Abstract:
The assertion that girls and boys in single-sex schools outperform their peers attending coeducational schools was investigated in this study with particular reference to physics achievement. The relationship between the school, the home and the student's performance in physics was also explored tentatively using multilevel analysis. The average home background (called socioeducational level in this study) was found to contribute towards student achievement to a greater extent when compared with such school effects as school type and the sex composition of the school. The importance of the use of the multilevel model in estimating microparameters such as sex differences, as well as macroparameters such as school type, is illustrated in this study by the significant influence of the school aggregate variable, socioeducational level. This is a preliminary report describing some of the school and home effects influencing student performance using multilevel statistical methodology. The relationship between student achievement in science and socioeducational factors, such as home background and school environment has been the subject of a great deal of interest in the school effectiveness debate. In particular, the common claim of enhanced achievement of students attending single-sex schools is the subject of this paper. While there is a perception that student performance is enhanced within the independent single-sex school, little concrete evidence has been provided that this is the case. When student achievement in single-sex and coeducational schools has been compared, studies often neglect to account for other factors such as home background and attitudes. In a recent article, Jones (1990) pointed out the importance of parental income, parental education, subject choices and teacher attitudes impacting on student achievement. In this study, parental occupation and education and the number of books in the home were combined into a single measure termed socioeducational level. The gradual merging of single-sex schools into coeducational schools is taking place in America, Britain and Australia with little discussion by the general community of educators. Coeducation is considered by some educators and parents to be more equitable and to represent the real world in which girls and boys will have to spend their future lives (Willis & Kenway, 1986), although this view has been challenged by other researchers such as Sarah, Scott and Spender (1980), Rossiter (1982), Howe (1984), Mahony (1985), Rowe (1988) and Jones (1990). Steedman's (1983) findings of higher science achievement of girls in single-sex schools than coeducational schools and higher science achievement of boys in coeducational schools than single-sex schools in England conflicts with research by Lee and Bryk (1986) in the United States and Carpenter and Hayden (1987) in Australia. The latter researchers found that girls and boys attending private single-sex schools had significantly higher academic achievement than students attending government coeducational schools. However, the research available is often confounded by socioeconomic factors and the school environment. Educational researchers often fail to adequately address the fact that single-sex schools are often private and have higher socioeconomic groups of students attending them. The higher achievement of students in the single-sex schools could be simply an artefact of the higher performance of students from upper class backgrounds attending private schools. The purpose of this study was to use secondary analysis of a large Australian database known as the Second International Science Study to examine the role of student, school and home factors in explaining student differences in science achievement and attitudes, particularly the single-sex school environment. Although sex differences in science achievement have been investigated in previous large scale studies (Keeves, 1973; Comber & Keeves, 1973; Kelly, 1978), these researchers did not account for the stratification of the sample by state and school type (government, Catholic and independent) and the multilevel nature of the data consisting of students nested within schools. A distinctive methodological feature of the present research is that it employed methodology which accommodated both the complex sample design and the multilevel nature of the data. This paper presents some results of a multilevel analysis of the Second International Science Study, revealing the strength of the socioeducational level measure in predicting student performance irrespective of the school type. Second International Science Study Data for this study was obtained from the Australian database of the Second International Science Study, a cross-country study of science achievement, student attitudes, teacher characteristics and school environment. This large database provided a wide range of variables and a stratified sample design not normally possible in ordinary survey analysis. The Second International Science Study was undertaken during 1980 to 1984 to provide an overview of science education across 24 countries (data collected in 1983). Although more than eight years old, the data still provide the largest reliable source of science education information available in Australia. It was designed to provide a basis for informed debate about the nature and content of school science education that would best suit the needs of the students and societies to which they belong. The study was conducted under the auspices of the International Association for the Evaluation of Educational Achievement (IEA) in association with educational research institutes in each country. In the Second International Science Study, only multiple-choice items were used, rather than written or open-ended questions and practical laboratory based exercises. It was possible that these types of items could have been biased favouring those types of students who do well on questions which require selection, rather than production, of a response from the student without prompting. There has also been some suggestion that multiple-choice items may be biased in favour of male students, when compared with female students (Harding, 1979; Murphy, 1982; Murphy, 1988; Bolger & Kellaghan, 1990; Mazzeo, Schmitt & Bleistein, 1991). Drawing inferences from this type of science test item may not be valid, if science achievement is usually assessed using a mixture of multiple choice, open-ended or other types of questions.