Parenting in the South Seas. Educational Skills in Remote Archipelagos of French Polynesia 1

L’articolo presenta i risultati di una ricerca esplorativa sulle pratiche e gli at-teggiamenti genitoriali in tre arcipelaghi periferici della Polinesia francese: le Marchesi, le Australi e le Isole Tuamotus. L’obiettivo era identificare le tipologie di genitorialità in contesti isolati utilizzando la Scala di valutazione delle compe-tenze genitoriali ( Échelle des Compétences Éducatives


Introduction
Children success in adapting to their school and social environment is significantly influenced by interactions between individual factors -such as intellectual or physical disabilities -and factors originating from the immediate environment: the interactive ecosystem in which children grow up. Among the factors that predict a child's school results and successful development, a growing number of research works has been confirming the pivotal role played by parenting skills (Frick, Christian, Wootton, 1999;Gryczkowski, Jordan, Mercer, 2010). As shown by Terrisse and Larose (2001), children with disabilities display more resilient attitudes if raised in supportive family environments; conversely, children with no special needs may experience difficulties if family environment is unsupportive. Pourtois (1979) found that 84% of the variance in school performance is attributable to the family environment: the parents' behaviors, attitudes, personality traits and intellectual capacity but also their social status and workplace culture. Evans and his team (2012) have recently demonstrated a link between unfavorable (inconsistent and punitive) parenting practices, mother's beliefs, and problematic child's temperament (shy, antisocial, or aggressive).
While American neo-behaviorist approaches have systematically identified distal variables, in particular the parents' socioeconomic and educational levels, as predictive of success or failure at school (Waxman, Padron, 1995;Garnier, Stein, Jacobs, 1997), ecological approaches, particularly as followed by Canadian researchers, have been more concerned with the study of proximal variables, such as parental attitudes and practices (Deslandes, 1996(Deslandes, , 2005Larose et al., 2004;Terrisse, 1996). This paper summarizes the results obtained in the framework of a research project aimed to describe parental attitudes and practices in remote archipelagos of French Polynesia, a French overseas territory in the Pacific Ocean 7 . We worked with a coherent sample of 120 parents of children aged between 8 and 10, without disabilities, living in the Marquesas, the Australs, the Tuamotus and the Gambier Islands. To identify self-reported parental practices, we used a psychometric tool developed by Bernard Terrisse and his colleagues, the Parenting Skills Evaluation Scale (Échelle des Compétences Éducatives Parentales, henceforth ECEP. Terrisse, Rouzier, 1986;. 7 The PrEEP project (Pratiques éducatives enseignantes et parentales en Polynésie française, Educative practices of parents and teachers in French Polynesia) was allowed a financial aid by the French Ministry of the Overseas Territories. For a detailed description of the project see Ailincai and Delcroix (2017). The first results of the survey presented in this article have been briefly diffused (in French) in Ailincai et al. (2021).
We hope that our work will contribute to the actual debate about the predictive factors of a child's favorable long-term development and the opportunities offered by existing approaches for the prevention of possible school difficulties, especially in overseas, postcolonial, or isolated contexts.

