A longitudinal study of Australian senior secondary school achievement

Brian C. Hemmings
Charles Sturt University - Riverina, Australia

---

This study involves the development and testing of a theoretical model consisting of a causal sequence of 13 constructs that influence senior secondary school achievement. The constructs are drawn from studies of both secondary school and tertiary institution students. A sample of students from 10 schools was surveyed three times during an 18 month period to gather data on each of the constructs. A path analysis was carried out to test the developed theoretical model. The results of this analysis confirmed that the theoretical model was adequate and appropriate in explaining and predicting senior secondary school achievement.

---

Introduction

Between 1967 and 1992 the retention rate in the post-compulsory school years in Australia more than trebled, with particularly sharp rises over the previous decade (Australian Bureau of Statistics, 1993). This rise in retention rates has led to changes in the post-compulsory school population: larger numbers of students are staying on to Years 11 and 12 and a more diverse group of students is returning to these two senior years (Ainley & Sheret, 1992; Batten, 1989; Ward, 1988). At present, approximately 80 per cent of students complete Years 11 and 12 (Sweet, 1995).

Because of the rapid increase in student numbers and changes in the type of students continuing their senior secondary school studies, little information is available about the factors which affect the educational outcomes of this new student (mass) populace. The data which do exist are based on studies conducted during the 1980s with a more academic (elite) student cohort.

The most well-known of these studies was conducted by Carpenter & Hayden (1985), who found that the main predictor of academic success was the expectation of good results. They identified two other good predictors of post-compulsory school success, namely, teacher support to proceed to higher education and attendance at a non-government school. These two determinants, in the current educational climate, should be less important given that: one, over 50 per cent of secondary school leavers accept post-secondary college and tertiary institution enrolment offers (Report of the Australian Education Council Review Committee, 1991); and two, approximately two thirds of all the Year 12 enrolments are accommodated in government schools (Ainley & Sheret, 1993). It should be noted that Carpenter & Hayden (1985) argued strongly that a measure of previous academic achievement should be included in a model of secondary school academic achievement. Boyle, Start & Hall (1989) did not take up this suggestion when using Year 10 English and Mathematics achievement as dependent variables in their study. These researchers showed that General Information Intelligence was the best predictor of school achievement, and that the explained variance accounted for in Year 10 achievement ranged from 25 to 35 per cent after combining General Information Intelligence with specific motivational measures.

A large corpus of Northern Hemisphere literature has evolved which examines the relationship of particular constructs to academic achievement. Many of these studies have focused on the academic attainments of primary school students and junior secondary school students. Early British studies, for example, highlighted an association between family environmental factors and school performance (Swift, 1967), and a strong correlation between academic motivation and school attainment (Entwistle, 1969). Writing in an Israeli context, Eshel & Kurman (1991) indicated that factors such as perceived academic ability and father's educational level were determinants of primary school academic success. Brown & Steinberg (1991), drawing from a broad North American student population, established that high school achievement was affected by a mixture of family, peer, and school influences. Moreover, Khayyer (1986) reported that gender was related to school achievement in Iran.

A study which concentrated on the academic attainments of final year school students was undertaken by Maqsud (1983). Using a sample of Nigerian Form Four students (average age of 16.73 years), he found that four independent variables viz., socio-economic background, locus of control, intelligence, and self-esteem had significant effects on academic achievement. The results reported by Duran & Weffer (1992) were also important since the relationships among final-year school achievement and several variables were explored. They found that the academic performance of their sample of Mexican-American students was influenced by pre-high-school attainment, academic skill development, curriculum studied, and commitment shown to school-related tasks.

Therefore, many studies support the contention that a combination of personal and environmental factors impacts on school achievement. However, many of the findings need to be viewed with some caution: first, some studies do not focus on final year students; second, many of the reported findings come from Northern Hemisphere sources; third, the Australian studies were conducted during the 1980s when a minority of the student populace completed post-compulsory schooling; fourth, the relationships between and among a variety of variables are not well-understood; and fifth, very few longitudinal studies have been carried out to ascertain how particular constructs act in a sequential way on academic achievement. These points highlight the need for a comprehensive theoretical model which can explain and predict post compulsory school achievement. The aim of this study was to develop and test such a causal model of post-compulsory school achievement.

The theoretical model

In the previous section individual factors which influence school achievement were identified. To be able to predict final year school achievement, however, requires the development of a theoretical model that incorporates some of these factors in combination and suggests their interrelationships.

