Issues in the multi-cultural assessment of parent–child interaction: An exploratory study from the starting early starting smart collaboration

Abstract

A national, multi-site study of behavioral health services integration developed a parent–child interaction assessment tool and culturally anchored videotape protocol. Representatives from programs serving Chinese, Native American¹, Latin-American, African-American, and Anglo-American families discussed cross-cultural issues in parenting and developed a set of guiding principles for the assessment of parent–child interaction, resulting in a revised Parent–Child Observation Guide (PCOG: Bernstein, Percansky, & Hans, 1987 [Bernstein, V. J., Percansky, C., and Hans, S. L. (1987, April). Screening for social–emotional impairment in infants born to teenage mothers. Paper presented at the biannual meeting of the Society for Research in Child Development, Baltimore, MD]) and program-specific laboratory protocols. Data from ratings of videotapes of 683 preschoolers and their primary caregivers are presented. Across ethnic groups, parental sensitivity correlated with child involvement and parental discipline correlated with child compliance. Mean PCOG factor scores differed between ethnic groups, and correlations between PCOG factors and independent measures of child social skills and family environment differed across groups, suggesting that different aspects of parent–child interaction may have more salience in different ethnic groups. The collaborative process and how “best practice” was applied to the development of the PCOG and videotape protocol as well as strengths and limitations of the PCOG are discussed.

Introduction

Starting early starting smart (SESS) is a multi-site collaborative research study designed to evaluate the effectiveness of integrating substance abuse and mental health services into early childhood and medical settings serving impoverished families and their young children. The collaboration consists of 12 regionally and ethnically diverse programs across the country. The two premises of the program were that: 1) substance abuse and mental health prevention and intervention services would more likely be used by parents of young children if provided in comfortable, familiar settings such as community health clinics and early childhood programs; and 2) the availability of such services would improve a variety of parent and child outcomes including parent–child interaction. The SESS collaboration has been described elsewhere (Starting Early Starting Smart Collaborative, 2001).

The present article reports data on 683 families from seven early childhood education programs participating in SESS (5 head start and 2 preschool/day care centers) serving preschool age (3–6) children including Chinese immigrants, Spanish-speaking Latin-Americans, Native Americans, African-Americans, and rural Anglo-American families.

The SESS steering committee spent 9 months, from September 1997 through June 1998, designing the evaluation and selecting instruments to be used in baseline data collection. The committee consisted of an interventionist and researcher from each site, two parent representatives, as well as three representatives from the Substance Abuse and Mental Health Services Administration (SAMHSA), two from Casey Family Programs, and one from the Data Coordinating Center. Much discussion focused on the roles that the primary caretaker, the family, and the broader environment play in the child's development. Salient to the committee deliberations was the widely replicated finding that a nurturing relationship between the child and his or her primary caretaker protects the child from the powerful negative influences of poverty and its associated risk factors, such as family conflict, parental substance abuse, and community violence (Bernstein & Hans, 1994, Rutter, 1990, Werner & Smith, 1992). Hence, the steering committee decided that assessment of parent–child interaction should be included in baseline data collection.

A challenge emerged from the discussion around instrument selection for assessing parent–child interaction: how can parent–child interaction be assessed in a way that is rigorous enough to be used for cross-site research purposes but that is also culturally appropriate? Among cultures as diverse as the ethnic groups participating in SESS, child-rearing goals, beliefs and practices vary widely (Greenfield, 1994, Lynch & Hanson, 1992, McCubbin et al., 1998).

One approach to the issue of multi-cultural assessment is universalist (Whiteside-Mansell, Bradley, Little, Corwyn, & Spiker, 2001), in which certain core parenting constructs are believed to be relevant across cultures. These dimensions often include parental sensitivity to the child's needs and behavior (De Wolff & van IJzendoorn, 1997, Whiteside-Mansell et al., 2001), parent socialization of the child to cultural norms of behavior (Baumrind, 1996), and the child's attachment to the primary caregiver (Attachment across Cultures, 2003, Posada et al., 2002).

