Relationships among mental health status, social context, and demographic characteristics in Taiwanese aboriginal adolescents: A structural equation model

Authors


Chung-Ping Cheng, PhD, Department of Psychology, Kaohsiung Medical University, no. 100 Shih-Chuan 1st Road, Kaohsiung City 807, Taiwan. Email: cpcheng@mail.psy.kmu.edu.tw

Abstract

Abstract  The purposes of this study were to examine the relationships among mental health status, demographic characteristics, and social contexts, including family conflict and support, connectedness to school, and affiliation with peers who exhibit delinquent behavior and who use substances, among Taiwanese aboriginal adolescents. A total of 251 aboriginal junior high school students in an isolated mountainous area of southern Taiwan were recruited, and the relationships among mental health status, demographic characteristics, and social contexts among them were examined using a structural equation model (SEM). The SEM revealed that family conflict and support had direct influences on mental health status and connectedness to school. Family conflict had a direct relationship with affiliation with peers who use substances, and family conflict and support were both indirectly linked with affiliation with peers who exhibit delinquent behavior and who used substances; these were mediated by a poor mental health status. Female and older age were directly linked with a poor mental health status and were indirectly linked with a greater number of peers who exhibit delinquent behavior and who use substances via the poor mental health status. Disruptive parenting was directly linked with affiliation with peers who use substances. The authors suggest that those who devise strategies to improve aboriginal adolescents’ mental health and discourage substance use should take these relationships among mental health, demographic characteristics, and social contexts into account.

INTRODUCTION

The mental health of aboriginal adolescents is of major concern to researchers, mental health practitioners, and educators in Taiwan and around the world. Several empiric studies have suggested that aboriginal adolescents may have more serious mental health problems than adolescents in the general population.1–3 Higher rates have been found with respect to suicide,4 feelings of sadness and hopelessness,5 conduct disorders,1 and aggression.6 However, not all aboriginal adolescents are drawn into a hopeless situation with poor physical and mental health; in fact, a high proportion of aboriginal youths reported that they are physically and emotionally healthy.7 There must be a complex interplay between personal and environmental risk and protective variables in determining the mental health outcomes for aboriginal adolescents.

Of the constellation of forces that influence adolescent mental health, the most fundamental are the social contexts in which adolescents are embedded,8 and the family and school contexts are among the most critical.9 Parent-family social support, called family support,6 family caring,10 family function,11 or family connectedness,7 may be particularly important as a protector from developing psychopathologies in adverse situations among minority and indigenous populations.6,12 Feelings of connectedness to school have also been significantly associated with adolescent emotional health.9,10 Although aboriginal adolescents were found to have a higher level of family adversity than nonaboriginal ones,6 few studies have examined the social contexts and their relationships with mental health for aboriginal adolescents.12,13 Meanwhile, previous studies on this issue have been primarily concerned with American Indian youth, who may be different from Taiwanese counterparts in many respects. Further studies are needed to examine the social contexts and their relationships with mental health for non-American aboriginal adolescents.

Substance use has long been considered one of the major health problems for aboriginal adolescents.1,6,14 Taiwanese aboriginal adolescents are more likely than the general adolescent population to have alcohol use disorders according to the revised 3rd edition of the Diagnostic and Statistical Manual.15,16 Affiliation with peers who engage in substance use17,18 and who exhibit delinquent behaviors19 can increase the risk of substance use in adolescents. It is noteworthy that poor mental health,20,21 a high level of family conflict,19 and impoverished family support19,22 have also been associated with adolescent use of substances. Because adolescent development is the result of multi-systematic interactions,23 these individual and environmental factors might not only have direct influences on adolescent drug use, but may also have indirect influences through their interactions with other factors.24 However, the effect of family conflict and support and mental health on the extent to which aboriginal adolescents affiliate with peers who use substances and who exhibit delinquent behavior is still underevaluated.

The aboriginal minorities in Taiwan consist of 12 different tribes with a total population of 456 364, accounting for approximately 2% of the population of Taiwan.25 Most Taiwanese aborigines live in the central mountain region and in the eastern coastal area. Although Taiwan has undergone vigorous industrialization and economic development in recent decades, the aboriginal population has generally not benefited. Instead it has suffered adverse impacts, particularly an inferior socioeconomic status and the destruction of individual traditional tribal social organizations. In general, the average household income of aborigines is far below that of the average for Taiwanese farmers.26,27 The health status of Taiwanese aborigines is also not as good as that of the rest of the Taiwanese population,28 with an average life expectancy 6–8 years shorter than that of the general Taiwanese population.29 Reducing health disparities between aborigines and the general population is a major national health objective.

