—Te Tari Tatau, 2004) Borrows et al. 705 number of Pacific births in New Zealand and is representative of the major Pacific ethnic groups (Samoan, Cook Island Maori, and Tongan). It was estimated that a cohort of 1,000 would provide sufficient statistical power to detect moderate to large differences after stratification for major Pacific ethnic groups and other key variables. Eligibility criteria included having at least one parent who self-identified as being of Pacific ethnicity and a New Zealand permanent resident. Thus, non-Pacific mothers (including indigenous Maori) were eligible for the study in cases where the infant’s father was of Pacific descent. Detailed information about the cohort and procedures is described elsewhere (Paterson et al., 2006; Paterson et al., 2008). All procedures and interview protocols for the PIF study were granted ethical approval from the National Ethics Committee. PIF Study Instrument A wide range of demographic, social, psychological, and health information was gathered in relation to the newborn infant and his or her parents at 6 weeks postpartum using individual interviews of mothers conducted in their homes. Items elicited details relating to household structure, education and employment, ethnic and cultural identification, length of residency in New Zealand, language use and fluency, child health and development, infant nutrition, infant sleeping, use of health services (such as family planning and pregnancy), childcare arrangements, parent childhood experiences, parental health and mental health, partner relationships, family finances, housing, transport, and church and leisure activities. In all, information on 941 variables of interest was gathered in the home interview, which lasted approximately 1.5 hours. Acculturation Measure Despite the importance of acculturation and its relevance for policy makers in plural societies, assessment of this concept remains problematic and no widely accepted measurement methods are available (Arends-Toth & van de Vijver, 2006). The acculturation measure chosen for the PIF study was an adaptation of the GEQ (Tsai et al., 2000). This scale included elements consistent with the current status of theory on the psychological responses to acculturation (Arends-Toth & van de Vijver, 2006; Berry, 2006; Cabassa, 2003). Moreover, the GEQ embodies elements of individual perceptions of characteristics of the island societies of origin and the New Zealand receiving society, it measured adoption and maintenance strategies from a bidimensional perspective, and it has been widely applied internationally. Although questioned more recently (Kang, 2006), a bidimensional scale was chosen because: Linear assimilation models continue to dominate public health research despite the availability of more complex acculturation theories that propose multidimensional frameworks, reciprocal interactions between the individual and the environment, and other acculturative processes and . . . the rare use of multidimensional acculturation measures and models has inhibited a more comprehensive understanding of the association between specific components of acculturation and particular health outcomes. (Abraído-Lanza, Armbrister, Flórez, & Aguirre, 2006, p. 1) With a demanding and lengthy study questionnaire, scales had to be abbreviated and adapted so that we would not lose participants in future measurement waves. To suit the specific purposes of the PIF study, the scale of Tsai et al. (2000) was further abbreviated and adapted, thereby developing the New Zealand (NZACCULT) and Pacific (PIACCULT) versions of the GEQ (Appendix). The original 38-item GEQ scale was reduced to 11 items on a pragmatic minimalist basis but included key items reflecting five of the six specific cultural dimensions identified by 706 Journal of Cross-Cultural Psychology 42(5) Tsai et al. (2000) and reflected the two fundamental issues of interest: (a) maintaining one’s heritage, culture, and identity and (b) relative preference for having contact with, and participating in, the larger society (Berry, 2006). Also important in selecting items was a concentration on items that were likely to apply to the complete respondent population (Van Nieuwenhuizen, Schene, Koeter, & Huxley, 2001). Included were questions relating to the specific cultural dimensions of language, social affiliation, activities, exposure in daily living, and food. The sixth dimension, pride in culture, was excluded as it was considered that this aspect was better accommodated by other questions in the measure that reflected and accommodated some aspects of this dimension. Some specific items were excluded because they bore little relevance to Pacific life in New Zealand, for example listening to radio in a Pacific language, as such services were not widely available at that time. We thus excluded items that seemed from knowledge of mainstream New Zealand culture and New Zealand Pacific culture as having less relevance (face validity) than for the American/Chinese population for which the GEQ scale was originally designed. The scale was further adapted to include a small number of items considered of particular cultural relevance in New Zealand. Two questions relating to social affiliation but not included as such in the original GEQ scale were exploring issues relating to contact with Pacific family and relatives and attendance at church, both of which were considered important in a Pacific context in New Zealand society. Similarly, inclusion of sport as a particular recreation was included because of the perceived importance of Pacific youth involvement in New Zealand sport and its importance in the context of the wider New Zealand society. The PIF study research group believed that measurement of acculturation as used in crosscultural psychology, but distinct from qualitative anthropologically and socially oriented cultural descriptions, was an important and relevant concept in the context of the longitudinal study on which we were embarking. This was an additional consideration in adapting an existing validated measure that included relevant domains and against which we had an existing reference standard to compare. Because of project constraints, it was not possible to pilot the measure we developed against the longer version of the GEQ—hence the inclusion in this article of the retrospective reliability and validity comparisons. The measure was developed to make it appropriate and relevant to Pacific peoples and New Zealand society as a whole and so as to provide reasonable approximations of the acculturation process for this population. Clear face validity for this combined scale was revealed by both the pre-study participant focus groups and the advice received from the study’s Pacific Advisory Board—this advice being integral to all substantive decisions on study content. Subsequent results from other PIF research (Abbott & Williams, 2006; Low et al., 2005; Paterson et al., 2007) demonstrated that the acculturation variable measured from these scales was a persistently strong associate for a range of health and social