Operationally, within a universalist approach, if a particular measure has equivalent factors when used with different cultural groups, then internal reliability allows cross-cultural comparisons. The universalist approach most often uses a standardized instrument and protocol. This approach has been criticized, however, because the factors may have different cultural meanings (even if the factor structure of an instrument is the same). Hence a particular parenting construct may differentially affect the child's development. Also, different ethnic groups may respond differently to a standardized protocol based, for example, on their perception of authority (Garcia Coll & Magnusson, 1999) or culturally anchored child-rearing practices (Rothbaum, Weisz, Pott, Miyake, & Morelli, 2000). Furthermore, applying a research paradigm or assessment instrument developed in one culture to another is not considered best practice in cross-cultural measurement (Shweder, 1995).

The SESS collaborators also had concerns about the universalist approach, in particular that the methodology would erroneously portray poor and minority parent–child relationships as deficient. Their fears were founded in a history of child development studies, recently criticized (Connors & Donnelan, 1998, Garcia Coll et al., 1995, Greenfield, 1994, Joe, 1994, Laosa, 1999, Parke, 2000, Rogoff & Morelli, 1989), in which ethnic minority families were described as limited, deprived, or deviant rather than different. As Greenfield (1994) states “For too long, minority child development has been viewed exclusively as a series of responses to negative environmental forces, such as poverty, discrimination, and slavery…Rather than focus exclusively on the maladaptive nature of particular socialization practices for the dominant society, it is time to consider and understand their adaptive roles in their culture of origin (pp. 11–14).” The study sites did not want to participate in another investigation that would reinforce negative stereotypes through the use of culturally biased measures. Their familiarity with social service providers' misinterpretation of ethnic parenting practices as abusive or neglectful caused further concern (Fadiman, 1997, Roer-Strier, 2001).

Contradictory findings from research on authoritarian parenting represent an example of ethnic minority concerns about the universalist approach. Mainstream investigators often report authoritarian parenting as maladaptive. An authoritarian parenting style that includes corporal punishment has been reported to increase children's externalizing behavior, and hence, many experts suggest that spanking adversely affects child development (Prevent Child Abuse America, 2003, Straus, 2000). Gershoff (2002), in a meta-analysis, found that corporal punishment was consistently associated with a variety of problematic child outcomes across studies. In contrast, however, some studies have even suggested that corporal punishment can be a protective factor for some non-Anglo children (Baumrind, 1996). For African-American families living in dangerous communities, strict parenting can prevent children from becoming involved with violent people or the anti-social activities that may be close at hand (Deater-Deckard et al., 1996, Gonzales et al., 2000, Mandara & Murray, 2000, Mason et al., 1996, Whaley, 2000). For Chinese families, strict parenting, including corporal punishment, is considered part of proper training and has been associated with positive child outcomes (Chao, 1994). Because of such differences in meaning of parenting behavior across cultures, some developmentally oriented investigators have advocated a culturally anchored approach in which research describes the normal range of relationships within a given culture and how those relationships are linked to child outcomes (Hughes & Seidman, 2002).

Considering these issues related to environmental risk and multi-cultural dimensions of parenting, the SESS steering committee called for the establishment of a multi-cultural workgroup. Representatives from each ethnic group would have input into the design of a parent–child interaction assessment instrument and observation protocol.

This article first describes the literature on parenting practices among the various cultural groups participating in SESS, the issues that arise when non-mainstream cultural practices interact with those of the mainstream, and how these issues affect the multi-cultural assessment of parent–child interaction. Further, it outlines a set of guiding principles that the SESS multi-cultural workgroup negotiated for approaching the assessment of parent–child interaction and in developing an assessment tool, the revised Parent–Child Observation Guide (PCOG: Bernstein, Percansky, & Hans, 1987). Next it describes how different sites modified the set of activities to be videotaped in order to make the activities appropriate for different ethnic groups. The psychometric properties of the PCOG are presented along with a description of procedures to maintain inter-rater reliability. To examine the construct validity of the PCOG, baseline data are presented for the various ethnic groups, along with a description of within group variations in parent–child interaction.

Finally, this article reports the relation between the Parent–Child Observation Guide (PCOG) and measures of child behavior and family environment. We hypothesized that over the sample as a whole, parents who were warm, sensitive, and invested in teaching their children proper ways of behaving would have children who were more socially competent and have fewer behavior problems. This pattern of dyadic behavior was expected to be related to other desirable features of the home and family environment as measured by the HOME scales (Caldwell & Bradley, 1984) and the Conflict Tactics Scale (CTS; Straus, 1979) as well as the child's social competence on the Preschool–Kindergarten Behavior Scales (PKBS; Merrell, 1994). As noted above, parents in different cultures may prioritize different goals for their children and hence focus on different aspects of parenting. Our exploratory hypothesis was that within each culture we expected different patterns of correlations to appear between the measures of parent–child interaction and the child social behavior and family environment measures assessed independently. We expected these differences to be interpretable post hoc in terms of the literature review of cross-cultural parenting practices.