The theoretical model, set out in Figure 1, links 13 different constructs in a temporal and causal sequence. Additionally, the model is assumed to be recursive and the linkages hypothesised to be linear and additive. As depicted by the model, senior secondary school achievement known as Tertiary Entrance Rank (TER), is affected by: (a) eight exogenous variables (viz., family background, age, gender, locus of control, academic integration, social integration, goal commitment 1, and school commitment 1) measured midway through Year 10; (b) late Year 10 school achievement; (c) two variables, labelled 'needs accommodation' and 'expectation versus reality', which purport to measure a student's academic and social transition from junior (Year 10) to senior (Year 11) secondary school; and, (d) modified assessments of goal and school commitment (goal commitment 2 and school commitment 2) taken at the end of Year 11.

Figure 1: A theoretical model of Year 12 school achievement

The constructs included in this new theoretical model were suggested by the literature previously reviewed and by additional literature pertaining to secondary and post-secondary school persistence and attrition. This latter body of literature was consulted because it is theory-driven and boasts a series of different causal models, including those of Bean (1982), Hemmings & Hill (1991), Pittman (1991), and Stage & Richardson (1985); and, because scholastic achievement is strongly linked to dropout behaviour (see e.g., Kandel, Raveis & Kandel, 1984; Lovitt, 1991; Sheret, Ainley & Paxman, 1989). Each construct employed in the model is discussed below. It must be pointed out, however, that the studies reviewed were not selected on the basis of their contribution to an understanding of the developmental concerns and factors which influence students' decisions. This means that some factors such as family background may remain important throughout the student's school life, but the kinds of concerns which are manifest at the Year 10 and Year 11 transition period may not persist into Year 12.

Family background

Previous research has indicated that the characteristics of a student's family background, such as father's educational level and socio-economic status, can affect academic achievement (Eshel & Kurman, 1991; Maqsud, 1983). The support or lack of support for study which students receive from their parents can impinge on the decision to leave or stay on at school (Department of Employment, Education and Training, 1989; Kysel, West & Scott, 1992). Ainley, Sheret & Paxman (1989) underscored the role that parental expectation plays in shaping youths' decisions to drop out of school. They showed that high parental expectation is correlated positively and significantly with continuing secondary school enrolment. More recently, Patrikakou (1996) has demonstrated that both expectations are instrumental in raising the academic achievement of adolescents with and without learning disabilities. The evidence available points to the fact that a family background construct should be included in a study of school achievement.

Age

Several North American studies revealed that age can be a contributing factor in the decision to drop out of school (Cairns, Cairns & Neckerman, 1989; Pallas, 1984). Although the accumulated evidence is minimal, the indication is that over-age students, that is, those older than the average of their class, are more inclined to drop out of school. By logical extension it can be argued that over-age students tend to be at-risk and thus are more likely to receive poor grades. Perhaps students who are older because they have repeated a class/classes, are more susceptible to low grades and to dropping out, as they lack the support commonly afforded by an ongoing age-cohort of students. Future research might profit from including age as an independent variable in studies of school achievement.

Gender

In an Iranian context, gender can impact on school achievement (Khayyer, 1986). It has also been included as a variable in many Australian studies relating to school persistence and attrition. In most instances, it has been either an antecedent variable (Braithwaite, 1987) or a variable controlled to discover which other variable differentiated the sexes (Apps, 1981). The findings of these studies tend to be equivocal, indicating that differences and similarities between the sexes with respect to certain school outcomes need further exploration.

Locus of control

Maqsud (1983) has reported that locus of control can be linked to school attainment. Locus of control can also play a crucial role in the school staying and leaving decision. Ekstrom, Goertz, Pollack & Rock (1986), for example, were able to demonstrate that at-risk high school students tend to attribute their school successes to external factors such as luck and ease of task; whereas, good students appear to attribute their successes and failures to factors such as effort within their control. Although there are limited findings relating self-attributions with specific school outcome measures, it does seem that locus of control is a personality variable worthy of attention in this area of study.

Academic integration

Academic integration is determined primarily by the students' perception of their academic performances and level of intellectual growth. Although this construct was first used by Tinto (1975) in his conceptualisation of the college dropout process, it is closely related to school-based variables such as quality of school life (see e.g., Ainley, Foreman & Sheret, 1991; Pittman, 1991), learning environments (see e.g., Cawthron & Craig, 1980; Coppell, 1986; du Bois-Reymond, 1988), and curriculum offerings (see e.g., Batten, 1989; Duran & Weffer, 1992; Strahan, 1988). The findings from Carpenter & Hayden's (1985) seminal work add support for the use of a construct such as academic integration in a study of academic success.

Social integration

This construct was utilised by Spady (1970) and Tinto (1975) in their theoretical models of college withdrawal. They construed it as the extent to which a student is incorporated into the social fabric of the college environment, and they operationalised the construct by measuring items such as the students' satisfaction with their peer relationships and the frequency of contact with faculty members outside class. The social integration construct can be modified to fit the secondary school situation, involving exchanges with peers, teachers, and other school personnel. Evidence shows that 'social integration' not only affects secondary school persistence and attrition (see e.g., Bos, Ruijters & Visscher, 1990; Coladarci, 1983; Wehlage & Rutter, 1986), but also impacts on school achievement (see e.g., Brown & Steinberg, 1991; Carpenter & Hayden, 1987).