Parenting, education, and child development
Skills are usually defined in relation to the ability to perform an action with good results: they are based on knowledge and experience acquired in a particular field of activity. The notion of skill has been entering in education and training studies since the 1970, thanks to the seminal works of Lester Thurow (1972) and Nobel Prize Joseph Stiglitz (1973), two economist that worked to understand the impact of education on economic development. While at first this notion was applied exclusively to teachers, other classroom actors and pupils, it is now increasingly being applied to parents and domestic education dynamics (Sellenet, 2009). According to Larivée and his colleagues (Larivée, Terrisse, Pithon, 2005;Larivée, Terrisse, Richard, 2013), informal and family education is built not only on knowledge and technical abilities, but also by appropriate attitudes towards learners. Madge (1983) and Trudelle (1992) highlighted, for indeed, that parents -as like as professional educators -tend to meet children needs mobilizing their savoir vivre and their physical, social, emotive, and intellectual capacities. There is a longstanding consensus about the principle according to which the quality of adults' interventions is based on their educational skills, and it influences children behaviors shaping their attitudes and feelings (Hersey and Blanchard, 1978). Pourtois and Desmet (2000) consider the development of parental skills as an evolutionary process implying an interactive dynamic -between individuals, natural environments, and socio-cultural contexts -subject to irregularities (satisfactions and disappointments) and influenced by endogenous family dynamics (parents' characters, marital relationship, socioeconomic status, social networks. Terrisse, 1996), but also by cultural values (educational ideologies, educative goals, and conception of social success. Ailincai, Jung, Alì, 2012;Alì, 2016a) 8 . Knowledge and skills shape parental practices, that Terrisse and Larose (2009) consider as adaptative and evolutive traits of the parent's personality. Parental skills and practices determine the way parents propose, justify, evaluate, and legitimize knowledge claims according to their educative style 9 (and their educational ideologies), both in everyday and epistemic activities (idest, activities aimed to generate knowledge. Pekrun et al., 2017): in short, the parenting style. Over the years, the scientific examination of parenting skills has led researchers to identify different parenting styles (a first systematization was proposed by Deslandes and Potvin, 1998), to observe a close relationship between parenting skills and parental self-esteem (Mash, Johnston, 1983;Main, Goldwin, 1984) and to detect that parent's self-efficacy 10 is linked with flexible parenting practices and democratic parenting style (Trudelle, 1992). Balance in top-down and bottom-up domestic interactions enhance child's initiative, curiosity, critical attitude, originality, respect for others, 8 A first attempt to describe parental ideologies in French Polynesia, based on an ethnopsychological approach, was presented by Levy (1975). More recently, Alì (2021) has been exploring educational ideologies among family and educators in the Marquesas Islands, using an ethnohistoric perspective and concluding that their conception of school and social success is nowadays shaped by global dynamics and labour market's needs (see also . Other researchers have been working on parenting in Polynesia during COVID-19 pandemic (Cadousteau et al., 2021). 9 The notions of educative and educational style appear since the 1970s to summarize the way the educator educates (Leichter, 1973) and they have not to be confused with the more polemical notion of learning style (the way a student learns. For a synthesis of the debate, see Norman, 2009). 10 According to Bandura (1997), the individual's belief in their capacity to execute a task and to achieve a goal (see also Rodebaugh, 2006). individual determination, facilitating the child's moral and cognitive development (Baumrind, 1966(Baumrind, , 1967Lautrey, 1995) and several authors have demonstrated that suggestive and empowering styles -akin to democratic practices -facilitate children ability to learn (Ailincai, Alì, Alby, 2018;Ailincai, Weil-Barais, Caillot, 2005).
Unlike parental practices, the notion of parental attitude is less concrete. It defines the parent's way of relating to the child and their ability to react (positively or negatively) to a situation (Schaefer, Bell, 1958). Social psychologists Fishbein and Ajzen (1975) proposed a theory of reasoned action to explain the correlations between attitudes and probabilities of producing or structuring behaviors. In this theoretical model, attitude has a predictive value with respect to practices (moderated by the sense of control that the individual has over his or her own behaviors). Rotter (1966) proposes the concept of locus of control to denote an individual's tendency to consider a phenomenon as the result of their own behavior (internal locus), or as controlled by factors over which they have no influence (chance, destiny, institutions, etc.) and occurring independently of the individual's own actions (external locus). To understand those dynamics and the general operating mode of parenting, in recent decades parental ideologies, skills, practices, and attitudes have been deeply explored and a growing corpus of research has been consecrated to detect, identify, measure, and analyze protective and critical factors in children development.

Measuring parenting skills and practices
Based on an interactional approach, Terrisse (1996) places the variables that can influence parenting skills and child development within an ecosystemic architecture. These factors are combined in a horizontal structure on three levels, based on Belsky's model (1984), and grouped according to Bronfenbrenner's criteria of ecological systems of children development (1979): the first level correspond to the Bronfenbrenner's micro-systems of primary socialization (the family, the school); the second level, is the mesosystem (where the microsystems interact) and the third level is the macrosystem (cultural and social values, institutions, global markets). Therefore, factors influencing parenting skills and their evolution in a done context seem to be related with parents' personality, their sense of competence, their values, their attitudes, and their child's situation (health and behavior).
Nevertheless, the accuracy of data gathered from self-administered questionnaires is often questioned because of various forms of bias well knew by social psychologists such as social desirability bias, researcher bias or respondent bias (Nederhof, 1985). As observed by a large number of researchers, saying and doing are two activities that are not necessarely coherent and it exists a gap separating self-reported and actual practices (Kochanska, Kuczynski, Radle-Yarrow, 1989;Dekovic, Janssens, Gerris, 1991;Deslandes, 1996). A more comprehensive research approach should to be based on both factor (declared and observed practices), including systematic observations, ethnographic fi eldwork, behavior sampling and critical discourse analysis but this kind of research scheme needs of a large scale programme management, a consistent team of researchers (and fi eldworkers) and a substancial fi nancial aid that is often out of scope in peripheral, isolated or marginalized envornments. Most of the work we cited in this section are consacrated to urban contexts in industrialized countries but our work would be contribute to this debate presenting the results of a research aimed to analyze declared parenting practices (and educational discourses among parents) in a postcolonial context, French Polynesia: a former colony of the French colonial Empire nowadays integrated to the administrative structure of the French Republic as an overseas territory.

