Validating Instruments Applied In Studying the: Relationship between Religious Commitment, Peer Relations and Behaviour of Adolescents Implications for School Counselling

The main purpose of this study is to validate the three instruments applied in studying the relationship between Religious Commitment, Peer Relations and Behavior of secondary school adolescents (Mwita, 2010). The greatest challenge in understanding the behavior of people is the adequate measurement of abstract construct (Hinkin, 1989). Thus confirmatory factor analysis was conducted, and factors derived from all the three instruments were then explored by orthogonal rotations. Three factors were extracted from the behavior items, two from the peer relations items and two from the religious commitment items. Reliability scores between the constructs indicated statistically significant difference on the latent factors of all the three instruments. Constructs were further validated by determining Average Variance Extracted (AVE) where high scores of AVE (between 0.8 & 0.9) and high factor scores of ≥.4 (Hair et al., 2010) were obtained. Finally, constructs were validated by applying SEM which proved the validity of all the constructs by obtaining good fit indices.

of multiple behaviors of the students of International Islamic School (IIS) in Malaysia, who originated from more than 60 countries around the world and also belonged to the group of elites and/or professional parents. Thus they are exposed to multiple models within their home environments, school environments and their life styles. Some researchers declare a biological difference between male and female brain in the development of cognitive skills (Maubach & Morgan, 2001), however Rafik and Lualdi (1990) stated that evidence are scant and inconclusive. Some of the researchers even claim for an existence of gender differences in motivation and academic achievement.
A Study conducted by Finn, Pannozo, and Achilles (2003) indicated that students with antisocial behaviors were coming late to class, annoying peers or/and interfere with their work, verbally or physically abusive to teachers and frequently need to be reprimanded. Almost the same picture was observed at the IIS (Mwita et al., 2010).
In the Qur'an, there are various warnings pertaining to wrong doers, and various wonderful promises pertaining to the followers of the right path, i.e., those who do what God wants them to do and leave what God wants them to leave is clearly explained in many verses of the Qur'an. Astonishingly, some of the people choose to go astray especially adolescents who are learning by imitating others-peers, friends, parents and some famous people around them (Bandura, 1986). Despite of the elaborated consequences in the Qur'an, yet some of the Muslim adolescents tend to imitate and support each other on such forbidden behaviors. This indicates a high demand of spiritual counseling to Muslim youths.
Helping these adolescents in making appropriate decisions and in selecting the right peer groups, counselors can apply Bandura's Social Interaction Theory which explains that peer selection and peer influence effects are both important in the formation and continuation of normal peer relationships as well as deviant peer groups (Bandura, 1986). In guiding Muslim adolescents, counselors may also add some Islamic Approaches by applying Qur'an and practices of Prophet Mohammad, who is considered the best role model by all the Muslims. Deviant peer involvement is known to be one of the several factors contributing to antisocial relationships. The Qur'an says: Evil (Sins and disobedience to Allah) has appeared on land and sea because of what the hands of men have earned (by oppression and evil deeds) that He (Allah) may make them taste a part of that which www.scholink.org/ojs/index.php/wjer World Journal of Educational Research Vol. 5, No. 4, 2018 343 Mohammad's practices in counseling and guiding these adolescents. However, all counselors can be encouraged to use Bandura's model of reciprocal causation shown in Figure 1 which explains that behavior, cognition and other personal factors, and environmental influences, all operate as interacting determinants that influence each other bidirectional. This theory can also be applied in preventing or rectifying psychological problems during adolescence that can be a precursor of later psychiatric disorders during young adult hood as also stated by Ferdinand, Stijnen, Verhulst, and Van Der Reijden (1999). Counselors can also apply Bandura's Social Learning Theory which emphasizes the importance of observing and modeling the behaviors, attitudes and emotional reactions of others. It explains that, one learns how new behaviors are performed, and uses this stored information as a reference or guide for action (Bandura, 1986).  Mwita et al. (2010), conducted a survey to 103 students and their 7 teachers from 6 grades. The students age range was between 11 and 16 years, with an average age of 14 years old. Selection of the respondents was through a convenient sampling, which was done primarily on the basis of respondents availability and willingness to respond, as suggested by Shaughnessy, Zechmester, and Zechmester (2006). In their study (Mwita et al., 2010) used three types of questionnaires whereby all items were measured on a 1-5 point Likert scale, ranging from "strongly disagree to strongly agree".


