The evolution of agriculture has occurred across four major phases, each shaped by technological and social interruption. Agriculture 1.0 was characterised by manual labour and animal power, while Agriculture 2.0 emerged through mechanisation during the Industrial Revolution. Agriculture 3.0, associated with the Green Revolution, introduced chemical fertilisers, irrigation, genetic selection, and early automation to improve productivity. More recently, the sector has transitioned towards Agriculture 4.0, marked by the integration of digital technologies including artificial intelligence, robotics, precision agriculture, and data-driven decision-making systems ([11]).

Technological innovation has meant that agricultural occupations now require higher levels of scientific literacy, technological capability, and adaptive problem-solving ([10]; [20]). This shift is reflected in industry workforce analyses, which emphasise the need for future-ready workers capable of integrating technical, digital, and adaptive skills within agricultural systems ([3]). As a result, workforce preparation has become increasingly dependent on formal education systems capable of developing both technical and transferable skills. Vocational Education and Training (VET) is well placed as a key mechanism for preparing a skilled workforce aligned with evolving industry needs ([23]; [4]; [12]).

However, despite these increasing demands, the agricultural workforce has historically relied on informal and workplace-based learning pathways. In 2021, approximately 49% of individuals employed in Australian agriculture held no post-secondary qualifications, compared with 29% across the broader workforce ([1]). This reflects a longstanding reliance on experiential “on-the-job” skill development rather than formalised training ([6]).

As agricultural systems become more technologically advanced, concerns emerge regarding the extent to which Agriculture Technical and Vocational Education and Training (ATVET) aligns with these evolving skill requirements. Evaluating the quality of ATVET is therefore critical in determining whether vocational training systems are effectively preparing a workforce capable of operating within increasingly complex and digitalised production environments.

In Australia, VET is delivered through Registered Training Organisations (RTOs), including Technical and Further Education (TAFE) institutes, private providers, universities, and schools offering VET in Schools (VETiS). Training is based on nationally endorsed training packages that define competency standards aligned with industry requirements. The quality of VET provision is regulated through a compliance-based framework, with the Australian Skills Quality Authority (ASQA) assessing providers against national standards using a risk-based approach ([6]).

One component of this regulatory system is the Student Outcomes Survey (SOS), which provides national data on learner satisfaction. Satisfaction measures capture learners’ perceptions of teaching quality, assessment practices, learning resources, and institutional support, and are often used as indicators of training quality ([13]). However, reliance on satisfaction as a proxy for quality remains contested.

The concept of quality in vocational education is inherently multidimensional, encompassing not only learner experience but also labour market outcomes, workplace learning conditions, and alignment with industry needs ([8]; [19]). From an integrative learning perspective, effective vocational education requires the synthesis of theoretical knowledge, practical skills, and authentic workplace experience ([25]). Similarly, recent international empirical research demonstrates that training quality is shaped by multiple interacting factors across institutional and workplace contexts ([7]; [9]).

Systemic evaluation frameworks provide a structured approach to capturing this complexity. The Context-Input-Process-Product (CIPP) model conceptualises educational quality across multiple dimensions, including contextual relevance, institutional inputs, delivery processes, and training outcomes ([24]; [21]; [26]). Within this framework, learner satisfaction represents an indicator of the process dimension, but must be interpreted alongside broader measures such as completion rates and employment outcomes.

Despite increasing attention to VET quality, empirical evaluations of ATVET remain limited. In particular, little is known about whether national student satisfaction data can provide meaningful insight into the quality of agricultural training programs.

This study is guided by the Context-Input-Process-Product model (CIPP) evaluation framework developed by Daniel [24]. The CIPP model conceptualises educational quality as a multidimensional construct comprising contextual relevance (Context), institutional and learner characteristics (Input), delivery and learning experiences (Process), and educational or workforce outcomes (Product). Within vocational education, this framework has been widely applied to evaluate training effectiveness across both institutional and workplace settings ([21]; [26]).

The selection of the CIPP model reflects the complexity of Agriculture Technical and Vocational Education and Training (ATVET) within the context of Agriculture 4.0. As agricultural industries become increasingly dependent on technological capability, adaptive expertise, and workplace integration, evaluating training quality requires consideration of multiple interacting dimensions rather than reliance on a single indicator of performance. While learner satisfaction provides insight into students’ perceptions of teaching quality, assessment practices, learning resources, and institutional support, it primarily reflects the process dimension of educational quality.