Parenting practices in part reflect larger cultural values and goals for children (LeVine, 1988). “Familial values foreshadow the content and model of cultural transmission…Whereas some—especially Western (individualistic)—cultures emphasize academic, technological, or cognitive modes of social integrations, others—especially African—cultures place primary emphasis on social–affective socialization” (Nsamenang & Lamb, 1994; pp.133–134). This contrast between cultures that encourage child independence such as those in Northern European and North American countries and cultures that encourage child interdependence such as those in African, Asian, and Latin American countries has often been described in the literature on socialization (e.g., Greenfield, 1994, Triandis, 2001).

Interdependent (collectivist) cultures most often have agrarian or Confucian roots (Greenfield, 1994, Hoffman, 1988, LeVine, 1988). Interdependent cultures share the view that a person's life takes on meaning through relationships with others (Nsamenang & Lamb, 1994, Weaver & White, 1997). Parents, other adults, and even older siblings socialize children to share responsibility and resources within the family and community. Children are raised to share and to be obedient, respectful (especially of elders), hard working, helpful, cooperative, and honest. Additionally, it is considered inappropriate for children to be fearful, inquisitive, independent, assertive, or to move away from the family and community when they become adults.

In the United States the cultural norms of the majority are Anglo-American, middle class and derive from a Western European/American heritage. The cultural goals are different from those of interdependent cultures (Hoffman, 1988). Parents want their children to become adults who are independent and economically self-sufficient—to move out rather than stay or return home to help contribute to the sustenance of the family. Anglo-American parents value their children's autonomous behavior, ability to play independently and solving problems on their own, and good verbal and creative abilities. Additional valued child characteristics include the child being happy, generous, assertive, curious and inquisitive, academically achieving, and capable of forming positive relationships with others outside the family (Rubin, 1998). Child-rearing practices for young children viewed as supporting these goals and characteristics include separate sleeping arrangements and structured bedtime, child-proofing the house so the child can explore without restriction, responsiveness to children's vocalizations, feeding on demand, maternal speech that draws the child's attention to exploring objects, distal verbal interaction and eye contact without physical contact, and encouragement to play independently. Most parents adhere to the goal of having the child be able to separate from them without distress. With more mothers of young children in the work force, these child-rearing goals now have economic salience. Structured meal and bedtime schedules, toilet training, separate sleeping arrangements, and day care attendance in infancy and preschool as a young child may be thought of as child-rearing practices that encourage the child to manage separation, a major step toward self reliance and independence (Hanson, 1992, Richman et al., 1988b).

Latin-American parents use different strategies and respond to different aspects of their children's behavior than Anglo-American parents. Young children are overtly cherished and receive a great deal of attention from siblings, extended family and friends. Latin American people value family closeness and transmit this expectation via co-sleeping, and frequent physical and verbal expressions of affection. They appear permissive and indulgent on the one hand, but insist on overt displays of respect and obedience to extended family and community members. When Mexican parents are asked to describe an intelligent child, they list characteristics such as obedient, cooperative, socially competent and motivated. They value family and positive relationships with others over expression of independence and personal achievement (Delgado-Gaitan, 1994, Zayas & Solari, 1994). Puerto Rican mothers have been characterized as valuing demonstration of affection and respect from their children. They respond positively with warmth and affection (Harwood, 1992). They are more likely to be directive and to exhibit physical control with their infants than Anglo-Americans, but their infants are no less securely attached (Carlson & Harwood, 2003).

For Native American children, enculturation (the child's connection to her or his Native American identity and traditions) is considered a central goal of child-rearing practices (Zimmerman, Ramirez, Washienko, Walter & Dyer, 1998). Storytelling and oral tradition are some ways that Native American children learn about their traditions and themselves in relation to others. Joe (1994) expressed the concern that non-Native American researchers believe “…that strong adherence to tribal culture is a strong negative force that prevents Native Americans from being... like other members of mainstream society (p. 108).” However, many Native Americans believe that traumatic historical events aimed at “mainstreaming”, such as educating Native American children in boarding schools, have stripped tribal members of their connection to their families and community and placed their children at great risk for poor development (Long & Nelson, 1999, Weaver & Brave Heart Yellow Horse, 1999). The responsibility for caretaking, teaching, and transmitting tribal norms and values is shared by the parents, extended family, and community members (Cross, 1998). Cooperation with others is emphasized over individual achievement. The learning style of Native American children has been described as more holistic and relational as opposed to the linear and sequential style more typical of Anglo-American children (Tharp, 1994).

