Utilizing a cross-sectional survey design [46], a total of 296 families were sampled using a multistage random sampling technique from a population of 11 districts in Padang City, West Sumatra, Indonesia. The age range of respondents was 15 years - < 64 years. In terms of age, 4 respondents (1.4%) were 20 years old, 24 respondents (8.1%) were 21 - 40 years old, 264 respondents (89.2%) were 41 - 60 years old, and 4 respondents (1.4%) were < 60 years old. In terms of educational background, 24 respondents (8.1%) were primary school graduates, 39 respondents (13.2%) were junior high school graduates, 96 respondents (32.4%) were senior high school graduates, 13 (4.4%) had a diploma, 90 respondents (30.4%) had a bachelor degree, 31 (10.5%) had a master degree, and 3 (1.0%) had a doctoral degree. Based on the occupations, 78 respondents (26.4%) were civil servants, 157 respondents (53.0%) were housewives, 21 respondents (7.1%) were self-employed, 40 respondents (13.5%) and 21 respondents (7.1%) were private employees. The distribution of respondents based on the number of children is as folloes: 218 respondents (74%) had 1-3 children, 73 respondents (25%) had 4-6 children, and 5 respondents (2%) had 7-9 children. Lastly, in terms of Mamak, 240 respondents (81%) had 1-3 Mamak, 52 respondents (18%) had 4-6 Mamak, and 4 respondents (1%) had 7-8 Mamak.

This research employed a questionnaire comprising four variables: matrilineal care, adolescent behavior, family environment, and community environment. The questionnaire consisted of 147 items with a 5 point Likert scale design assessing the frequency of items, from 'always', 'often', 'sometimes', 'never', to 'never at all’ (Table 1).

The questionnaire was distributed to participants using a Google Form. Before the questionnaire was distributed, it was tried out on mothers not included in the actual research sample and validated by three professors and a lecturer colleague who were experts in fields related to research variables. The Government of West Sumatra Province, Indonesia, had given official written approval to carry out this research. In addition, participants in this research had given their consent before filling out the questionnaire voluntarily. Withdrawal of participation from the research was also allowed at any time. The data collection was carried out for twenty days.

This research used a theoretical model analysis with Partial Least Square-Structural Equation Modeling (PLS-SEM) and was assisted by SmartPLS 3.0 version 3.2.7 software [47]. PLS-SEM is a variant-based technique that has been widely used in the increasingly popular tourism, business, and technology research [48]. However, the technique is rarely used to explain the behavior of adolescents and parenting in families. Structural equation modeling based on Variance PLS-SEM can be a suitable methodological alternative for testing covariance-based structural equation theory [49, 50].

This research refers to a general guideline for using PLS-SEM with a two-step technique, namely evaluating the Measurement Model (Outer Model) and designing the Structural Model (Inner Model). These two steps were conducted to analyze a theoretical research model [51, 52].

Evaluation of the feasibility of the measurement model and structural model has been carried out to ensure reliability and validity. Evaluation of the measurement (outer model) has been conducted to find out that each indicator in each variable has a good relationship with the other indicator. The outer model in Structural Equation Modeling-Partial Least Square (SEM-PLS) is evaluated through four stages, namely:

Structural Model (inner model) is built in four stages, namely by looking at the results of the values of R-Square, Multicollinearity, F-Square (F2), Q-Square (Q2), and Goodness-of-Fit (GoF). The description of the test results of each test component is as follows:

3.2.5.. Results of Goodness of Fit (GoF)

Based on the data obtained from the previous SmartPLS output, the average AVE value was 0.843, and the average value of R2 was 0.911. After obtaining the average AVE and R2 values, the next step was to calculate the Goodness of Fit value based on the formula below.

The GoF result was 0.876, which was greater than 0.38. Thus, it can be concluded that the model built had a good Goodness of Fit. After testing R-Square, Q-Square, and Goodness of Fit, the model formed has been considered as robust [25, 20]. The outer model and inner model outputs of the SEM-PLS model that had passed the testing stages and found to be robust are as follows:

Fig. (1) shows that hypothesis testing can be done by determining the t-statistic value and the probability value. To test the hypothesis in this research, a significance level of 5% was used so that the t-statistic value used was 1.96. The criteria for accepting a hypothesis are based on t-statistics; if the t-statistic > 1.96, then the hypothesis is accepted and vice versa. Furthermore, the criteria for rejecting or accepting the hypothesis are based on probability; if p-value < 0.05, Ha is accepted.

Fig. (1). PLS-SEM model output.

Based on the bootstrapping test in the SmartPLS software shown in Fig. (1), the t-statistic value for the variables of family environment and matrilineal parenting with respect to adolescent behavior and the variables of family and community environment with respect to matrilineal parenting had shown a value above 1.96. Meanwhile, the variable of community environment for adolescent behavior had exhibited a t-statistic value below 1.96. Hypothesis testing in this research used a one-tailed test, which indicates whether it has a positive or negative effect. The acceptance or rejection of the hypothesis can be seen from the bootstrapping report in the Path Coefficient presented in Table 5.

Table 5. Path coefficient.

