Multivariate Probit Model Analysis of the Factors Influencing Smallholder Farmers’ Choice of ICT Tools: A Case Study of Mpumalanga, South Africa

Abstract

This study examined factors influencing smallholder farmers’ decisions to use Information and Communication Technologies (ICTs) for agricultural information in Mbombela Local Municipality, Mpumalanga, South Africa. Data were collected from 308 respondents through a quantitative cross-sectional survey using a structured questionnaire, with systematic sampling to select participants. Multivariate probit regression identified factors affecting ICT tool choices. Analysis revealed a significant positive relationship between gender, age and language use with smallholder farmers’ preference for using radio. Factors like farm size, off-farm income, and language positively influence the choice of basic cell phones. In contrast, educational level, marital status, and electricity supply negatively influence the choice to use radio and basic cell phones. Network connectivity and ICT awareness positively influence TV use, while household size and ICT costs have a negative effect. Educational level and ICT awareness positively influenced the use of computers and smartphones, whereas age, gender, off-farm income, electricity supply, farm size, household size and network connectivity had a negative influence. When smallholder farmers have access to multiple ICT tools, they can select the most beneficial combination for improving crop productivity. To maximise ICTs’ potential, policymakers should promote inclusive ICT access, awareness and training tailored to farmers’ needs, focusing on affordability, connectivity and literacy to support agricultural information dissemination.

1. Introduction

The agriculture sector is vital to South Africa’s economy, contributing around 15% to the GDP and supporting a large portion of the population’s livelihoods [1]. It plays a crucial role in economic development, reducing poverty and protecting natural resources [2,3]. However, agricultural development is hindered by information disparities, which can be addressed by meeting smallholder farmers’ information needs [4]. When dealing with farm-related problems to improve productivity, smallholder farmers rely heavily on relevant, timely and correct information as a major decision-making tool [5]. There are various categories of agricultural information, including climate or weather, agronomy, biotechnology, and chemical or organic inputs [6]. For decades, smallholder farmers have relied on interpersonal communication channels such as extension agents and local networks to access agricultural information [7,8]. Local networks include successful fellow farmers, neighbours, friends, social gatherings and relatives [9]. Efobi et al. [9] also indicated that the likelihood of smallholder farmers relying on informal and interpersonal contacts for agricultural information is likely due to these contacts being readily available, frequently interacting, and being cheap and easy to obtain. Shemfe & Modirwa [7] and Zodidi & Mudhara [10] also indicated that an increased number of agricultural information sources utilised by smallholder farmers is likely due to farmers’ tendency to not depend on a singular information source, which may originate from concerns regarding the accuracy and accountability of their information requirements. Although there are multiple sources of agricultural information, smallholder farmers still face challenges in obtaining and effectively utilising information to make informed decisions [11].

Challenges such as limited extension agents, poor infrastructure at the community level, and inadequate funding impede the effective dissemination of agricultural information [11]. To address these challenges, there is a need for alternative strategies, such as utilising communication technologies in agricultural extension services [12]. Agricultural ministries in African countries have made efforts to increase smallholder farmers’ access to agricultural information and services by incorporating ICT into agricultural information delivery [13,14]. Ologundudu and Eniola [15] explain that ICT refers to any device, tool, or application that facilitates the transfer and collection of data through interaction or transmission, thereby supporting communication. ICT includes both Internet-based and mobile technologies that operate on wireless networks, encompassing devices such as radios, televisions, landline phones, mobile phones, and computers. According to Sakthivel [16], ICT tools can be categorised into traditional ICT, such as radio and television, and modern ICT, such as telephones, computers and the Internet. The integration of ICT tools has become crucial for providing smallholder farmers with access to technical and market information, improving efficiency and productivity across agricultural value chains, particularly in developing African countries [17,18]. ICT tools have been shown to be an effective means of disseminating agricultural information to the farming community. The technique facilitates the sharing of knowledge among smallholder farmers, which increases crop yields by making information on cutting-edge techniques, sustainable practices, and methods for reducing input and output costs available to farmers [19].

In South Africa, the crop yields of smallholder farmers are being restricted by the current slow growth of the local economy, as well as the high costs of agricultural inputs, water scarcity, and limited access to agricultural information [20]. These challenges have limited the ability of smallholder farmers to implement sustainable intensification strategies in agriculture and enhance productivity, potentially worsening food insecurity [21,22]. Improving rural farmers’ income and productivity hinges on timely access to relevant and accurate information [20,23,24,25]. ICT tools emerged as more conventional pathways of disseminating agricultural information that help overcome communication gaps and are also considered a viable option for improving advisory services in the agricultural sector across Africa [19]. The South African government has promoted the use of ICT in agriculture, with initiatives like the 2011 Extension Recovery Plan (ERP) by the Department of Agriculture, Forestry and Fisheries (DAFF) aimed at enhancing extension services and agricultural information availability. Liebenberg [26] highlighted the ERP’s focus on providing ICT tools such as radios, laptops, and phones to improve agricultural productivity and access to information. Despite these efforts, numerous rural areas in South Africa continue to grapple with low ICT adoption and limited access to agricultural information, particularly the timely and relevant information required by smallholder farmers [8,27,28].

While many rural areas in South Africa still face challenges with ICT adoption, some smallholder farmers have begun using tools such as radio and mobile phones to access agricultural information, with mobile voice calls playing a significant role in facilitating market access and extension services [27]. However, the adoption of more advanced technologies like mobile apps and internet-based tools remains limited, particularly in rural areas, due to a range of barriers. These include low levels of digital literacy, high costs of data and airtime, unreliable network coverage, and limited electricity access [29]. Despite the broader national efforts to improve ICT access, in Mbombela, Mpumalanga Province, smallholder farmers show a growing awareness of the potential of ICT tools, yet actual usage remains low [30]. Research indicates that while mobile phone ownership is relatively widespread, these devices are primarily used for basic functions rather than accessing digital agricultural services [27]. Factors such as lack of training, limited exposure to ICT applications in agriculture, and scepticism about the relevance or trustworthiness of digital information contribute to low adoption rates [30,31]. Furthermore, the absence of structured support systems such as localised training programmes, extension services using ICT platforms, and community-based digital literacy initiatives continues to hinder the effective integration of digital tools into smallholder farming practices in the region [27,30].

