Band 13, Heft 2 - Special issue: Statistical analyses in human developmental research: Illustrations based on Chinese studies / Editors: Daniel T.L. Shek, Rachel C.F. Sun

In social sciences, it is common to hypothesize that latent factors (e.g., psychological well-being) underlie the observed variables (e.g., depression and risk behavior). Hence, it is important to examine the nature of latent variables, the inter-relationships among such variables, and their associations with other predictors and outcome variables. These latent variable-related issues can be well addressed by adopting the approach of structural equation modeling. Apart from describing the use of structural equation modeling in research on human development, this paper also presents the assumptions underlying structural equation modeling, steps of model construction and model assessment, and both the strengths and limitations of this method in human development research. Finally, some examples using structural equation modeling in the Chinese contexts are also illustrated.

In the context of intervention research, it is always important to ask whether the growth of the participants in the intervention group is superior to that of the control group participants. Although this question can be answered by analyses based on traditional generalized linear models, there are many problems associated with such analyses. Alternatively, individual growth curve (IGC) modeling analyses based on linear mixed methods (LMM) are proposed. Although LMM are available in generic computer software, such as SPSS, there is a paucity of user-friendly manuals and illustrations. To help researchers utilize SPSS in creating IGC models, we present in this paper the basic concepts of IGC modeling and of the process of using SPSS to conduct IGC.

Individual growth curve (IGC) analyses are often used to examine growth curves of participants under different conditions, such as differences in the growth trajectories in an experimental group and a control group. Although researchers can use Linear Mixed Models (LMM) in SPSS to conduct IGC analyses, growth curves are not automatically generated from SPSS output, and researchers have to plot growth curves based on the information derived from the SPSS output. To facilitate the plotting of individual growth curves to aid the interpretation of LMM results, an EXCEL program was written by the researchers. Examples and step-by-step illustrations are presented in the paper regarding how to use this EXCEL program to plot growth curves based on the SPSS output.

The purpose of this paper is to demonstrate the process of using AMOS to test first- and higher-order confirmatory factor analysis (CFA) models. We performed the analyses with the AMOS 17.0 statistic package, a very user-friendly program for structural equation modeling. In this paper, we describe the concepts, theories, and basic steps of conducting CFA as well as provide a general introduction to the software AMOS. The process of conducting two different types of CFA within the framework of AMOS program (first-order CFA and higher-order or hierarchical CFA) are illustrated based on the data collected from 604 secondary school teachers involved in the Project P.A.T.H.S. in Hong Kong. The factor structure of a subjective outcome evaluation form developed to assess program implementers’ subjective evaluation about the project was examined.

One pre-condition to use scale or subscale scores of an instrument to compare different groups on a specific construct is the establishment of factorial invariance (i.e., factor structure is the same across groups). Despite its importance, testing for factorial invariance across groups has been rarely conducted in human development research. In this paper, we describe the importance of conducting factorial invariance tests as well as the related basic theoretical and methodological issues. We then illustrate the process of testing factorial invariance across independent groups using AMOS 17.0. In the demonstration, we examine the factorial invariance of a subjective outcome evaluation form (Form B), designed to be a three-factor structure, in three groups of potential program implementers in Hong Kong (n=2007). As evidenced by the multigroup factorial invariance found in this study, the three-factor model of Form B was cross-validated across different groups of program implementers.

Factor analyses are often used to examine latent variables underlying a set of observed variables. Although exploratory factor analyses are commonly used, the findings are not definitive and there are few useful methods to determine superiority of different exploratory factor analysis models. In contrast, confirmatory factor analyses are used to test models defined in an a priori manner and different goodness-of-fit indicators are available. However, confirmatory factor analyses are complex if one uses the traditional syntax commands in LISREL. In this paper, primary factor and hierarchical confirmatory factor analyses as well as factorial invariance were examined based on the data collected in the Project P.A.T.H.S. in Hong Kong. It is expected that through a step-by-step approach, researchers and students can understand the basic procedures in performing confirmatory factor analyses using SIMPLIS commands in LISREL.

This paper illustrates the procedure of testing full latent variable models using AMOS. Based on a sample of 4106 secondary school students in Hong Kong, the relationships among family functioning, positive youth development, and internet addiction were tested with the AMOS 17.0 program. Several competing models were examined and compared. The results revealed that both positive youth development and family functioning predicted internet addition among adolescents negatively. Higher level of family functioning also had indirect effects on students’ internet addictive behaviors through partial mediation of positive youth development. This study highlights the importance of promoting positive youth development and strengthening family functioning in reducing internet addiction in Hong Kong secondary school students.

