Validation of the Adolescent Concerns Measure (ACM): Evidence from Exploratory and Confirmatory Factor Analysis
Ang, Rebecca P., Chong, Wan Har, Huan, Vivien S., Yeo, Lay See, Adolescence
Some early researchers have argued that adolescence is a period of heightened "storm and stress" and that it is universal and inevitable (e.g., Hall, 1904). Anthropologists, led by Margaret Mead (1928), opposed this view by describing non-Western cultures in which adolescence was neither stormy nor stressful. Contemporary researchers have argued for a modified view: evidence supports the existence of some degree of storm and stress but it is important to recognize that there are individual differences among adolescents in the extent to which they exhibit storm and stress and that there are cultural variations in its pervasiveness (Arnett, 1999; Buchanan, Eccles, Flanagan, Midgley, Feldlaufer, & Harold, 1990; Holmbeck & Hill, 1988).
Researchers worldwide have investigated concerns of adolescents. In Western countries such as the United States, Canada, the United Kingdom, and Australia, peer relationships, family problems, school and education, personal, and other social issues appear to be salient concerns among adolescents (Boehm, Schondel, Marlowe, & Manke-Mitchell, 1999; D'Andrea, Daniels, & Gaughen, 1998; Gillies, 1989; Springer, 1998; Violato & Holden, 1988). For example, Boehm et al. (1999) studied reasons for teens' usage of a peer listening phone service at various sites in the United States. Peer relationships (46% to 60%) and family problems (10% to 20%) were the most frequently discussed issues (Boehm et al., 1999). Other less consistently cited, but relevant adolescent concerns include finance, health, sexuality, drug use, pregnancy, AIDS, and sexually transmitted diseases (e.g., Boehm et al., 1999; Gillies, 1989).
Similar types of adolescent concerns were also reported in Asian countries. For example, Hui (2000, 2001) found seven dimensions of adolescent concerns among Hong Kong students: physical appearance and friendship, psychological well-being, family problems, school-related problems, study concerns and the future, peer relationship problems, and maladjusted behavior. Five areas of concerns were identified in a study of Singapore adolescents which include school, personal/self, peer, recreation, and family issues (Isralowitz & Ong, 1990). While many of these concerns were similar to those found in Western countries, some differences emerged. Given the strong emphasis on education and academic excellence, findings of studies conducted in Asian countries appear to point toward concerns relating to school adjustment, grades, and meeting the expectations of themselves, parents, and teachers to be primary adolescent concerns (Ang & Huan, 2006; Hui, 2000; Isralowitz & Ong, 1990). Furthermore, because of the collectivistic nature of Asian cultures and the emphasis on filial piety, education, and proper behavior, adolescents' school-related concerns are inextricably linked with family and personal concerns (Ang & Huan, 2006; Gloria & Ho, 2003; Yeh & Huang, 1996). Consequently, Asian adolescents' concerns about school will invariably be more closely linked with their personal and family concerns as compared to those of adolescents from Western countries.
For adolescents from the Middle East (e.g., Alzubaidi, Upton, & Baluch, 1998, for Yemen; Friedman, 1991, for Israel), in addition to expressing the typical concerns reviewed for adolescents from Western and Asian countries, they reported concerns with national/army service and existential issues such as politics, economics, religion, and observing tradition. Given the political, religious, cultural, and economic climate of countries in the Middle East, it is not surprising that adolescents from these countries report unique concerns about military and existential issues, in addition to the typical concerns of adolescents already reviewed. Taken together, international research on adolescent concerns suggest that while core concerns remain similar, there is some variation in specific concerns due to difference in culture, belief systems, and unique political or societal features such as threat of war or level of unemployment (Bennett, Klein, & Deverensky, 1992; Porteous, 1985).
Different researchers have classified adolescent concerns in various ways. Most report a four-factor structure (e.g., D'Andrea et al., 1998; Springer, 1998; Violato & Holden, 1988). Springer (1998), for example, conceptualized adolescent concerns as comprising four separate yet interdependent domains of family, school, peer, and individual problems (particularly depression). Specifically, he developed a 40-item inventory and validated the instrument using 339 adolescents (Springer, 1998). D'Andrea et al. (1998) designed and tested their 28-item worry survey on 495 African American youths. Like Springer (1998), D'Andrea and colleagues (1998) focused on four areas that typify the concerns of youths during adolescence; these include concerns regarding peers, family, various social and moral issues, and those related to personal well-being. Violato and Holden (1988) also reported a four-factor model for adolescent concerns but in their view, the four factors consisted of personal self, social self, future and career, and health and drugs. Other researchers have reported six-factor, seven-factor, and 13-factor models for adolescent concerns. Hui (2000) found a seven-factor model with a sample of 2,103 Hong Kong adolescents. Friedman (1991) identified six major categories of concerns among 1,645 ninth-grade and eleventh-grade Israeli adolescents, while Millar and Gallagher (1996) found 13 categories of concerns among 3,983 post primary students in Northern Ireland.