Research context
French Polynesia consists of many widely dispersed islands in the Pacifi c Ocean, spread over an area equivalent to that of Europe (see Fig.1). It consists of fi ve archipelagos with a total of 118 islands: the Society Islands (242 726 inhabitants), the most populous archipelago, including the main island, Tahiti, situated 17,000 km to the south-west of Europe; the Marquesas (9 346 habitants); the Austral Islands (6 965); and the Tuamotu and Gambier Islands, combined in the same administrative district (16 881). Our study was carried out on the last four archipelagos: scarcely populated (by autochthonous communities), more isolated from the urban values of the capital, Papeete (in Tahiti), and still animated by a traditional way of life. Their rural environment is characterized by an economy of subsistence mainly based on agriculture, breeding, sheep farming and artisanal fishing. Their isolation is both geographical and social: the Marquesas consist of mountainous islands, located approximately 1,400 km north-east of Tahiti; the Austral Islands are located more than 500 km south of Tahiti; to the east, the Tuamotu-Gambier Islands ensemble is composed by 76 atolls scattered over an area of two million square kilometers. These archipelagos are accessible from Tahiti by air (the main islands dispose of airports and some remote islands of simple landing strips) but the local transports are guaranteed only by the sea. Even if French Polynesia is connected to the global network, in remote archipelagos the internet connection is unstable and exposed to the vagaries of the climate: for local communities, ultimately, the integration to the global village is still far to be achieved.
While the average monthly household income in French Polynesia is 398,900 XPF 11 (equivalent to 3,343 €) and 497,200 XPF (4,166 €) in urban Tahiti, we observe much lower incomes in remote archipelagos (Benoit, 2017;Champion, 2018). The average monthly income per household in 2015 was 299,200 XPF (2,507 €) in the Marquesas Islands, 235,800 XPF (1,976 €) in the Austral Islands, and 319,000 XPF (2,673 €) in the Tuamotu-Gambier archipelagos, with many households consisting of several families living under one roof.
French is the official language -vehiculated by school, media and administrative lifebut every archipelago is characterized by its own native language. There are seven distinct Polynesian languages spoken in French Polynesia: Tahitian, Pa'umotu, North Marquesan, South Marquesan, Mangarevan, Rapa and Austral (Charpentier, François, 2015). Tables 1  and 2 show the overall linguistic situation of the archipelagos where our study was carried out. Over the years, local educational authorities have been dealing with the dispersion of the school population, especially in scarcely populated remote islands, and with the goal to assure the functioning of the educative system, even in the most isolated district and even for a small number of pupils. Nowadays, French Polynesia dispose of a public school system based on 203 primary schools and 46 secondary schools attended by more than 51,000 students (DGEE, 2020). The private system disposes of 34 schools: 20 primary schools and 14 secondary schools, most of them linked with religious congregations 12 .

Research method
This research was aimed to better understand parenting dynamics in some remote archipelagos of French Polynesia. We choose to focalize on two groups of variables: • At a microsystemic level, the proximal variables: parental attitudes and practices.
• At a macrosystemic level, the distal variables: parents' socioeconomic status, level of education, families' vulnerability index.
Finally, we looked for tendencies and correlations between the two groups of variables, interpreting and discussing the results we obtained by taking into the account the peculiar situation of the research context.
Prior to the deployment of research activities, a formal approval was obtained from the ethics committee of University of French Polynesia. Written informed consent was obtained from the participants prior to the survey. We first informed the education authorities in Polynesia, then the directors of the schools, as well as the teachers and the parents. The most important step was obtaining the consent of the parents who participated in the study. All documents were then anonymized so that no names could identify the parents.