Accordingly, this study conceptualises student satisfaction as one component of a broader ATVET quality system. Contextual variables such as field of education and provider characteristics, input variables including student demographics and qualification levels, and product indicators such as completion and occupational outcomes are incorporated to support a more comprehensive interpretation of training quality. The conceptual relationship between these dimensions informed both the analytical structure of the study and the interpretation of findings.

This study, therefore, addresses the following research questions:

RQ1: Do student satisfaction levels differ between agriculture and non-agriculture VET programs?

RQ2: Which aspects of training delivery contribute most strongly to satisfaction among ATVET students?

RQ3: How do demographic and contextual factors influence satisfaction within agricultural VET programs?

The data for this study were obtained from the National Centre for Vocational Education Research (NCVER), which commissions the Social Research Centre (SRC) at the Australian National University (ANU) to administer the annual National Student Outcomes Survey (SOS). This month-long survey collects information on the satisfaction and experiences of students undertaking, part completing or having graduated from vocational education and training across Australia during the preceding year. The survey employs stratified sampling procedures designed to produce nationally representative estimates of VET student outcomes across Australian states and provider types.

De-identified unit record files (URFs) for the years 2020 to 2024 were accessed through NCVER. These individual datasets were imported into SPSS and merged to produce a consolidated file representing a total of 746,678 respondents across the five-year period. The merged NCVER unit record file contained 746,678 respondent records across the 2020-2024 survey years. Following removal of partial and non-responses for satisfaction variables, the analytic sample used for satisfaction analyses comprised 587,389 valid responses. Within this sample, Agriculture students (ASCED FOE 0501 and 0503) represented approximately 1.2% of respondents (n = 6875), with the remainder classified as Non-agriculture. Consistent with NCVER’s published reporting approach, short-course participants were excluded. Two main respondent groups were considered: those who had completed their qualification at the time of the survey (completers), and those who had withdrawn or were still enrolled (part-completers). The time period for this study was selected as it was the only time frame available that utilised the national survey instrument in comparative forms. While this period incorporates the COVID-19 global pandemic, analysis evaluated any inconsistencies over the timeframe.

Programs were classified using the Australian Standard Classification of Education (ASCED) framework. This framework classifies educational areas by their various sector grouping, subject areas and occupations within each sector. Each respondent was coded as belonging either to the Agriculture or Non-agriculture category according to their reported Field of Education (FOE). Respondents were classified as Agriculture where their reported Field of Education was coded to ASCED 0501 Agriculture or ASCED 0503 Horticulture and Viticulture. All other ASCED fields were classified as Non-agriculture. This classification included Agricultural Science, Animal Husbandry, Wool Science, Horticulture and Viticulture programs. Forestry, Fisheries, Aquaculture, Environmental Studies and Conservation-related fields were excluded because they fall outside ASCED 0501 and 0503 and were not considered production agriculture for the purposes of this study, as demonstrated in Table 1.

Table 1. Inclusion of Field of Education detailed codes in “Agriculture” cohort (adapted from [2]).

Each record was coded according to whether the respondent had completed or not completed their program of study at the time of survey administration. A review of the encoded identifiers confirmed that there were no duplicate respondents, supporting the assumption of independent responses across years.

Data relating to program enrolments and completions were obtained from the most recent National Centre for Vocational Education Research (NCVER) publications ([15]), covering the period from 2019 to 2023. These datasets were filtered using the same four-digit Field of Education (FOE) codes described above to enable consistent identification of Agriculture and non-agriculture programs. Completion to enrolment ratios were calculated as the proportion of recorded completions relative to total enrolments within each sector, allowing comparison between agricultural training programs and the broader VET population. Because enrolment and completion counts do not track the same student cohort over time, the measure should be interpreted as a completion-to-enrolment ratio rather than a true longitudinal completion rates for a tracked cohort.

In addition to program completers, the analysis included a group classified as part-completers. This category comprises respondents who had not completed their qualification at the time of survey administration and includes both individuals who had formally withdrawn from their training and those who remained enrolled but had not yet reached completion. As such, the category represents a heterogeneous population of learners at different stages of program participation. However, the available dataset does not clearly differentiate between those who have permanently withdrawn from training and those who remain enrolled but have not yet completed their qualification. Consequently, the data do not provide a comprehensive picture of the experiences of students who have actively exited training.