Based on the literature review, at present, there are limited measures assessing adolescent concerns using appropriate statistical procedures for the development and validation of questionnaires. With the exception of Violato and Holden's (1988) 14-item questionnaire, all other scales were constructed with sole reliance on the use of exploratory factor analysis (EFA). The EFA approach has been criticized for having statistics rather than theory determine the structure of scale scores and for not adequately assessing error (Gorsuch, 1983; Thompson & Daniel, 1996). Dickey (1996) argued that it is therefore important to note that EFA itself cannot be used as the basis for a final determination regarding an underlying construct, because the analysis is designed to maximize the amount of variance within the current variable set and subsequent analyses with other data sets may not reproduce the same factor structures. Given the limitations of existing instruments reviewed, it was therefore necessary to develop an empirically validated adolescent concerns inventory for use with Asian adolescents employing both EFA and confirmatory factor analysis (CFA) procedures.
STUDY 1: EXPLORATORY FACTOR ANALYSIS AND INITIAL VALIDATION
The purpose of Study 1 was to generate an initial pool of items for a scale to measure the construct of adolescent concerns, to conduct an EFA to assess the factor structure of the scale items, and to investigate the initial estimates of internal consistency and convergent validity of ACM scores.
Based on the literature review, using the broadest conceptualization available, adolescent concerns centered around four primary domains: concerns relating to family, friends, self, and school. An initial pool of 54 items was generated to tap into these four facets of concerns among adolescent students. The response format for the ACM is a Likert-type scale ranging from 1 (Strongly Agree) to 4 (Strongly Disagree). Items were scored such that higher scores indicated endorsement of a greater degree of adolescent concern.
A total of 619 adolescents (288 males and 330 females; 1 individual did not provide gender information) from a secondary school in Singapore participated in the study. The sample consisted of students from Grades 7 and 8, with participants' ages ranging from 12 to 17 years (M = 13.52, SD = 0.67). Self-reported ethnic identification for the sample was as follows: 91% were Chinese, 2.4% were Indian, 3.9% were Malay, 2.4% endorsed Others (all other ethnic groups not listed), and 0.3% did not provide information on ethnicity.
The preliminary ACM. The initial version of ACM consisted of 54 items which measured adolescent concerns among adolescent students in four domains: family, peer, personal, and school. Higher scores indicated endorsement of a greater degree of adolescent concern.
Behavior Assessment System for Children--Self Report of Personality (BASC-SRP). The BASC adolescent self-report form (Reynolds & Kamphaus, 1992) was used with only the following three subscales administered: Relations with Parents (8 items), Interpersonal Relations (16 items), and Attitude to School (10 items). Scores from the Relations with Parents subscale (e.g., "My parents trust me") measures the perception of being important in the family, the degree of parental trust and concern, as well as positive regard toward parents. A high score indicates positive adjustment and positive parent-child relations. Scores from the Interpersonal Relations subscale (e.g., "I'm good at making new friends") measures the degree to which an individual has good social relationships and friendships with peers. A high score indicates positive adjustment and good interpersonal relations. Scores from the Attitude to School subscale (e.g., "I hate school") measures maladjustment in terms of feelings of alienation, hostility or dissatisfaction regarding school. A high score indicates relative dissatisfaction with school, and a poor attitude toward school-related matters. Responses to each of these items on the BASC self-report subscales were made using a True/False format. The reliability estimates for the scores of these subscales in the present study were: Relations with Parents (.82), Interpersonal Relations (.85), and Attitude to School (.82). The BASC-SRP has been correlated with several established instruments providing documentation of the validity of BASC-SRP's scores (e.g., Reynolds & Kamphaus, 1992; Sandoval & Echandia, 1994).
Reynolds Adolescent Adjustment Screening Inventory (RAASI). The RAASI (Reynolds 2001) was used with only the Emotional Distress subscale (10 items) administered. Scores from the Emotional Distress subscale (e.g., "I worried about a lot of things") measures feelings of excessive anxiety and worry, dysphoric mood, crying behavior, and general distress. The RAASI items use a 3-point response format with items scored from 1 (Never or almost never) to 3 (Nearly all the time). The response format assesses the frequency of signs and symptoms of problems related to emotional distress with higher scores reflecting greater levels of emotional distress. The Cronbach alpha for the present sample was .82. Scores from the RASSI Emotional Distress subscale have shown expected relationships with scores from other established scales measuring similar constructs (Reynolds, 2001).
Consent and Procedure
In Singapore, permission to conduct research and data collection is typically granted by the school Principal. Approval was sought and obtained for the researchers to conduct the research investigation at the school prior to data collection. The purpose of the study was explained to the students and consent to participate was obtained from all students involved. Participation was strictly voluntary and students' responses were kept confidential. Students were also informed that they could refuse or discontinue participation at any time without penalty. All questionnaires were administered in English. No translation is needed as English is the language of instruction for all schools in Singapore.