Study population
The data was collected from 120 parents 13 -of pupils aged between eight-and nine-years old -living in the Marquesas, the Austral, the Tuamotu and Gambier archipelagos. Participants were asked to complete the questionnaire during a school meeting 14 . Most of them (75%) were not able to complete the task autonomously and needed an assistance to translate the questions in local languages (because of their scarce knowledge of French).

Measurement tool
The ECEP tool is a Parenting Skills Evaluation Scale. The first version was developed in 1986 but we used the last available version (Terrisse and Larose, 2009) 15 . The scale uses 47 questions to explore three themes: 12 The role played by religion and religious congregations in French Polynesia culture and social organization (and the impact of congregational private schools) has been deeply analysed by Saura (2013). 13 Both biological and legal parents. In French Polynesia but also in other Oceanian archipelagos the practice adoption is widespread and regulated by the fa'a'amu customary law (Rigo, 2004;Rollin, 1974). In Polynesia, faʼaʼamu is generally managed within the same kin network, privileging family members (older siblings, aunts, or uncles) as adoptive parents (Asselin, 2021).
14 In the framework of the PrEEP Project (see footnote n.7), some of the parents participating in this survey were also interviewed and filmed at their home, with their children. The activity was aimed to better understand the gap between their educational discourses and family praxis. Results were recently published by Ailincai and Sramski (2020). 15 There are different versions of the ECEP tool according to the age of the children. For our research, we used the version destinated to parents with children aged 6-9.
• parental attitudes, measured crossing the values of Rigid/Flexible and Distrust/Trust polarities; • parental practices, based on Normative/Elaborative and a Strict/Liberal polarities; • parental locus of control, made up of five separate items.
The items related to the locus of control were introduced to the ECEP tool in the 2009 version to refine the interpretation of the scores for attitudes and practices. According to the theories on sense of control (Fishbein, Ajzen, 1975;Rotter, 1996), parents indeed perceive their child's behavior as the result of both endogenous and exogenous dynamics: on one hand, their parental performance and on the other hand, the role played by children agency or by external agents of socialization (at the school or belonging to the neighborhood). Therefore, parents focused on endogenous dynamics (or internalities, adopting an internal locus of control) tend to be more proactive than those who display an external locus of control (externalities, more focused on endogenous causes). Integrated to the ECEP tool, this item allows to better understand how parents value their parental role and the impact of their educational activity 16 .
To evaluate the efficacy of the ECEP tool in Polynesian context, previously to the main survey, we carried out a preliminary trial test with a sample of six parents. This trial test allowed us to detect some vulnerabilities in the original version of the tool needing some adaptations to avoid any misunderstanding with Polynesian participants. We reformulated some questions (privileging simplified French vocabulary) and we choose to make easier the input process, asking the parents to tick a box indicating whether they agreed or disagreed with the statement rather than entering values of 0 or 1 for agreement or disagreement. Also, we adapted some questions to Polynesian context: • Using the school levels according to the French education system; • Taking into the account the monthly rather than the annual income; • Using XPF rather than Euro.