The purpose of this research was to gain insights into the relationships among the mental health status, levels of family conflict and support, connectedness to school, and extent of affiliation with peers who exhibit delinquent behavior and who use substances among Taiwanese aboriginal adolescents. The authors extended previous research by creating a structural equation model (SEM) that integrates the above constructs into a more general framework. The authors proposed the following hypotheses based on prior research: (i) family conflict and support have direct influences on mental health status; (ii) family conflict and support are indirectly linked with mental health status as mediated by connectedness to school; (iii) family conflict and support have direct influences on affiliation with peers who exhibit delinquent behavior and who use substances; (iv) family conflict and support are indirectly linked with affiliation with peers who exhibit delinquent behavior and who use substances as mediated by mental health status; and (v) demographic characteristics, including gender, age, parents’ marital status, the family’s social status, parents’ alcohol consumption behavior, and religiosity have influence in fostering a poor mental health status, and, therefore, influence adolescents’ choices of peers who exhibit delinquent behavior and who use substances as mediated by a poor mental health status.

MATERIALS AND METHODS

Sample

Aboriginal adolescents from the three junior high schools in the mountainous region of southern Taiwan were recruited for this study. This mountainous region is characterized by a low population density (5–13 persons/km2), inconvenient transportation, and a lack of medical resources. Adolescents in this region live a very different lifestyle from those living in other areas of Taiwan. For example, there is no after-hours school campus, theater, shopping center, or fast food restaurant in this area. Very few adolescents in this area have personal computers or access to the Internet in their homes. There is no senior high school in this region, and those who want to continue their education after graduating from junior high school have to make a long daily trip to attend senior high school or move away from their hometown. The majority of the aboriginal population living in this mountain region is from the Bunun tribe, which typically has a nuclear-family structure and are Protestant or Catholic. Many young Bunun have moved into metropolitan regions for employment in industrial factories, and the middle-aged and elderly who have remained in the hometowns are farmers.

In April 2002, there were 269 aboriginal adolescents in the three junior high schools. First, the authors contacted their parents to obtain consent for their children’s participation in this study. Then, the authors explained the purpose and procedure of this study to the adolescents in their classes and encouraged them to participate. The protocol was approved by the Institutional Review Board of Kaohsiung Medical University. All subjects provided written informed consent.

Assessment

Symptom Checklist-90-Revised (SCL-90-R)

The SCL-90-R was designed for use as a descriptive measure of psychopathology.30 The Chinese version of the SCL-90-R was used to measure the mental health status with a 5-point rating scale ranging from ‘not at all’ (0) to ‘extreme’ (4).31 Its validity in assessing neurotic symptomatology of patients and psychopathology of adolescents in Taiwan has previously been established.31,32 A higher total score on the SCL-90-R indicates that a subject has a poorer mental health status.

Adolescent Family and Social Life Questionnaire (AFSLQ)

This study used five 4-point subscales in the AFSLQ to evaluate adolescents’ levels of family conflict (four items), family support (three items), the proportion of peers who exhibit delinquent behavior (seven items) and who use substances (five items), and connectedness to school (four items),33 with Cronbach’s α-values of 0.65 and 0.85 in this study. Higher subscale scores indicate a more severe level of family conflict, a lower level of family support, lower connectedness to school and have greater numbers of peers who exhibit delinquent behavior and who use substances. Delinquent behaviors include the behaviors of aggression to people, destruction of property, deceitfulness or theft and serious violations of rules. Substances refer to alcohol, tobacco, areca quid, sedatives/hypnotics and all illicit drugs. The AFSLQ also assesses parents’ alcohol consumption behaviors, in which consumption of alcohol three times or more per week is labeled ‘regular alcohol consumption’.

Social Status Rating Scale (SSRS)

The researchers evaluated social status of adolescents’ family based on the SSRS, which classify social status into five classes according to education level and occupation of their main caregivers.34 For statistical purposes, class I to class IV are labeled a ‘high social status’ and class V is considered a ‘low social status’.

The authors also ascertained the salient demographic factors, including gender, age, involvement in religious activities, and parents’ marital status.