Goal commitment 1

The construct, goal commitment 1, was employed by Tinto (1975) and referred to the student's commitment to college graduation. It can be argued that the construct is adaptable to the secondary school situation, representing a student's commitment to secondary school graduation and his/her future plans. Support for this line of argument can be found in cognate studies of student aspiration and expectation where evidence suggests strongly that secondary school stayers and leavers can be differentiated according to their degree of educational and vocational aspiration/expectation (see e.g., Cobb, McIntire & Pratt, 1989; Poole, 1978; Stoessiger, 1980; Wehlage & Rutter, 1986). There is also evidence available from Duran & Weffer's (1992) study that 'goal commitment' is associated with high school academic attainment.

School commitment 1

School commitment 1 is another construct drawn from Tinto's (1975) conceptualisation. He named his construct 'institutional commitment' and viewed it as the commitment students feel towards their tertiary institution. Analogously, school commitment 1 stands for the commitment students have for their current secondary school. There is consistent evidence from a number of studies that students who are fearful of, anxious about, lack interest in, or simply dislike 'school' are much more inclined to drop out (Ekstrom et al., 1986; Kysel et al., 1992). On the other hand, students who enjoy school and its associated experiences are more likely to persist with their secondary school studies (Batten, 1989; Pittman, 1991). Significantly, some of the rules and regulations of secondary schools have been shown to have an effect on the decision of high school students to leave before graduating. Wehlage & Rutter (1986) found that a proportion of high school dropouts felt that school discipline methods were both inappropriate and unjust and influenced, in part, their early school departure decision. An inference to be drawn from the above set of results is that school commitment is potentially related to a more specific school outcome measure viz., school achievement.

Year 10 school achievement

It is hypothesised that the previous eight constructs influence a student's school achievement at the end of Year 10. Moreover, the theoretical model portrays Year 10 school achievement as a critical construct in that it has both direct and indirect influences on Year 12 achievement. Completion of Year 10 studies represents the end of four years of junior secondary schooling and marks the transition to two further years of non-compulsory schooling. A view commonly held by secondary school teachers and administrators is that Year 10 school achievement is strongly linked to final school attainment at the end of Year 12. Although Carpenter & Hayden (1985) argued strongly for the inclusion of a prior school achievement measure in Australia, research evidence has not been forthcoming. There is some evidence, however, from a North American context that school achievement is associated with previous school achievement (Duran & Weffer, 1992).

Needs accommodation

This construct is based on features of the Neumann and Finaly-Neumann (1989) quality of learning experience schema. Unlike other tertiary persistence and attrition models, their schema took into account the seniority/experience of the student and, as well, considered the student's 'perception' of his/her college environment. As a consequence, factors relating to the usefulness of resources/facilities and the flexibility of programmess offered were built into their theoretical model. Since secondary school students could be viewed also as moving from a junior rank (Years 7 to 10) to a more senior position (Years 11 and 12), it can be reasoned that some of the factors which Neumann and Finaly-Neumann (1989) suggested are crucial to the college dropping out decision, might be applicable to the senior secondary school situation. If a Year 11 student's academic and social needs are met by the school, then his/her commitment to specified educational and vocational goals should be strengthened.

Expectation versus reality

Expectation versus reality is a construct drawn from an idea formulated by Winteler (1986), who claimed that college students during a point in their study programmes contemplate seriously whether or not the academic standards set might be achieved. He termed this personal assessment 'realisation or frustration of study expectations'. Because senior secondary school students need to reassess their school realities, particularly during the transition phase from Year 10 to Year 11, it can be argued that secondary school students appraise whether or not their academic and social expectations are being met and that this appraisal strongly influences their decisions regarding future educational and vocational pursuits.

Goal commitment 2

Students revise their initial commitment to secondary school graduation and future plans (goal commitment 1) as a result of academic and social experiences during both the Year 10/11 transition period and the latter part of Year 11. This revised commitment should affect directly Year 12 school achievement. The results of the Duran & Weffer (1992) study provide support for this proposition.

School commitment 2

The experiences gained during the Year 10/11 transition period and Year 11 also influence a student's commitment to school, that is, school commitment 2. This construct denotes a student's revised commitment to his/her current school.

Method

Design

In order to achieve the aim of developing and testing a theoretical model which explains and predicts senior secondary school achievement, a longitudinal research design was adopted to allow for a periodic assessment of a cohort of students from Year 10 through to the possible completion of Year 12.

Participants

The study involved 817 Year 10 students (368 males and 449 females) in 10 co-educational government secondary schools in New South Wales (NSW), Australia. During the course of the study, however, the effective sample size decreased from 817 to 352 students by the end of Year 12 because of such effects as attrition and relocation. At the beginning of the study the average age of the Year 10 students was 16.3 years.