Impact of socioeconomic variables
The ECEP tool address the socioeconomic variables most widely used in research literature to evaluate vulnerability: family income, parent's educational level, housing type, leisure practices, neighborhood network and social support (Ligon, Schechter, 2003;Gallardo, 2020). We interpreted these items taking into the account the peculiar context of remote archipelagos and avoiding any generalization with an abstract "Polynesian context". Adopting a more contextualized approach, we can explain some local specificities. Whereas the relative poverty threshold 17 for the Westernized zones of French Polynesia (Tahiti, Bora Bora or Moorea) is indeed 67,395 XPF/month (565 €/month), in remote archipelagos -where rural lifestyles are rather different, shaped by a mode of production based on subsistence farming, self-consumption, and barter - 16 Recently, a team issued by the Department of Psychology at Emory University (Atlanta) and the Bristol Medical School found that the greater value assigned by parents to externalities, the greater the increased risk of the child's adverse behaviour as rated by teachers (Nowicki et al., 2018). The risk was generally greatest if both parents adopted an external locus of control and lowest if both were focused on internal locus. Their work confirm that the locus of control plays a pivotal role influencing parental attitudes and practices. 17 Economists measure poverty in absolute or relative terms: "in countries with a very low level of development, it is often the absolute poverty threshold that is measured, idest the poverty threshold is based on the notion of a basic minimum for survival. In developed countries, poverty is measured in relative terms, idest the poverty threshold changes with movements in the country's standard of living" (Tepava, Vucher-Visin, 2005, p.1). the threshold is reduced to half (or less) : 33,544 XPF/month (281 €/month) in the Marquesas, 26,168 XPF/month (219 €/month) in the Austral Islands, and 31,158 XPF/month (261 €/ month) in Tuamotu Gambier archipelagos (Herrera and Merceron, 2010). The relative value of self-produced food sources exceeds expenditure by 22% in the Marquesas, by 23% in the Austral Islands, and by 15% in Tuamotu-Gambier (Benoit, 2017). In these archipelagos, two housing types coexist: the traditional rural fare 18 and the social housing unit. Family structure is nuclear, and most households are relatively small (with a predominance of single-parent and one-child families), the average number of occupants per household being 3.5 (ISPF, 2021).
As proposed by Terrisse et Larose (1999), to contextualize our results, we measured the family environment. We used their Index of Disadvantage (ID), based on the core variable of monthly family income (scored on a four-point scale, as in Table 3). To calculate ID, family income is associated with four secondary variables (each scored with 1 point): parents' educational level 19 (1 point for higher than upper secondary), their employment (1 point if at least one parent hold a permanent job), family structure (1 point for two-parents and 0 points for single-parent families) and number of children (1 point for families with 1 or 2 children; 0 for families with 3 children and more).

MI≤50,000 XPF
Family ID values are therefore distributed on a scale from 0 (very disadvantaged) to 8 points (not disadvantaged). Families with a score of 7 or 8 points are considered as not particularly vulnerable.

Results
A total of 125 parents accepted to participate to our survey: mainly mothers (121 participants), with only 24 fathers. Five questionnaires were not included in the results because of a failure to reply to more than one question and the final research sample was based on 120 valid questionnaires: • 97 were issued by mothers and 23 by fathers.
• 38 were issued from parents living in the Marquesas, 57 from the Austral, and 25 from Tuamotu-Gambier Islands.

Family structure and organization
Most of parents participating to our study lived below the relative poverty threshold (82.5%), disposing of a monthly income less than 250,000 XPF (2,095 €), well below the average of French Polynesia as a whole (398,930 XPF or 3,343 €. Champion, 2018). The incidence of single-parent families was relatively high (15%) compared to the average of French Polynesia as a whole (6%). Most of the households hosted three children or more (68%) and less than one third (32%) hosted 1 or 2 children: a distribution close to the average of French Polynesia as a whole (29%, as reported by the Institute of Statistics of French Polynesia. ISPF, 2015). Globally, our sample is characterized by a low socioeconomic status (see Tab Austral 11 20 26 Tuamotu-Gambier 5 14 6 Total n=120 38 (32%) 41 (34%) 41 (34%)

33
Austral 13 44 Tuamotu-Gambier 0 25 Total n=120 18 (15%) 102 (85%) Table 3. Socioeconomic data Our sample is distributed across five age groups, with most parents (52%) aged 41 or more. The proportion of unemployed parents (31%) corresponds to the unemployment rate of the active population in Polynesian remote archipelagos (30.6%, as calculated by the ISPF, 2015). Furthermore, only one third of our sample (38%) declared a permanent formal employment: the others survive with informal or occasional jobs. A considerable part of the parents participating to our study (61%) declared they never achieved the compulsory education (see Tab. 5): this rate is relatively close to that of all the remote archipelagos combined (53%) but higher than that of French Polynesia as a whole (40%). The results we obtained show that three quarters of our sample live in a poor or very poor socioeconomic situation (with and ID less to 4 points), 22% in a moderately poor situation (ID of 5 or 6 points), and only 3% of the parents are living in a favorable family context (ID of 7 or 8 points). Even if our sample it is not representative of the demographic distribution of the general population (with an overrepresentation of mothers relative to fathers), it is still statistically significant and coherent with the socioeconomic situation of Polynesian remote archipelagos (according to the data issued by the ISPF).