indicators. Assessment of Acculturation This was undertaken using the classical adaptation and acculturation strategies model described by Berry (1980, 2003, 2006). The model describes four distinct dimensions, with two parts to each dimension depending on whether the acculturation strategy is freely adopted by the individual or minority group or imposed by the dominant culture. The strategies are as follows: (a) Separation (minority group or individual choice) or segregation (dominant society preference or force), (b) integration (minority group or individual choice) or multiculturalism/ pluralism (dominant society preference or force), (c) assimilation (minority group or individual choice) or melting pot/pressure cooker (dominant society preference or force), and (d) marginalization/deculturation (minority group or individual choice) or exclusion/ethnocide (dominant society preference or force). Borrows et al. 707 Selection of Maternal and Infant Risk Factors To assess the association of acculturation and maternal and infant risk factors likely to result in poor infant health outcomes, a variety of relevant maternal and infant variables that may provide insights into such links were extracted from the extensive PIF variable dictionary. The risk factors chosen and included for analyses were (a) maternal factors considered to place the baby at higher risk—namely, unplanned pregnancy, single mother without partner, mother perpetrator of severe interpartner violence, and mother clinically depressed (Edinburgh Post-natal Depression Score > 12), and (b) direct infant health risk factors likely to result in poor long-term outcomes— namely, small for gestational age, exposed to maternal smoking in utero, exposed to alcohol in utero, attended/admitted to hospital, not immunized at 6 weeks, and not exclusively breastfed. All factors were chosen taking into account known maternal and infant risk factors for avoidable morbidity and mortality (Ministry of Health & Ministry of Pacific Island Affairs, 2004). Some of the identified risk factors were included because they were widely considered very important by stakeholders in terms of Pacific health in New Zealand (e.g., single parents without partner and maternal depression). The factor relating to maternal perpetration rather than victimization of severe intimate partner violence was included because an earlier article from the study had identified cultural alignment as significantly associated with maternal perpetration of violence but not victimization. Some infant health and health-related variables were excluded, as they were highly correlated with other variables (e.g., mother currently smokes as compared to exposed to maternal smoking in utero). Others were excluded because there were too few cases. For example, the APGAR score at birth was excluded because only 28 cases in the cohort met a clinically significant low score (< 8 at 5 minutes post-birth), although it has a demonstrated relationship with longer term health outcomes, educational achievement, and social stability (Oreopoulos, Stabile, & Walld, 2007; Weinberger et al., 2000). Statistical Analysis Each of the respondents was individually scored on both the NZACCULT and PIACCULT scales and allocated to one of the categorical model classes dependent on whether their individual score fell above or below the median of the full group: namely, Low New Zealand—High Pacific (Separator), High New Zealand—High Pacific (Integrator), High New Zealand—Low Pacific (Assimilator), and Low New Zealand—Low Pacific (Marginalisor). Subsequent analysis was carried out in terms of this categorization. To investigate, (a) aims and (b) all risk factors were simultaneously associated with the 4- leveled acculturation variable (taking separators as the reference category) using a binomial generalized estimating equation (GEE) model. Because the risk factors are without natural order and have different binary distributions, an unstructured covariance matrix was adopted for the GEE model. Two separate GEE models were run: (a) an unadjusted model that consists of main effects corresponding to the acculturation variable and risk factors, and their interactions, and (b) an adjusted model that consists of main effects corresponding to the acculturation variable and the risk factors, and their interactions, together with selected sociodemographic variables: mother’s age, ethnicity, highest educational qualification, and household income. Estimated marginal odds ratio (OR) means associated with the four-levelled acculturation variable overall risk factors were calculated and reported to provide a global measure of the effect of acculturation. The robust Huber-White sandwich estimator of variance was used to calculate standard errors and confidence intervals. GEE statistical analyses were performed using Stata/IC 10.0 for Windows (Stata Corp, College Station, TX, USA), and a significance level of α = 0.05 was used to determine statistical significance for all tests. 708 Journal of Cross-Cultural Psychology 42(5) The NZACCULT and the PIACCULT were tested for reliability (internal consistency) using Cronbach’s α. Following Tsai et al. (2000), we analyzed aspects of validity in two ways: First, we measured the correlations between average cultural orientation (as measured by the scales) and a recognized standard index of acculturation (length of residence in New Zealand); second, the mean scores on each of the modified scale items were calculated for participants who migrated to New Zealand—less than 2 years ago, between 3 and 5 years, between 6 and 10 years, more than 10 years, and in addition those who were born in New Zealand. In line with Tsai et al. (2000), we predicted that if the PIACCULT was a valid measure of cultural orientation, then Pacific people who migrated recently to New Zealand would report (a) speaking a Pacific language more, (b) understanding a Pacific language better, (c) being more exposed to Pacific culture, (d) being more affiliated to Pacific peoples, and (e) participating more in Pacific activities than longer term migrants, who in turn would report higher Pacific orientation than those born in New Zealand. Conversely, if the NZACCULT measure was a valid measure of orientation to New Zealand culture, New Zealand–born Pacific people and those who had been resident in New Zealand for a longer period would report (a) speaking English more, (b) understanding English better, (c) being more exposed to New Zealand culture, (d) being more affiliated to non-Pacific peoples, and (e) participating more in New Zealand activities. Connected line plots of mean scores of the 11 acculturation questions for NZACCULT and PIACCULT scales by years resident in New Zealand, together with a superimposed lowess curve (a nonparametric estimator of the mean function), were used to graphically demonstrate this relationship. Analysis of variance was used to statistically test these suppositions, along with post hoc tests including Tukey’s honestly significant difference multiple comparison test and Welch’s robust test of equality of means. Results In total, 