This context is particularly critical in Mbombela, where agriculture plays a significant role in the livelihoods of rural households, yet access to timely and relevant agricultural information remains constrained [30]. According to Statistics South Africa [32], only about 24% of households in rural Mpumalanga have reliable internet access, and less than 15% of smallholder farmers utilise the internet or mobile apps for farming purposes. This highlights a significant digital divide in the region, particularly for those engaged in agriculture. ICT tools, when effectively adopted, can bridge the information gap by providing real-time weather forecasts, market trends, pest and disease alerts, and improved access to extension advice. These factors are often inaccessible through traditional means and are crucial for increasing productivity as well as climate resilience [18,33]. In the Mbombela region, where smallholder farmers face challenges like climate variability, market fluctuations, and high input costs, the strategic use of ICTs can be crucial for improved decision-making, increased productivity, and enhanced resilience. However, without specific measures to address digital inclusion, these farmers risk being left behind, potentially worsening existing inequalities and hindering the positive impact ICTs could have on rural development [30,31].

Research around the globe shows that socioeconomic, institutional, and psychological factors significantly impact farmers’ ability to access and use agricultural information available on ICT tools, which is crucial for improving livelihoods [34,35,36,37,38]. According to Mukasi [39], regardless of the acknowledged advantageous use of ICTs in agricultural development worldwide, research in this area remains limited in South Africa. Several studies have made attempts in South Africa to document the utilisation of ICT tools within the agricultural sector specifically in remote areas. Recent studies focused on the use of ICT tools for accessing agricultural information and possible challenges encountered by smallholder farmers [7,10,27,40,41,42,43]. However, these advancements in research underscore a significant knowledge gap in the literature, revealing a limited examination of how socioeconomic, institutional and psychological factors of smallholder farmers influence their choice to use different ICT tools for accessing agricultural information in South Africa. There is also a dearth of information, and no study has documented the contribution of these factors to farmers’ decisions to use ICT tools employing multivariate probit regression. Therefore, for the first time, this study attempts to comprehensively investigate the factors that influence smallholder farmers’ choice to use ICT tools for agricultural information in Mbombela Local Municipality, Mpumalanga, South Africa.

This study aims to fill this gap and contribute to the body of knowledge by focusing on the influence exerted by several factors on smallholder farmers’ choice to use ICT tools. This study departs from prior research conducted in South Africa in several aspects. Firstly, to the best of the researcher’s knowledge, this study distinguishes itself as a pioneer investigation in the country to comprehensively and empirically examine factors influencing smallholder farmers’ choice to use ICT tools for agricultural information access. This will guide the development, dissemination and implementation of agricultural information, as it will consist of ICT tools most accessible to farmers, important factors that influence the choice to use such tools, and mitigating strategies to reduce the effects of said factors. Additionally, it distinguishes between traditional and advanced ICT tools, providing nuanced insights into the differential drivers of adoption across ICT types. It offers context-specific policy recommendations that reflect both infrastructural and socio-economic barriers to ICT adoption, especially in rural agricultural settings. This objective-orientated study will show that academic findings can be translated into policy recommendations to improve strategies of sustainable agriculture.

Secondly, it surpasses preceding endeavours by conducting a multivariate probit analysis between the five main ICT tools (radio, TV, computer, basic cell phone and smartphone) commonly used by smallholder farmers and socioeconomic, institutional and psychological factors. The findings will inform agricultural extension and advisory services stakeholders, local government and policymakers to improve ICT adoption level. This could be achieved through the initiation of awareness programmes tailored to the information needs and contexts of smallholder farmers. Additionally, it could be achieved through the implementation of strategies that will ease the adoption as well as usage of tools such as the subsidy approach.

The main objective of this study was to assess the determinants shaping smallholder farmers’ choices of ICT tools. The specific objectives were to describe the socio-economic characteristics of smallholder farmers in the study area and to identify ICT tools available to smallholder farmers in the study area.

The rest of the paper is organised as follows: Section 2 discusses the relevant materials and methods (study area, sampling procedure and sample size, data collection procedure and analysis) used in the paper. Results are presented in Section 3, focusing on the socio-economic characteristics of the smallholder farmers, ICT tools available within the context of the smallholder farmers and determinants of the farmers’ preference for ICT tools. This is followed by the discussion of the results presented in Section 4. The conclusion and policy implications are presented in the final section of the paper, Section 5.

2. Materials and Methods

2.1. Description of the Study Area

The study was conducted in the Mbombela Local Municipality, located in the Ehlanzeni district of Mpumalanga Province, South Africa (Figure 1). The municipality spans 7139 square kilometres and is divided into four regions: Northern, Central, Southern, and Eastern [44]. It has a subtropical climate with mild winters, summer rainfalls, and an average annual rainfall of 300 mm–500 mm. Key crops grown in the area include macadamia, vegetables, citrus, grains, avocados, and nuts, which are important for both household consumption and income generation [45]. The population of Mbombela is 695,913, with a majority residing in peri-urban and rural areas [46]. Within the Ehlanzeni district, Mbombela Local Municipality is considered the top crop-producing municipality [47]. The agricultural sector in the municipality has a potential for growth through new farming techniques addressing land limitations and weather challenges [44]. In the municipality there is limited and outdated infrastructure that needs upgrading, which hampers ICT access, especially in rural areas [48]. Despite these challenges, there is a move towards greater ICT integration in agriculture, as smallholder farmers in the province leverage a variety of ICT tools to improve their knowledge and farming practices. Smallholder farmers use radio, TV, and mobile phones, alongside mobile applications like WhatsApp, Twitter, Facebook, and Instagram, to enhance their knowledge and practices. Agricultural extension workers also use ICT to directly share valuable information with smallholder farmers [30,31,49].

2.2. Research Design

The study was conducted using a quantitative research approach and cross-sectional survey design, which is commonly used to investigate a broad range of social issues and challenges [50]. A quantitative design is employed to collect and analyse the data in numerical form. It can help identify trends and averages, predict future outcomes, test causal relationships, and generalise findings to a larger population [51]. A cross-sectional survey involves collecting data from a population or a representative subset at a single point in time, allowing researchers to analyse the prevalence and relationships of specific variables within that timeframe. This design is particularly useful for exploring associations between different factors without requiring long-term follow-up [50]. In the context of this study, the cross-sectional survey enabled the researcher to gather relevant data from farmers regarding their ICT usage for agricultural information at one point in time. Together, the quantitative approach and cross-sectional design are well-suited for this research, as they allow for the systematic examination of various influencing factors on smallholder farmers’ decisions to adopt ICT tools. This approach supports a comprehensive understanding of patterns and relationships that can inform policy and development strategies aimed at enhancing ICT adoption in agriculture.

