Thesis – Adoption of Precision Agriculture Technologies Among Australian Farmers

Abstract

Precision agriculture (PA) technologies offer Australian broadacre farmers substantial gains in input efficiency, yield and environmental stewardship, yet uptake of the more advanced tools remains uneven. This thesis extract examines the determinants of variable-rate technology (VRT) adoption among grain producers in the eastern and southern cropping zones. A sequential mixed methods design combined a survey of 350 broadacre farmers with 18 semi-structured interviews, and a binary logistic regression modelled the odds of VRT adoption. Farm size, tertiary education, adequate on-farm connectivity, trust in data handling and engagement with a private agronomist significantly increased the odds of adoption, while operator age and a strong perception of cost as a barrier reduced them. Connectivity emerged as the single largest contextual constraint, consistent with concerns raised in the 2021 Regional Telecommunications Review. The findings indicate that sustained extension effort, targeted co-investment and improved regional connectivity would narrow the persistent adoption gap identified by ABARES and GRDC.

Introduction

Precision agriculture encompasses a suite of spatial technologies, including global navigation satellite systems, yield monitors, variable-rate application controllers and remote sensing, that allow producers to manage variability within paddocks rather than treating each field as uniform (Barnes et al. 2019). For Australian broadacre agriculture, which operates on thin margins across large and climatically variable landholdings, the appeal is considerable. More precise placement of seed, fertiliser and chemical reduces input costs, lifts partial factor productivity and limits off-target losses to soil and waterways. ABARES (2023) reports that the most productive quartile of Australian grain farms has continued to extend its lead over the rest of the sector, and digitally enabled input management is widely regarded as one mechanism behind this divergence.

Adoption of PA in Australia is nevertheless uneven. Basic guidance and automated steering are close to ubiquitous in the grains industry, yet the tools that capture the largest agronomic value, in particular variable-rate technology and integrated data-driven decision support, remain confined to a minority of operations (GRDC 2022; Llewellyn and Ouzman 2019). This pattern, in which an enabling technology diffuses rapidly while the higher-order applications built upon it lag well behind, defines the adoption gap that motivates the present study. The gap matters because the productivity and environmental returns of PA accrue disproportionately from the advanced layer rather than from guidance alone, a distinction central to the shift from precision agriculture to what Leonard et al. (2017) term decision agriculture.

This thesis extract examines why the adoption gap persists and which farm, operator and contextual factors distinguish adopters from non-adopters. Three research questions guided the study:

1. What is the current level of uptake of key PA technologies among Australian broadacre grain farmers?
2. Which farm, operator and contextual factors predict adoption of variable-rate technology?
3. How do farmers themselves account for their decisions to adopt or defer PA?

The scope is limited to broadacre grain production in the eastern and southern cropping zones, and VRT is used throughout as the marker of advanced adoption because it demands the data infrastructure, agronomic interpretation and capital commitment that basic guidance does not.

Literature Review

Diffusion of innovations and technology acceptance

Two complementary theories frame the analysis. Diffusion of innovations theory holds that the rate at which a technology spreads depends on five perceived attributes: relative advantage, compatibility, complexity, trialability and observability (Rogers 2003). PA scores highly on relative advantage but poorly on complexity and observability, since the payoff from variable-rate application is often diffuse and difficult to attribute in a single season. The Technology Acceptance Model (TAM) narrows the behavioural focus to two beliefs, perceived usefulness and perceived ease of use, that mediate the relationship between external conditions and the decision to adopt (Davis 1989). Its later extension into the Unified Theory of Acceptance and Use of Technology adds facilitating conditions and social influence, both of which are salient in tight-knit farming communities (Venkatesh et al. 2003). Together the two traditions suggest that adoption is driven not only by objective advantage but by how usefulness and effort are perceived under local constraints.

Barriers to adoption

The international evidence identifies a recurring set of barriers: high capital cost and uncertain payback, limited digital skills, weak connectivity, data interoperability problems, and distrust over who controls farm data (Pathak, Brown and Best 2019; Pierpaoli et al. 2013). In the Australian setting, connectivity is a distinctive constraint. The 2021 Regional Telecommunications Review documented persistent gaps in mobile and broadband coverage across agricultural regions and characterised the demand for reliable regional connectivity as a step change rather than an incremental need (RTIRC 2021). Because much PA value depends on real-time data transfer between machinery, sensors and cloud platforms, poor connectivity undermines the perceived usefulness of the whole technology stack, not merely one component.

Research gap

While the drivers of PA adoption have been reviewed extensively overseas, fewer Australian studies model adoption multivariately using primary data, and connectivity is rarely tested as a formal predictor alongside behavioural beliefs. Existing national work has mapped uptake and profitability (GRDC 2022; Llewellyn and Ouzman 2019) but has not integrated survey estimates of contextual constraint with farmers’ own accounts of the adoption decision. This study addresses that gap for the broadacre grains sector.