Instrumentation

Two data collection instruments were used, namely, questionnaires and a student database. Three questionnaires, which employed a variety of question types (e.g., Likert scales and dichotomous responses), were designed to gather data on the constructs dealing with junior and senior secondary school students. Questionnaire #1 sought information about eight constructs (i.e., family background, age, gender, locus of control, academic integration, social integration, goal commitment 1, and school commitment 1) mid-way through Year 10. Questionnaire #2 was designed to measure two variables, needs accommodation and expectation versus reality, at the beginning of Year 11; while, Questionnaire #3 explored goal commitment 2 and school commitment 2 at the end of Year 11.

A student database was created to monitor the progress of the sample and to ascertain which students remained in the same secondary school, and thus were available for future survey. Additionally, the database was used by the participating schools to record the Year 10 and Year 12 school achievements of their respective students.

Procedure

Questionnaire #1 was distributed among the 10 secondary schools together with a covering letter detailing how these questionnaires were to be administered. A total of 844 questionnaires was returned but 27 could not be used for a variety of reasons (N=817). The two most common reasons for rejections were that no name/identification was provided or only several pages had been completed. Once responses to Questionnaire #1 were added to a data file, school principals were asked to update the student database by providing a complete set of English, Mathematics, and Science Year 10 Reference Test results.

Only those Year 11 students who had completed Questionnaire #1, and were still enrolled at the same secondary school, were asked to respond to Questionnaire #2. A total of 574 questionnaires was returned and all of these questionnaires were useable. At the end of Year 11, the same group of students was asked to complete Questionnaire #3. A total of 484 questionnaires was received and again all questionnaires were useable.

Throughout the data-collection period, the student database was updated to determine whether students had transferred to another school, had left secondary school, had repeated a school year, or had continued to study with the same student cohort. Unfortunately, the progress of several students could not be recorded fully because a school-based student coding system was lost subsequent to the retirement of the school principal who had maintained the student database. These students were therefore deemed 'missing'. The progress status of the participating students to this point is provided in Table 1.

A final student database update required school principals to furnish the results of the 1993 Year 12 Higher School Certificate (HSC) examination. In most cases, students gain admission to university and other tertiary institutions on the basis of the HSC results. These results were added to the student database during February 1994. It needs to be noted that a substantial number of students, who completed Questionnaire #3, left school during Year 12 to seek work.

Questionnaire data screening

Data from the three questionnaires were coded and computer analysed using the SPSS programs LIST and FREQUENCY (SPSS, 1988). The LIST procedure was used to check that the raw data matched the computer printouts, and the FREQUENCY analysis permitted a screening of the data to detect any out-of-range values and to check the plausibility of the means and standard deviations. Additionally, the coefficient of variation was calculated to determine whether it was greater than 0.0001 for all measured variables. These preliminary analyses showed that the data had been entered accurately and that there were no obvious signs of 'suspect' data. Students who did not respond to pertinent questionnaire items were excluded, in most instances, from particular analyses requiring the inclusion of those data. Where missing values occurred for less important items, however, these cases were retained by using mean value substitution (Tabachnick & Fidell, 1989).

Data combination and reduction

Although some 'variables', notably, age, gender, locus of control, Year 10 school achievement, and Year 12 school achievement were considered important in their own right (single-status variables), many others did not warrant separate status and were combined into 'variable clusters' deemed to represent the total domain being measured (Baumgart & Johnstone, 1977). A principal components analysis was used to establish whether or not a relationship existed among the combined variables, and if the variable cluster could be reduced to a smaller, more economical 'set of variables' (Gorsuch, 1983; Norusis, 1985). All the variable clusters were subjected to a principal components analysis using the SPSS program FACTOR (SPSS, 1988). The factors were extracted using the scree test and an eigenvalue specification of 1.0 plus, and, in most instances, were rotated using the varimax criterion (Kim & Mueller, 1978). The factors forming the variable sets are presented in Table 2 below.

Age was determined by calculating the student's age in months as of June 1, 1991 (i.e., midway through Year 10) and locus of control was computed by summing individual item responses on a 16-item scale. This scale was developed from instruments used by Ashkanasy & Gallois (1987) and Lefcourt, von Bayer, Ware & Cox (1979). The reliability of the scale was determined by using the SPSS program RELIABILITY (SPSS, 1988) and was found to be acceptable (Cronbach's alpha = .70). Year 10 school achievement was derived from a composite of three different state wide Year 10 Reference Test results; while, Year 12 school achievement was represented by a TER which consisted of an aggregate score of internal and external items of assessment. A TER is expressed in percentile form.