2.3. Target Population

The target population of this study consisted of smallholder crop farmers within the Mbombela Local Municipality, specifically those from rural communities in the area. This group included both smallholder farmers who use ICT tools such as radio, TV, computers, and mobile phones for accessing agricultural information, as well as those who do not. This particular group was registered with the Department of Agriculture, Rural Development, Land and Environmental Affairs (DARDLEA) by the time the study was conducted. The study used the list of registered smallholder farmers as a sampling frame to select study participants.

2.4. Sampling Technique

A systematic sampling technique was employed to select units from the targeted population. Systematic sampling is a probability sampling method where study units are chosen from a larger population starting from a randomly selected point, with a consistent, periodic interval between selections. This interval, known as the sampling interval, is calculated by dividing the population size by the desired sample size [51,52]. The study used the systematic sampling technique to select 308 smallholder crop farmers from the sample frame. This method involved randomly selecting a starting point, referred to as the first subject. For this study, the number two was chosen as the starting point [52]. To determine the sample interval, the study applied the recommended formula, which resulted in 1343/308 = 4. Thus, four was used as the constant difference between consecutive numbers in the sequence. Sampling occurred at an interval of four, meaning that after counting three names of smallholder farmers in the sample frame, starting from the selected point, the next individual was included in the sample [52]. This technique was employed to ensure that everyone in the target area had an equal chance of being selected and to prevent biases related to factors such as gender and distance.

2.5. Sampling Size

Using Slovin’s formula, the sampling units were determined. In this study, there was a 95% likelihood that the sample results represent the true characteristics of the population within a specified accuracy range, leaving only a 5% chance that they do not. The total number of 308 units was required to be sampled in this study. The formula used to produce this total was generated using the formula illustrated below.

$$\mathit{n}={\displaystyle \frac{\mathit{N}}{\left(\mathbf{1}+\mathit{N}{\mathit{e}}^{\mathbf{2}}\right)}}$$

$$\mathit{n}={\displaystyle \frac{\mathbf{1343}}{\left(\mathbf{1}\mathbf{+}\mathbf{1343}{\left(\mathbf{0.05}\right)}^{\mathbf{2}}\right)}}$$

• n = 308;
• n = sample size; N = total population size; e = constant margin of error (0.05)

2.6. Data Collection Procedure

A structured questionnaire was used to gather data for the study. With a structured questionnaire, the questions are prepared in advance, ensuring that all respondents are asked the same ones in the same order. The questionnaire consists of questions with predefined answers, which may be either closed or prompted. One advantage of using structured questionnaires is that they require less cognitive effort from respondents, as they minimise the need for critical thinking to answer the questions. This often leads to higher response rates and more accurate data. Therefore, the researcher finds it easier to code and analyse the responses [53]. During the survey, the researcher and trained enumerators personally administered the questionnaire to respondents and remained present throughout the process to address any questions or concerns that arose.

2.7. Data Analysis

Descriptive statistics were employed to summarise and interpret the data collected in the study, using the Software for Statistics and Data Science version 18.0 (STATA 18.0). Descriptive statistics involve the use of numerical measures such as means, standard deviations, frequencies, and percentages to present a clear and concise summary of data. These methods provide an overview of patterns, trends, and variations within a dataset, without making inferences beyond the data at hand [51]. In this study, descriptive statistics were specifically applied to analyse the distribution of smallholder farmers’ socio-economic characteristics and the ICT tools available in the study area. The mean scores were used to represent the central tendencies of responses on Likert-scale items, providing insight into the average perception or experience of respondents. Frequencies and percentages were applied to summarise variables, such as types of ICT tools used and demographic information. This approach facilitated a comprehensive understanding of the respondents’ profiles and the prevailing ICT conditions in the study area.

The study used a multivariate probit regression model to analyse the factors influencing smallholder farmers’ choice to use ICT tools. The multivariate probit model is a statistical technique used to estimate the probability of multiple, potentially correlated binary outcomes [54]. To model ICT tool choice, a univariate binary model (logit or probit) could be suitable due to the dichotomous nature of the dependent variable [55]. However, using separate binary equations for each ICT tool may cause bias, as it ignores the correlation of error terms. To address this, the study employed the Multivariate Probit (MVP) model, which simultaneously regresses multiple correlated binary equations against a single set of explanatory variables. This approach avoids statistical bias and inefficiency, allowing for a deeper understanding of ICT tool selection and their interrelationships [55]. In the context of this study, smallholder farmers may adopt multiple ICT tools (radio, mobile phones, television, computer) simultaneously, and these adoption decisions are likely interdependent. The MVP model is therefore appropriate, as it captures the correlation between the unobserved factors influencing the adoption of each ICT tool, providing more efficient and unbiased estimates than separate univariate models. This approach enhances the robustness of the analysis and aligns with the study’s objective of understanding the factors influencing multiple ICT adoption decisions among smallholder farmers. Empirically, the model can be expressed in the following equations:

$$Yi=Xi\mathrm{ᵦ}+ϵi$$

where Yi is the latent (unobserved) variables representing the propensity to use different ICT tools (radio, TV, basic cell phone, computer, smartphone). Xi is a vector of farmers’ characteristics (independent variables). These include gender, age, educational level, marital status, household size, years of farming experience, farm size, cooperative membership, off-farm income, network connectivity, electricity supply, cost of ICT, language use, ICT literacy and ICT awareness. β1, β2, β3, β4, β5 are vectors of coefficients to be estimated for each ICT tool. These coefficients indicate how each farmers’ characteristic affects the likelihood of using a specific ICT tool. ϵi1, ϵi2, ϵi3, ϵi4, ϵi5 are vectors of error terms for each ICT tool’s latent variable. These are assumed to follow a multivariate normal distribution with mean 0 and covariance matrix Σ, where the covariance structure allows for correlation between the errors. This model accounts for the fact that the choice to use one ICT tool might influence or be correlated with the choice to use another. The independent variables were derived from a review of past studies on the usage of ICT tools [37,56,57,58,59]. The dependent variable is a dummy variable, which takes a value of 1 for the choice of radio, TV, computer, basic cell phone and smartphone, and 0 otherwise.

