John Keeves
Permanent URI for this collection
Browse
Recent Submissions
1 - 6 of 15
-
Item
-
ItemThe multilevel analysis of students' achievement in learning the Chinese language.(Shannon Research Press, 2001-12)A three-level hierarchical or multilevel model is employed to examine the factors that influence students' achievement in learning the Chinese language. The factors might be many or various, such as school factors, factors related to teachers, parents and peers. This study, however, only examines student level factors. The first level of the three-level multilevel model is the within-student level; the second level is between student level; and the third level is between-class level. The results show that different factors at different levels of the model have effects on the students' achievement. Interaction effects are also observed between some variables and the achievement.
-
ItemThe development of scales to measure students', teachers' and scientists' views on STS.(Shannon Research Press, 2001-03)The starting point for this work on the development of scales was an existing instrument concerned with Views on Science, Technology and Society (STS) which had been prepared in Canada. This Australian study developed scales to measure views towards science, technology and society, and, it was necessary initially to specify scores to the alternative responses or views for each of the statements included in the scales used in this study. The initial scores or codes for the scales were based upon preliminary analysis and the researcher's judgment derived from a review of the literature. Subsequently, a validation study used the opinions of experts to confirm the numerical codes assigned to the responses. It was also necessary to test the items in each of the scales to see whether the model of a unidimensional scale was consistent with recorded data. It was possible to show that by using the numerical codes, the chosen items fitted well their respective scales. Once the three scales (a) effects of Society on Science and Technology (Society), (b) the effects of Science and Technology upon Society (Science), and (c) characteristics of Scientists (Scientists), had been specified and items were identified that satisfied the requirement of unidimensionality, it was possible to calibrate the three scales and the items within them using the partial credit model for Rasch scaling. The construction and calibration of these three scales permitted an investigation to proceed that involved the accurate measurement of students', teachers' and scientists' views on STS. [ Author abstract]
-
ItemAccountability of teachers and schools : a value-added approach.(International Education Journal, 2006-06)Currently, there has been substantial interest, in Australia and internationally, in policy activities related to outcomes-based educational performance indicators and their link with growing demands for accountability of teachers and schools. In order to achieve a fair comparison between schools, it is commonly agreed that a correction should be made for lack of equity. It is argued that student performance is influenced by three general factors: the student background, classroom and school context, and identified school policies and practices. In this article the effects of these three factors on science achievement among students in Canberra, Australia have been addressed. The effects are discussed with reference to Type A, Type B, Type X, and Type Z effects. Type A effects are school effectiveness indicators controlled for student background. Type B school effects are controlled for both student background and context variables. Type X effects are estimated with student effects, context effects and non-malleable policy effects controlled for. Finally, Type Z effects invoke school effectiveness indicators, controlled for student, context, and all identified policy effects. [Author abstract]
-
ItemSuppressor variables and multilevel mixture modelling.(Shannon Research Press, 2006-06)A major issue in educational research involves taking into consideration the multilevel nature of the data. Since the late 1980s, attempts have been made to model social science data that conform to a nested structure. Among other models, two-level structural equation modelling or two-level path modelling and hierarchical linear modelling are two of the techniques that are commonly employed in analysing multilevel data. Despite their advantages, the two-level path models do not include the estimation of cross-level interaction effects and hierarchical linear models are not designed to take into consideration the indirect effects. In addition, hierarchical linear models might also suffer from multicollinearity that exists among the predictor variables. This paper seeks to investigate other possible models, namely the use of latent constructs, indirect paths, random slopes and random intercepts in a hierarchical model. [Author abstract]
-
ItemSome problems in the analysis of cross-national survey data(Shannon Research Press, 2006-06)In this article three emergent problems in the analysis of cross-national survey data are raised in a context of 40 years of research and development in a field where persistent problems have arisen and where scholars across the world have sought solutions. Anomalous results have been found from secondary data analyses that would appear to stem from the procedures that have been employed during the past 15 years for the estimation of educational achievement. These estimation procedures are briefly explained and their relationships to the observed anomalies are discussed. The article concludes with a challenge to the use of Bayesian estimation procedure, while possibly appropriate for the estimation of population parameters would appear to be inadequate for modelling scores that are used in secondary data analyses. Consequently, an alternative approach should be sought to provide data on the performance of individual students, if a clearer and more coherent understanding of educational processes is to be achieved through cross-national survey research. [Author abstract, ed]