1,708 mothers were identified, 1,657 invited to participate, 1,590 (96%) consented to a home visit, and of these, 1,477 (93%) were eligible for the PIF study. Of those eligible, 1,376 (93%) mothers giving birth to 1,398 infants (22 pairs of twins) of which 680 (49%) were female participated at the 6-week interview. As non-Pacific mothers were eligible if the child’s father was Pacific, some 107 non-Pacific mothers and 1,269 Pacific mothers participated at the 6-week interview. Island-specific ethnic distributions in the cohort were approximately representative of the ethnic distribution and economic and social characteristics of the main ethnic Pacific population in New Zealand (Table 1). However, they do not reflect the proportions of populations from the islands of origin largely because Cook Island Maori, Niueans, and Tokelauans, unlike Samoans and Tongans, qualify automatically for New Zealand citizenship. Cultural Orientation In total, 445 (35%) of the sample was categorized as separators, 231 (18%) as integrators, 342 (27%) as assimilators, and 242 (19%) as marginalisors. The group was subdivided on a median split-half, and the means, medians, and dispersions of the PIACCULT and NZACCULT scales (N = 1,258) were PIACCULT: M = 43.7, SD = 7.32; Median = 45.0; Interquartile range = 11; and NZACCULT: M = 34.2, SD = 7.78; Median = 35.0; Interquartile range = 12. Ethnic group differences within the overall group in relation to cultural alignment are outlined in Table 2. All investigated risk factors were simultaneously associated with the acculturation variable using a binomial GEE model. Table 3 includes the percentage of poor outcomes for each risk factor and the unadjusted (OR) and associated 95% confidence intervals (95% CI) for the acculturation classifications derived from this model. Perusal of Table 3 reveals considerable heterogeneity in the estimated ORs between acculturation classifications over the considered Borrows et al. 709 risk factors. For example, compared to separators, the ORs associated with infant exposure to alcohol during pregnancy was 2.58 for integrators, 14.62 for assimilators, and 6.98 for marginalisors. For infants born small for their gestational age, the estimated ORs were 0.88 for integrators, 1.47 for assimilators, and 1.68 for marginalisors. In this GEE model, the main effect variables corresponding to acculturation and the risk factors were significant (both p < .001), as was their interaction (p < .001). To provide a global measure of the effect of acculturation over the 10 investigated risk factors, the estimated marginal OR means associated with the four-levelled acculturation variable was calculated and reported in Table 4. In the unadjusted analysis, integrators, assimilators, and marginalisors had significantly higher estimated marginal OR means than separators (all p < .001). Furthermore, assimilators and marginalisors had significantly higher estimated marginal OR means than integrators (p = .004 and .007, respectively), but no significant difference was observed between assimilator and marginalisor participants (p = .86). When the GEE analysis was repeated with the addition of selected sociodemographic variables, including mother’s age, ethnicity, highest educational qualification, and household income, there remained considerable heterogeneity in the estimated adjusted OR between acculturation classifications over the considered risk factors but some dampening in their effect sizes compared to the unadjusted ORs. This dampening can be seen in Table 4, which also includes the estimated marginal adjusted OR means associated with the four-levelled acculturation variable. Again, integrators, assimilators, and marginalisors had significantly higher estimated marginal adjusted OR means than separators (all p < .001). However, assimilators and marginalisors had estimated marginal adjusted OR means that were no longer significantly higher than integrators (p = .06 and .23, respectively). As before, there was no significant difference in estimated marginal adjusted OR means between assimilators and marginalisor participants (p = .50). In the adjusted GEE analysis, there was a significance difference in estimated risk factor ORs between ethnic groups (p < .001), with Tongan mothers having an OR of 1.32 (95% CI: 1.15, 1.51), Cook Island Maori mothers having an OR of 1.50 (95% CI: 1.29, 1.74), Niuean mothers having an OR of 1.65 (95% CI: 1.32, 2.05), and other Pacific mothers having an OR of 1.93 (95% CI: 1.48, 2.51) compared to their Samoan counterparts. However, there was no significant interaction between the acculturation classifications and mother’s ethnicity (p = .40), suggesting that the effect of acculturation and ethnicity are independent important factors. Reliability and Validity of the Acculturation Instruments Cronbach’s α of 0.81 and 0.83 were obtained for the NZACCULT and the PIACCULT scales, respectively—values that are acceptable. The length of residence in New Zealand was significantly Table 2. Acculturation Classifications by Ethnicity Acculturation Classifications Separators Integrators Assimilators Marginalisors Ethnicity N % N % N % N % Samoan 304 47.4 151 23.5 125 19.5 62 9.7 Tongan 115 40.8 48 17.0 61 21.6 58 20.6 Cook Island 15 6.6 17 7.4 103 45.0 94 41.0 Niuean 4 6.8 10 16.9 26 44.1 19 32.2 Other 5 10.9 5 10.9 27 58.7 9 19.6 All 443 35.2 231 18.4 342 27.2 242 19.2 710 Table 3. Percentage of Risk Factor Poor Outcomes and Unadjusted OR and Associated 95% Confidence Intervals (95% CI) for the Acculturation Classifications Derived from a Binomial Generalized Estimating Equation (GEE) Model With Unstructured Covariance Matrix Separators Integrators Assimilators Marginalisors Risk Factors N % ORa % OR 95% CI % OR 95% CI % OR 95% CI Maternal Unplanned pregnancy 1,256 55.9 1.00 62.3 1.31 0.94, 1.81 71.3 1.97 1.46, 2.66 61.8 1.28 0.93, 1.77 Single without partner 1,258 15.3 1.00 19.5 1.33 0.88, 2.02 25.1 1.85 1.30, 2.64 20.2 1.40 0.93, 2.10 Perpetrator of severe IPV 1,070 9.0 1.00 21.4 2.81 1.71, 4.62 25.8 3.61 2.32, 5.60 27.9 4.12 2.58, 6.59 Depressed (EPDS > 12) 1,253 10.8 1.00 12.7 1.21 0.74, 1.97 18.1 1.82 1.21, 2.74 26.7 2.99 1.98, 4.52 Infant Small for gestational age 1,130 8.0 1.00 7.4 0.88 0.45, 1.72 11.7 1.47 0.87, 2.49 12.2 1.68 0.96, 2.92 Exposed to maternal smoking in utero 1,257 10.6 1.00 23.8 2.63 1.71, 4.03 35.4 4.60 3.16, 6.69 29.8 3.56 2.36, 5.36 Exposed to alcohol in utero 1,258 0.7 1.00 1.7 2.58 0.57, 11.6 9.1 14.62 4.43, 48.2 4.5 6.98 1.93, 25.3 Attended/admitted to hospital 1,258 9.9 1.00 12.1 1.30 0.79, 2.14 10.8 1.10 0.69, 1.75 17.8 1.96 1.25, 3.08 Not immunized at 6 weeks 1,258 19.9 1.00 29.0 1.65 1.14, 2.38 29.5 1.69 1.22, 2.35 31.8 1.88 1.32, 2.69 Not exclusively breastfed 1,258 45.5 1.00 47.6 1.11 0.81, 1.53 52.9 1.35 1.02, 1.80 53.3 1.37 1.00, 1.88 OR = Odds Ratio; CI = Confidence Interval; EPDS = Edinburgh Post-natal Depression Score. a. Reference category. Borrows et al. 711 correlated with average scores on the NZACCULT (r = 0.58) and the PIACCULT (r = –0.45), both p < .001. That is, the more oriented participants were to New Zealand culture