	Original Sample (O)	Sample Mean (M)	Standard Deviation (STDEV)	T Statistics (|O/STDEV|)	P Values
Family Environment -> Adolescent Behavior	-0.281	-0.285	0.029	9.581	0.000
Family Environment -> Matrilineal Parenting	-0.152	-0.154	0.026	5.842	0.000
Community Environment -> Adolescent Behavior	-0.032	-0.030	0.048	0.662	0.508
Community Environment -> Matrilineal Parenting	-0.818	-0.816	0.024	33.840	0.000
Matrilineal Parenting -> Adolescent Behavior	0.700	0.700	0.054	12.858	0.000

The family environment is the first and major aspect that influences the behavior of adolescents. They spend more time in the family environment so family plays a significant role in developing their personality. The family greatly determines the good and bad behavior and the personality of adolescents [54]. Although, there are also other influential factors.

Based on the output path coefficients mentioned in Table 5, the variable of family environment had been shown to have a positive and significant influence on adolescent behavior. This was evidenced by the t-statistic value of 9.581, which was greater than 1.96, and the significance was recorded at alpha 5% (P-value < 0.05). Thus, the hypothesis accepted H_¹. In other words, the family environment had been found to exert a significant positive influence on adolescent behavior.

This result is in line with the findings in the research conducted by Krauss, Orth and Robins, which explains that various conditions in the family environment form the behavior of adolescents [52]. In addition, good communication and support provided by the family have a significantly positive influence on adolescent attitudes and behavior [55-57]. An unsupportive family environment for adolescents, such as a conflict between parents and violence in the family, initiates psychological disorders and affects the development of adolescent behavior [29, 58, 59]. Therefore, the family environment greatly determines the adolescent behavior.

The family environment also has an influence on parenting style in the family. The parenting style of families living in villages is different from those who live in cities [60, 61]. Likewise, the socioeconomic status of each family has an influence on parenting in the family [62-63].

Based on the output path coefficients listed in Table 5, the variable of family environment had been found to have a positive and significant influence on matrilineal parenting. This has been evidenced by the t-statistic value of 5.842, which was greater than 1.96, and the significance was set at alpha 5% (P-value < 0.05). Thus, the hypothesis accepted H₁.

This research reveals that a good family environment formulates the patterns of parenting in the family [64-65]. Interaction and cooperation in the family between parents and uncle (mamak) help to form better parenting. This provides many benefits for the family, but it is also very complex because both the parents and the uncle (mamak) must simultaneously carry out their roles as caregivers, and maintain relationships with each other [66-68].

Community is one of the educational environments that strongly influence the personal development of adolescents. Besides the family, the community is the second natural environment known by adolescents. They need support from the surrounding environment in their development process [69, 70].

Based on several research findings, it is evident that the community environment influences the character of adolescents. Relationships with peers also influence the character of adolescents. However, some adolescent behaviors are not influenced by their peers [71, 72].

Based on the output path coefficients shown in Table 5, the variable of community environment had been found to exert a negative and insignificant influence on adolescent behavior. This is evidenced by the t-statistic value of 0.662, which was less than 1.96, and the significance was set at alpha 5% (P-value < 0.05). Thus, the hypothesis accepted H₀.

The findings of this research reveal that the community environment does not have a significant influence on the behavior of adolescents, because the development of adolescents' personalities is mostly influenced by the family environment. Therefore, the community does not have a significant influence on adolescents' behavior.

Matrilineal parenting that involves parents and uncle (mamak) in the family is strongly influenced by geographical area, norms and culture of the local community, social partners, activities in the community, and the mass media [73-74]. Likewise, the lifestyle in the community also influences parenting [75].

Based on the output path coefficients shown in Table 5, the variable of community environment had been found to exert a positive and significant influence on matrilineal parenting. This has been evidenced by the t-statistic value of 33.840, which was greater than 1.96, and the significance was set at alpha 5% (P-value < 0.05). Thus, the hypothesis accepted H₁. In other words, the community environment had been found to have a positive influence on matrilineal parenting.

Parenting is related to the ability of a family or household and community to provide attention, time, and support to meet the physical, mental, and social needs of growing children. There are three parenting patterns, namely authoritarian, authoritative, and permissive parenting. Each of them affects the emotional, behavioral, and mental aspects of children [76].

The parenting styles received by children influence their behavior and personality [25]. Parents are the most basic role model in the family. If parents behave violently in the family, then children tend to imitate it. Likewise, if parents behave well in the family, the children also tend to behave well. Adolescents receive teachings and instructions given by parents [75]; therefore, the support of parents is favored by adolescents in their development. In matrilineal parenting, parenting is not only provided by parents but also involves mamak [51].

Based on the output path coefficients provided in Table 5, the variable of matrilineal parenting had been found to have a positive and significant influence on adolescent behavior. This has been evidenced by the t-statistic value of 12.858, which was greater than 1.96, and the significance was recorded at alpha 5% (P-value < 0.05). Thus, the hypothesis accepted H₁. In other words, matrilineal parenting had been shown to have a positive influence on adolescent behavior.