Exploratory Factor Analysis
Principal components analysis with an oblique rotation (e.g., pro-max) was performed on the scores of the 54-item ACM. An oblique rotation was used because we expected the factors to be correlated. We based the decision about number of factors to retain on a combination of methods (e.g., parallel analysis, eigenvalue > 1.0, scree plots) as well as conceptual clarity, interpretability, and theoretical silence of the rotated factors, and simple structure. Parallel analysis has consistently been shown to be superior to other factor-retention rules in terms of extracting the correct number of factors in Monte Carlo studies (Zwick & Velicer, 1986). In the present study, parallel analysis and the other methods used to determine factor retention indicated the same number of factors to be retained for the final solution. Our goal was to have the smallest number of possible factors and for each item to load on only one latent factor. Items should preferably load greater than .4 on the relevant factor and less than .4 on all other factors (Stevens, 1996). Of the 54 items, 30 were dropped from subsequent analyses because these items either had very low communalities, loaded greater than .4 on multiple factors, or did not have a factor loading of at least .4 on any factor. These procedures resulted in a 24-item instrument which accounted for a total of 47.78% of the variance in ACM scores.
As expected, the rotated factors had scores that were correlated (see Table 1). The factor pattern and factor structure coefficients are presented in Table 2, along with communalities ([h.sup.2]) of the measured variables. All 24 items had communalities of at least .30 and above.
The first factor consisted of 9 items, was labeled Family Concerns, and accounted for 25.95% of the variance. The first factor contained items that reflect concerns of adolescents about their family including relationships with parents and siblings. The second factor consisted of 5 items, was labeled Peer Concerns, and accounted for 8.85% of the variance. The second factor contained items that reflect concerns of adolescents about their friends and other students in school. The third factor consisted of 6 items, was labeled Personal Concerns, and accounted for 7.60% of the variance. The third factor contained items that reflect various domains of personal concerns such as worries, feelings of anger and hopelessness, and thoughts about dying. The fourth factor consisted of 4 items, was labeled School Concerns, and accounted for 5.38% of the variance. The fourth factor contained items that reflect concerns of adolescents related to academic issues such as the ability to keep up with lessons, confidence, grades, and overall coping. The percentage of variance refers to variance-accounted-for post rotation. Whenever factors are correlated, structure coefficients are also important aids in interpretation (Thompson, 1997; Thompson & Borrello, 1985). Large structure coefficients were obtained for all measured variables on all factors, which is consistent with the moderate correlation between the scores of the rotated components.
We computed estimates of internal consistency using Cronbach's coefficient alphas. Scores obtained from the 24-item ACM had a Cronbach alpha of .87. The internal consistency estimates for the four factors were as follows: Family Concerns (9 items; [alpha] = .87), Peer Concerns (5 items; [alpha] = .75), Personal Concerns (6 items; [alpha] = .70), and School Concerns (4 items, [alpha] = .60). These Cronbach alpha estimates appear adequate for general research purposes although the alpha value for the School Concerns subscale is less than desirable (Nunnally & Bernstein, 1994).
We used RAASI's Emotional Distress subscale (Reynolds, 2001) and three of BASC-SRP's subscales (Reynolds & Kamphaus, 1992) to provide estimates of convergent validity for the ACM scores. Among the various ACM subscales, we expected scores from the Family Concerns subscale to be the strongest predictor for scores from BASC's Relations with Parents subscale. Similarly, among the four ACM subscales, we expected scores from the Personal Concerns subscale to be the strongest predictor for scores from RAASI's Emotional Distress subscale. Finally, we expected the School Concerns subscale to be the strongest predictor among the four ACM subscales for scores on BASC's Attitude to School subscale.
A total of five multiple regression analyses were performed, and the [R.sup.2] values for the five regression models investigating the impact of Family, Peer, Personal, and School Concerns on Relations with Parents, Interpersonal Relations, Emotional Distress, and Attitude to School were .51, .40, .41, and .14, respectively. Because beta weights ([beta]) are affected by collinearity while structure coefficients ([r.sub.s]) are not, Thompson and Borrello (1985) argued for the importance of reporting and interpreting both beta weights and structure coefficients in regression results.
As hypothesized, scores from Family Concerns ([beta] = -.63; [r.sub.s] = -.98) emerged as the strongest predictor compared with Peer Concerns ([beta] = -.04; [r.sub.s] = -.46), Personal Concerns ([beta] = -.06; [r.sub.s] = -.41), and School Concerns ([beta] = -.11; [r.sub.s] = -.55) for scores on Relations with Parents. Likewise, scores from Peer Concerns ([beta] = -.49; [r.sub.s] = -.92) emerged as the strongest predictor compared with Family Concerns ([beta] = -.02; [r.sub.s] = -.50), Personal Concerns ([beta] = -.14; [r.sub.s] = -.50), and School Concerns ([beta] = -.17; [r.sub.s] = -.62) for scores on Interpersonal Relations. Also as expected scores from Personal Concerns ([beta] = .43; [r.sub.s] = .92) emerged as the strongest predictor compared with Family Concerns ([beta] = .12; [r.sub.s] = .57), Peer Concerns ([beta] = .12; [r.sub.s] = .50), and School Concerns ([beta] = .07; [r.sub.s] =.55) for scores on Emotional Distress. Contrary to our expectations, School Concerns ([beta] = .16; [r.sub.s] = .78) was the second largest (not the strongest) predictor compared with Family Concerns ([beta] = .21; [r.sub.s] = .85), Peer Concerns ([beta] = .02; [r.sub.s] = .47), and Personal Concerns ([beta] = .09; [r.sub.s] = .60) for scores on Attitude to School.