Table 1. Characteristics of hypothesised dependent and independent variables.

Variables	Unit	Description	Expected Sign
Dependent variable
ICT tools choice	Binary	1 if a farmer chooses to use ICT tools, 0 otherwise
Independent variable
Gender	Categorical	1 if a male, 0 otherwise	+/−
Age	Continuous	Number of years of participants	−
Marital status	Categorical	1 if married, 0 otherwise	+/−
Educational level	Continuous	Number of years spent in formal school	+
Farming experience	Continuous	Number of years practicing farming	+/−
Farm size	Continuous	Total area under crop production	+/−
Household size	Continuous	Number of people in a household	+/−
Cooperative membership	Binary	1 if a member of the cooperative, 0 otherwise	+
Source of income	Categorical	Sources of income of participants either on-farm or off-farm	+
Network connectivity	Binary	1 indicates stable network connectivity, while 0 otherwise	+
Electricity supply	Binary	1 if the participant has a proper electricity supply, 0 otherwise	+
Cost of ICT tools	Binary	1 if the cost of ICT is considered affordable, 0 otherwise	+/−
Language use	Binary	1 if able to comprehend the language used on ICT tools, 0 otherwise	+/−
ICT literacy	Binary	1 if the participant is ICT literate, 0 otherwise	+
ICT awareness	Binary	1 if the participant is aware of ICT tools, 0 otherwise	+

describes the characteristics of hypothesised dependent and independent variables in the usage of ICT tools.

2.8. Ethical Consideration

The study attained ethical clearance from the University of Mpumalanga Ethics Committee, under the protocol reference number: UMP/Ntsoane/219040508/MAGR/2024. It adhered to all essential ethical guidelines required for conducting research and fully complied with the ethical standards outlined in the University of Mpumalanga’s code of ethics. The study was mindful of the diverse expectations and experiences of the respondents during data collection. Confidentiality and the well-being of the respondents were prioritised throughout the process. Anonymity of the participants was maintained by ensuring that no personally identifiable information was disclosed. Participants were fully informed of the research protocols and procedures prior to their involvement. Participants were made aware of the purpose of the study, the associated risks and their rights as participants. Furthermore, participants were assured that all responses would be used solely for research purposes and would not be used against them in any other context.

3. Results

3.1. Socioeconomic Characteristics of the Respondents

Table 2. Frequency distribution of respondents according to their demographic characteristics. This table presents data on the socioeconomic characteristics of respondents, focusing on their gender, age, education level, marital status, farming experience, farm size, household size, cooperative membership, and source of off-farm income.

Socioeconomic Characteristics	Frequency	Percentage
Gender
Male	113	36.7%
Female	195	63.3%
Age
18–35	65	21.1%
36–55	111	36%
55 and above	132	42.9%
Educational level
No formal education	87	28.2%
Primary	51	16.6%
Secondary	74	24%
Matriculated	50	16.2%
ABET	26	8.4%
Higher certificate and above	20	6.5%
Marital status
Single	99	32.1%
Married	125	40.6%
Divorced	26	8.4%
Widowed	58	18.8%
Farming Experience
≤5 years	101	32.8%
6–10 years	113	36.7%
11–20 years	60	19.5%
≥21 years	34	11.0%
Farm size
≤5 ha	186	60.4%
6–10 ha	117	38.0%
≥11	5	1.6%
Household size
≤5 ha	92	29.9%
6–10	144	46.8%
≥11	72	23.4
Cooperative membership
Yes	227	73.7%
No	81	26.3%
Source of off-farm income
Social relief of distress	48	15.6%
State old-age pension	55	17.9%
Child support grant	34	11%
Salary	7	2.3%
None	163	53%

offers an in-depth review of the socioeconomic characteristics of respondents engaged in farming, with a focus on understanding their potential to access and use ICT tools in agriculture. The study includes the frequencies and percentages for each variable. From the total sample, 113 respondents are male, representing 36.7%, while a larger proportion of 195 respondents (63.3%) are female. This gender imbalance indicates that women constitute a large portion of the sample, which could reflect their increased involvement in farming. In terms of age distribution, the sample comprises 65 individuals aged 18–35, accounting for 21.1% of the respondents, 111 individuals aged 36–55 (36.0%) and 132 respondents aged 55 and above, representing the largest age category at 42.9%. The distribution indicates that an older population, with 42.9% of participants aged 50 or older, dominates farming. The educational level of respondents revealed that 87 individuals (28.2%) have no formal education, which is the largest group within this category. This cohort is followed by 74 respondents (24.0%) with secondary education, 51 (16.6%) with primary education, and 50 (16.2%) who have matriculated. Additionally, 26 respondents (8.4%) have attended ABET (Adult Basic Education and Training) programmes, and only 20 (6.5%) possess a higher certificate or above. The frequency of respondents without formal education or only primary schooling together accounts for 44.8%, whereas a higher frequency of respondents to have received some form of formal education accounts for 55.2% of the respondents. When examining marital status, 125 respondents (40.6%) are married, reflecting the largest group, while 99 (32.1%) are single, 58 (18.8%) are widowed, and 26 (8.4%) are divorced. The relatively high frequency and percentage of married individuals further reflect the older demographic makeup of the sample.

The results further revealed that the highest proportion of respondents, 113 individuals (36.7%), have between 6 and 10 years of experience, closely followed by 101 (32.8%) with 5 years or less. This indicates that a significant portion (69.5%) are relatively new or mid-term participants in farming. Fewer participants have 11–20 years (19.5%) or 21 years and above (11.0%) of experience, suggesting that long-term farming experience is less common among this group of farmers. Farm size analysis indicated that 186 respondents (60.4%) were farming on 5 hectares or less. Another 117 (38.0%) operate on 6–10 hectares, while only 5 respondents (1.6%) manage farms of 11 hectares or more. This pattern reflects the predominance of small-scale farming in the sample and may indicate limited production capacity. Regarding household size, 144 respondents (46.8%) come from households with 6–10 members. A total of 92 individuals (29.9%) report living in households of 5 or fewer, while 72 respondents (23.4%) belong to larger households of 11 or more members. These figures indicate that the majority (70.2%) of households are relatively large, which could suggest increased labour availability, higher household consumption demands and more pathways for information access. Cooperative membership is notably high, with 227 respondents (73.7%) being members of cooperatives, while only 81 (26.3%) are not. This implies a strong presence of organised group participation, which may help facilitate access to extension services. Lastly, the source of off-farm income shows a heavy reliance on social support. In total, 163 respondents (53.0%) report having no source of off-farm income, highlighting more dependence on farming. Among those who do, the most common sources are the state’s old-age pension (17.9%), social relief of distress (15.6%), and child support grants (11.0%). Only seven respondents (2.3%) earn a salary, suggesting that formal employment is rare within this population.