This limitation is important when interpreting the results. Students who withdraw from training may do so for a variety of reasons, including dissatisfaction with teaching quality, assessment practices, program relevance, institutional support, or broader personal and employment circumstances. Research on vocational education has shown that learner dissatisfaction and disengagement are often associated with increased likelihood of withdrawal or non-completion ([6]; [13]). If individuals who withdraw are less likely to respond to post-training surveys, their perspectives may be underrepresented in the dataset. As a result, the available satisfaction data may disproportionately reflect the experiences of students who remained engaged with training or who completed their qualification. This potential underrepresentation of withdrawn students limits the ability of the present analysis to fully identify weaknesses in training delivery or other factors that may contribute to program discontinuation.

Survey participants responded on a five-point Likert scale ranging from very satisfied to very dissatisfied. Likert responses were collapsed into three categories (satisfied, neutral, dissatisfied) following NCVER reporting conventions to facilitate comparison with published national results. Partial and non-responses were removed, yielding a total of 587,389 valid responses. Frequencies and percentages were generated using SPSS, with summary tables prepared in Microsoft Excel.

Analyses were conducted using the unweighted unit record data. The purpose of the study was comparative analysis between Agriculture and Non-agriculture cohorts rather than estimation of population parameters. While NCVER weighting procedures are used for official national reporting, unweighted analyses preserve observed respondent frequencies and are appropriate for examining associations between variables. Results should therefore be interpreted as reflecting relationships within the survey sample rather than weighted national estimates.

Three main analyses were conducted:

Overall satisfaction comparisons between cohortsSatisfaction with specific program aspectsDemographic influences on satisfaction among Agriculture studentsOverall Satisfaction

To evaluate for differences in satisfaction between cohorts, a chi-square test of independence was employed to determine whether satisfaction distributions differed between groups. Chi-square tests were used due to the categorical nature of the satisfaction variables and the large sample size of the dataset. Due to the very large sample size of the SOS dataset, statistically significant chi-square results should be interpreted cautiously. Large datasets can detect extremely small differences that may not represent meaningful practical effects. For this reason, and because contingency tables exceeded a 2 × 2 table, effect sizes were measured using Cramer’s V(φc), which determine the substantive strength of associations. The strength of association was interpreted using the thresholds outlined in Table 2.

Table 2. Effect size and strength of association using Cramer’s V ([22]).

The satisfaction indicators used in the analysis are derived from summary variables in the NCVER Student Outcomes Survey dataset. These variables are constructed from Likert-scale survey items measuring learners’ perceptions of different aspects of their training experience. Table 3 summarises the constructs used in the analysis and the corresponding survey items from which they are derived. Chi-square tests were again used to determine whether responses differed between Agriculture and Non-agriculture students, and Cramer’s V was used to interpret effect size.

Following NCVER reporting conventions, responses were recoded into broader satisfaction categories for comparative analysis across training fields. While the original responses are measured on ordinal Likert scales, recoding was undertaken to enable cross-tabulation and comparison with national NCVER reporting formats. Because the SOS variables are provided as validated summary indicators by NCVER, internal reliability statistics are not reported within the URF dataset.

Table 3. Satisfaction summary variables (adapted from [14]).

The third stage of analysis examined demographic and contextual influences on satisfaction among Agriculture students. Overall satisfaction scores were cross-tabulated against demographic and institutional factors listed in Table 4. Each variable was analysed using chi-square tests to determine significance and Cramer’s V to assess association strength.

Table 4. Description of demographic and descriptive grouping variables (adapted from [14]).

The demographic results are presented by group to enable comparison across population subcategories and to identify patterns relevant to satisfaction within agricultural training programs.

To clarify how the conceptual framework was operationalised in the empirical analysis, Table 5 maps the key dimensions of the Context-Input-Process-Product (CIPP) evaluation model to the variables available within the NCVER Student Outcomes Survey dataset and associated enrolment and completion data. This approach allows student satisfaction measures to be interpreted as indicators of the training process dimension, while contextual and outcome indicators provide a broader perspective on ATVET program quality.

This operationalisation enables the analysis to examine student satisfaction within a broader framework of ATVET quality while maintaining consistency with the research questions guiding the study. As the SOS dataset was not originally designed as a comprehensive program evaluation instrument, the variables available represent proxy indicators of the CIPP dimensions rather than direct measures of all elements of vocational training quality.

Table 5. Operationalisation of the CIPP evaluation model in the analysis of ATVET quality.

Figure 1 presents completion-to-enrolment ratios for the period 2019-2023. Across the study period, agricultural programs recorded lower completion-to-enrolment ratios (24% - 29%) than the overall VET population (30% - 33%). Completion rates for structured and contextualised training pathways, such as the AgCAREERSTART program, have been reported at substantially higher levels (exceeding 80%) ([17]).