In the present study, both beta weights and structure coefficients from the analyses provided consistent information with regard to interpretation of the influence of specific predictor variables. In general, results indicated that adolescents' domain-specific concerns correspondingly predicted domain-specific problems. The strongest predictor for adolescents' attitude to school was concerns related to family followed by concerns related to school. While this finding was unanticipated, it is not surprising given the importance parents and families place on education in the Singapore or larger Asian context. There is some research evidence suggesting that Asian socialization practices emphasize the need to succeed educationally because academic achievement is perceived as one of the few avenues for upward mobility, thus the significance that individuals and families attribute to academic success is intensified (Ang & Huan, 2006; Gloria & Ho, 2003; Sue & Okazaki, 1990).
STUDY 2: CONFIRMATORY FACTOR ANALYSIS AND TEST-RETEST RELIABILITY
There are two main purposes for Study 2. The first was to test the factor structure of the scores obtained from the 24-item ACM that was determined in Study 1 via an EFA procedure, with an independent sample, through the use of a CFA procedure. Students who participated in Study 2 did not participate in Study 1. The second purpose was to examine the stability of ACM scores.
Participants were 811 Grade 7 and Grade 8 students from two secondary schools in Singapore. For the CFA analysis, the authors randomly divided the sample into two, permitting the testing of our model on two separate data sets through two-fold cross-validation. Sample A consisted of 405 adolescents (209 males and 196 females). The students' ages ranged from 11 to 17 years (M = 13.62, SD = 0.92). Self-reported ethnic identification for sample A was as follows: 71.4% of the participants were Chinese, 6.7% were Indian, 18.5% were Malay, and 3.5% endorsed Others (all other ethnic groups not listed). Sample B consisted of 406 adolescents (214 males and 192 females). The students' ages ranged from 12 to 16 years (M = 13.66, SD = 0.81). Self-reported ethnic identification for sample B was as follows: 70.2% of the participants were Chinese, 8.1% were Indian, 18.7% were Malay, 2.4% endorsed Others (all other ethnic groups not listed), and 0.5% did not report information pertaining to ethnicity.
Of the 811 students, a subset of 322 students (153 males and 169 females) was used to examine the stability of ACM scores across time. These students were from Grades 7 and 8, and their ages ranged from 12 to 16 years (M = 13.48, SD = 0.74). Self-reported ethnic identification for this subsample was as follows: 77.6% of the participants were Chinese, 5% were Indian, 13.4% were Malay, and 4.1% endorsed Others (all other ethnic groups not listed).
ACM. The 24-item ACM was found in Study 1 to have four subscales, Family Concerns (9 items), Peer Concerns (5 items), Personal Concerns (6 items), and School Concerns (4 items).
Consent and Procedure
The procedures used for obtaining consent, participation, and questionnaire administration were similar to those of Study 1. For the subsample of 322 students, they completed the ACM at Time 1 and completed the ACM again two weeks later (Time 2).
Confirmatory Factor Analysis
We used CFA to test the stability of scores from the four-factor 24-item ACM using EQS Version 6.1 (Bentler, 2004). The hypothesized multidimensional model identified in Study 1 consisted of four first-order latent variables representing four subscales, with each variable having 9 (Family Concerns), 5 (Peer Concerns), 6 (Personal Concerns), and 4 (School Concerns) indicators respectively. Each item (measured variable) was constrained to load only on one factor. In addition, correlated errors and other post hoc model respectification were not permitted. The hypothesized multidimensional model (four correlated first-order factors) was compared against a hierarchical model and a competing one-factor model. The hierarchical model assumes a single second-order global adolescent concerns factor underlying the covariation among the four correlated first-order factors. The competing one-factor model is unidimensional and assumes that all 24 items reflect a single, global adolescent concerns factor.
Multiple fit indices provided by EQS were examined to provide an evaluation of model fit for multidimensional, hierarchical, and competing 1-factor models. The Satorra-Bentler rescaled [chi square] ([SB.sub.[chi square]; Satorra & Bentler, 1998) are reported as analysis revealed that the data violated the multivariate normality assumption; therefore robust maximum likelihood estimation was employed in CFA to correct for this violation. The [SB.sub.[chi square] has been found to perform consistently well across small, moderate, and large sample sizes; hence, researchers have recommended its use for nonnormal multivariate data (Curran, West, & Finch, 1996; Hu, Bentler, & Kano, 1992). Other fit indices used to assess the adequacy of model fit included the comparative fit index (CFI), the incremental fit index (IFI), and the root mean square error of approximation (RMSEA) and its confidence intervals. Although a value of .90 for the CFI and IFI indices has served as a rule-of-thumb lower limit cutoff of acceptable fit, a value of .93 is expected of models considered to be well-fitting (Bryne, 1994). RMSEA values of less than .06 indicate a good fit, and values as high as .08 indicate a reasonable fit (Hu & Bentler, 1999). Final assessment of fit for all models was based on [SB.sub.[chi square] and its related robust fit indices (CFI, IFI, and RMSEA) but for the sake of completeness, the uncorrected [chi square] statistic is also reported.