3.2. ICT Tools Available to the Respondents

Table 3. Distribution of respondents according to available ICT tools. The table below presents the frequency of use for various ICT tools as reported by respondents. The frequencies are categorised into three distinct usage patterns: frequently, occasionally, and not at all. Each ICT tool is also assigned a mean score, reflecting its average usage frequency, and ranked accordingly.

	Frequency of Using ICT Tools			Ratings
ICT Tools	Frequently	Occasionally	Not at All	Mean Score	Ranking
Radio	15(40.5)	22(59.5)	0(0.0)	1.41	3rd
TV	1(25.0)	3(75.0)	0(0.0)	1.25	4th
Smartphone	89(67.4)	43(32.6)	0(0.0)	1.67	1st
Basic cell phone	8(44.4)	10(55.6)	0(0.0)	1.44	2nd
Computer	0(0.0)	0(0.0)	0(0.0)	N/A	N/A

presents the frequency and intensity of ICT tool usage among 191 respondents who reported using such tools to access agricultural information from a total population of 308. The ICT tools evaluated include radio, television (TV), smartphone, basic cellphone, and computer. Respondents were asked to rate their use of each tool as frequent, occasional, or not at all, and mean scores were calculated to determine the overall usage intensity and ranking. The results reveal that the smartphone is the most widely and intensively used ICT tool, with 89 respondents (67.4%) reporting frequent use and 43 (32.6%) using it occasionally. No respondents reported not using smartphones at all, and the mean score of 1.67 confirms that smartphones are the top-ranked ICT tool (first) for accessing agricultural information. This high usage suggests strong penetration of smartphones in the farming community, likely due to their multifunctionality, portability and accessibility to a wide range of agricultural apps, weather updates, market prices and extension services.

The basic cell phone ranks second, with a mean score of 1.44. It is frequently used by 8 respondents (44.4%) and occasionally by 10 (55.6%), with again no one reporting non-use. Though less advanced than smartphones, basic cell phones still serve as important tools for communication, voice-based agricultural advice, and SMS updates. The radio is ranked third, with 15 respondents (40.5%) using it frequently and 22 (59.5%) using it occasionally. No respondents indicated that they do not use it at all. The mean score of 1.41 reflects steady but slightly less intensive use compared to phones. The radio remains a valuable, low-cost, and easily accessible medium for disseminating agricultural extension programmes, weather information, and government announcements, especially in rural areas where literacy may be a barrier to text-based information. In contrast, the television (TV) is used frequently by only one respondent (25.0%) and occasionally by three (75.0%), with a mean score of just 1.25, ranking it fourth. This indicates very limited use of TV as a source of agricultural information, potentially due to less flexibility in programming, fewer specialised agricultural broadcasts and lack of electricity access in rural households. The computer is not used at all by any respondent, as all usage ratings are zero. This result reflects a complete lack of computer use for accessing agricultural information among this group of farmers. In conclusion, the interpretation of this table reveals a clear hierarchy of ICT tools based on actual use among farmers. Smartphones are the most frequently and widely used, followed by basic cell phones and radio, while TV use is minimal and computers are entirely unused.

3.3. Sources of Agricultural Information Used by Respondents

Table 4. Distribution of respondents according to different sources of agricultural information in the study area. This table outlines the sources of information used by respondents, categorising them into two primary groups: personal locality sources and cosmopolite sources. The table includes the frequency as the number of respondents, the percentage of respondents who use each source, the mean score (indicating the typical frequency of use), and the ranking of each source based on its mean score. The data offers conclusions about the relative importance and usage patterns of various information sources among the respondents.

Sources of Information	Frequency (No of Respondents)	Percentage	Mean Score	Ranking
Personal locality sources
Farmer to Farmer	32	27.35%	0.55	2nd
Friends	0	0%	0	4th
Cosmopolite Sources
Public extension worker	71	60.68%	1.21	1st
Community-Based Organisations (CBOs)	14	11.97%	0.24	3rd

provides insights into the sources of agricultural information used by 117 respondents who do not use ICT tools. Instead, these individuals rely on traditional, non-digital channels, categorised as either personal locality sources (interactions within the local community) or cosmopolite sources (external or more formal information providers). The table shows the frequency of respondents using each source, their percentage share, and a mean score representing how intensively each source is used, which contributes to their ranking. Fellow farmer is the most frequently used local source, with 32 respondents (27.35%) indicating that they obtain agricultural information from other farmers in their community. It has a mean score of 0.55, making it the second-ranked source. This suggests that peer-to-peer learning and knowledge exchange remain important, especially in contexts where formal information is limited or digital access is absent. Zero respondents (0%) reported receiving agricultural information from friends, giving this source a mean score of zero and placing it fourth (last) in the ranking. This indicates that while friends may be present in respondents’ social networks, they are not perceived as a reliable or relevant source for farming-related guidance. Public extension is the most relied-upon source among non-ICT users, with 71 respondents (60.68%) identifying extension officers as their key information providers. The mean score of 1.21 places this source first in the ranking, indicating a high intensity and frequency of use. This highlights the central role of public agricultural extension services in bridging the information gap for farmers without access to digital tools. Extension officers likely serve as trusted, trained intermediaries offering practical advice, training, and updates on government support or innovations. Community-Based Organizations (CBOs) are used by 14 respondents (11.97%). CBOs are a less common but still present source of information. Their mean score of 0.24 positions them third in the ranking. While CBOs may not be as widely used as extension services, they may still play a role in facilitating access to agricultural information through collective structures. This data reveals that among the 117 non-ICT-using respondents, the majority rely on human-based, direct-contact sources for agricultural information. The most prominent source is the public extension service (60.68%), showing the continued importance of government or institutional support where digital channels are inaccessible. Farmer-to-Farmers is the second-most used source (27.35%), emphasising the value of informal, experience-based knowledge-sharing in rural communities. CBOs provide limited support, and friends are not used at all for farming information.