In human developmental research, researchers always look at the effects of predictors on criterion variables, such as the influence of family on adolescent developmental outcomes. In more complex models, intervening variables such as mediators and moderators are also intrinsic to predictive relationships. Although such relationships can be examined by traditional multiple regression analyses using the ordinary least square approach, there are many limitations associated with such analyses. Alternatively, structural equation modeling represents a more powerful approach that can enable researchers to look at the relationships among three or more predictor, intervening, and criterion variables. In this paper, the influence of perceived family functioning (predictor variable) on adolescent consumption of pornographic materials (criterion variable) via positive youth development (intervening variable) was examined via LISREL in a study conducted in the context of the Project P.A.T.H.S. in Hong Kong.

This study examined the influences of family functioning and positive youth development on the intention among adolescents to engage in sexual behavior using a longitudinal design involving two waves of data. Using a cross-lagged approach, analyses using structural equation modeling showed that positive youth development qualities at Time 1 predicted the intention to engage in sexual activity in the future at, Time 2, via family functioning at Time 2. However, positive youth development at Time 2 did not mediate the effect of family functioning at Time 1 on the intention to engage in sexual behavior at Time 2. Family functioning at Time 2 was concurrently associated with the intention to engage in sexual activity at Time 2. This study highlighted the importance of family functioning and positive youth development in reducing risky sexual behavior of early adolescents in Hong Kong.

An adapted version of the Tier 1 Program of the Project P.A.T.H.S. developed in Hong Kong was implemented in four secondary schools in East China, with the financial support of the Tin Ka Ping Foundation. A quasi-experimental design with the inclusion of both experimental and control schools was adopted. Results showed that the experimental schools and control schools were basically equivalent in terms of pretest measures. For the posttest measures, analyses showed that students in the experimental schools performed better than did students in the control schools. Alternative analyses using between-subjects ANCOVAs showed similar effects. Results also revealed that high proportions of the respondents had positive perceptions of the program and the instructors, with 96.3% of the respondents regarding the program as helpful to them. The present study provides preliminary support for the effectiveness of the adapted Project P.A.T.H.S. in mainland China.

This study examined the program effectiveness of a 20-h training program of the P.A.T.H.S. (Positive Adolescent Training through Holistic Social Programs) project using subjective outcome evaluation data. Quantitative data were collected from 670 participants attending training workshops conducted in the second year of the extension phase of the P.A.T.H.S. project. Results showed that the training program was received well by the respondents. Favorable and positive perceptions were observed for the program, instructors, participants, and administrative arrangement. Multiple regression analyses showed that perceived program qualities and the instructors predicted overall satisfaction. The paper illustrates the use of several basic statistical techniques in the evaluation context.

The Tin Ka Ping P.A.T.H.S. project was designed to promote positive youth development in junior secondary school students in China. At the end of the first year of implementation, 1049 students from four schools completed a subjective outcome evaluation form that measured the participants’ views of the program, instructors, and effectiveness of the program. Multiple regression analyses showed that perceived program and instructor qualities predicted perceived benefits of the program. At the same time, 21 teachers were invited to respond to a subjective outcome evaluation form assessing their perceived views of the program, themselves, and benefits of the programs. Results based on the data collected from the program participants and implementers generally showed that different stakeholders had positive perceptions of the program, instructors, and benefits of the program. The present findings provide support to the effectiveness of Secondary 1 curriculum of the Tin Ka Ping P.A.T.H.S. project from the perspectives of both program participants and program implementers.

Although a preoccupation with material possessions is a growing concern in Chinese adolescents, there is a lack of validated measures of adolescent materialism in different Chinese contexts. Based on a thorough literature review, the Chinese Adolescent Materialism Scale (CAMS) was developed and administered in 1658 Chinese secondary school students. Results showed that the CAMS was internally consistent. Factor analyses showed that three factors were intrinsic to the scale. Consistent with expectations, the CAMS scores were significantly related to measures of morality, egocentrism, spirituality, and empathy. Analyses of responses to the CAMS items revealed that a significant proportion of respondents displayed materialistic beliefs.

Although contemporary adolescents are commonly described as “egocentric” and “self-centered”, there is a severe lack of validated Chinese measures of adolescent egocentrism. Based on a thorough literature review, the 14-item Chinese Adolescent Egocentrism Scale (CAES) was developed and administered to 1658 Chinese secondary school students. Factor analyses showed that two factors were intrinsic to the scale and the related subscales as well as the overall scale had high internal consistency. Consistent with expectations, CAES scores were significantly related to measures of morality, egocentrism, spirituality, and empathy. Analyses of the responses to the CAES items revealed that a significant proportion of respondents showed self-centered characteristics.