The Modified Child Rating Questionnaire (Mwita et al., 2010), was adapted from the Child social behaviors. Strayers (1985) had 52 Items but in the modified version (2010), 5 items were not used and only 47 items were used. In designing his questionnaire, Strayer (1985) adapted the Prosocial Behavior Questionnaire of Weir, Stevenson, and Graham (1980) and also the Affect Expression Questionnaire of Buck (1977). Strayer (1985) reported that the 2 stated questionnaires had been previously used to measure the social and antisocial behavior. Therefore, after the removal of 5 items from Child rating Questionnaire Strayer (1985), Mwita et al. (2010) used it in assessing the social and antisocial behavior of IIS students. The Cr. Alpha for Social Items was 0.86 and for anti-social items Cr.
Alpha 0.06. Therefore anti-social items can be considered unreliable, thus became one of the reasons for validating the instruments applied in a study conducted by Mwita et al. (2010) as also suggested by Hikin (1998), who recommended factor analysis in refining a new scale.  University, USA. The Index of Peer Relations (IPR) is a short, easily administered, and simply scored instrument which was designed to assess the severity or magnitude of an individual's problem in experiencing peer relationships. Hudson (1993) described The IPR as a short-form, 25-item self-report questionnaire that is administered to individual adults and young adults over the age of 12 years, which is also the age range of IIS students (12 and 18 years old). In their study Mwita et al. (2010) minimized it to 20 items and the Cr Alpha 0.86 was obtained for social peer relations items, on the contrary only 0.06 Cr Alpha was obtained for anti social items. This indicated either invalid items or due to its negative nature. This result is one of the reasons which prompted Mwita et al. (2010) to validate the three instruments. Hikin (1998), recommend factor analysis in refining a new scale, thus authors of this study decided to conduct an Exploratory factor analysis so as to validate the constructs and confirm them with SEM which is one of the best statistical tool in validating instruments.
Based on literature reviews including theories, field observations, focus group discussions and statistical results, Mwita et al. (2010) assumed that religious commitment influences behavior on the other hand behavior determines peer relationships, where positive behavior influences social peer relationship and negative behavior influences antisocial peer relationships.
In our current study, we designed a theoretical model (Figure 4), based on the assumptions made by  Mwita et al. (2010). Exploratory Factor Analysis was carried-out so as to validate the instruments used in studying the relationship between religious commitment, peer relations and behavior of adolescents.
This is based on the suggestion made by Antonakis and Dietz (2010) who demonstrated that a theoretical approach and the use of poor procedures can result in biased validity estimates. And also according to Hikin (1998), who recommend factor analysis in refining a new scale. This is the missing gape in the study of (Mwita et al., 2010). Therefore the main purpose of the present study was strictly confirmatory of the assumption made after completion of the first study through exploration of the latent factors in the Religious Commitments, Peer Relations and Positive Behavior Instruments as well as the Structure equation modeling.  Table 1. Bartletts test of sphericity was significant for all the constructs, which indicated that the correlation matrix was not an identity matrix. instruments had a significance of 0.00, this suggested that factor analysis was appropriate and the sample size was adequate for meaningful factorability (Pett, Lackey, & Sulivan, 2003). Thus, the choices of PCA and Varimax rotation were made based on the exploratory nature of this study and the low correlation among the latent factors (Pett et al., 2003;Green, Neil, & Salkind, 2005). Furthermore, the solution of oblique rotation was identical to that of orthogonal rotation.
The number of factors was based on the analysis of Kaiser Normalization criteria (Pett et al., 2003), screed plots and theoretical frame works in the related study on religious commitment, peer relations & behavior. Researchers chose the three factor orthogonal solution for behavior factors; two factor orthogonal solutions for religious commitment factors; and two factors for peer relations factors.
Interpretation was based on simple structure convergence, item loadings and conceptual clarity. From 7 constructs, behavior Items were reduced to 3 factors, i.e., morality, confidence and human relations. All three factors had Eigen values of more than one. The first one had Eigen value 16.1, explained by 64% variance; the second one had Eigen value 1.1, explained by 4.5% variance and the third one was 1.0 explained by 4.2% variance, making a total of 73% of Variance explained.
To begin with peer relations had four constructs but having only one construct that had factors which were ≥.4 the other three had very weak factors, so it was decided to set the rotation for two components.
Therefore three items were having more than 1 Eigen value, but only two constructs were   Therefore, while the mean and Cronbach Alpha identified the seven constructs of behavior to be internally consistent for behavior items, three constructs as consistent for religious items and four constructs consistent for peer relations items, but the exploratory factor analysis found out that the factors were not clearly separable. Thus, more rigorous structural based approach of Confirmatory Factor Analysis (CFA) subsequently provided stronger evidence in supporting the constructs as uni-dimensional in representing single concepts (Anderson & Gerbing, 1988) of the CFA model. The factors (for all the 3 tools) and their factor loadings are detailed in Tables 2-4.