and the less oriented they were to Pacific culture was correlated with the number of years that they had resided in New Zealand. However, PIACCULT and NZACCULT scales are not strongly correlated (r = –0.33). Analysis of variance by group supported the predictions noted previously with regard to the validity of the NZACCULT and PIACCULT scales. It revealed significant differences among the five New Zealand residency groups for 9 of the 11 items on both the NZACCULT and PIACCULT scales (Table 5). Generally, increasing mean item values on the NZACCULT scale were observed with increasing length of New Zealand residency for migrants, with respondents born in New Zealand exhibiting the highest item scores (Figure 2a). A converse pattern (Figure 2b) was observed for the PIACCULT scale. Church attendance on the NZ scale and Pacific sports participation on the Pacific scale failed to discriminate significantly between the five NZ residency groups. Larger effect sizes were observed for speaking and understanding language and being brought up and being familiar with the relevant language and customs than friendship and external social activities. Discussion The PIF study was designed to research issues of identified relevance to the New Zealand Pacific community. Community consultation undertaken to establish relevant dimensions for the protocols and advice received from our Pacific Advisory Board reinforced the perspective that maintenance of original Pacific culture was a relevant and positive dimension to good health outcomes in community perceptions. The Association Between Mother and Infant Health Variables The first aim of the study was to investigate the association between mother and infant health variables that might act as infant risk indicators and adaptation to living in New Zealand. The classical acculturation conceptual model (Berry, 1980) was applied to achieve this aim. On the basis of accumulated evidence in the literature, it would be expected that those categorized as integrators (high NZ, high PI) would have good or very good health outcomes, separators (high PI, low NZ) would have good or reasonable outcomes, assimilators (low PI, high NZ) would have reasonable outcomes, and marginalisors (low PI, low NZ) would have poor outcomes. Table 4. Estimated Marginal OR Means Associated With the Four-Levelled Acculturation Variable Over All 10 Risk Factors From Two Separate Binomial Generalized Estimating Equation (GEE) Regression Models Separators Integrators Assimilators Marginalisors GEE model ORa OR 95% CI OR 95% CI OR 95% CI (i) Unadjusted 1.00 1.56 1.25, 1.94 2.39 1.98, 2.88 2.33 1.91, 2.83 (ii) Adjusted 1.00 1.53 1.23, 1.91 2.03 1.66, 2.48 1.84 1.50, 2.26 OR = Odds Ratio; CI = Confidence Interval. (i) An unadjusted model that consists of main effects corresponding to the acculturation variable and risk factors and their interactions. (ii) An adjusted model that consists of main effects corresponding to the acculturation variable and the risk factors and their interactions, together with selected sociodemographic variables: mother’s age, ethnicity, highest educational qualification, and household income. a. Reference category. 712 Journal of Cross-Cultural Psychology 42(5) Brought up NZ way Familiar with NZ customs Understanding of English Have non-Pasifika friends Friends speak English Participate in NZ sports Speak English Have non-Pasifika contacts Eat non-Pasifika food See western-trained doctors Non-Pasifika church attendees 1 2 3 4 5 Mean acculturation scores 0-2 years 3-5 years 6-10 years >10 years NZ born New Zealand residency A Table 5. Analysis of Variance Results Comparing Five New Zealand Residency Groups (0 to 2 Years, 3 to 5 Years, 6 to 10 Years, > 10 Years and New Zealand Born) on Item Scores of the PIACCULT and NZACCULT Scales Item F p Partial Eta-Squared PIACCULT Scale I was brought up the Pasifika way 69.8 < 0.001 0.181 I am familiar with Pasifika practices and customs 45.3 < 0.001 0.126 I can understand a Pasifika language well 61.8 < 0.001 0.164 I have several Pasifika friends 3.7 0.005 0.012 Most of my friends speak a Pasifika language 33.3 < 0.001 0.096 I participate in Pasifika sports and recreation 1.1 0.370 0.003 I speak a Pasifika language 120.2 < 0.001 0.276 I have contact with Pasifika families and relatives 8.1 < 0.001 0.025 I eat Pasifika food 17.6 < 0.001 0.053 I visit a traditional Pasifika healer . . . 13.2 < 0.001 0.040 I go to a church mostly attended by Pasifika people 27.1 < 0.001 0.079 NZACCULT Scale I was brought up the NZ way 135.1 < 0.001 0.300 I am familiar with NZ practices and customs 105.1 < 0.001 0.250 I can understand English well 70.6 < 0.001 0.183 I have several non-Pasifika friends 61.0 < 0.001 0.162 Most of my friends speak English 79.2 < 0.001 0.201 I participate in NZ sports and recreation 27.6 < 0.001 0.080 I speak English 112.1 < 0.001 0.262 I have contact with non-Pasifika families and relatives 33.2 < 0.001 0.095 I eat non-Pasifika food 8.0 < 0.001 0.025 I visit Western-trained doctors 2.6 0.037 0.008 I go to a church mostly attended by non-Pasifika people 1.6 0.183 0.005 Figure 2a. Connected Line Plot Of Mean Scores of the 11 Acculturation Questions of NZACCULT Scale for Participants Over the Years They Had Been Resident in New Zealand (NZ), Together with the Lowess Curve (Dashed Line) Borrows et al. 713 Although our findings showed a clear direction for these relationships, they were not in the expected direction in terms of the majority of the existing acculturation literature, although, as indicated previously, there have been some exceptions (Ataca & Berry, 2002; Berry, 2006; Jones et al., 2002). The association between maintenance of constructive health behaviours and existence and maintenance of aspects of original society social and cultural practices has also been noted in the ethnocultural qualitative literature and the paediatric and nursing literature (Callister & Birkhead, 2002; Gurman & Becker, 2008). Several studies have also documented this apparent epidemiologic paradox, with better outcomes occurring among disadvantaged immigrant people (Liu, Chang, & Chou, 2008). However, unlike this study, some of these studies focus their analysis on a single acculturation related factor, such as length of residence (Hawkins, Lamb, Cole, & Law, 2008) or ethnicity (Gould, Madan, Qin, & Chavez, 2003), rather than a validated or reliable measure of acculturation and fail to adjust for important risk factors and confounders. Within this cohort, the marginalisor, assimilator, and integrator groups had poorer outcomes in terms of all the measured infant-related health risk factors except for the risk factor, small for gestational age. In this isolated case, the integrator group OR was smaller than that for the reference separator group. Overall, our findings showed a clear gradation of risk indicators from a low-risk position held by the reference separator group to the much-increased OR of each risk factor for both the assimilator and the marginalisor groups, with the assimilator and the marginalisor groups showing no significant difference. As noted earlier, there was considerable heterogeneity in the estimated OR between acculturation classifications over the considered risk factors. However, in terms of the identified maternal risk factors, three factors could be identified as having greater risk ORs across the acculturation categories other than the reference separator group—namely, the mother being the perpetrator of severe interpersonal violence, association with maternal smoking in utero, and Brought up Pasifika way Familiar with Pasifika customs Understand a Pasifika language well Have Pasifika friends Friends speak a Pasifika language Participate in Pasifika sports Speak a Pasifika langauage Have Pasifika contacts Eat Pasifika food Visit Pasifika healers Church mostly Pasifika 1 2 3 4 5 Mean acculturation scores 0-2 years 3-5 years 6-10 years >10 years NZ born New Zealand residency B Figure 2b. Connected Line Plot of Mean Scores of the 11 Acculturation Questions of PIACCULT for Participants Over the Years They Are Resident in New Zealand (NZ), Together with the Lowess Curve (Dashed Line). 