Procedures and statistical analysis

The authors invited the adolescents to complete the SCL-90-R, AFSLQ, and the questionnaire for demographic factors based on the explanations of the research assistants and under their direction. Those who refused to participate were allowed to leave the blank questionnaires on their desk. All participants received a gift at the end of the assessment. The interrelationships among mental health, family, peer, school, and demographic characteristics were analyzed using the SEM. The LISREL software program (Scientific Software International Inc., Lincolnwood, IL, USA) was used for parameter estimation, testing of the adequacy of the model, and evaluating the adequacy between the data and the model that indicates the extent of agreement between the observed data and the covariance matrix estimated from the model.35 A P-value of <0.05 was considered statistically significant. The authors also relied on several statistics to evaluate the goodness-of-fit of the model including the non-normed fit index (NNFI), the comparative fit index (CFI), the root mean square error of approximation (RMSEA), and the standardized root mean residual (SRMR).36 The authors also used the bootstrapping approach suggested by Bollen and Stine to test the stability of the result from small sample.37

RESULTS

In total, 251 adolescents (93.3%) completed all questionnaires and the interview. Those who were absent from class (n = 12), who refused to participate or whose parents refused to allow them to participate (n = 4), or who could not understand the content of the interview due to below-average mental function (n = 2) were excluded. According to school records, there was no difference in the gender ratio between the groups who participated and did not participate in this study. The demographic characteristics, mental health status, and social contexts among the 251 aboriginal adolescents are shown in Table 1 and the correlations among the variables are listed in Table 2. Those adolescents with higher levels of family conflict or lower levels of family support had a poorer mental health status, lower connectedness to school, and a greater number of peers who use substances. The adolescents with lower levels of family support had greater numbers of peers who exhibit delinquent behavior. Those adolescents with a poor mental health status had greater numbers of peers who use substances and who exhibit delinquent behavior. Girls were more likely to have a poor mental health status. Those adolescents who were older and had disruptive parenting were more likely to have a greater number of peers who use controlled substances. The directions of the correlations among these variables were consistent with the authors’ predictions.

Table 1.  Demographic characteristics, mental health status, and social contexts among the 251 aboriginal adolescents
 n (%)Mean (SD)
  1. SCL-90-R, Symptom Checklist-90-Revised; SD, standard deviation.

Gender (boys)125 (49.8) 
Age (year) 14.8 (1.0)
Infrequent attendance of religious activities 78 (31.1) 
Disruptive parenting 68 (27.1) 
Low socioeconomic status219 (87.3) 
Regular alcohol consumption by parents 42 (16.7) 
Total SCL-90-R scores 77.5 (49.1)
Family conflict 2.0 (0.4)
Family support 1.9 (0.8)
Connectedness to school 1.9 (0.5)
Affiliation with peers who exhibit delinquent behavior 0.6 (0.6)
Affiliation with peers who use substances 1.1 (1.0)
Table 2.  Correlations between variables
 123456789101112
  • *

    P < 0.05;

  • **

    P < 0.01;

  • ***

    P < 0.001.

  • 1, family conflict; 2, family support; 3, Symptom Checklist-90-Revised scores; 4, connectedness to school; 5, affiliation with peers who use substances; 6, affiliation with peers who exhibit delinquent behavior; 7, age; 8, gender; 9, religiosity; 10, disruptive parenting; 11, socioeconomic status; 12, parents’ regular consumption of alcohol.

11.000           
20.158*1.000          
30.280***0.268***1.000         
40.149*0.302***0.0821.000        
50.168**0.176**0.189**0.0891.000       
60.1050.167**0.204**0.132*0.563***1.000      
70.140*0.0730.118−0.0020.138*0.0131.000     
8−0.106−0.063−0.342***0.188**−0.0120.0010.137*1.000    
9−0.0110.082−0.0090.0630.071−0.0140.0680.192**1.000   
100.0410.131*0.0450.0390.179**0.0640.197**0.0740.1141.000  
11−0.0380.013−0.0420.0630.0020.037−0.021−0.0490.0500.126*1.000 
120.0600.081−0.0190.137*−0.0260.0280.0370.215**0.137*0.135*0.0431.000

Since the data are from different schools, the authors conducted a multigroup SEM analysis to examine the homogeneity of covariance matrices of the three schools. The result reveals no significant difference between these schools (χ2 = 169.69, d.f. = 156, P = 0.21, n = 251). From the point of the variables concerned, there is no evidence to conclude that the three schools are heterogeneous. Analyses followed are based on pooled data. Then, the authors estimated the hypothesized model by the maximum likelihood method using the LISREL 8.7 procedure. As shown in Table 3, the initial model (model I) had 8 degrees of freedom and a χ2 value of 102.640 (P < 0.001). This represents a poor fit to the data. The goodness-of-fit for model I also revealed the same conclusion. Considering the modification index and theoretical plausibility, the authors modified model I by adding the path from peers exhibiting delinquent behavior to peers using substances (model II), and this resulted in a significant improvement over model I (χ2 (1) = 95.645; P < 0.001). To further simplify model II, the authors deleted the 29 insignificant paths from demographic factors to endogenous variables, and the final model (model III) fit the data very well (χ2 (36) = 29.734; P = 0.760; RMSEA <0.001; normed fit index, 0.914; NNFI, 1.041; CFI, 1.000; SRMR, 0.041; goodness-of-fit index, 0.981; and adjusted goodness-of-fit index, 0.959). For the small sample size, following the bootstrapping procedure proposed by Bollen and Stine,37 100 samples were resampled from the original sample, and the results revealed that a large proportion (90%) of the P-values associated with χ2 were above 0.05. Bootstrapping goodness-of-fit indices confirmed the stability of the result (RMSEA <0.05, proportion = 97%; SRMR <0.08, proportion = 100%; NNFI >0.95, proportion = 100%; CFI >0.95, proportion = 88%) and suggested that the data-model fit in the model III was not subject to sampling variation of small sample size.