Model Building
After validating the instruments, basic models were proposed and examined with SEM techniques. The analysis adopted a one way approach for analyzing models, where confirmatory measurement models were specified prior to simultaneous estimation of the measurement and the structural model. In a process of exploring the most appropriate models, a theory-driven approach of model comparison strategy was applied, and competing models were generated on the basis of alternative formulations of the underlying theory (Hair et al., 2010). In evaluating the alternative models, consideration was given to the overall fit measures based on a number of fit indices namely Normed chi-square index (cmindf), Comparative Fit Index (CFI) and Root Mean Square Error of Approximation (RMSEA). The component fit measures came from parameters estimates (Bollen, 1989), which included the squared multiple correlation (R²) for each pair of relationship.
Overall fit indices of individual model and component fit measures were examined to check whether any of the three models would be rejected or not. As indicated within the models, all models were fit and accepted, which also answers our three research questions (pp. 8-9). The acceptable models in terms of overall fit and component fit were then compared and evaluated in terms of parsimony indices, and all indicators were found to have significant loadings. The measurement of Normed chi-square was recommended by Bentler and Bonnet (1980) to be greater than 0.9 in a good fitting model. This means that, according to chi-square obtained from our study, all three models are considered to be good fitting models. As for Religious Commitment the chi-square was 1.459, for Peer Relations 1.629 and for behavior model 1.864.
The Comparative Fit Index (CFI) which was originated by Bentler (1980), takes into account the sample size (Byrne, 1998), it performs well even when the sample size was small (Tabachnick & Fidell, 2007) and Fan et al. (1999). Thus we included it as part of the fit index in this study due to the fact that we had a small sample size (<200). Normal values for this measurement are between 0.0 and 1.0, with values closer to one indicating good fittings (Hooper et al., 2008 RMSEA was one of the selected measurement because, it indicates how well the model, with unknown chosen parameter estimates would fit the population covariance matrix (Byrne, 1998). Due to its sensitivity to the number of estimated parameters in the model, it is regarded as the most informative fit indices (Diamantopoulos & Siguaw, 2000). Thus we also used it to assess our three instruments.

Discussion
Exploratory as well as confirmatory factor analysis was selected in running the present study because, they allow the reduction of a set of observed variables of the original set of observations, thus providing evidence of construct validity (Guadagnoli & Velicer, 1988;Bahaman Abu Samah, n.d.).
Several reductions of components were carried out and in the final reduction, latent factors for each instrument were extracted through an exploratory factor analysis with orthogonal rotations.
Two factors were extracted from the religious commitment variables, two from the peer relations variables and three from the positive behavior variables. Comparison of scores by levels of behavior, peer relations and religious commitment of Secondary School Adolescents indicated statistically significant difference in scores on the latent factors of the three Instruments. The scores were then validated by measuring the factor loadings. Convergent validity (a set of variables that presume to measure variables), was tested by using Average Variance Extracted (AVE) where high scores of AVE were obtained (between 0.8 & 0.9). And also high factor loadings of ≥.4 were obtained. Hair et al.
Since SEM is mostly used as a confirmatory technique rather than an exploratory technique, it was used to confirm the construct validity of the three instruments which were used to assess the relationship between religious commitment, peer relations and adolescent behavior (Mwita, Ssekamanya, & Md. Noor, 2010

Conclusion
The strength of this study was the ability to examine the hypothesized factor models and to validate the results through structure equation modeling for the three instruments that were measuring the religious commitment, peer relations and behavior of the adolescents (Mwita et al., 2010). All adequate fit, for the factor models were indicated across all the three tools. All items were reliable with standardized loadings greater than 0.5. Therefore, all the three tools are considered valid and can be used by School counselors in studying the students behavior.
Thus The results of this study will not only contribute to the literature and researches done on behaviors, but might also allow the introduction of valid instruments that can be used by School Counselors and counseling students of Malaysia in identifying and rectifying issues on adolescent behaviors, peer relationships as well as religious commitment, especially Islamic religion. Counselors may be urged to apply Bandura's reciprocality Model while counseling these Muslim youths.