Investigators of socialization in African-American families have identified child-rearing practices that can be traced to African roots. At an early age, for example, young children contribute to the child care of their younger siblings. In describing West-African socialization of children, Nsamenang and Lamb (1994) state, “Socialization…is organized to teach social competence and shared responsibility within the family system and ethnic community (p. 137).” In Africa, as well as in the United States with African-American families, language and social interaction focuses on important people in the children's lives (Richman et al., 1988a). As Blake (1994) states, “The traditional cultural emphases of the African-American includes interdependence, extended family, and personal expression (p. 189).” As noted earlier, authoritarian parenting among African-Americans is intended to ensure child survival both in the immediate environment and in mainstream society.

In Chinese culture, children are raised to value familial obligation and to respect parents and elders. Chinese immigrant child-rearing practices are more parent-directed and less child-centered than those of Anglo American families (Rothbaum, Morelli, Pott, & Liu-Constant, 2000). Chinese immigrant parents emphasize school achievement in their children because it brings honor to the family (Chao, 2000). In describing the Chinese immigrant parents, one SESS collaborator stated, “There has been a tendency to pathologize cultural differences that are observed in the way that Asian parents interact with their children—they do not emphasize individuation (a central goal of Western childrearing), rather they emphasize connection (between parent and child).” Cooperation and interpersonal harmony are valued. Self-assertion, conflict, and forceful emotional expression are to be avoided both among adults and children. Chinese parents demonstrate less overt affection and enthusiasm directly to their children than do Anglo-Americans. They are more likely to endorse the use of physical punishment and are more involved with issues of safety and protection than Anglo-American parents. These parenting beliefs are aligned with the Chinese child displaying behavioral inhibition, a desirable trait in Chinese culture (Chan, 1992, Chao, 1995, Ho et al., 1999).

Discussing cultural differences is useful in order to alert the investigator to biases in evaluation/assessment methodology. Drawing distinctions, however, can lead to overgeneralization or stereotyping. Making conclusions based on cultural patterns can result in desensitization to the range of individual differences between cultures that share a common history and within a given culture (Garcia Coll et al., 1995, Greenfield, 1994, Harkness, 1992, Long & Nelson, 1999, Ogbu, 1994). Many Anglo-American mothers, for example, have been found to value a child's being interdependent as well as independent (Raeff, 2000). Joe (1994) notes that investigators often fail to consider individual differences when generalizing research findings, projecting dysfunctional behavior to an entire ethnic group rather than recognizing maladaptive behavior as one point on a continuum of behaviors displayed by ethnic group members. From a research perspective, examining factors related to within cultural group variation becomes equally as important as portraying between group differences when striving to avoid generalizations that lead to cultural stereotyping.

Adding to the complexity of conceptualizing how the culture of origin relates to socialization are intracultural differences, such as urban vs. rural or working class vs. middle class (Gorman-Smith et al., 2000, Howrigan, 1988, Oloko, 1994). The assimilation of immigrant and ethnic minorities into mainstream American culture further complicates generalizations based on ethnic differences. Length of time in the United States may decrease some ethnic differences. Important beliefs about parenting among African-Americans are more similar to those of Anglo-Americans than they are to those of first generation African immigrants (Whiteside-Mansell et al., 2001).

How does one empirically bridge the gap between the culturally anchored and universalist viewpoints? One approach is to posit universal dimensions of parent–child behavior and examine their validity both cross-culturally and within particular cultures. Few large-scale, multi-cultural studies of parenting practices, however, have been conducted either cross-nationally or within the United States. Almost all of these have used the Home Observation for the Measurement of the Environment (HOME; Caldwell & Bradley, 1984) as the assessment of parenting. Generally, the methodology has been to compare and contrast the factor structure and construct validity of the HOME scale data in different cultures.