Results

Bivariate analysis

A correlation analysis was conducted to examine the relationships between the 'single-status' variables and variable sets defined in the previous section. This analysis was performed using the SPSS program CORRELATION (SPSS, 1988) and the results are presented in Table 3. An examination of the Pearson product-moment correlation coefficients revealed that the direction of the measures was as expected and that, with some exceptions the predictor variables were significantly associated (p<.001) with the TER (dependent) measure. As predicted, there were significant correlations between Year 12 school achievement (TER) and having a supportive family environment (r=.265), being over-age (r=-.207), feeling a sense of academic growth (r=.241), expressing a firm commitment to education/vocational goals at Year 10 (r=.346), and showing a strong commitment to one's school at Year 10 (r=.170). Moreover, and in line with predictions, those students who gained a high TER were more likely to have performed well in their Year 10 studies (r=.792), to consider that they had coped well with the transition from junior secondary to senior secondary school (r=.237), and to have been firmly committed to academic/vocational pursuits at the conclusion of Year 11 (r=.491).

In short, the bivariate statistics accorded with the findings of studies on school achievement and the relationships hypothesised earlier in this paper. Each of the reported correlation coefficients, however, represents an isolated relationship between a pair of variables. A more important question to be asked concerns the relationships when all variables are considered simultaneously. This question can be answered through the use of multiple regression equations in path analysis.

Multivariate analysis

The reliability of multiple regression results is dependent on sample size. Approximately 15 subjects per independent variable are needed for a reliable regression (Stevens, 1986) and this criterion was met in this study. As noted in a variety of sources (see e.g., Norusis, 1985; Tabachnick & Fidell, 1989) the conditions of singularity, univariate outliers, and multivariate outliers can pose difficulty for a multiple regression. No evidence of the first two conditions was found; however, using the SPSS program REGRESSION (SPSS, 1988) on the dependent variable TER, and remaining 13 'variables' as predictors, three multivariate outliers were identified and deleted from further analyses. Multivariate outliers were evaluated as chi-square divided by degrees of freedom equal to the number of predictor variables. In this specific case, the Mahalanobis distance cutoff value was 34.528 at p<.001.

A path analysis, effected through multiple regression techniques, was used to test the adequacy of the theoretical model posited to account for the factors which influence Year 12 school achievement. The first step in the path analytic procedure was taken by using the SPSS program REGRESSION (SPSS, 1988). A series of regressions was run yielding standardised (beta) weights which can be considered as path coefficients (Heise, 1975). The next step, following these determinations, involved the construction of a path model (see Figure 2). Although the path model explained 67.5 per cent of the variance in Year 12 school achievement, with three predictor variables contributing, one variable, namely school commitment 2, had a negative path coefficient with the dependent variable. Further investigation of this unanticipated, but significant result, revealed evidence of multicollinearity, so that this path coefficient was subsequently omitted.

Figure 2: Path model of Year 12 school achievement (TER)

As shown in Figure 2, the school achievement of students at the end of Year 10 had a significant direct effect (P14,9=.721) on TER. As well, the goal commitment of these students, at the end of Year 11, had a direct effect (P14,12=.172) on TER. Further, particular measures taken during the earlier part of Year 10 (viz., family background, age, academic integration, and goal commitment 1) impacted directly and significantly on Year 10 student achievement. Interestingly, all the endogenous variables explained 20.2 per cent of the variance in Year 10 school achievement. In general, the other relationships represented in Figure 2 appear to be consistent with those hypothesised.

So that this claim could be investigated further, an additional step in the path analytic process was required: based on the results of the first set of regressions, nonsignificant variables with low path coefficients were removed from a second round of regressions. These analyses produced a parsimonious or 'reduced' model (see Figure 3) displaying only those hypothesised paths that were shown to be nonzero (Heise, 1975). The criterion for retaining a path for the reduced path model was p<.05 using a two-tailed test.

Figure 3: Reduced path model of Year 12 school achievement (TER)

As a result of the 'model reduction', four variables (viz., gender, locus of control, social integration, and school commitment 1) and seven paths were removed. It came as no surprise that the paths (causal links) between gender and Year 10 school achievement and locus of control and Year 10 school achievement were relatively weak, given that the research evidence linking these respective variables is limited and questionable. However, the causal links between social integration and Year 10 school achievement, school commitment 1 and Year 10 school achievement, Year 10 school achievement and needs accommodation, and Year 10 school achievement and school commitment were also weak, despite the fact that some prior studies had suggested a stronger relationship. Even though four variables were deleted in the 'model reduction' exercise, all other variables and most hypothesised paths were retained. The reduced path model explained 65.4 per cent of the variance in TER with school achievement at the end of Year 10 and goal commitment 2 having significant direct effects (P14,9=.724 and P14,12=.153).

*p<.001(2-tailed)

Discussion

The theoretical model developed and tested was successful in highlighting a series of factors that impacted progressively on senior secondary school achievement. Moreover, the findings of the study shed light on the interrelationships and relative importance of these factors. Of even greater significance, this study developed a sound conceptualisation of the senior secondary school achievement process.