CFAs were conducted on the scores of the 24-item ACM. Results are summarized in Table 3. As expected, the [SB.sub.[chi square] yielded values that were substantially lower than the uncorrected [chi square] statistic for all models. Both the hypothesized multidimensional model and the hierarchical model had acceptable fit indices for the data from both samples A and B. In comparison, data from both samples A and B suggested that model fit was poor for the competing 1-factor model. The Language Multiplier Test indicated the presence of a correlated error between items 10 and 11, which when respecified would substantially improve the model fit. However, in general respectification of correlated errors for the purpose of achieving a better-fitting model is not an acceptable practice unless the respecification makes both substantive as well as statistical sense (Byrne, 1994). Hence the authors decided against performing any post hoc model fitting procedures. The results of the confirmatory factor analyses provided preliminary support for the factor structure of the ACM scores established in Study 1.
Internal Consistency and Test-Retest Reliability
The Cronbach alpha estimates for the ACM scores on Study 2's 811 students were as follows: Total (sample A = .85, sample B = .86), Family Concerns (sample A = .86, sample B = .88), Peer Concerns (sample A = .67, sample B = .71), Personal Concerns (sample A = .64, sample B = .64), and School Concerns (sample A = .62, sample B = .64).
For the subsample of 322 students, the two-week test-retest reliability coefficients for the scores on the 24-item ACM and the scores on Family Concerns, Peer Concerns, Personal Concerns, and School Concerns subscales were .78, .79, .61, .64, and .62, respectively.
SUMMARY AND GENERAL DISCUSSION
The objective of this investigation was to construct and validate scores on a self-report inventory to measure Asian adolescents' concerns. Findings from EFA conducted in the first study (N = 619) indicated that the ACM scores have four factors that were labeled Family Concerns, Peer Concerns, Personal Concerns, and School Concerns, respectively. This four-factor structure of ACM scores was confirmed via CFA in the second study (N = 811) using a two-fold cross-validation procedure. Both the multidimensional model and hierarchical model of adolescent concerns had an acceptable fit with the data while the competing one-factor model did not, providing additional support for the multidimensional and hierarchical models of ACM scores. Cronbach alpha and test-retest reliability estimates for ACM total and subscale scores appear adequate for general research purposes (Nunnally & Bernstein, 1994). The 4-item ACM School Concerns subscale has Cronbach alpha estimates in the low .60s, which is lower than desired. This could in part be attributed to the small number of items in that subscale.
Based on preliminary research evidence, ACM scores do appear to exhibit reasonable levels of convergent validity in the sample examined. In general, as expected, domain-specific concerns predicted domain-specific problems. Adolescent concerns related to family emerged as the strongest predictor of the quality of adolescents' relations with their parents. Similarly, adolescent concerns related to peers emerged as the strongest predictor of adolescents' social relationships and friendships with peers. Adolescent concerns related to the self emerged as the strongest predictor for adolescents' personal emotional distress. The only unanticipated finding was that the strongest predictor for adolescents' attitude toward school was concerns related to family followed by concerns related to school. While unexpected, this finding was not surprising because of the emphasis parents and families place on education and schooling in Singapore and in the larger Asian context (Ang & Huan, 2006; Yeh & Huang, 1996). In addition, because of the collectivistic nature of most Asian families, adolescents' school-related issues are inevitably linked with family concerns (Gloria & Ho, 2003).
The emergence of the four factors (Family Concerns, Peer Concerns, Personal Concerns, and School Concerns) in the present study is consistent with the literature on adolescent concerns. Most report a four-factor structure for adolescent concerns (e.g., D'Andrea et al., 1998; Springer, 1998; Violato & Holden, 1988) with slight variations of content within specific factors. Other less commonly reported factor structures include six-, seven-, and 13-factor models for adolescent concerns (e.g., Friedman, 1991; Hui, 2000; Millar & Gallagher, 1996). Previously, most investigations on factorial structures of adolescent concerns have relied exclusively on EFA approaches which yield results that are sample specific and may not reproduce the same factor structures in subsequent data sets. The present study extends previous research by investigating the factorial structure of adolescent concerns using both EFA and CFA approaches with an Asian Singapore sample.
It is interesting to note that while the broad four-factor classification of adolescent concerns is consistent with the larger international body of research in this area, it is also important to highlight some unique variations within these factors that are particularly relevant for Asian adolescents. During the scale construction phase of the study, the authors generated a pool of items to tap into the four hypothesized domains of family, peer, personal, and school concerns. Based on a thorough literature review, in the original conceptualization, the item "I have confidence in myself" (see item 23, Table 2) was classified within the domain of "Personal Concerns." However both EFA and CFA analyses indicated that the item clearly belonged to the School Concerns factor. This finding is not surprising in an Asian adolescent student sample and is consistent with the view that education is important for individuals and families in Singapore and Asia at large (Ang & Huan, 2006; Gloria & Ho, 2003). This finding suggests that in this sample, confidence in oneself is closely related to, and hence interpreted by adolescents as confidence in one's school-related abilities and adjustment.