3.4. Determinants Shaping Smallholder Farmers’ Choice of ICT Tools

The multivariate probit regression model results focus on the relationship between various socio-demographic and institutional factors and the likelihood of using different ICT tools (radio, TV, computer, basic cell phone, and smartphone) among your sample of 308 respondents (

Table 5. Multivariate probit results for determinants influencing farmers’ choice among various ICT tools.

Variables	ICT Tools
	Radio	TV	Computer	Basic Cell Phone	Smartphone
	Coefficient (S.E)	Coefficient (S.E)	Coefficient (S.E)	Coefficient (S.E)	Coefficient (S.E)
Gender	0.350 (0.207) *	0.240 (0.342)	−0.545 (0.217) ***	−0.115 (0.171)	−0.156 (0.210)
Age	0.511 (0.157) ***	−0.202 (0.265)	−0.530 (0.220) **	0.124 (0.136)	−0.872 (0.168) ***
Marital status	−0.133 (0.105)	−0.264 (0.229)	0.178 (0.154)	−0.124 (0.096)	0.060 (0.119)
Educational level	−0.343 (0.072) ***	−0.102 (0.118)	0.182 (0.070) ***	−0.185 (0.059) ***	0.064 (0.067)
Farming experience	0.117 (0.114)	−0.011 (0.222)	−0.099 (0.161)	−0.111 (0.102)	−0.040 (0.140)
Farm size	−0.057 (0.177)	0.138 (0.282)	−0.106 (0.210)	0.380 (0.152) ***	−0.527 (0.225) **
Household size	−0.011 (0.132)	−0.439 (0.270) *	−0.131 (0.204)	−0.051 (0.113)	−0.156 (0.153)
Cooperative membership	−0.195 (0.236)	0.037 (0.398)	−0.129 (0.298)	−0.208 (0.197)	−0.319 (0.256)
Off-farm income	0.047 (0.028) *	0.025 (0.044)	−0.058 (0.034) *	0.053 (0.024) **	−0.003 (0.030)
Network connectivity	0.768 (0.559)	2.389 (0.680) ***	−3.150(95.380)	0.264 (0.536)	−0.840 (0.816)
Electricity supply	−0.219 (0.354)	−0.737 (0.528)	−0.947 (0.568) *	−0.144 (0.306)	0.340 (0.448)
Cost of ICT tools	−4.736 (190.409)	−6.931 (143.252)	−1.531 (99.802)	−0.284 (0.924)	−0.913 (0.936)
Language use	1.738 (0.331) ***	−0.350 (0.559)	−0.115 (0.552)	1.192 (0.261) ***	0.242 (0.329)
ICT literacy	0.326 (0.262)	0.309 (0.452)	−0.384 (0.380)	0.213 (0.229)	0.580 (0.289) **
ICT awareness	−0.118 (0.215)	1.056 (0.398) ***	0.907 (0.279) ***	−0.292 (0.184) *	1.361 (0.238) ***
Constant	−0.297 ( 190.412)	−3.881 (143.261)	1.324 (138.057)	−1.572 ( 1.311)	1.051 (1.618)
Log Likelihood	−474.246
No of Obs	308
Wald chi2 (75)	307.52
Prob > chi2	0.000

Notes: ***, ** and * are significant at 1%, 5% and 10% levels, respectively. Values in parentheses are standard errors. Bold values are significant variables.

). The multivariate probit model jointly estimates the likelihood of using five ICT tools, considering that these choices may be correlated. The model is statistically significant, as indicated by the Wald chi²(75) = 307.52, p < 0.001, showing that the predictors collectively influence ICT tool choice. A log-likelihood of −474.25 suggests the model fits reasonably given the complexity of the data.

The results revealed that several socioeconomic and demographic factors significantly influence smallholder farmers’ choice to use ICT tools for accessing agricultural information. These include gender, age, marital status, educational level, farming experience, farm size, household size, cooperative membership, off-farm income, network connectivity, electricity supply, cost of ICT tools, language use, ICT literacy and ICT awareness.

Among these, age was found to positively influence the use of radio (p < 0.01) but negatively influence the adoption of computers (p < 0.05) and smartphones (p < 0.01). This reflects a generational divide in ICT use, where traditional platforms remain central for older farmers, yet younger farmers are more engaged with modern digital tools. Gender effects also emerged, with males more likely to use radio (p < 0.1) and less likely to use computers (p < 0.01). This finding highlights the persistence of gender gaps in digital literacy and access and also suggests that men continue to favour more conventional channels, while barriers remain in their uptake of advanced ICTs. In contrast, higher levels of education reduce the likelihood of using radio (p < 0.01) and basic cell phones (p < 0.01) and increase the likelihood of computer use (p < 0.01). This demonstrates that education facilitates the transition from basic to advanced ICTs, with more educated farmers favouring complex devices that can process and deliver a wider range of information.

Results further found that larger farm holders are more likely to use basic cell phones (p < 0.01) and less likely to adopt smartphones (p < 0.05). This suggests that farm scale may shape ICT preferences, with larger producers gravitating towards simpler communication tools, while smaller producers exhibit greater uptake of smartphones. Household size negatively influences the choice to use TV (p < 0.1). This may indicate that larger households face competing demands on resources, influencing access to certain ICTs such as television. Off-farm income positively influences the choice to use radio (p < 0.1) and basic cell phones (p < 0.0 but negatively influences computer use (p < 0.1). These patterns suggest that while supplementary income supports reliance on low-cost ICTs, it does not necessarily translate into greater investment in advanced technologies.

Additionally, network connectivity positively influences TV use (p < 0.01), highlighting the role of infrastructure availability in shaping ICT access. In rural contexts, where connectivity is often unreliable, better network access may enhance the appeal of television not only for entertainment but also as a trusted medium for agricultural information. Electricity supply negatively influences computer use (p < 0.1). This indicates that access to electricity alone does not automatically translate into greater computer use; rather, it interacts with social, economic, and skill-based factors that determine whether farmers can effectively integrate such devices into their information-seeking practices. Language use positively influences the use of radio (p < 0.01) and basic cell phones (p < 0.01). This underscores the accessibility of these platforms for farmers who communicate primarily in local languages, which may limit engagement with more text-intensive and English-dominated ICTs. ICT literacy positively influences smartphone use (p < 0.05). This finding suggests that farmers who possess the skills and confidence to operate digital tools are more inclined to adopt smartphones as information and communication devices. As they involve navigating applications, internet use and interactive features. ICT awareness positively influences the use of TV, computers, and smartphones (all p < 0.01), although it negatively influences basic cell phone use (p < 0.1). This reflects the role of awareness in shaping farmers’ perceptions of what technologies are useful and beneficial for accessing agricultural information. Awareness expands farmers’ knowledge of the opportunities that advanced ICTs offer, encouraging a shift away from basic devices toward tools with greater capacity for delivering diverse and detailed information. In this way, awareness functions as a cognitive driver of ICT adoption, influencing not only whether farmers use ICTs but also which technologies they prioritize.