Parental attitudes
The ECEP tool allowed us to identify the parental attitude of participants to our study and to determine the prevalent attitude according to the archipelago (see Table 7). Attitudes are grouped in three major categories: controlling, moderately normative and laxist attitude. According to the number of points issued by the ECEP, it is possible to associate every parent with a typological profile. The data we obtained show that 56% of the parents adopt a controlling or extremely controlling attitude (10 to 13 points). For them, the pillar of domestic life is parental authority and the educational relationship with their children need to be based on a rigid hierarchy. Their educational style involves direct interventions, negative rather than positive feedback, reprimands, and little openness towards dialogue, explanations, and communication (Terrisse, Larose, 1998) 20 . Less than half of the study population (43%) adopt a moderately normative attitude, combining authoritative with suggestive approaches: these parents actively stimulate their children with the aim to develop creativity and autonomy. The sub-scales associated with this item (aimed to measure the degree of intensity of an attitude on a Rigid/Flexible and Distrust/Trust polarity) indicates that in our sample, parents declare extreme attitudes, with an overwhelming majority of parents demonstrating extreme rigidity in education and extreme distrust towards their children (see Tab. 8).

Rigid/Flexible sub-scale polarity
Extreme rigidity (score 4-5) Moderate attitude (score 6) Extreme flexibility (score 7-8) Tuamotu-Gambier (n=25) 23 92% 2 8% 0 0% 20 A deeper analysis of this normative educational style and its implications on child development has been proposed by Ailincai, Weil-Barais and Caillot (2005). About his prevalence in postcolonial contexts, see Alì and Ailincai (2019). We crossed these results with the data concerning the locus of control following the procedure proposed by Terrisse and Larose (2009) and according to other paradigmatic works (Campis, Lyman, Prentice-Dunn, 1986;Lefcourt, 1991). Most of our sample (61%) show a strong concern on internalities (scores of 10 to 15 on the internal-external control scale), correlated with a very directive attitude (56%) or with a moderately normative attitude (43%). One third of the parents (38%) show a balanced level of control, granting a fair value to both internal and external causes (represented in the diagram by scores of 16 and 17): this balance is generally correlated with moderately normative attitudes (scores 14-16 on the attitudes scale) but it may also be associated with temporarily rigid behavior in response to a specifi c situation of loss of control (see Fig. 2).

Fig. 2 -Parental attitudes and locus of control
Only one participant shows a locus of control extremely focused on externalities (a score higher than18 points). As suggested by Terrisse and colleagues (2000), this parental typology is often associated with rather fatalistic and laxist attitude and with a scarce involvement in children schooling. However, the results we obtained seem to indicate that parenting among Polynesian remote islanders is mostly characterized by a supervisory, authoritarian, controlling attitude considering children development a "family business". Finally, we related the variance in attitudes and locus of control to the Index of Disadvantage (ID). The Chi-squared test (χ2) does not reveal any clear dependence between attitudes and the ID (p > 0.05) 21 suggesting that, in small-size remote contexts (such as the Polynesian islands), local communities has been developing over the years (and over the centuries, sometimes) shared educational attitudes -shaped by local habits, socioeconomic ecosystems, and ideologies -adopted by the whole community (integrating all the social sectors and statuses) 22 .

Parental practices
The ECEP tool was originally designed to identify parental practices and to define parenting typologies based on three categories: • parents privileging formal practices; • parents privileging moderately normative practices; • parents privileging liberal practices.
• Based on this typology, we were able to determinate the parental practices privileged in every archipelago (see Tab. 9).
Formal practices score 16 to 20 Moderately normative score 21 to 24 Liberal score 25 to 32 Total n=120 0 0% 5 4% 115 96% Table 9. Parental practices according to geographic residence In our sample, attitudes are all relatively represented in all the archipelagos, but the situation with declared practices is slightly different. An overwhelming majority of the parents participating to our study is associated with an extremely liberal parental profile (96%). This typology is close to the negligent and detached interactive style described by Ailincai, Weil-Barais and Caillot (2005), characterized by the absence of parental supervision and scarce educative interactions in the domestic environment. According to the ECEP manual developed by Terrisse and Larose (2009), this profile is associated with a low level of educative feedback, and laxist attitudes, and it may have a negative impact on the balanced development of children sense of self-esteem and self-efficacy. In our sample, parents combine elaborative practices with strictness (see Tab. 10), showing scarce interest in child's activities but being disposed to exercise authority, either by explanations (discussing with the child), or by threats (with no explanation nor discussion). 21 Looking for statistical correlation, we choose to adopt the theoretical value of the threshold of probability the most accepted in social and human sciences (p < 0.05). 22 Alì (2016a; has recently suggested a similar hypothesis, based on comparative study of autochthonous communities in remote villages of the Amazon rainforest and in a Marquesan Island (see also . Our results suggest that, in remote Polynesian archipelagos, parental practices are made up of extremely elaborative combined with strict behaviors closely associated with an internal locus of control (as in Fig. 3).  Terrisse and Larose (2009). Th e bold arrows show the correlation we found in our sample.

