School Values: A Comparison of Academic Motivation, Mental Health Promotion, and School Belonging With Student Achievement

Abstract

School vision and mission statements are an explicit indication of a school's priorities. Research has found academic motivation, mental health promotion, and school belonging to be the most frequently cited themes in these statements. The present study sought to examine whether these themes relate to student academic achievement, as indicated by National Assessment Program — Literacy and Numeracy (NAPLAN) scores. A stratified sample of 287 secondary schools in Victoria, Australia was analysed using two language analytic approaches: qualitative emergent coding and supervised lexical analysis. The highest academic scores occurred when mental health promotion was included, though results depended to some extent on the analytic approach and the level of aggregation. Results do suggest that explicitly prioritising both academic performance and mental health is beneficial. Further, the study provides an approach for using language analysis to investigate multilevel constructs in schools.

Keywords::

• academic achievement
• mental health promotion
• school belonging
• lexical analysis
• mixed methods

Notes

1 LIWC allows users to take a set of text and calculate the relative frequency that each dictionary occurs. New dictionaries, such as those we created here, can be added using the LIWC infrastructure. While this automatic process speeds up the coding process, it can also result in many false positives (words and phrases are present but represent the wrong context) and false negatives (missing relevant words that were not included in the lexica). To minimise automatic errors, we chose to use NVivo for processing the data, allowing us to consider the context that words and phrases occurred.

2 It is unclear what statistical significance means in this context. Language is noisy, such that the unweighted estimates are often not powerful enough to detect significance with language, whereas with very large samples, most estimates will be statistically significant but not necessarily meaningful (Kern et al., 2016). Bootstrapping provides an estimate of accuracy of the effects and indicates the stability of the results. Thus, we focus on effect sizes and confidence intervals and the pattern of results rather than on significance testing, but also report p values for completeness.

3 The general patterns and effect sizes were consistent in both the weighted and unweighted estimates, although for the unweighted estimates, confidence intervals are wide, suggesting less accuracy in the estimates, so we focus on the weighted estimates here (see Supplement tables for unweighted estimates).