Some limitations of the present study warrant comment. Given the relative importance of school concerns in the Singapore and larger Asian context, the four-item factor does not adequately capture the complexity of this domain of concern among adolescents. Several items that the authors generated for the school concerns domain at the scale construction phase were eliminated due to cross-loadings on multiple factors or low commonalities after being subjected to the EFA procedure, resulting in a four-item school concerns factor. It is possible that the school concerns factor may be a multidimensional concept, hence necessitating a much broader item bank than was used in the current investigation.
Only three subscales from BASC and one subscale from RAASI were used to provide initial estimates of convergent validity for ACM scores. Clearly further research to establish the validity of ACM scores using a variety of measures is still very much needed. The present investigation used two independent Singapore school-based adolescent samples and hence the findings should not be generalized beyond these samples until current findings have been replicated in samples from other settings as well as across nationalities and cultures. Future work could consider examining measurement invariance of ACM scores with respect to gender and culture.
In summary, the present investigation has reported preliminary evidence for the reliability and validity of the obtained scores from the 24-item ACM. This study has extended research on the measurement of adolescent concerns in the literature by using both EFA and CFA approaches in the initial development of such a measure. Notwithstanding the need for additional research, it is hoped that the ACM will be a useful tool for psychologists, psychiatrists, counselors, social workers, and other mental health professionals involved in understanding the concerns of adolescent students in Asia.
Alzubaidi, A., Upton, G., & Baluch, B. (1998). Expressed concerns of Yemeni adolescents. Adolescence, 33, 193-208.
Ang, R. P., & Huan, V. S. (2006). Academic Expectations Stress Inventory (AESI): Development, factor analysis, reliability and validity. Educational and Psychological Measurement, 66, 522-539.
Arnett, J. J. (1999). Adolescent storm and stress, reconsidered. American Psychologist, 54, 317-326.
Bennett, A., Klein, C., & Deverensky, J. L. (1992). A developmental examination of adolescent fears. Canadian Journal of School Psychology, 8, 69-79.
Bentler, P. M. (2004). EQS 6 structural equations program manual. Encino, CA: Multivariate Software.
Boehm, K. E., Schondel, C. K., Marlowe, A. L., & Manke-Mitchell, L. (1999). Teens' concerns: A national evaluation. Adolescence, 34, 523-529.
Buchanan, C. M., Eccles, J. S., Flanagan, C., Midgley, C., Feldlaufer, H., & Harold, R. D. (1990). Parents' and teachers' beliefs about adolescents: Effects on sex and experience. Journal of Youth and Adolescence, 19, 363-394.
Byrne, B. M. (1994).Structural equation modeling with EQS and EQS/Windows: Basic concepts, applications, and programming. Thousand Oaks, CA: Sage Publications.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.
Curran, P. J., West, S. G., & Finch, J. F. (1996). The robustness of test statistics to nonnormality and specification error in confirmatory factor analysis. Psychological Methods, 1, 16-29.
D'Andrea, M., Daniels, J., & Gaughen, K. J. S. (1998). Testing the validity and reliability of the Worry Survey among urban African American youth. Measurement and Evaluation in Counseling and Development, 30, 170-182.
Dickey, D. (1996). Testing the fit of our models of psychological dynamics using confirmatory methods: An introductory primer. In B. Thompson (Ed.), Advances in social science methodology (Vol. 4, pp. 219-227). Greenwich, CT: JAI Press.
Friedman, I. A. (1991). Areas of concern and sources of advice for Israeli adolescents. Adolescence, 26, 967-976.
Gillies, P. (1989). A longitudinal study of the hopes and worries of adolescents. Journal of Adolescence, 12, 69-81.
Gloria, A. M., & Ho, T. A. (2003). Environment, social, and psychological experiences of Asian-American undergraduates: Examining issues of academic persistence. Journal of Counseling and Development, 81, 93-106.
Gorsuch, R. L. (1983). Factor analysis (2nd ed.). Hillsdale, NJ: Erlbaum. Hall, G. S. (1904). Adolescence: Its psychology and its relation to physiology, anthropology, sociology, sex, crime, religion and education (Vols. 1 & 2. Englewood Cliffs, NJ: Prentice-Hall.
Holmbeck, G., & Hill, J. (1988). Storm and stress beliefs about adolescence: Prevalence, self-reported antecedents, and effects of an undergraduate course. Journal of Youth and Adolescence, 17, 285-306.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indices in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1-55.
Hu, L., Bentler, P. M., & Kano, Y. (1992). Can test statistics in covariance structure analysis be trusted? Psychological Bulletin, 112, 351-362.
Hui, E. K. P. (2000). Personal concerns and their causes: Perceptions of Hong Kong Chinese adolescent students. Journal of Adolescence, 23, 189-203.