4. Discussion

4.1. Smallholder Farmers’ Socio-Economic Characteristics

The results on gender highlight that women make up a larger part of the farming group, suggesting that more women are involved in farming activities and decision-making related to farming in this particular area. These findings concur with those of Agholor et al. [30] and Lubisi & Agholor [60], who discovered that in Mpumalanga, South Africa, smallholder farmers are more likely to be female (74.7%) (55.4%) than their male counterparts. The study of Janavi et al. [61] indicates that gender significantly influences innovation adoption, as men tend to use social media more frequently than women. The difference may be attributed to the generally lower engagement of women with technology. Results show that most of the farmers are older, with nearly 43% being 55 or older. This finding could suggest that younger people are less involved in farming, while older adults do most of the farming work. This finding is consistent with those of Agholor et al. [30] and Shemfe & Modirwa [7], who determined that elderly people were more engaged in farming than their younger counterparts. However, Zondo & Ndoro [49] reported that smallholder farmers aged 20–39 are the primary participants in agricultural activities and are also able to utilise social media as a component of an ICT tool. Moreover, the results indicate that the majority of smallholder farmers surveyed have achieved a certain level of formal education, with most having completed at least high school. This conclusion is in line with previous research showing that a higher percentage of smallholder farmers have completed secondary education [49,60]. Muhammad et al. [57] corroborated these results and found that among farmers surveyed, 55.0% had completed primary school, 12.5% had completed secondary school, 7.5% had completed university education, and 25.0% had not completed any formal education at all. This suggests that a large percentage of smallholder farmers in the area are literate, which supports their ability to use ICT to improve their farming practices. A well-educated farmer will find it easier to navigate and make use of the new information and technology available through ICT.

Furthermore, the findings indicate that married individuals, rather than other categories, primarily dominate the agricultural activities in the region. The results are also consistent with the findings of Shemfe & Modirwa [7], who found that the vast majority of smallholder farmers in the Northwest were married, rather than those who were unmarried and divorced. Tambo et al. [62] found that married couples tend to adopt ICT more than single males and females, possibly due to their shared decision-making process. Similarly, other studies indicate that married farmers use information and communication technologies more than those who are single, divorced, or widowed [11,57,63]. It was also noted that the majority of smallholder farmers are still relatively new in the sector, with nearly 70% having 10 years or less of experience. Mishra et al. [64] noted that smallholder farmers with extensive farming experience tend to be more progressive, adaptable, and skilled at evaluating new technology, which helps them make more efficient decisions. Mishra et al. [64] also emphasized that these farmers are keen to improve agricultural practices, enabling them to learn more about applying ICT in farming. Ghosh et al. [65] argued that more farming experience leads to lower ICT technology use. This may be due to the fact that experienced farmers feel more confident in their existing knowledge, while beginners in farming who require additional information tend to use ICT to acquire it. It also shows that most farmers work on small pieces of land. The results align with Zondo and Ndoro [49], who highlighted that a significant proportion of smallholder farmers possessed land measuring 5 hectares or less. The amount of land a farmer cultivates determines and shapes the usage of ICTs in seeking information from various sources [40,66].

Additionally, the majority of the farming households are medium-sized (6–10 people), and only a small number of households are very large (11+ people). In relation to this study, Tijjan et al. [67] discovered that the household distribution is broad, which in turn provides additional labour for farming. Furthermore, the large size of a household may facilitate the acquisition of additional agricultural information through a variety of ICT tools. Dyanty et al. [68] align with this finding by indicating that Mbombela smallholder farmers’ households vary between 6 and 10 members, implying substantial availability of family labour for agricultural activities. Leng et al. [69] and Muhammad et al. [57] corroborated this by demonstrating that households with more family members may use various ICT tools available inside the home, hence promoting the adoption and utilisation of ICT in their agricultural practices. Results further show that most smallholder farmers are part of a cooperative. The results are corroborated by research by Efobi et al. [9], which indicated that in their study region, the majority of respondents (89.2%) were affiliated with various cooperative organisations, whereas 10.8% were not members of any cooperative society. This suggests that the substantial membership of cooperatives may provide a means to obtain agricultural information and insights regarding the advantages of utilising ICT tools. The results show that over half of farmers rely solely on farming income, while others receive government aid or social grants. The findings align with Feju et al. [40], who contended that the primary objective of most farmers is to cultivate in order to sustain their livelihoods. As demonstrated by Tamako et al. [70], farmers rely on a variety of economic streams to keep their families afloat. They further showed that pension payouts are the primary source of income for 57.1% of the smallholder farmers surveyed, followed by 26.9% of smallholder farmers relying on government social grants to support their families. Most smallholder farmers were able to receive these subsidies, as their monthly household incomes were between R1500 and R3500 [71].

4.2. ICT Tools Available to Smallholder Farmers

The study’s results reveal that the most commonly used ICT tools among the respondents are mobile-based devices (smartphones and basic cell phones), followed by traditional media (radio and TV). The results align with the findings, emphasising that 85% of South Africans possess a mobile phone, highlighting its significance for agricultural businesses, particularly in communication as well as advertising [72,73]. Muhammad et al. [57] affirm that mobile phones and radios are the predominant ICT tools utilised by farmers, due to their accessibility, portability and affordability. The findings align with those of Levi et al. [11], who discovered that in certain communities, television surpassed radio in popularity due to its audio-visual characteristics. However, farmers still underutilise it for acquiring essential agricultural information, primarily viewing it as a medium for entertainment. Luqman et al. [74] corroborate the findings about the frequency and scope of ICT tool utilisation in agriculture, indicating that respondents use ICT for less than 2 h daily. They expressed in an informal talk that they are information searchers who utilise ICTs to obtain the necessary information. Contrary to smartphone being the most used, Feju et al. [40] found that minimal application usage was observed among farmers who accessed mobile-based devices such as smartphones. Out of the total number of those involved, only 22.2% of farmers were found to employ mobile applications in their agricultural operations. WhatsApp (11.2%), Facebook (8.1%), LinkedIn (1.2%), YouTube (1.1%), mobile finance (1.1%), and GPS (1.1%) are all utilised by farmers. The Khula app (2.3%), GPS treble (1.1%), and Tillo app (1.2%) were also mentioned by the participants as agricultural mobile applications. Other studies indicated that farmers’ utilisation of mobile-based applications in agricultural operations was minimal. This is due to farmers possessing knowledge of a specific extant agriculture mobile-based application; however, the application is not critical to their farming system [75,76,77].