Table 3.  χ2 and goodness-of-fit indices for the structural equation models
Modelχ2DFPΔχ2ΔDFPRMSEANFINNFICFISRMRGFIAGFI
  1. Model I, initial model; Model II, model I with the influence of affiliation with peers who exhibit delinquent behavior and those who use substances included; Model III, model II with 29 insignificant paths having been deleted.

  2. AGFI, adjusted goodness-of-fit index; CFI, comparative fit index; GFI, goodness-of-fit index; NFI, normed fit index; NNFI, non-normed fit index; RMSEA, root mean square error of approximation; SRMR, standardized root mean residual.

I102.640 80.000   0.2020.702−1.8050.6600.0680.9450.461
II 6.995 70.44695.645 10.0000.0000.9801.0001.0000.0240.9950.949
III 29.734360.76022.739290.7880.0000.9141.0411.0000.0410.9810.959

The final SEM (Fig. 1) confirmed the authors’ first hypothesis that a high level of family conflict and a low level of family support had direct influences on a poor mental health status (β = 0.20 and 0.21, respectively). However, although a high level of family conflict and a low level of family support had influences on low connectedness to school (β = 0.12 and 0.30, respectively), no linkage was found between connectedness to school and the mental health status. The authors’ second hypothesis that a high level of family conflict and a low level of family support were indirectly linked with a poor mental health status as mediated by low connectedness to school was not supported. The final model also revealed that a high level of family conflict had a direct relationship with affiliation with a greater number of peers who use substances (β = 0.11), but it had no direct relationship with affiliation with peers who exhibit delinquent behavior. Meanwhile, family levels of support had no direct relationship with affiliation with peers who exhibit delinquent behavior or who use substances. However, a high level of family conflict and a low level of family support were both indirectly linked with affiliation with a greater number of peers who exhibit delinquent behavior, as mediated by a poor mental health status (indirect effects, 0.046 and 0.042, respectively). A high level of family conflict and a low level of family support were also indirectly linked with affiliation with a greater number of peers who use substances, as mediated by a poor mental health status and a greater number of peers who exhibit delinquent behavior (indirect effects, 0.025 and 0.023, respectively).

Figure 1.

The standardized coefficients of the LISREL structural equation model. Only significant coefficients (P < 0.05) are depicted.

Among the demographic characteristics, female gender was directly linked with a poor mental health status (β = −0.33). Female gender was also indirectly linked with affiliation with a greater number of peers who exhibit delinquent behavior via a poor mental health status (indirect effect, −0.072), as well as indirectly linked with affiliation with a greater number of peers who use substances via a poor mental health status and affiliation with a greater number of peers who exhibit delinquent behavior (indirect effect, −0.053). Older age was directly linked with a poor mental health status (β = 0.12) and indirectly linked with a poor mental health status, as mediated by a high level of family conflict and a low level of support (indirect effect, 0.036). Older age was also indirectly linked with affiliation with a greater number of peers who exhibit delinquent behavior via a poor mental health status (indirect effect, 0.032), as well as indirectly linked with affiliation with a greater number of peers who use substances via a poor mental health status and affiliation with a greater number of peers who exhibit delinquent behavior (indirect effect, 0.034). Disruptive parenting was directly linked with affiliation with a greater number of peers who use substances (β = 0.14). A family’s social status, parents’ alcohol consumption patterns or religiosity was not found to be linked with the mental health status, affiliation with peers who exhibit delinquent behavior or those who use substances.

