Discipline social identification, study norms and learning approach in university students

Abstract

Adopting a deep approach to learning is associated with positive academic outcomes. In the current paper, we extend this analysis in a university context by recognising that learners are not isolated individuals, but share important social identifications with others. Using online surveys at an Australian university, we examine the effects of discipline social identification and educational norms on the adoption of learning approaches. Students from a range of academic disciplines indicated their social identification with their discipline, their perceptions of peer norms within their discipline of study, and what their own learning approaches were. Results demonstrate a significant role of discipline-related social identification in predicting learning approaches, even after controlling for personal factors and quality of teaching. Moreover, perceived norms moderated this effect. Students’ approaches to learning are affected not simply by their salient self-concepts, but by their salient discipline-related self-concepts and the norms embodied in the learning environment.

Keywords:

• discipline-related social identity
• learning approach
• norms
• teaching quality
• higher education

Notes

1. Discipline was measured using a free-response item worded ‘What is your main area of study (i.e. your chosen major, the discipline – e.g. Psychology, Biology etc)?’. A wide range of responses were given and coded into 20 categories (approximate percentages of the full sample below): accounting (9%), actuarial/statistics (3%), archaeology/anthropology (2%), Asia-pacific (1%), astronomy/physics (5%), biology/zoology (12%), history/ philosophy/sociology (5%), economics (3%), literature/film/art/music (3.5%), languages (5%), business/commerce/marketing/management (4%), chemistry (3%), information technology (3%), finance (2.5%), development/environmental science (2%), electronic/engineering/mechanic/mechatronics (6%), international relations/politics (10%), law (10%), medical science (1%) and psychology (15%). Owing to the structure of this question, responses were highly variable and in many cases more than one response was given. This does not affect the participants’ identification with their self-identified ‘field of study’ but it does prevent us from examining systematic effects across disciplines.However, we thank two anonymous reviewers for the suggestion that we may be able to analyse this data using a broader domain categorisation. To explore this, the disciplines were organised into 6 broad categories (approximate percentages of the full sample below: health & natural science (18%), business & economics (23%), arts & languages (14%), social science & history (21%), mathematics, physical sciences and engineering (18%) and law (7%)). We then performed analyses on the 48% of participants (n = 141) who fit neatly into one discipline area. We dummy coded the six domains into five variables and entered them in the first block of the sequential regressions in order to see if there were any effects for discipline grouping even prior to accounting for any other variables. Neither the total effect of the block (representing the main effect of discipline grouping) nor any of the individual grouping codes was significant. Therefore, there was no indication that the discipline-grouping variable was acting as a predictor of learning approach in this data, and we did not consider it further.

2. To explore the possibility that there were substantive differences between these two-year-level cohorts, independent samples t-tests were used to compare the means on the key variables of interest (identification, both learning approaches, learning approach norms, teaching quality and conscientiousness) between year levels. As none of the differences were significant, the year-level cohorts were collapsed and treated as one sample.

3. In simple models, the causal order of variables can be reversed without changing the implied covariation matrix, and therefore, the reciprocal relations are not informative. In the current case, however, the reciprocal model is not a simple reversal of all relations in the model, and the patterns of covariates are somewhat different, suggesting that the reciprocal models here represent substantively different, and separately testable, models of the underlying covariation matrix.