Normative/Elaborative polarity (score 9 to 18)
Th is unusual situation was not described by Terisse and his colleagues, who generally have been associating liberal attitudes with a focus on externalities: a parenting typology facilitating the development of an enfant-roi 23 personality. We suggest that this new typology could be associated with specific local contexts (with a strong cultural identity), where customs and traditions assign explicit and ritual responsibilities to parents (facilitating the development of an internal locus of control) but where actual practices are shaped by a disadvantageous and vulnerable socioeconomic environment (such as in postcolonial, peripheral and marginalized contexts). As with attitudes, we did not find any possible correlation between practices and the ID but, in this case, we can explain these results because of the extreme polarization of the data around the items: on the one hand, an ID shared by most of the sample, and on the other, a homogenous typology of parenting practices.

Parental attitudes and practices
The last results we obtained thanks to the ECEP tool was to correlate parental attitudes and practices. The standard deviation indicates little dispersion of the values around the sample averages, and therefore a close correspondence between practices and attitudes (see Tab. 11). This low standard deviation suggests a high degree of homogeneity in practices and attitudes, which could be attributed both to the highly specific nature of the island context (its isolation and the role of the local culture) and to a general situation of vulnerability, confirmed by the data on ID. In other words, in these "closed contexts" where everyone knows each other, people tend to maintain a coherent behavior (and parents' attitudes are coherent with their practices) due to the relative absence of privacy, the discrete presence of the community control, and the risk to be exposed to social stigma. 23 The notion of enfant-roi (the "child king"), widespread in the Francophone research community, is difficult to translate. According to Simone Korff-Sausse (2007), from a psychological point of view, it defines a child who harbours the illusion of infantile omnipotence (developing the Freudian concept of "His Majesty, the Baby", as in Freud 1914. See also Szwec, 2007). If the illusion develops into a pathological condition, the enfant-roi becomes a "child-tyrant" (Renier, Schrod, 2008). During the XX th century, sociologists and anthropologists applied this notion to those human communities where children spent most of their daytime escaping the control of adults. About its use and misuse in a Polynesian context, see Troadec (1996) and Alì (2021).

Discussion
Evaluating parenting attitudes and practices with the ECEP tool allow us to identify some local trends, incoherencies and specificities typifying the context of Polynesian remote archipelagos.
Firstly, the entire study population adopt liberal parenting practices, with most of the parents being extremely elaborative. Coherently, their attitudes are moderately normative but, paradoxically, also strongly polarized and extremely rigid. In their domestic environments, children are relatively autonomous but, when the situation became too difficult to control, parents de not hesitate to formulate threats and to impose sanctions, without providing explanations: a parental compensatory reaction to re-establish the order at home (as demonstrated by Nelson et al., 2009). According to Chang and his colleagues (2004), coexistence of extremely rigid attitudes and liberal practices among parents are predictive of developmental difficulties in the child. Nevertheless, our parents show a certain confidence in their parental skills and tend to feel they control the impact of their parental activity and the future effects on children development: in fact, an overwhelming majority of our sample indicated an internal and midrange locus of control (99%). According to the Lefcourt's typology (1991), parents focused on internalities, such as in our case, are more predisposed to reconsider their convictions and their parenting mode. For indeed, they are more incline to participate in parental awareness programs promoting child-friendly practices 24 .
The results obtained thanks to the Index of Disadvantage show a situation of general disadvantage and economic depression. The homogenous nature of the sample (a rather rural population, ethnically homogeneous, belonging to the same autochthonous population, in a context of geographical isolation and remoteness) could explain the tendency to express a similar pattern of practices. The situation is slightly different in other South Seas archipelagos, especially where population ID variance is more significant (see, for indeed, the case of Tahiti in Cadousteau et al., 2021). The multivariate analysis of the external correlations shows that, for our sample, the only socio-economic variable influencing parents' behavior is their age (and only on the Strict/Liberal index): actually, parents aged 36 and older display a mainly moderately normative attitude, whereas extremely strict practices are found in younger parents.