The HOME is a semi-structured interview centered on regularities of experience in the child's life and combines both parent-report and observations of parent behavior, the family's home, and the neighborhood. The HOME does not assess the child's contribution to the interaction. In a review of cross-national studies of the HOME in both independent and collectivist cultures, Bradley, Corwyn, and Whiteside-Mansell (1996) found that HOME factor structure and construct reliability data in countries that emphasized individualism as a child-rearing goal were comparable to those from studies done in the United States. In collectivist countries, the factor structure was less likely to replicate American patterns. When contrasting the results from the two types of cultures the authors state, “In general, there seemed to be greater cross-cultural equivalence for items assessing cognitively stimulating aspects of the environment than for items assessing socioemotional support…Evidence (for collectivist cultures) attesting to the cultural equivalence (and validity) of HOME subscales was far less plentiful and compelling (p. 251).”

Within North America, these cross-national HOME factor structure and validity findings were replicated. In a large-scale factor-analytic study of children born preterm, the HOME factors for Anglo-Americans and African Americans were similar and corresponded to the published subscales, although only 2/3 of the items were shared by the two groups. The HOME factor structure for more recently immigrated Latin-Americans differed considerably from both Anglo- and African-Americans. There was more similarity in factor structure for cognitively oriented than socioemotional items across the three ethnic groups (Bradley, Mundfrom, Whiteside, Casey, & Barrett, 1994). In a multi-site North American study, the pattern of construct validity for the HOME also differed by ethnic group (Bradley et al., 1989). Family socioeconomic status ranged from poverty to middle class and was matched across ethnic group. The “correlations between HOME scores and social status indices were generally higher for Whites. Total HOME scores, for instance, had correlations ranging from .4 to .6 with social status variables for Whites, compared with .0 to .3 for Blacks and essentially zero for Mexican-Americans (pp. 222–223).” A similar pattern across cultures emerged; for example total HOME scale scores were related to a 3-year child IQ with correlations of .42 for Whites, .50 for Blacks, and .10 for Mexican-Americans. To summarize these multi-cultural studies using the HOME, subscales directed at cognitive development showed more predictive validity across cultures than those directed at social–emotional development. The HOME was generally less ecologically valid with families from collectivist, interdependent cultures than with those from individualistic, independent cultures.

Funding was secured from Casey Family Programs to form a workgroup to provide a multi-cultural perspective for the SESS collaborative parent–child interaction study. Two Asian representatives (one investigator and one interventionist) from the SESS site serving Chinese families, one Native American and one Anglo-American investigator representing the Native American Site, one African-American investigator and one Anglo-American investigator (from two inner city Head Start programs serving predominantly African-Americans), and one Anglo-American interventionist from a predominantly Anglo-American program attended the SESS video multi-cultural workgroup session held over three days in March, 1999. The workgroup discussed the following set of questions from a culturally anchored perspective. These questions derive from what has been identified as fundamental to best practice in multi-cultural research and culturally sensitive intervention (Lieberman, 1990, Pope-Davis et al., 2001):

• How do parents, family and community members show children they are valued?

• How do children show parents that they know they belong?

• What are some central child-rearing goals for your ethnic group?

• How do children act when they are or are not complying with these goals?

• What kind of interaction between parent and child is associated with good for the child's development in your community?

• What kind of interaction between parent and child is not good for development?

• How would you decide that a particular parent–child interaction (an example is used) is problematic or within the normal range? What makes it that way? How come?

The result of this discussion was the following set of guiding principles:

An assessment tool developed from the study of middle class Anglo-Americans may not be appropriate for assessment of poor or non-Anglo Americans without reliability and validity analyses for the groups to be studied. Ethnic or cultural differences identified with such a tool may reflect reliable between-group differences on the measure, but likely are not ecologically valid (Shweder, 1995). The instrument and/or protocol itself may need to be altered to achieve “functionally equivalent” data and experience for the different ethnic and minority groups.

Because of the impact of one's own culture, training, and experience many cultural “outsiders” are not qualified to identify the behavioral range of maladaptive to adaptive (or poor to good) in a culture different from their own. Cultural “insiders” are individuals who have insight and understanding about the range of poor to good in their culture of origin. The mainstream investigator needs to involve the “insider” in all phases of the research study, including instrument development, field testing, subsequent revision, and interpretation of the results (Ho et al., 1999, Segall et al., 1998, Tharp, 1994, Weaver, 1997). Efforts should be made to match the ethnicity of parent–child dyads and video coders (Gonzales, Cauce, & Mason, 1996).