714 Journal of Cross-Cultural Psychology 42(5) exposure to alcohol in utero. The latter two risk factors could be recognized as negative adaptation associated with undesirable but widespread socio/cultural behaviours in the host society: alcohol consumption by women is not considered appropriate behaviour in traditional Pacific societies but is sometimes linked to tolerated private and sometimes aggressive male behaviours (Ministry of Health: Sector Analysis, 1997). Similarly, interpartner violence has been consistently linked to excessive alcohol consumption (Leonard, 2000; Paterson et al., 2007). Such sociocultural behaviours appear to provide evidence of negative adaptation of risk-taking host society behaviours by all groups other than those who hold strongly to traditional values and behaviours in the new society. Conversely, it is possible that the more private corporal health factors such as birth control, breast feeding, and attitudes to immunization are more deeply imbedded psychological rather than recently adopted sociocultural behaviours (Ward & Leon, 2004), which are subject to slower (less extreme) pace of change. Detailed analysis of these is beyond the scope of this article, as further research will be required to clarify the complex relationships between each of these identified risk factors within a revised and more complex acculturation model. Is Strong Cultural Alignment to the Original Culture Associated With Better Outcomes? In terms of the second aim of the study, we found that when the two dimensions of the acculturation measure NZACCULT and PIACCULT were separately and simultaneously considered, they provided evidence to support the current Pacific cultural and New Zealand official dogma. That is, when Pacific cultural orientation is high, it has a protective effect; however, this effect is reduced in the presence of a high New Zealand orientation. Existing empirical studies show that at the time of migration, people are at special risk for adoption of negative health risk practices (Carballo & Nerukar, 2001; Prior et al., 1987; Salmond et al., 1985), and at the time of birthing, mothers are doubly at risk for maintenance or adoption of negative health practices (Carballo & Nerukar, 2001). The results presented in this article suggest that there may be something protective in the process of maintaining original cultural habits toward good health behaviours. For example, it is logical to assume that responsible parenthood would enhance prospects of successful adaptation to the new society. Although the two high PI orientation groups (separators and integrators) did not differ significantly on the mean overall PI scale, there was considerable heterogeneity between individual items. The separators scored significantly higher than the integrators on scale items relating to custom and active use of a Pacific language, and these (especially church attendance) are still important and relevant parts of strong Pacific identity in New Zealand. These items measure traditional Pacific values and reflect the strength of immediate family bonds through which these young mothers traditionally obtain crucial childbearing and childraising support. Pacific cultures have strong existing culturally bound positive traditions toward birthing and family welfares (Abel et al., 2001; Barclay, Aiavao, Fenwick, & Papua, 2005). It could be that those in the separator group have the full advantage of strong family and community associations within a culture of origin that enhances responsible traditional behaviour and allows consideration of selected new society behaviours that are considered advantageous. In this critical arena of maternal and infant risk, these findings provide evidence of the benefit of maintaining strong cultural ties especially where the transition to the new societies systems is not fully developed. When the relationships were examined in light of selected sociodemographic variables, there was no significant difference in estimated marginal adjusted OR means between assimilator and marginalisor groups, except the extent to which the assimilators report some negative healthrelated practices such as smoking and alcohol consumption during pregnancy. Although individual Borrows et al. 715 socioeconomic status is accommodated in the adjusted analysis, the majority of the PIF cohort resides in South Auckland, which has a high proportion of the most deprived economic areas as outlined in the New Zealand Atlas of Socioeconomic Difference (Crampton, Salmond, Kirkpatrick, Scarborough, & Skelly, 2000). The extent to which the negative health risk practices in the assimilator group are reflecting or dependent on this relatively poorer socioeconomic setting within the dominant subregional culture is an interesting question. These communities, in themselves multicultural, might also be considered marginalized in terms of mainstream New Zealand social culture. In this context, the different modes of acculturation become different social determinants. This article is a first step in exploring and providing some evidence to refute the melting pot as a preferred hypothesis. Significant differences in estimated risk factors between ethnic groups were found, with Tongan, Cook Island Maori, Niuean, and other Pacific mothers all having higher risk than their Samoan counterparts and relatively different proportions in each of the acculturation groups. The larger numbers of the Samoan community could explain the greater number of individuals in the separator category than might be expected from comparable studies. As is shown in Table 2, Samoans made up 51% of the cohort and also had the highest proportion of participants classified as separators. This also suggests that having strong and numerous bonds to identify with may have a protective influence in terms of positive health outcomes in this particular New Zealand setting. Where these bonds are weak (e.g., small numbers for specific island ethnic group or for those who choose assimilation or marginalized acculturation strategies), some