Malaysian youth's propensity towards ambidextrous learning: an in-depth analysis of behavioural insights dimensions

Across Asia, countries are reforming vocational education to equip students with versatile, industry-relevant skills. Ambidextrous learning—balancing exploration and exploitation—integrates rigorous technical training with intellectual study to prepare adaptable graduates. In Malaysia, the 2020 curriculum revision expanded access to vocational electives in upper secondary school to support pathways into TVET-focused institutions. Despite growing employer recognition, TVET enrolment remains low due to persistent stigma, social norms, limited awareness and knowledge, and issues of self-efficacy. The study investigates whether an ambidextrous learning intervention can positively shape students’ perceptions and intentions toward TVET-related careers. Research objectives: (i) compare pre- and post-test effects of an ambidextrous learning programme on TVET intention, TVET knowledge, and self-efficacy; (ii) examine whether TVET knowledge and self-efficacy predict TVET intention.

Behavioural Insights: Drawing from psychology, cognitive, and social sciences, Behavioural Insights (BI) approaches have effectively modified behaviours across domains and are used in policy design. This study’s intervention adopts BI principles to influence youth decisions about TVET. Self-efficacy: Prior work links vocational self-efficacy to exploration, persistence, classroom participation, and reduced procrastination among TVET students. Low self-efficacy undermines persistence and interest, while higher self-efficacy predicts greater interest in TVET. Hypothesis 1: Self-efficacy positively predicts TVET intention. TVET knowledge: Lack of accurate information reduces participation and narrows career perceptions. Providing correct, comprehensive TVET information reduces misconceptions and supports informed choices. Hypothesis 2: TVET knowledge positively predicts TVET intention.

Design: Randomised controlled trial with no-treatment control group; pre- and post-tests administered with a 2–3 month interval. Participants: N=480 (240 intervention, 240 control), 16-year-old Form 4 students from eight government academic schools across four Peninsular Malaysia regions (Northern, Central, Southern, Eastern). Sampling: Cluster random sampling; approximately 60 students per school (two classes, ~30 each) split between intervention and control. Intervention: Ambidextrous TVET Junior programme (half-day, ~5 hours), developed using Behavioural Insights, integrating Theory of Planned Behaviour and Social Cognitive Theory. Target field: automotive. Seven modules across three phases (knowledge/comprehension; application/practice; evaluation): TVET World, Mental Focus, Workplace Safety, See and Practise, Explore and Trial, Exploit and Response, and Reflection. Methods included problem-solving, hands-on activities, games, and imitation exercises to influence perceptions, build knowledge, and strengthen self-efficacy. Measures: Self-report questionnaire—demographics (4 items), TVET knowledge (6 items), TVET intention (4 items), self-efficacy (4 items); Likert-type scales (5-point). Reliability: TVET knowledge α=0.968; TVET intention α=0.838; self-efficacy α=0.809. Procedures: School-level scheduling; informed consent obtained. Analysis: Normality via skewness/kurtosis (|values|<2.0 deemed normal); independent and paired t-tests for mean differences; structural equation modelling using SmartPLS with standard bootstrapping (500 samples) to test direct effects of self-efficacy and TVET knowledge on TVET intention (post-test).

• Pre-test: No significant differences between intervention and control groups on TVET intention, TVET knowledge, or self-efficacy (all p>0.05). Example pre-test means: TVET intention Actual 2.10 vs Control 2.12; TVET knowledge Actual 2.09 vs Control 2.06; Self-efficacy Actual 2.45 vs Control 2.43.
• Post-test: Significant differences favoring the intervention group across all outcomes: • TVET intention: Actual 4.35 vs Control 2.32; t=45.153, p<0.001. • TVET knowledge: Actual 4.31 vs Control 2.26; t=53.395, p<0.001. • Self-efficacy: Actual 4.33 vs Control 2.43; t=44.350, p<0.001.
• Structural model (post-test, N=480, bootstrapped 500): • Self-efficacy → TVET intention: β=0.399, t=10.080, p<0.01 (H1 supported). • TVET knowledge → TVET intention: β=0.546, t=14.008, p<0.01 (H2 supported).
• TVET knowledge was the strongest predictor of TVET intention, surpassing self-efficacy.

The ambidextrous learning intervention, designed with Behavioural Insights and hands-on, problem-based activities, produced significant gains in students’ TVET knowledge, intention, and self-efficacy. Allowing participants to explore and practise vocational tasks (e.g., wiring a car horn circuit) enhanced cognitive engagement and efficacy, aligning with established pathways for behaviour change. Clear exposure to TVET fundamentals—concepts, clusters, and career trajectories—appears to shift attitudes and subjective norms, countering misconceptions that deter enrolment. Despite strong effects, delivery constraints (half-day sessions within school hours) limited the depth and pacing for some participants. Future research should vary delivery formats and tailor interventions to different student streams, prior vocational experience, and school types to optimise impact and generalisability.

TVET knowledge and self-efficacy significantly shape students’ intentions to pursue TVET, with knowledge exerting the strongest influence. Many students lack sufficient, accurate information about TVET concepts, clusters, and career pathways, impeding informed decision-making. Educational initiatives should prioritise comprehensive, accessible information and guided exposure to TVET to improve attitudes and subjective norms and reduce stigma. The study demonstrates that an ambidextrous, Behavioural Insights–based intervention can effectively increase TVET-related knowledge, self-efficacy, and intention. Future work should refine delivery modalities and broaden sectoral coverage to sustain and scale these gains.

• Intervention duration was constrained to a half-day (~5 hours) within school hours, limiting depth and pacing; some participants were uncomfortable completing all activities in the available time.
• Sessions were scheduled separately across schools due to logistics, which may introduce contextual variability.
• The intervention content specifically targeted the automotive occupational field, which may limit generalisability across all TVET clusters; future studies should test across diverse fields and student characteristics (study streams, vocational experience, school types).