This common ground becomes the starting point for a nonjudgmental discussion of the various cultural scripts that underlie different parenting styles. The challenge becomes to apply this construct given local, culturally diverse variations that arise when answering two important questions: “How do parents, family, and community members show children they are valued?” and “How do children show parents that they know they belong?”

An instrument must be able to generate data able to account for within group variation—ideally with a range from poor to good—and be related to child outcomes (Cocking, 1994, Stevenson-Hinde, 1998). This article describes the within group construct validity of the measure ultimately developed by the SESS collaborators.

The first goal of the workgroup was to develop a parent–child interaction assessment instrument. Attendees brought at least two pilot videotapes of parent–child interaction that, using their clinical judgment, represented a range in the quality of parent–child interaction. The videotapes were discussed in terms of the above questions. Collaborators found it useful to have “outsiders” ask questions based on their observations of the interactions. Representatives from the same ethnic group, the “insiders”, reported that the questioning was helpful in gaining a deeper understanding of their own cultural practices, particularly of behaviors they had taken for granted. The resulting discussion of the videos and parenting practices was similar to the descriptions of parenting in diverse cultures summarized above in the review of the literature.

The workshop attendees then examined item-by-item a tool for assessing parent–child interaction that had demonstrated reliability and construct validity when used with varied ethnic groups including African-Americans (e.g., Bernstein & Hans, 1994, Voight et al., 1996, Wakschlag & Hans, 1999), Dominican immigrants (Bejarano, 2001, Nusbaum, 2000) and in a multi-ethnic study that included Anglo-Americans, African-Americans, and Latinos (O'Connell, 1996). The age-specific versions include: newborn 0–3 months; infant 4–15 months, toddler 16–36 months, early childhood 3–8 years, school age 9–12 years of age, and adolescent 13–17 years of age. For the current study the early childhood version was used as the starting point for the workgroup deliberations.

The PCOG is a tool that can be used flexibly to code interaction in a variety of contexts including everyday situations. Rather than using time or event coding, the PCOG uses global judgment to code each item with respect to the entire interaction. Because parents and children influence one another (Bell, 1968, Sameroff, 1975), the PCOG goes beyond rating solely the parent's behavior and includes the child's as well. Based on resilience research and Goldberg's (1977) concept of mutual competence, the PCOG addresses the question: “What type of communication between parent and child is good for development—both of the child and the parent?”—with the answer: “Any interchange in which the child and parent feel secure, valued, successful, happy, or enjoy learning together.” The answer is operationalized by child items focusing on initiative, involvement with parent, expression of positive feeling, communication and learning with parent. Similarly, parent items focus on balancing the child's self-expression and socialization (discipline), responding to the child's activity and interests, caring feelings shown to the child, and helping the child learn new skills and to communicate. The items, being in question form, are dichotomous (yes = observed and no = not observed). Generally speaking, the various studies using the different age-based versions of the PCOG have yielded similar factors. For the child two factors have been reported: positive involvement and negative emotional expression (or noncompliant behavior as the child grows older). For the parent two basic factors have been identified: sensitivity and teaching (teaching splits into 2 factors as the child turns age 3—teaching and effective discipline).

The workgroup members chose to redefine or expand some of the items. For example, one item on the initial PCOG draft asked “Does child make eye contact with the parent?” Representatives from the Chinese site said that it is relatively uncommon for children and parents to interact with eye-to-eye contact. They suggested that more examples of nonverbal child-involvement behavior (e.g., reaching up toward the parent without looking or, while playing, positioning self in a position oriented toward the parent without directly looking) would be important indicators of the child's involvement—albeit less obvious. The item was rewritten to “Does the child act connected to the parent?” In the coding manual entry for the “act connected” item, making eye contact, reaching for the parent without looking, and orienting toward the parent in space were all listed as examples of the child acting connected.