Hui, E. K. P. (2001). Hong Kong students' and teachers' beliefs on students' concerns and their causal explanation. Educational Research, 43, 279-284.
Isralowitz, R. E., & Ong, T. H. (1990). Singapore youths: The impact of social status on perceptions of adolescent problems. Adolescence, 25, 357-362.
Mead, M. (1928). Coming of age in Samoa. New York: Morrow.
Millar, R., & Gallagher, M. (1996). The "Things I Worry About" scale: Further developments in surveying the worries of postprimary school pupils. Educational and Psychological Measurement, 56, 972-994.
Nunnally, J. C., & Bernstein, I. H. (1994), Psychometric theory (3rd ed.). New York: McGraw-Hill.
Porteous, M. A. (1985). Developmental aspects of adolescent problem disclosure in England and Ireland. Journal of Child Psychology and Psychiatry, 26, 465-478.
Reynolds, C. R., & Kamphaus, R. W. (1992). Behavior Assessment System for Children Manual. Circle Pines, MN: American Guidance Services.
Reynolds, W. M. (2001). Reynolds Adolescent Adjustment Screening Inventory Professional Manual. Lutz, FL: Psychological Assessment Resources.
Sandoval, J., & Echandia, A. (1994). Behavior Assessment System for Children. Journal of School Psychology, 32, 419-425.
Satorra, A., & Bentler, P. M. (1988). Scaling corrections for chi-square statistics in covariance structure analysis. ASA 1988 Proceedings of the Business and Economic Section (pp. 308-313). Alexandria, VA: American Statistical Association.
Springer, D. W. (1998). Validation of the Adolescent Concerns Evaluation (ACE): Detecting indicators of runaway behavior in adolescents. Social Work Research, 22, 241-250.
Stevens, J. (1996). Applied multivariate statistics for the social sciences (3rd ed.). Mahwah, NJ: Erlbaum.
Sue, S., & Okazaki, S. (1990). Asian-American educational achievements: A phenomenon in search of an explanation. American Psychologist, 45, 913-920.
Thompson, B. (1997). The importance of structure coefficients in structural equation modeling confirmatory factor analysis. Educational and Psychological Measurement, 57, 5-19.
Thompson, B., & Borrello, G. M. (1985). The importance of structure coefficients in regression research. Educational and Psychological Measurement, 45, 203-209.
Thompson, B., & Daniel, L. G. (1996). Factor analytic evidence for the construct validity of scores: A historical overview and some guidelines. Educational and Psychological Measurement, 56, 197-208.
Violato, C., & Holden, W. B. (1988). A confirmatory factor analysis of a four-factor model of adolescent concerns. Journal of Youth and Adolescence, 17, 101-113.
Yeh, C. J., & Huang, K. (1996). The collectivistic nature of ethnic identity development among Asian-American college students. Adolescence, 31, 645-662.
Zwick, W. R., & Velicer, W. F. (1986). Comparison of five rules for determining the number of components to retain. Psychological Bulletin, 99, 432-442.
This study was supported by the Academic Research Fund (RI 5/04 YLS) from National Institute of Education to Lay See Yeo.
Rebecca F. Ang, Division of Psychology, School of Humanities and Social Sciences, Nanyang Technological University, Singapore.
Wan Har Chong, Vivien S. Huan, and Lay See Yeo, Psychological Studies Academic Group, National Institute of Education, Nanyang Technological University, Singapore.
Requests for reprints should be sent to Rebecca P. Ang, Division of Psychology, School of Humanities and Social Sciences, Nanyang Technological University, Nanyang Avenue, Singapore 6397908. E-mail:firstname.lastname@example.org
Table 1 Factor Correlations Factors 1 2 3 4 1. ACM--Family Concerns -- 2. ACM--Peer Concerns .38 -- 3. ACM--Personal Concerns .30 .23 -- 4. ACM--School Concerns .41 .34 .36 -- Note. ACM = Adolescent Concerns Measure. For all correlations, p < .01. Correlations range between .23 and .41 with Cohen's d effect size estimates ranging between .47 and .89. The magnitude of the correlations was moderate and corresponded to. effect sizes in the medium to large range (Cohen, 1988). Table 2 Rotated Factor Pattern and Structure Matrices for the ACM and Communalities of the Measured Variables Factor 1 Factor 2 Item P S P S Factor 1: Family Concerns 1. 1 can talk to my parents about my problems. .74# .75 -.01 .30 2. My parents trust me. .72# .73 .09 .36 3. The rules in my family are fair. .70# .70 -.04 .25 4. My parents understand me. .78# .79 -.02 .31 5. 