negative health practices of the dominant society may be freely adopted. This could explain why excess alcohol consumption during pregnancy is characteristic of the assimilators who are most closely tied to negative cultural practices of the wider society but less strongly associated with those in the marginalisor category. The crude ethnic acculturation differences are also partly explained by the findings of the reliability/validity results. These confirm that Pacific people who migrated recently to New Zealand are less oriented to New Zealand mainstream culture and those who migrated to New Zealand less recently have had greater opportunity for exposure to mainstream New Zealand behaviour and lifestyle concepts (Figures 2a and 2b). The Cook Island and Niuean participants in this study have a longer (if still relatively recent) migration history than those of Samoan and Tongan ethnicity. Hence, Cook Islands and Niuean participants have greater proportions in the integrator and marginalisor categories than is the case for those from Samoa or Tonga (Table 2). However, although the univariate analysis provides support for the thesis that the differences between acculturation groups is mediated by the ethnic group differences, there was no significant interaction between the acculturation classifications and mothers’ ethnicity in the adjusted GEE model. This suggests that the effects of acculturation and ethnicity are independent important factors. The finding that separators are at lower risk run counter to many of the studies that have examined acculturation strategies in nondominant cultural groups. In most such studies, preferences for integration are expressed over the other three strategies (Berry, 2006). Integrative strategies seem to be preferred at a societal level (Hjerm, 2000), but there are subtleties (Arends-Toth & van de Vijver, 2003), and exceptions have been found in indigenous groups and in some cases in lower socioeconomic immigrant groups in some settings, for example Turks in Canada (Ataca & Berry, 2002). This raises the question as to why preference for integration in this cohort would not be associated with the best outcomes given that most studies in the acculturation literature have produced results pointing in this direction. General community and subregional social and economic factors may be influencing the positive association between adherence to traditional culture and health outcomes with the relative collective disadvantage of those who attempt to adopt assimilation or an integration cultural strategy in the setting of an economically deprived area. That is, are the wider regional cultural examples and imperatives themselves marginal to 716 Journal of Cross-Cultural Psychology 42(5) the economically advantaged mainstream? This may mean that assimilation and marginalisor groups identified in this study are in fact themselves aligned with the predominant subregional economically deprived culture and share the negative prospects and health outcomes of that subregional culture. In this case, it is possible that marginalization and assimilation are failed outcomes of regional group rather than individual cultural integration. These findings also underscore the need for acculturation research to incorporate the possibility of more than two cultures or regional subcultures into the explanatory framework and to examine the extent to which ethnocultural identities are contextually bound (Persky & Birman, 2005). Aside from location in disadvantaged neighbourhoods, these findings raise the question as to whether New Zealand society limits the opportunities for Pacific people to be exposed to ethnic groups other than the range of minority Pacific ethnicities. That is, is this an ethnic ghetto? As is shown in the description of the place of Pacific people in contemporary New Zealand society, there is little doubt that opportunities for pursuing migration strategies of choice have been available to Pacific communities. The PIF findings that the separator group has better outcomes are consistent with Sam (2006a), who found that immigrant youth who preferred assimilation and integration had a higher risk of engaging in health-compromising behaviour, such as smoking and drinking alcohol, than their peers who preferred separation. It is also important to recognize that these results are in line with the historical views of acculturation scholars, including Berry (2003), who points out that it is not inevitable that intergroup contact will proceed uniformly through a sequential process to ultimate assimilation. Flannery et al. (2001) also noted that insights generated by a bidirectional model hold the promise of correcting melting-pot assumptions and promoting political sensitivities among ethnicities and as such fit explicitly in terms of the social determinants theories for explaining the epidemiology of health outcomes. Recent theory and research offers a deeper insight as to the multidimensional nature of acculturation and its components than that incorporated in the general model we and others have used. As noted previously, it is possible that the advantages or disadvantages of one or another mode of acculturation may vary according to broad dimensions such as sociocultural and psychological adaptation (Ward & Leon, 2004), and in relation to the domain or competence under study, such as self-esteem, social competence, and behaviour and skills and experience. However, most significantly, advances in the theory of measurement of acculturation and related cross-cultural relationships (Boski, 2008) point out that integration, in terms of Berry’s model of acculturative attitudes or strategies, and as used for the framework for this analysis, operates within a limited concept of integration and in a sense is acultural and as such might be interpreted as a measure of double social identity. The abbreviated scales used for this analysis (PIACCULT and NZACCULT) were not designed to distinguish these sophisticated and important contexts in measurement of integration and acculturation—for example, (a) integration as a cognitive-evaluative merger of two cultural sets or (b) integration and functional (partial) specialization in life’s public and private domains (Boski, 2008). In terms of the former, the fact that little differentiation in poor outcomes for the assimilator and marginalisor groups suggests that Boski’s value placement concepts could hold true and that for some fully individually and socially functioning individuals, values oriented toward single culture separation rather than some overlapping entity may prove preferable. In terms of the second of these integration models, there is the possibility that the individual responses to the two subscales were mediated by an essentially private response to the Pacific orientation in the context of language, families, and way of life but an alternative public response to the New Zealand orientation when responding in the context of English being widely used (and of necessity understood) in the context of external employment and social and public life