Methodology

Research design and sample

A sequential mixed methods design was adopted, in which a quantitative survey established the predictors of adoption and a qualitative interview phase explained the statistical patterns from participants’ perspectives. The survey sample comprised 350 broadacre grain farmers drawn from grower-group membership lists and regional field-day registrations across New South Wales, Victoria, South Australia and southern Queensland, weighted to approximate the farm-size distribution reported in ABS agricultural commodity data (ABS 2022). The realised response rate was 41 per cent. A maximum-variation subsample of 18 survey respondents, spanning small and large enterprises and both adopter and non-adopter groups, completed semi-structured telephone interviews of 40 to 55 minutes.

Conceptual model and measures

The dependent variable was binary adoption of variable-rate technology (adopter or non-adopter). Predictors were selected to represent the farm, the operator and the contextual environment, and were organised into the conceptual model shown in Figure 1, which integrates the behavioural beliefs of TAM with the contextual constraints emphasised in the Australian literature. Continuous predictors were farm size (per 1,000 ha) and operator age (per decade); categorical predictors were tertiary education, adequate on-farm connectivity, a strong perception of cost as a barrier, trust in data handling, and engagement of a private agronomist.

Analysis and ethics

Survey data were analysed using descriptive statistics and binary logistic regression, with results reported as adjusted odds ratios and 95 per cent confidence intervals. Interview transcripts were analysed thematically. The study was approved by the administering university’s Human Research Ethics Committee; participation was voluntary with written informed consent, and identifying details were removed so that individual enterprises could not be recognised.

Results

Sample characteristics and technology uptake

Respondent and farm characteristics are summarised in Table 1. The sample skewed towards experienced operators, with 76 per cent aged 45 years or older, and towards mid-sized to large enterprises. Uptake confirmed the adoption gap: automated steering was near universal at 86.0 per cent, whereas variable-rate technology reached only 40.0 per cent of respondents, a figure broadly consistent with earlier national estimates for the grains sector (Llewellyn and Ouzman 2019).

Table 1: Respondent and farm characteristics, and precision agriculture uptake (n = 350)

Determinants of adoption

The logistic regression results are reported in Table 2. Seven predictors were entered, and the model was a good fit to the data (model chi-square (7) = 96.4, p < 0.001; Nagelkerke R-squared = 0.34; Hosmer-Lemeshow p = 0.62), correctly classifying 74.0 per cent of cases. Adequate connectivity produced the largest effect, more than doubling the odds of adoption, followed by tertiary education and trust in data handling. Operator age and a strong perception of cost as a barrier reduced the odds of adoption, the latter almost halving them.

Table 2: Binary logistic regression predicting variable-rate technology adoption (n = 350)

Note. Continuous predictors are centred at the sample mean (age at 50 years, farm size at 2,000 ha). Reference categories are secondary or trade education, inadequate connectivity, no strong cost barrier, no trust in data handling, and no agronomist engaged.

The practical meaning of the model is best conveyed through a predicted probability. For an operator aged 40 years (one decade below the mean, so the centred value is -1.0) managing 4,000 ha (2.0 units above the 2,000 ha mean), university educated, with adequate connectivity, trust in data handling and an engaged agronomist, and who does not regard cost as a strong barrier, the log-odds are obtained by summing the relevant coefficients:

logit(p) = -0.850 + 0.350(2.0) + (-0.248)(-1.0) + 0.765 + 0.986 + 0.636 + 0.548 = 3.03

Converting this log-odds to a probability gives:

p = 1 / (1 + e^(-3.03)) = 1 / (1 + 0.048) = 0.95

The model therefore assigns this profile a 95 per cent probability of VRT adoption, against a baseline of roughly 30 per cent for an otherwise average operator lacking connectivity, tertiary education and advisory support. The contrast underlines that adoption is not the product of any single factor but of a cluster of enabling conditions acting together.

Interview themes

The 18 interviews explained these patterns in farmers’ own terms. Four themes recurred. First, unreliable connectivity was described as a daily friction that eroded confidence in cloud-dependent tools, echoing the systemic gaps recorded in the Regional Telecommunications Review (RTIRC 2021). Second, capital cost was framed less as an absolute obstacle than as an uncertain payback: several non-adopters said they could not confidently forecast a return within a realistic replacement cycle. Third, data trust was material, with farmers wary of how machinery and platform providers might use paddock data. Fourth, trusted intermediaries mattered, and respondents who engaged an agronomist or drew on peer networks and GRDC-supported grower groups were markedly more confident interpreting variable-rate prescriptions.