While the findings of the study provide overall support for the theoretical model being tested, there are many individual aspects of the data and the theoretical model worthy of closer consideration. For example, when the theoretical model was tested, the results of the quantitative analyses showed a lower level of importance than expected for some of the variables e.g., social integration, school commitment 1, and school commitment 2. The strength of relationships depends on both the timing of the observation and the number of questionnaire items relating to each of the variables. In future studies, the number of items might be increased and the data gathered at more appropriate times throughout a school year to measure these particular variables. Additionally, school commitment might not be an effective measure given that many Australian government school students usually do not have a wide choice of secondary schools, unlike their British and North American counterparts, because of travel, zoning, and financial restrictions.

Although the discussion in this paper relates only to quantitative results, qualitative data were also collected and analysed. The results of that analysis have been reported elsewhere (see e.g., Hemmings & Hill, 1995; Hemmings, Jin, & Low, 1996). These data supported the developmental nature of the decision-making process of senior secondary school students in a manner consistent with a person-environment interaction framework; that is, students made decisions about career aspiration, motivation, subject choice, and pattern of social interaction, but the importance of these decisions in relation to Year 12 school achievement varied across time.

The results of the current research show that the best predictor of Year 12 academic achievement is previous school achievement as measured by a composite of Year 10 results. This finding accords with the commonly-accepted wisdom of students and teachers that students who gain high grades in Year 10 can expect to continue to do well in Years 11 and 12. Teachers in effect operate as if there are few exceptions to this generalisation: they advise students to select subjects during the final (post-compulsory) two school years on the basis of their Year 10 Reference Test results, even if it means dropping a level of study to match the Year 10 achievement. Given that the strength of the relationship between Year 10 school achievement and Year 12 school achievement is so strong, it might be argued that a student's academic future is virtually decided by the time he or she sits examinations during Year 10. Has the die been cast? Not only should future studies test the relationship between TER and Year 10 Reference Test results, but, they should examine the effects of pre-Year 10 achievements on TER.

The developed theoretical model adds potentially to the developmental and theoretical understanding of the longitudinal nature of secondary school achievement and can serve as a basis for deriving interventions. From a theoretical perspective, the findings of the study demonstrate that the proposed theoretical model does have a useful level of explanatory power. Kerlinger (1986, p. 616) notes that 'it is surprising that [theoretical] models can be and are successfully tested given the complexity and even delicacy of the undertaking'. The present study was successful in bringing together a number of key variables into a theoretical model. The reduced model which resulted from this study is consistent with the developmental nature of the decisions which students make during their senior secondary school years, where one concern is replaced by another or subsumed within a larger one. This model, comprising nine constructs, has the potential to make a significant contribution to the literature on senior secondary school achievement. From a practical perspective, the findings of this study suggest that secondary school staff members, including school administrators, teachers, career advisers, and school counsellors, need to be aware of critical periods in the transitions which take place during the senior secondary school years; they should be sensitive to the individual needs of students; and, they should offer guidance to the wide range of students, who now comprise the secondary school student population, in relation to the 'variables' embedded in the theoretical model. In particular, these staff members need to coordinate and carry out periodic student needs assessments which take into account those specific factors which are deemed relevant by the theoretical model. During Year 10 and the early phases of Year 11, school personnel should question the initial school and goal commitments of their students and monitor the perceptions of their students about academic (and social) integration. Additional monitoring of the extent to which their students' academic and social needs are being accommodated and expectations fulfilled should occur during the remainder of Year 11.

The theoretical model, although robust and comprehensive, is relatively simple to use in the workplace, because it provides a sound platform for discussions among students, parents, and secondary school staff members and permits the formation of action plans. School personnel, using student and parent input, for example, could develop individualised educational plans which target the needs and increase the motivation of individual students, thus promoting overall student achievement.

The extent to which the results and conclusions may be generalised to other students and settings will be determined ultimately by further studies; it needs emphasising, however, that the NSW government school system educates the great majority of students in NSW and that both non-government and other Australian state/territory school systems closely parallel the NSW government school system. Conceivably, researchers might apply the theoretical model to similar school systems such as those in British and North American settings. Although this study has filled a void in the literature, the field is still open for further investigation.

Acknowledgements

Work on this paper was partially supported by a grant from Charles Sturt University. I would like to thank Doug Hill, Russell Kay, Laine Martinsons, and anonymous reviewers for their comments and assistance with the preparation of this paper.

References

Ainley, J., Foreman, J., & Sheret, M. (1991). High school factors that influence students to remain in school. Journal of Educational Research, 85, 69-80.

Ainley, J., & Sheret, M. (1992) Program through high school: A study of senior secondary schooling in New South Wales. ACER Monograph No. 43. Hawthorn, Victoria.