Limitations of this study
Even if the results we obtained allow us to identify a parenting typology associated with Polynesian isolated contexts, they are limited to parents' perceptions and declarations (and refer to self-reported parental practices). A methodological triangulation of the research, with other corpus, collected by more qualitative methods (observations, recordings, interviews) would allow a better understanding of this phenomena. As seen in the first sections of this paper, these corpuses exist, focused in both declared and real practices: it will be up to the researchers to compare and critically analyze the data.
Moreover, a growing number of systematic research carried out in French Polynesia has been widening this field of research, describing parent-child interactions, and contributing to identify educative praxis in family environment. In recent years, researchers have been observing a general trend towards a mainly directive parental practice: Ailincai and Sramski (2020) described epistemic activities in Rurutu families, Marchal and Ailincai (2019) observed systematically the reading activities in Tahitian families, and Alì (2016a) sampled educative interactions in the Marquesas, discovering an increase of directive practices on working days, when parents go to work, and children go to school.
Concerning the ECEP, the scientific literature highlights the possible biases inherent to all psychometrics and attitude measurement scales. We presented at the beginning of the article the possible biases related to the use of the ECEP and our interventions to adapt the questionnaire to the context of this study. Even if this tool has been the subject of several validations, we are conscious of the limits of self-assessments and we suggest the use of complementary techniques which would allow a more precise assessment of the phenomenon studied: observation (discrepancy between saying and doing); observation of non-verbal and extralinguistic acts (gestures, postures, expression, prosody); collect the children's point of view; collection of teachers' point of view; full interviews. These limits and perspectives related to measurement scales reinforce the need for methodological triangulation.

Epistemological outcomes
We are also going to highlight some strengths of this study: firstly, the data collection was carried out thanks to the mobilization of partners in the field, known and trusted by the population we surveyed. Secondly, the project team was composed of several local researchers who grew up in the islands we studied, speaking local languages (Tahitian, Puamotu, Rurutu and Marquesan), experienced with Polynesian culture and able to provide an emic point of view 25 . This mixed team, combining interpretative views from inside and outside the community we studied, allowed a detailed and comprehensive examination of parental attitudes and practices.
Also, our research contributed to better understand some peculiar dynamics proper to rural and postcolonial contexts, where most of scientific literature has been focusing on urban and industrialized contexts.

Conclusions and recommendations
Even if the remote archipelago of French Polynesia we studied seems to be an isolated case, we need to consider that, mutatis mutandis, they share a significant number of socioeconomic variables with similar postcolonial, peripheral, and marginalized contexts: a colonial past, a situation of general economic stagnation, a high level of concentration of resources, the presence of autochthonous communities, local customary law, and identity revindication. Those factors exert a certain influence on the diffusion of hybrid educational ideologies adapting local habits and value systems with administrative top-down policies and injunctions The parenting mode we observed among Polynesian remote islanders was observed too in other similar contexts where adults seems to be more prone to liberal parenting, allowing children an unsupervised autonomy combined with moments of strictness and rigid behaviors (New, 1994;LeVine, Norman, 2001;LeVine, New, 2008;Alì, 2016b) 26 . Far to be fatalist, the parenting typology we observed in our sample is mainly focused on internalities and conscious of the role played by domestic environment on children development.
The ECEP scale confirmed his efficacy as a research tool. In our preliminary phase, we detected some difficulties, especially in the questions' formulation. Some modifications were needed to adapt the original version (in French language) for Polynesian users: simplifying the vocabulary and better explaining the instructions. Nevertheless, even if the tool is an excellent way to detect local parenting typologies or family fragilities, it is limited to self-declared practices and a more comprehensive approach need to associate ECEP psychometric tool with fieldwork, in situ systematic observations and behavior sampling.
We hope that our work may contribute to the emerging debate about postmodern parenting: our case of study confirmed that, even in remote and isolate contexts, parental practices are shaped by macrosystemic issues, fashioning local educational modes. These praxis -result of a hybridization between the colonial heritage and the native culture -are often indecipherable by foreigners but they respond to a systemic logic, they consent to adapt community lifestyle to the local socio-economic and natural environment, and ultimately, they allow people to survive and to live in such peculiar contexts.
Finally, concurring with the results obtained by other studies on school-family ties, which advance the idea that parental practices and attitudes are predictive of favorable long-term child development, we hope that our research will be an useful tool to design and develop more contextualized parental education programs, adapted to the specificities of overseas, postcolonial or isolated contexts.