1 get along with my father. .64# .64 -.02 .24 6. My family respects my feelings and opinions. .76# .74 -.08 .24 7. I get along with my mother. .68# .67 -.03 .24 8. I get along with my brother(s) /sister(s). .47# .53 .08 .30 9. All in all, I like my family. .67# .70 -.02 .28 Factor 2: Peer Concerns 10. My friends respect me. .12 .39 .72# .75 11. My friends care about me. -.01 .29 .79# .76 12. I get along with other students in the school. .05 .33 .68# .71 13. I have no problem making friends. -.15 .15 .57# .58 14. I have a lot of fun with my friends. -.10 .20 .75# .71 Factor 3: Personal Concerns 15. I have thought about dying. (R) .19 .32 -.05 .15 16. I worry about what others think about me. (R) -.08 .07 -.12 .02 17. I take a long time to decide on things. (R) -.23 -.02 -.09 .04 18. 1 feel hopeless about my situation. (R) .10 .31 .11 .29 19. I feel angry a lot of the time. (R) .06 .26 .14 .28 20. I am confused about what kind of person I am. (R) -.04 .20 .02 .20 Factor 4: School Concerns 21. I can follow the lessons in class. .07 .27 .03 .23 22. 1 am usually happy with my grades. .03 .22 -.O1 .18 23. I have confidence in myself. .14 .35 .10 .31 24. All in all, I cope well in school. .06 .26 -.01 .20 Factor 3 Factor 4 Item P S P S [h.sup.2] Factor 1: Family Concerns 1. 1 can talk to my parents about my problems. .04 .26 .00 .24 .56 2. My parents trust me. -.16 .09 .07 .26 .56 3. The rules in my family are fair. .02 .23 .03 .24 .49 4. My parents understand me. .03 .26 .05 .29 .63 5. 1 get along with my father. .01 .19 .01 .20 .41 6. My family respects my feelings and opinions. -.01 .21 .05 .26 .55 7. I get along with my mother. .01 .19 -.03 .18 .44 8. I get along with my brother(s) /sister(s). .04 .22 .07 .25 .30 9. All in all, I like my family. .13 .32 .01 .25 .51 Factor 2: Peer Concerns 10. My friends respect me. -.08 .13 -.01 .21 .58 11. My friends care about me. -.05 .13 -.04 .l7 .59 12. I get along with other students in the school. .04 .23 .02 .24 .51 13. I have no problem making friends. .07 .23 .19 .33 .38 14. I have a lot of fun with my friends. .06 .20 -.04 .16 .52 Factor 3: Personal Concerns 15. I have thought about dying. (R) .67# .66 -.16 .12 .48 16. I worry about what others think about me. (R) .58# .57 .12 .26 .35 17. I take a long time to decide on things. (R) .46# .51 .39 .45 .41 18. 1 feel hopeless about my situation. (R) .66# .69 -.07 .22 .50 19. I feel angry a lot of the time. (R) .67# .67 -.16 .13 .49 20. I am confused about what kind of person I am. (R) .63# .67 .14 .35 .47 Factor 4: School Concerns 21. I can follow the lessons in class. -.09 .17 .69# .69 .48 22. 1 am usually happy with my grades. .06 .26 .56# .59 .35 23. I have confidence in myself. -.01 .25 .57# .63 .43 24. All in all, I cope well in school. -.03 .23 .68# .69 .48 Note. P = Pattern coefficients. S = Structure coefficients. ACM = Adolescent Concerns Measure. [h.sup.2] = Communalities of the measured variables. Pattern coefficients with values of .40 or greater are in boldface. Note: Pattern coefficients with values of .40 or greater are in boldface are indicated by #. Table 3 Summary of Fit Indices From Confirmatory Factor Analyses [SB.sub. Model [chi square] df [chi square] CFI Sample A (n =405) Hypothesized multidimensional model 495.40 * 246 370.75 * .93 Hierarchical model 505.49 * 249 379.14 * .92 Competing 1-factor model 960.69 * 252 708.11 * .72 Sample B (n = 406) Hypothesized multidimensional model 477.90 * 246 372.23 * .93 Hierarchical model 481.27 * 249 372.92 * .93 Competing 1-factor model 996.51 * 252 755.17 * .70 Model IFI RMSEA RMSEA(C) Sample A (n =405) Hypothesized multidimensional model .93 .035 .028 - .043 Hierarchical model .93 .036 .028 - .043 Competing 1-factor model .73 .067 .061 - .073 Sample B (n = 406) Hypothesized multidimensional model .93 .036 .028 - .043 Hierarchical model .93 .035 .027 - .042 Competing 1-factor model .71 .070 .064 - .076 Note. [SB.sub.[chi square] = Satorra-Bentler rescaled [chi square]; CFI = comparative fit index; IFI = incremental fit index; RMSEA = root-mean-square error of approximation. RMSEA(C) = confidence intervals for root-mean-square error of approximation. * p < .01.…
Questia, a part of Gale, Cengage Learning. www.questia.com
Publication information: Article title: Validation of the Adolescent Concerns Measure (ACM): Evidence from Exploratory and Confirmatory Factor Analysis. Contributors: Ang, Rebecca P. - Author, Chong, Wan Har - Author, Huan, Vivien S. - Author, Yeo, Lay See - Author. Journal title: Adolescence. Volume: 42. Issue: 166 Publication date: Summer 2007. Page number: 221+. © 1999 Libra Publishers, Inc. COPYRIGHT 2007 Gale Group.
This material is protected by copyright and, with the exception of fair use, may not be further copied, distributed or transmitted in any form or by any means.