Discussion

The results answer the research questions and align the Australian evidence with established theory. Connectivity was the strongest determinant of adoption, which situates PA uptake as much within telecommunications policy as within agronomy. Because variable-rate systems depend on reliable data transfer, the coverage deficits documented by the Regional Telecommunications Review (RTIRC 2021) depress the perceived usefulness of the entire technology stack, exactly the belief that TAM identifies as pivotal (Davis 1989). Improving regional connectivity is therefore not a peripheral enabler but a precondition for narrowing the adoption gap.

Cost operated in the manner predicted by diffusion theory. It was the perception of uncertain payback, rather than price alone, that suppressed adoption, reflecting the weak observability and trialability of PA benefits that Rogers (2003) associates with slow diffusion. This finding supports policy instruments that reduce payback uncertainty, such as demonstration sites and co-investment, over blunt subsidies. The significant effect of engaging an agronomist reinforces the point: trusted advice compresses the complexity of the innovation and improves its observability, consistent with the international barriers literature (Pathak, Brown and Best 2019; Pierpaoli et al. 2013).

Trust in data handling emerged as an independent predictor, underlining that adoption is a socio-technical decision and not merely a financial one. As the sector moves from precision agriculture towards decision agriculture (Leonard et al. 2017), clarity over data ownership and interoperability will increasingly shape uptake. Finally, the role of extension is central. The concentration of adoption among operators connected to agronomists and grower groups suggests that publicly supported extension through GRDC, CSIRO and the innovation priorities set out by DAFF (2023) remains one of the most effective levers available, particularly for smaller and older operators who face the steepest belief-related barriers.

Conclusion

This thesis extract set out to explain the persistent gap between near-universal adoption of basic guidance and the limited uptake of advanced precision agriculture among Australian broadacre farmers. Using survey and interview data, it found that adoption of variable-rate technology is driven by a cluster of enabling conditions, of which regional connectivity, tertiary education, trust in data handling and access to trusted advice are the most influential, while operator age and perceived cost uncertainty act as brakes. The analysis contributes an Australian, multivariate account that positions connectivity as a formal determinant rather than a background condition. Its limitations, a cross-sectional design, self-reported measures and a sample weighted towards experienced operators, mean the estimates should be read as associations rather than causes. Even so, the policy implication is clear: closing the adoption gap identified by ABARES and GRDC depends less on the technology itself than on the connectivity, extension and trust that allow farmers to use it with confidence.

References

Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES) 2023, Australian farm survey results 2020-21 to 2022-23, Department of Agriculture, Fisheries and Forestry, Canberra.

Australian Bureau of Statistics (ABS) 2022, Agricultural commodities, Australia, 2020-21, ABS, Canberra.

Barnes, AP, Soto, I, Eory, V, Beck, B, Balafoutis, A, Sanchez, B, Vangeyte, J, Fountas, S, van der Wal, T and Gomez-Barbero, M 2019, ‘Exploring the adoption of precision agricultural technologies: a cross regional study of EU farmers’, Land Use Policy, vol. 80, pp. 163-174.

Davis, FD 1989, ‘Perceived usefulness, perceived ease of use, and user acceptance of information technology’, MIS Quarterly, vol. 13, no. 3, pp. 319-340.

Department of Agriculture, Fisheries and Forestry (DAFF) 2023, National statement on agricultural innovation, Australian Government, Canberra.

Grains Research and Development Corporation (GRDC) 2022, Precision agriculture adoption and profitability in the Australian grains industry, GRDC, Canberra.

Leonard, E, Rainbow, R, Trindall, J, Baker, I, Barry, S, Heath, R, Jakku, E, Laurie, A, Lamb, D, Llewellyn, R, Perrett, E, Sanderson, J and Skinner, A 2017, Accelerating precision agriculture to decision agriculture: enabling digital agriculture in Australia, Australian Farm Institute, Sydney.

Llewellyn, R and Ouzman, J 2019, Conservation cropping and precision agriculture adoption in the Australian grains industry, CSIRO, Canberra.

Pathak, HS, Brown, P and Best, T 2019, ‘A systematic literature review of the factors affecting the precision agriculture adoption process’, Precision Agriculture, vol. 20, no. 6, pp. 1292-1316.

Pierpaoli, E, Carli, G, Pignatti, E and Canavari, M 2013, ‘Drivers of precision agriculture technologies adoption: a literature review’, Procedia Technology, vol. 8, pp. 61-69.

Regional Telecommunications Independent Review Committee (RTIRC) 2021, 2021 Regional Telecommunications Review: a step change in demand, Commonwealth of Australia, Canberra.

Rogers, EM 2003, Diffusion of innovations, 5th edn, Free Press, New York.

Venkatesh, V, Morris, MG, Davis, GB and Davis, FD 2003, ‘User acceptance of information technology: toward a unified view’, MIS Quarterly, vol. 27, no. 3, pp. 425-478.