Ainley, J., & Sheret, M. (1993). Students in the senior secondary years. Unicorn, 19, 81-88.

Ainley, J., Sheret, M., & Paxman, M. (1989). Between the idea and the reality: The transition from Year 10 to Year 11. Paper delivered at the AARE Conference, Adelaide, December.

Apps, J.M. (1981). An analysis of factors leading to the decision to discontinue or continue after Year 10. Unpublished M.A. (Hons) thesis, Macquarie University.

Ashkanasy, N.M., & Gallois, C. (1987). Locus of control and attribution for academic performance of self and others. Australian Journal of Psychology, 39, 293-305.

Australian Bureau of Statistics (1993). Schools, Australia 1992. (Catalogue Number 4221.0). Canberra: ABS.

Batten, M. (1989). Year 12: Student's expectations and experiences. ACER Research Monograph No. 33. Hawthorn, Victoria.

Baumgart, N.L., & Johnstone, J.N. (1977). Attrition at an Australian university: A case study. Journal of Higher Education, 48, 553-570.

Bean, J.P. (1982). Conceptual models of student attrition: How theory can help the institutional researcher. In E.T. Pascarella (Ed.), New directions for institutional research: Studying sudent attrition, pp. 17-33. San Francisco: Jossey-Bass Limited.

Bos, K.T., Ruijters, A.M., & Visscher, A.J. (1990). Truancy, drop-out, class repeating and their relation with school characteristics. Educational Research, 32, 175-185.

Boyle, G.J., Start, K.B., & Hall, E.J. (1989). Prediction of academic achievement using the school motivation analysis test. British Journal of Educational Psychology, 59, 92-99.

Braithwaite, R.J. (1987). Influences affecting school students' staying or leaving decisions in the post-compulsory years. Curriculum and Teaching, 2, 3-15.

Brown, B.B., & Steinberg, L. (1991). Noninstructional influences on adolescent engagement and achievement. Madison, Wisconsin. (ERIC Document Reproduction Service No. ED 340 641).

Cairns, R.B., Cairns, B.D., & Neckerman, H.J. (1989). Early school dropout: Configurations and determinants. Child Development, 60, 1 437-1 452.

Carpenter, P.G., & Hayden, M. (1985). Academic achievement among Australian youth. Australian Journal of Education, 29, 199-220.

Carpenter, P., & Hayden, M. (1987). Girls' academic achievements: Single-sex versus coeducational schools in Australia. Sociology of Education, 60, 156-167.

Cawthron, E.R., & Craig, R.A. (1980). School and post school experience of rural youth in South Australia. Pivot, 7, 44-48.

Cobb, R.A., McIntire, W.G., & Pratt, P.A. (1989). Vocational and educational aspirations of high school students: A problem for rural America. Research in Rural Education, 6, 11-16.

Coladarci, T. (1983). High school dropout among native Americans. Journal of American Indian Education, 23, 15-22.

Coppell, W.G. (1986). To stay or leave -- That is the question. Sydney: NSW Department of Education.

Department of Employment, Education and Training (1989). The challenge of retention. Canberra: Curriculum Development Centre.

du Bois - Reymond, M. (1988). The life-world of Dutch youngsters: School and family. Paper presented at the Biennial Meeting of the Society for Research on Adolescence, Alexandria, Virginia, March.

Duran, B.J., & Weffer, R.E. (1992). Immigrants' aspirations, high school process, and academic outcomes. American Educational Research Journal, 29, 163-181.

Ekstrom, R.B., Goertz, M.E., Pollack, J.M., & Rock, D.A. (1986). Who drops out of high school and why? Findings from a national study. Teachers College Record, 87, 356-373.

Entwistle, N.J. (1969) Academic motivation and school attainment. British Journal of Educational Psychology, 39, 181-188.

Eshel, Y., & Kurman, J. (1991). Academic self-concept, accuracy of perceived ability and academic attainment. British Journal of Educational Psychology, 61, 187-196.

Gorsuch, R.L. (1983). Factor analysis. (2nd ed.). Hillsdale, New Jersey: Lawrence Erlbaum Associates, Publishers.

Heise, D.R. (1975). Causal analysis. New York: John Wiley and Sons.

Hemmings, B., & Hill, D. (1991). Challenge of post-compulsory schooling: Monitoring student perceptions. Catholic School Studies, 64, 45-49.

Hemmings, B., & Hill, D. (1995). To stay or leave: A discussion framework. Australian Journal of Career Development, 4, 20-23.

Hemmings, B., Jin, P., & Low, R. (1996). Testing a theoretical model: Australian high school student persistence and attrition. Journal of Research and Development in Education, 30, 10-21.