DISCUSSION

In this study, the authors tested a theoretical model of the relationships among mental health status, levels of family conflict and support, connectedness to school, and affiliation with peers who use substances and who exhibit delinquent behavior in Taiwanese aboriginal adolescents in a real context using the SEM, and found the model fit the data well. The authors found that levels of family conflict and support directly influenced the mental health status, which further confirmed the results of previous studies on the Native American adolescents.7,10 This result indicated that although the family, as a primary social unit, is not the only domain for adolescents’ livings,38 it still has a very important influence on the mental health status among aboriginal adolescents. Much of the emphasis of mental health interventions has been devoted to school or community centered efforts that focus solely on the knowledge, attitudes, or self-esteem of youths themselves, with only a few programs also involving parents or families.39 The SEM described in this study implies that mental health workers routinely need to assess the levels of family support and conflict and implement family oriented interventions. Contrary to the authors’ hypothesis, the effect of the levels of family conflict and support on mental health was not mediated through their influence on the connectedness to school. However, there were 12 aboriginal adolescents (4.5%) who were absent from class and, therefore, did not participate in this study, and the authors could not determine the pattern of association between their mental health status and connectedness to school.

The SEM in this study revealed that family conflict had a direct relationship with adolescents’ affiliation with peers who use substances. Although not directly related to affiliation with peers who exhibit delinquent behavior, levels of family conflict and support were both indirectly linked with affiliation with peers who exhibit delinquent behavior, as mediated by the mental health status, and was further linked with affiliation with peers who use substances. Adolescence is the period in which individuals try to construct an emancipated identity while still maintaining transactional relationships with their significant others, including parents.38 A high level of family conflict and a low level of family support may push adolescents away from their families and cause them to seek affiliation with peer groups to aid in achieving social adaptation. As aboriginal adolescents were found to have a higher level of family adversity than nonaboriginal ones and affiliation with peers who use substances can effectively predict adolescents’ substance-use behaviors,6,18 early monitoring of the possibility of use of substances should commence for adolescents who have grown up in families with high levels of conflict and low levels of support and who have a poor mental health status.

This study found that female aboriginal adolescents had poorer mental health than males, and this result was supported by the findings of Duclos and colleagues3 who reported that female American Indian adolescents were more likely than males to have major depressive and/or anxiety disorders and to have three or more psychiatric disorders. Except for an emotional disposition which might make female adolescents more sensitive to common events of everyday life,40 this study found that a high level of family conflict and a low level of family support might partially account for the association between female gender and a poor mental health status. This study also revealed that female aboriginal adolescents with a poor mental health status were more likely to have a greater number of peers who exhibit delinquent behavior and who use substances, indicating that caring about female aboriginal adolescents’ mental health conditions is important in preventing the future development of substance use.

Older age was directly associated with a poor mental health status among aboriginal adolescents in this study. In the process of growing up, aboriginal youths may encounter increasing numbers of uncomfortable events. For example, because the duration of obligatory education in Taiwan is 9 years, adolescents have to decide whether they will continue their education or find employment after graduating from junior high school. If choosing to continue their education, they may suffer from stresses related to passing the highly competitive entrance examination for senior high school. They may also worry about the fact that they will have to move away from their hometown to realize greater opportunities for education and employment. Meanwhile, many aboriginal adolescents begin to experience the distress caused by economic poverty and have to share their parents’ economic burdens from the time of middle adolescence. These negative events may result in a poor mental health status among older aboriginal adolescents. This study also indicated that the association between age and mental health status was mediated by a high level of family conflict and a low level of family support among aboriginal adolescents. From a developmental view, individuals in middle adolescence have stronger urges to construct an emancipated identity, and conflicts in their objective relations with their parents are likely to occur. Further studies are needed to survey how long the mediating effects of family conflict and support on the association between age and mental health status persist.

It was hypothesized that parents’ regular consumption of alcohol will exacerbate family conflicts and reduce family support, which can endanger adolescents’ mental health. Parents’ substance-using behaviors have also been found to increase the risk of adolescents’ substance use.41 However, parents’ alcohol consumption behaviors were not linked with aboriginal adolescents’ mental health or affiliation with peers who exhibit delinquent behavior or who use substances. Further study is required to determine whether this is a common phenomenon among Taiwanese aboriginal families.

Some potential limitations of this study should be considered. First, the cross-sectional nature of this study limited the authors’ ability to draw conclusions about the causal relationship between mental health status and the social contexts. Second, this study did not examine the mental health of aboriginal adolescents who had dropped out of school, although the proportion was fairly small according to the school records. Third, the small sample in this study limited the authors’ ability to draw conclusions about the relationships among mental health status, social context, and demographic characteristics in Taiwanese aboriginal adolescents. In particular, the majority of the aboriginal adolescents are from the Bunun tribe. Further studies are needed to compare the relationships among mental health status, social context, and demographic characteristics among the aboriginal adolescents of different tribes, as well as between aboriginal and nonaboriginal adolescents.

Ancillary