Figure 1. Comparison of completion-to-enrolment ratios (percentage of completions versus individual enrolments) for all VET courses and Agricultural programs (FOE-0501 and FOE-0503), 2019-2023 (Adapted from [15]).

Table 6 presents overall satisfaction rates expressed as within-group percentages. Across both completion groups, Agriculture students reported higher levels of satisfaction than Non-agriculture students.

Among completers, 93.8% of Agriculture students reported satisfaction compared with 90.4% of Non-agriculture students. Among part-completers, satisfaction was lower overall, with 86.9% of Agriculture students and 80.6% of Non-agriculture students reporting satisfaction.

Table 6. Student satisfaction by training aspect, program type, and completion status (percentages and differences), 2020-2024 (adapted from [14]).

The part-completer category includes both learners who had withdrawn from training and learners who remained enrolled but had not yet completed their qualification. Satisfaction outcomes for this group should therefore be interpreted cautiously, as the category combines both disengagement and normal progression through training pathways.

Across all measures, satisfaction was lower among part-completers than completers. The reduction in satisfaction between completers and part-completers was approximately 7 percentage points for Agriculture students and 9 percentage points for Non-agriculture students.

Chi-square analyses indicated statistically significant differences between Agriculture and Non-agriculture groups across most satisfaction measures; however, effect sizes were consistently negligible (φc < .02), indicating very weak associations.

Across training aspects, satisfaction with teaching and assessment was highest, while student support services were the lowest-rated component for both cohorts. Learning resources, facilities, and training location showed intermediate levels of satisfaction.

Figure 2 presents satisfaction rates within the Agriculture cohort across demographic and institutional variables. Overall satisfaction ranged between approximately 86% and 92% across all groups.

Figure 2. Percentage of survey respondent satisfaction in Agriculture-based qualifications by demographic classifiers, 2020-2024 (Adapted from [14]).

A statistically significant association was observed between gender and satisfaction (χ² (4, N = 6556) = 12.66, p = .013), although the strength of association was negligible (φc = .031).

Age group also showed a statistically significant association with satisfaction (χ² (2, N = 6556) = 8.23, p = .016), with a negligible effect size (φc = .035).

Disability status was associated with satisfaction (χ² (4, N = 6556) = 19.33, p < .001), although the strength of association remained weak (φc = .054).

No statistically significant association was observed between Indigenous status and satisfaction (χ² (4, N = 6556) = 0.53, p = .971), with a negligible effect size (φc = .009).

Remoteness showed no statistically significant association with satisfaction (χ² (8, N = 6556) = 12.78, p = .120), with a negligible effect size (φc = .031).

Level of qualification was statistically associated with satisfaction (χ² (15, N = 6875) = 46.02, p < .001), although the effect size was negligible (φc = .047).

Registered Training Organisation (RTO) type showed a statistically significant association with satisfaction (χ² (15, N = 6875) = 69.27, p < .001), with a weak effect size (φc = .058).

State of operation was also statistically associated with satisfaction (χ² (15, N = 6875) = 38.78, p < .001), although the strength of association remained negligible (φc = .043).

Across all demographic and institutional variables, effect sizes were small, indicating minimal practical differences in satisfaction outcomes.

Satisfaction of school-based Agriculture respondents by level of real workplace training.

Figure 3 shows satisfaction among Agriculture students by level of workplace learning participation. A statistically significant association was identified (χ² (15, N = 2558) = 451.039, p < .001), with a moderate effect size (φc = .242).

Note: χ² (15, N = 2558) = 451.039, p < .001, φc = .242.

Figure 3. Percentage of survey respondent satisfaction in school-based Agriculture qualifications by level of real workplace training included in the qualification, 2020-2024 (Adapted from [14]).

Programs incorporating workplace learning showed higher levels of satisfaction compared with those without workplace learning. However, increasing levels of workplace participation did not correspond to proportional increases in satisfaction.

Figure 4 presents satisfaction by post-training occupational destination. A statistically significant association was observed (χ² (9, N = 6875) = 663.881, p < .001), with a weak effect size (φc = .179).

Satisfaction was lower among respondents with unknown occupational destinations compared with those employed in the same or different industries.

Note: χ² (9, N = 6875) = 663.881, p < .001, φc of .179.

Figure 4. Percentage of survey respondent satisfaction in Agriculture-based qualifications by post-training occupational destination, 2020-2024 (Adapted from [14]).