Representatives from the Native American site suggested that less formal examples of teaching be included as part of observing the parents' behavior, since teaching in their community is less directive and more holistic and relational than the linear and sequential style typical of Anglo-American children (Tharp, 1994). Beyond teaching basic concepts, they suggested that story telling (e.g., relating incidents about family or community members) serves an important educational function in their community and should be included in coding the extent to which the parent interacts with the child in a teaching mode. These suggestions were included under expanded explanations for the items “Does the parent help the child learn basic concepts?” and “Does the parent help the child learn new skills?” Also, based on this discussion, an additional item was added to the first parent category about socialization, i.e., “Does parent/caregiver communicate values, morals and traditions to the child?” This item definition included the parent's encouraging the child's patience, generosity, humor, humility, sense of connection to ancestors, and tribal leaders, as well as discussion of participation in cultural activities. The PCOG version developed by the SESS workgroup will be described in more detail in the Method and Results sections below.

Another goal of the workgroup was to develop a protocol for videotaping parents and children that would have common elements across sites, but would include situations familiar to families from each site. Usually the videotaping was done in a room at the early childhood center at the site. However, practices differed at some sites. For the predominantly Latin-American site (see below), the taping was done at home. One of the workgroup members proposed a videotape protocol that included four situations, listed in sequential order:

• The parent is asked to have their child put a US$15.00 bag of real groceries away on shelves marked with pictures of the items organized by type (drinks, condiments, canned fruit, paper products, etc.). For the predominantly Latin-American site, groceries were put away in their usual place in the home.

• The parent is asked to play with the child “as you would at home” for 5 min. Age-appropriate toys are provided. These include: a story book, farm and farm animals, xylophone, child-size broom, real desk phone, toy doctor kit, 10″ rubber ball, 15″ tall baby doll, blanket, and bottle. The doll is matched to the ethnicity and gender of the child. At the predominantly Latin American site parents were asked to play as they usually would with their child using their own toys. After the 5-min play segment the caregiver is asked to have the child clean up the toys;

• The parent is asked to have the child choose a snack from the groceries on the shelf for the caregiver and the child. A table and two chairs are provided nearby for the caregiver and child to sit at during snack time. Snack time lasts approximately 5 min. At the conclusion of snack time the caregiver is asked to have the child clean up the table. A spray bottle filled with water and paper towels are placed nearby to aid in the clean up process;

• The caregiver is asked to have the child pack up the remaining groceries from the shelves into the grocery bag so the caregiver can take them home. The groceries also served as an incentive for participation in the videotaped observation. This task was not completed at the predominantly Latin-American site, since videotaping was done in the home.

Representatives from the SESS sites participating in the initial workgroup meeting requested various changes to the original videotape protocol in order to capture as much within-cultural group variability as possible. Participants explained that a rigidly standardized protocol could have the unintended effect of reducing variability (by increasing social desirability or resistance in certain ethnic groups). For example, Chinese parents may try to please the examiner at the expense of ignoring the child. The site representatives suggested that the examiner, caregiver and child have tea together at the San Francisco Study Site before turning on the videotape in order to relax the families and allow to them to behave more naturally on camera. The representatives from the Native American site suggested that families on the reservation would feel more comfortable sharing a picnic lunch together at the beginning of the protocol, rather than at the end, based on feedback from a community focus group. Mealtime is considered central to family interaction within the tribe. In addition, tribal representatives chose to substitute story telling or book-reading for the free play component of the protocol because the former is more consistent with what caregivers and children do in their community. The predominantly Latin-American site chose to collect the videotape data in the home rather than at an early childhood education center. The site representatives felt the Latin-American parents' behavior would be too strongly influenced by what they perceived as socially desirable and be too inhibited from interacting naturally with their child. This change was consistent with a protocol developed by Posada et al. (2002) for collecting parent–child interaction data in a Latin American country. Each study site chose the different foodstuffs to be put in the grocery bag, (e.g., fresh bananas instead of canned fruit).

The outcome of the SESS workgroup collaboration was a revised version of the PCOG with a scoring manual that included definitions and explanations for coding items that were deemed acceptable to all members of the workgroup. The workgroup also designed site-specific alterations to the videotape protocol in order to accommodate what representatives found to be appropriate for each particular ethnic group. Although the sequence may vary, for all ethnic groups the videotape protocol included the core activities of putting away groceries, play time and toy cleanup, and sharing a snack together. The resulting protocol instructions were translated and then back-translated into Cantonese and Spanish. The script was revised slightly when a direct translation resulted in a different meaning than in the English script. For example, the word “have” (as in “have your child pick up the toys”) has a different connotation when translated into Cantonese; have implies insistence. The word “ask” was used instead of “have” in order to avoid provoking different behaviors among Chinese speaking parents.