in a multicultural city such as Auckland. This concept of double response to identity might partly explain why some questions with seemingly high face validity proved problem items in terms of the validity testing. In the context of the private Pacific identity, sports is not a separate identity concept being bound up with normal social, community, and church life Borrows et al. 717 (McGregor & McMath, 1993), whereas for a New Zealand–oriented public response, the direction of response is very much affected by the part sport plays in the context of mainstream life and work and social exchanges. Within New Zealand, culturally bound supportive services have been developed over the last decade—for example, dedicated Pacific support unit in communities and hospitals. The efficacy of such services remains the subject of debate, but these initiatives show that central government is focused on pursuing an effective public institutional and societal strategy in areas of high ethnic concentration and demand. Traditional island cultures also have strong alternative community and church ties that provide support and education around childbirth (Barclay et al., 2005). It is acknowledged that a more sensitive measure is needed to elucidate the complex interaction between the individual’s preferred cultural identity and the accommodating multicultural society that has evolved in New Zealand. That is, a society that allows strong personal (internal) maintenance of values derived from the original island societies in family home and private life domains, which are protective of mother and infant, while functional specialization is enabled in public life domains such as work, education, and civic society (in this case, health services) from the concern and service efforts provided by the host society. The well-established services allow ample opportunity for effective (if selective) participation in most public life domains. Examination of these concepts in greater depth is beyond the scope of this current article but will be pursued in the future phases of the PIF longitudinal study. Is the Abbreviated Version of the GEQ a Valid and Reliable Instrument? The ancillary aim for this study was to demonstrate that the abbreviated version of the GEQ adopted for use in the PIF longitudinal study was both a

id and reliable instrument in the context of the range of health and social outcomes that were of principal interest for the PIF study. Our confidence in the selection of items was borne out by the psychometric analysis that showed very good internal consistency of the resultant abbreviated New Zealand (NZACCULT) and Pacific (PIFACCULT) scales. The use of these scales was justified in terms of testing our aims and appropriate for ongoing use for Pacific people in this longitudinal study and for similar epidemiological oriented studies in the future. To improve face validity, the scale was adapted to include a limited number of items assessing concepts considered important and central to New Zealand or Pacific culture. The analysis revealed that some of these items did not contribute significantly to the measure of cultural differentiation—hence, we were sacrificing internal consistency at the expense of content validity. Rather than remove them from the scales, we left them in place for they had different impacts in terms of the respective PIACCULT and NZACCULT scales and provided further insight into how the New Zealand and Pacific cultures view and accommodate such issues. In brief, these nondiscriminatory items provide insights into some of the differences in the Pacific versus New Zealand cultural view in the context of New Zealand society. They confirm that in a Pacific domain context, sport is not a single distinguishable variable in establishing Pacificness (McGregor & McMath, 1993); conversely, in a New Zealand domain context, church attendance is not a relevant variable as the wider New Zealand society and world view is more secularly oriented, with 65% of the New Zealand population nominating a religious affiliation as compared to 86% of Samoans and 90% of Tongan people who were affiliated with a religion (Statistics New Zealand—Te Tari Tatau, 2006b). Strengths of This Study There are some specific strengths of this study that deserve elucidation. First, the short but robust acculturation measure used was constructed so that the cultural orientation and change could be described and its impact could be quantitatively measured for inclusion in the ongoing explanatory 718 Journal of Cross-Cultural Psychology 42(5) models for healthy child and family development. This approach can be useful in the context of the universal modelling rationale for this longitudinal study, providing both insights for testing and explanation of the results as is the case in this initial study of the association of acculturation and maternal and infant health risk indicators. Despite having many salient features, including the ability to accommodate and appropriately model correlated binary data, GEE methods used here have not readily been adopted by behavioural researchers (Lee, Herzog, Meade, Webb, & Brandon, 2007). The approach also fits a modern epidemiological perspective for examining the impacts of relevant social and health determinants, in this case the mode of acculturation, and serves to enrich the literature in terms of the place of acculturation and acculturation strategies in the context of the wider psychosocial and epidemiological literature. Second, although this is a birth cohort, the island-specific ethnic distributions in the cohort are approximately representative of the ethnic distribution of the main ethnic Pacific population in New Zealand. This is unexceptional as a great majority of the Pacific population in New Zealand is located in the wider Auckland metropolitan area but still useful in terms of policy and planning for areas such as ongoing refinement of antenatal and birthing services and community health promotion activities such as immunization strategy, nutrition advice, and exercise programs. Specific Limitations There are four specific limitations of this study that need to be recognized: (a) Abbreviating the GEQ from a 38-item to 11-item scale was a necessary requirement for the PIF study to avoid lengthening an already long multidisciplinary questionnaire. The resultant bi-dimensional scales have proved robust and successful in the context of a general measure of acculturation for the epidemiological explanatory model used here and can continue to be used in this context. This is notwithstanding the limitations on the use of the median split method outlined in Arends-Toth and van de Vijver (2006), and the conclusions of Kang (2006), that lack of independence between ethnic and mainstream cultural orientations is partially due to specific scale format and that structural features commonly found in bi-dimensional acculturation instruments cause strong inverse associations between the two cultural orientations. Our analyses have shown that the PIACCULT and NZACCULT are not strongly correlated (–0.33) and show a wide