Kandel, D.B., Raveis, V.H., & Kandel, P.I. (1984). Continuity in discontinuities: Adjustment in young adulthood of former school absentees. Youth and Society, 15, 325-352.

Kerlinger, F.N. (1986). Foundations of behavioural research. (3rd ed.). Chicago: CBS Publishing Japan Ltd.

Khayyer, M. (1986). School failure and its relation to family background. Journal of Humanities and Social Sciences, 2, 73-85.

Kim, J., & Mueller, C.W. (1978). Factor analysis: Statistical methods and practical issues. London: SAGE Publications.

Kysel, F., West, A., & Scott, G. (1992). Leaving school: Attitudes, aspirations and destinations of fifth-year leavers in Tower Hamlets. Educational Research, 34, 87 105.

Lefcourt, H.M., von Baeyer, C.I., Ware, E.E., & Cox, D.J. (1979). The multidimensional-multiattributional causality scale: The development of a goal specific locus of control scale. Canadian Journal of Behavioural Science, 11, 286 304.

Lovitt, T.C. (1991). Preventing school dropouts: Tactics for at-risk, remedial, and mildly handicapped adolescents. Austin: Pro-ed.

Maqsud, M. (1983). Relationships of locus of control to self-esteem, academic achievement, and prediction of performance among Nigerian secondary school pupils. British Journal of Educational Psychology, 53, 215-221.

Neumann, Y., & Finaly-Neumann, E. (1989). Predicting juniors' and seniors' persistence and attrition: A quality of learning experience approach. Journal of Experimental Education, 57, 120-140.

Norusis, M.J. (1985). Advanced statistics guide: SPSSX. New York: McGraw-Hill Book Company.

Pallas, A.M. (1984). The determinants of high school dropout. Unpublished doctoral dissertation, The Johns Hopkins University.

Patrikakou, E.N. (1996). Investigating the academic achievement of adolescents with learning disabilities: A structural modelling approach. Journal of Educational Psychology, 88, 435-450.

Pittman, R.B. (1991). Social factors, enrolment in vocational/technical courses, and high school dropout rates. Journal of Educational Research, 84, 288-295.

Poole, M.E. (1978). Identifying early school leaving. Australian Journal of Education, 22, 13-24.

Report of the Australian Education Council Review Committee (1991). Young people's participation in post-compulsory education and training. Canberra: Australian Government Publishing Service.

Sheret, M., Ainley, J., & Paxman, M. (1989). Characteristics of Year 11 students who initially did not intend to stay on beyond Year 10. Paper presented at the Research Seminar with Tertiary Institutions, Sydney, December.

Spady, W.G. (1970). Dropouts from higher education: An interdisciplinary review and synthesis. Interchange, 1, 64-85.

SPSS (1988). SPSS-X user's guide. (3rd ed.). Chicago: SPSS Inc.

Stage, F.K., & Richardson, R.C. (1985). Motivational orientation within the Tinto model of college withdrawal. Paper delivered at the Annual Meeting of the Association for the Study of Higher Education, Chicago, March.

Stevens, J. (1986). Applied multivariate statistics for the social sciences. London: LEA, Publishers.

Stoessiger, R. (1980). School leavers in country areas. Education Department, Tasmania Research Study No. 55, Hobart.

Strahan, D. (1988). Life on the margins: How academically at-risk early adolescents view themselves and school. Journal of Early Adolescence, 8, 373-390.

Sweet, R. (1995). Vocational preparation: A new model for the new century. Australian Journal of Career Development, 4, 5-7.

Swift, D.G. (1967). Family environment and 11+ success: Some basic predictors. British Journal of Educational Psychology, 37, 10-21.

Tabachnick, B.G., & Fidell, L.S. (1989). Using multivariate statistics. (2nd ed.). Tokyo: Harper and Row, Publishers.

Tinto, V. (1975). Dropout from higher education: A theoretical synthesis of recent research. Review of Educational Research, 45, 89-125.

Ward, J. (1988). Increased retention to the senior years of high school: Some considerations. The Bulletin of the National Clearinghouse for Youth Studies, 7, 8-10.

Wehlage, G., & Rutter, R. (1986). Dropping out: How much do schools contribute to the problem? Teachers College Record, 87, 374-392.

Winteler, A. (1986). Differential validation of a path analytic model of university dropout. Paper presented at the Annual Meeting of the American Educational Research Association, San Francisco, April.

© 1996 Issues in Educational Research
Last revision: 21 Oct 2013. This URL: http://www.iier.org.au/iier6/hemmings.html
Previous URL: http://education.curtin.edu.au/iier/iier6/hemmings.html
Previous URL from 26 Jan 1998 to 2 Aug 2001: http://cleo.murdoch.edu.edu.au/gen/iier/iier6/hemmings.htm
HTML: Clare McBeath [c.mcbeath@bigpond.com] and Roger Atkinson [rjatkinson@bigpond.com]