New 32-item scale measures how ready small farmers are for digital transformation

Researchers have built and tested a questionnaire that measures, in a single instrument, both the mental and organizational skills that let farmers adopt digital technology and the stage of digital maturity they have actually reached. The tool, described in a study published in Frontiers in Sustainable Food Systems, was developed and validated with 68 potato farmers in south-central Chile, and its authors present it as a preliminary but promising diagnostic instrument for one of the least-studied corners of agricultural digitalization: the smallholder in an emerging economy.

The scale rests on two established theoretical frameworks. The first is Teece’s dynamic capabilities model, which breaks an organization’s adaptive capacity into three sequential abilities: sensing, or spotting opportunities and threats in the environment; seizing, or mobilizing resources to act on them; and transforming, or reconfiguring routines and structures to sustain the change. The second is Verhoef and colleagues’ staging of digital maturity, which runs from digitization, the conversion of analog information into digital formats, through digitalization, the redesign of business processes with digital tools, up to full digital transformation, in which technology reshapes the business model itself. Earlier work had examined digital literacy, adoption behavior, or market participation among farmers in isolation; the authors argue that no validated tool had combined capabilities and maturity stages within one framework with the individual farmer as the unit of analysis.

Building the instrument followed five stages. The team first grounded each dimension in the literature, then drafted 46 items written in language accessible to producers with varied schooling, deliberately replacing corporate jargon with farm-level concepts and using empowering, capability-oriented wording to counter documented psychosocial barriers among marginalized rural populations. A panel of four experts from academia, agribusiness consulting, and public agricultural development in three Latin American countries then reviewed relevance, clarity, and coherence, prompting reformulation of seven items and removal of two, leaving 44. The refined questionnaire was administered to the farmer sample for an exploratory factor analysis, and a confirmatory factor analysis followed using SmartPLS software.

The statistical pruning was substantial. Parallel analysis and model comparison supported retaining the theoretically specified six-factor structure, which explained 50.10 percent of total variance and fit better than a three-factor alternative that merged sensing and seizing into one undifferentiated factor. Items with weak sampling adequacy, low factor loadings, or negative effects on internal consistency were dropped one by one, and four further items were removed during reliability analysis. The final instrument contains 32 items across the six dimensions, rated on a five-point agreement scale, with an average completion time of roughly 15 minutes.

The psychometric results are strong in most places, with acknowledged weak spots. Cronbach’s alpha, composite reliability, and McDonald’s omega exceeded the conventional 0.70 threshold for every dimension; Guttman’s lambda 6 fell slightly short at 0.636 for seizing, a shortfall the authors attribute to that dimension’s small item count. Average variance extracted surpassed 0.50 everywhere except digitalization, where the value of 0.428 was deemed acceptable because composite reliability remained above 0.60. The Tucker-Lewis index for the exploratory solution, 0.867, sat marginally below the 0.90 benchmark. Discriminant validity held across all construct pairs by the HTMT criterion, though four of fifteen pairs had confidence intervals touching or exceeding 0.90, and one transforming item showed a slightly higher cross-loading on a neighboring construct but was retained to preserve conceptual coverage.

The substantive findings sketch a clear picture of where these farmers stand. Composite scores were highest for sensing (mean 3.75) and seizing (3.71), indicating broad recognition that digital technology matters and willingness to invest in it. Digitization scored 3.37, consistent with the near-universal use of smartphones, messaging applications, and digital photos for monitoring crops. Scores dropped sharply for the more advanced stages: digitalization averaged 2.47, with 77.9 percent of producers below the scale midpoint, and digital transformation averaged 2.73, with 55.9 percent below the midpoint. Transforming capability itself was moderate at 3.09, with 45.6 percent below the midpoint, suggesting that turning awareness into concrete changes in routines and business models remains the bottleneck. The authors stress that these are relative profiles against the scale midpoint, not validated classifications of individual farmers into maturity categories, since no normative thresholds yet exist.

The structural model estimated alongside the measurement model lent support to most of the hypothesized pathways. Sensing predicted seizing (beta 0.619), seizing predicted transforming (0.653), digitization predicted digitalization (0.639), digitalization predicted digital transformation (0.498), and sensing predicted digitization (0.441), all significant at p < 0.001. Two paths failed: the direct link from seizing to digitalization (beta 0.103, p = 0.273) and the hypothesized moderating role of transforming capability on the digitalization-to-transformation relationship (beta 0.054, not significant). The authors attribute these null results at least partly to the limited statistical power of a 68-person sample, particularly for interaction tests, and call for re-examination in larger studies. Even so, the model explained 60.1 percent of the variance in digital transformation, the most advanced maturity stage.

Which items fell out of the scale is itself informative. Items about listening to others before technological change, participation in regional technical assistance programs, sensors for monitoring crops and weather, automated irrigation, and training others in new technologies all performed poorly, patterns the authors read as reflecting small producers’ relative isolation, limited access to formal support systems, and constrained capacity to absorb complex technologies. The lowest-scoring retained items concerned management software and product tracking, while the highest concerned perceived value of technology, willingness to invest, and smartphone ownership. That contrast, high perceived value alongside low implementation, points to structural barriers in rural contexts such as infrastructure, financing, and digital skills rather than a lack of motivation.

The study’s limitations are stated plainly. The sample was a non-probabilistic purposive draw from an INDAP registry of 338 potato farmers in the Biobío Region, recruited at two PRODESAL extension workshops between August and September 2025, so unregistered farmers outside public programs are not represented. The cross-sectional, self-reported design cannot establish that farmers actually progress through maturity stages as capabilities develop; that would require longitudinal or experimental evidence. Item interpretability was checked only by expert judgment, without cognitive interviews with farmers, and the in-person group administration cannot fully exclude interviewer or social desirability effects. Multivariate normality was not met, which the authors cite as further justification for their PLS-SEM approach.

For now, the instrument’s practical use is as a comparative diagnostic rather than a classification system. The recommended scoring averages each dimension’s items into a six-point profile per farmer, interpreted against the scale midpoint to flag comparatively underdeveloped capabilities or maturity stages. Extension agents and policymakers could use such profiles to target support, and future applications might generate individual visual summaries for producers. The authors frame the tool as an initial step toward validated diagnostic instruments for digital transformation in small-scale agriculture, with replication needed in larger, more diverse samples and with the excluded contextual items, covering institutional support, community networks, and infrastructure, potentially re-entering future models as external determinants rather than individual capacity measures.

Subject of Research: Validation of a survey scale measuring digital transformation capabilities and digital maturity among smallholder potato farmers in Chile

Article Title: Design and validation of a measurement scale for digital transformation capabilities in agricultural settings

Article References: Niño-Amézquita, D., León-Castro, E., & Blanco-Mesa, F. (2026). Design and validation of a measurement scale for digital transformation capabilities in agricultural settings. Frontiers in Sustainable Food Systems, 10, Article 1950275. https://doi.org/10.3389/fsufs.2026.1950275

Image Credits: AI Generated

DOI: 10.3389/fsufs.2026.1950275

Keywords: digital transformation, agriculture, smallholder farmers, scale validation, dynamic capabilities, digital maturity, psychometrics, Chile, potato farming, survey instrument, PLS-SEM, rural digitalization