distribution of the means between the NZACCULT and the PIACCULT scales. This means that when responding to the Pacific-oriented scale, the tendency was to a more uniform and positive response than was the case with the New Zealand scale but not for those mother participants (≈40%) who were New Zealand born. It is also clear that other than the expected trends over time in relation to length of residency in New Zealand, no obvious differential exists in terms of the way in which the New Zealand–born as compared with island-born participants responded to the two questionnaires. (b) A more important limitation in relation to the use of this scale for this study is the inability to apply it in the contexts of more recent, complex, and richer acculturation models that have aroused interest elsewhere. These include, for example, domain-specific models (Arends-Toth & van de Vijver, 2006, 2007; Tsai et al., 2000) and specialized acculturation and integration concepts such as cognitive-evaluative, functional specialization, frame switching, and constructive marginalization models as summarized by the five-level model of the acculturation process postulated by Boski (2008). The approach adopted in the measurement used in this study carries an inherent risk that may remain fixed at the first level (acculturation attitudes) rather than moving on through cultural perception and evaluation to areas such as functional specialization and perhaps true multiculturalism, cultural heteronomy, and true autonomy of self. (c) The demonstrated difference in the means between the acculturation groups other than the separator group (Table 4), while significant, is probably insufficient in practical clinical terms to suggest that identification of at-risk individuals based solely on the acculturation scale used in Borrows et al. 719 this study would not be practical for direct clinical use in the health and social services. However, these findings can be used to highlight the areas of cross-cultural difference in perception of, and potential use of, health services by individuals caught between or outside cultures. It is this issue that needs to be addressed in health promotion and service terms so that the benefit or use of such services can be optimized. In addition, these findings suggest that cultural alignment should be considered for inclusion in explanatory epidemiological models and support the perspective that culture be given proper consideration in the clinical decision-making process. (d) Last, it is also important to recognize that this analysis is constrained by the nature of limitations common to longitudinal studies, with large multidimensional questionnaires resulting in lesser opportunity to drill down into multifaceted issues. This approach limits the degree to which the specific role of Pacific subcultures and their elements can be elucidated. For example, we were not able to investigate the impact of individual attitudes on mode of acculturation at this data collection point. Separator mothers may be inherently group or community aligned rather than more individually oriented and hence may be less likely to engage in potentially risky behaviour. We may be able to consider individual versus group personality behavioural characteristics of participants and the association with acculturation in later phases of the study. These findings provide support for the view that retaining and enhancing strong cultural links for Pacific immigrants is likely to have positive benefits. The acculturation measure proved robust and reliable as an overall measure. A clear association was shown between mode of acculturation and the group of maternal and infant risk factors, however this measure did not sufficiently reveal which of the infant and maternal outcomes were individually effective indicators of acculturation risk independent of the overall acculturation categories. Also, such detailed relationships may comprise a useful outcome only if the other subtleties of the acculturation process pointed to elsewhere in this article are properly accommodated. In particular, those subtleties related to attitudinal and behavioural responses in public and private domains and attitudes and behaviours in both the sociocultural and more personal psychological and corporal health realms. We acknowledge that it is not possible from this study to determine whether in terms of recent models of integrative acculturation strategies the findings presented here are in fact indicators of an effective New Zealand public integrative but not assimilative (melting pot) strategy. These findings raise questions about the stability of the relationships between culture and health risk factors; how reflections of disadvantage are maintained over time; at what speed post-migration changes take place; how these changes support, refute, or assist in better explaining current migration/ acculturation and health hypotheses such as the “immigrant health paradox” (Sam, 2006a); and what factors influence this, especially in relation to acculturative stress. Further planned work in the longitudinal PIF study will determine the durability of these findings and explore in more depth aspects of cultural contact between Pacific peoples and the wider New Zealand society and examine this in terms of degree of change, elements of the process that lead to cultural alignment remaining static or the rate of change over time, and ultimately the relationship between the cultural alignment of the parent(s) and the children in this birth and family cohort. This could add a significant dimension to the understanding of the modes of the classical acculturation model (Berry, 2003; Sam, 2006b) and the more recent explanatory models of levels of integration in the acculturation process (Boski, 2008). Conclusion Most descriptions of the acculturative processes, particularly exceptions to the assimilative norm (Ataca & Berry, 2002), are generally cross-sectional in nature. This initial analysis of acculturation in the context of this large-scale longitudinal epidemiological study (Paterson et al., 2008) provides a singular opportunity to explore these concepts over time in greater depth. In spite of 720 Journal of Cross-Cultural Psychology 42(5) current limitations, further research within the parent longitudinal study offers ongoing opportunity to unravel some of the nuances and impacts of cultural alignment, in terms of historical recognized models and modes of acculturation that are still rarely considered in a traditional epidemiological approach. This study, placing acculturation at the centre of interest and analysis, provides an interdisciplinary approach aimed at beginning the process of filling this deficit. “And most New Zealanders, whatever their cultural backgrounds, are good-hearted, practical, commonsensical and tolerant. Those qualities are part of the national cultural capital that has in the past saved the country from the worst excesses of chauvinism and racism seen in other parts of the world. They are as sound a basis as any for optimism about the country’s future.” (King, 2003, p. 520)