The finding that none of the farmers surveyed in Mbombela Local Municipality, Mpumalanga, use computers to access agricultural information aligns with broader research across South Africa and the African continent. Although existing literature often includes computers as potential ICT tools for smallholder farmers, actual usage in rural settings remains extremely limited [30,78,79]. Studies conducted in South Africa have shown that smallholder farmers predominantly rely on low-tech tools such as mobile phones, SMS, radio, and printed materials, largely due to challenges such as poor digital literacy, limited infrastructure, and the high cost of computers [30,41,49]. In Mpumalanga specifically, these constraints are evident, making the use of computers for agricultural information access unlikely [30,31]. Similarly, research across sub-Saharan Africa confirms that computer ownership and usage among rural farmers are minimal, with most ICT interventions and real-world practices focused on mobile-based solutions [80]. Therefore, while computers were initially included in this study’s theoretical framework based on existing literature, the study’s findings reflect the practical realities of the area, which are consistent with regional and continental trends in digital agriculture.

4.3. Sources of Agricultural Information Accessible to Smallholder Farmers

The findings show that formal agricultural information sources, such as public extension agents, are the most utilised among the respondents. This is followed by informal sources, including Farmer-to-Farmers, whereas friends are much less relied upon. In contrary, Ogidi [81] discovered that the majority of agricultural knowledge in rural farming communities is disseminated through interpersonal routes such as friends, relatives, and neighbours. Research conducted across varied African countries found that farmers sought agricultural information from a variety of sources. These included government agriculture extension agents, social gatherings, relatives, other successful farmers, primary agricultural cooperative banks (PACBs), groups representing farming associations, meetings with village leaders, representatives from nongovernmental organisations (NGOs), and agricultural exhibitions [82,83,84]. In support of this view, Efobi et al. [9] showed that smallholder farmers rely on informal and interpersonal contacts for agricultural information. This is likely due to these contacts being readily available, interacting frequently, and being cheap and easy to obtain. An increase in the number of information sources utilized by smallholder farmers correlates with a decreased likelihood of employing ICT, likely due to the respondents’ tendency to not depend on a singular information source, which may originate from concerns regarding the accuracy and accountability of their information requirements [7].

4.4. Factors That Influence Smallholder Farmers’ Choice of ICT Tools for Agricultural Information Access

The results of the multivariate probit regression indicate that gender significantly influences the choice of ICT tools among smallholder farmers. Male farmers are more likely to use radio than female farmers. Conversely, gender has a negative effect on computer use, with males less likely to use computers than females. According to Mdoda and Mdiya [85], the gender disparity in ICT tool usage appears linked to unequal access to resources, lower digital literacy among women, and entrenched cultural norms. These findings align with Wawire et al. [86], who emphasise that men have more access to agricultural information than women due to cultural practices that allocate most of the domestic responsibilities to women, leaving them with almost no extra time to allow them to pursue additional services related to farming.

The age of smallholder farmers was found to positively influence radio usage at the 1% significance level, suggesting that older individuals are more likely to use a radio. Additionally, age negatively affects the use of TV and smartphones. The results indicate that older individuals are less likely to choose to use these tools. Age significantly reduces the likelihood of choosing smartphones and TV at the 1% significance level. Mishra et al. [65] corroborate these findings, indicating that for each additional year of age among respondents, the likelihood of utilising ICT diminishes. This may be attributed to the decline in learning behaviour among farmers as they age; they tend to anticipate risk and choose simplicity. Some studies contend that the uptake of ICT is not directly influenced by age. However, it further concluded that the most likely innovators and early adopters of ICT are younger smallholder farmers [25].

The study found that educational level negatively influences the use of radio and basic cell phones at the 1% significance level, respectively. Conversely, higher education is positively associated with computer usage at the 1% significance level, which implies that individuals with higher education are less likely to use a radio and basic cell phone as they prefer modern forms of media. The findings align with Shemfe & Modirwa [7], who demonstrated a favourable statistical correlation between educational level and ICT usage. Indicating that more educated respondents are more likely to take advantage of and utilise ICT technologies efficiently to access crucial information that could enhance agricultural activity. An increase in educational attainment enhances ICT adoption by improving farmers’ ability to understand, access, and apply new technologies [41,87].

The study results found that large farm size are significantly associated with the use of basic cell phones at the 1% significance level. This may be due to the coordination of farm operations and communication with labourers, suppliers and markets. However, smartphone usage declines as farm size increases (5% significance level), possibly because larger farms demand more investment in inputs, leaving fewer resources for advanced ICT. Shemfe & Modirwa [7] suggest that larger farm sizes enable better ICT access through higher income and operational needs. In another view, Kalema [88] argues that farm size negatively influenced ICT adoption, as farmers with larger farms already have established market connections and rely less on ICT for outreach compared to those with smaller farms.

At the 10% significance level, family size revealed a significant negative influence on TV usage. This finding suggests that individuals from larger households are less likely to use TVs, possibly due to other family members watching entertainment channels rather than agricultural programmes. On the contrary, an increase in family size will yield an increase in the likelihood of the use of ICT tools. This suggests that as family size increases, more people in the household are exposed to ICTs, so they can enhance their lives and farming through using ICTs. Having a big family size is beneficial to smallholder farmers, as they can use some members of the family to train the smallholder farmers in using ICTs and to act as correspondents through ICTs, as smallholder farmers are busy working on the farm [85].

The study found that individuals with additional income sources are significantly more likely to use radio, computers and basic cell phones at 10% and 5% significance levels, respectively. This suggests that off-farm income increases access to ICT tools by providing more disposable resources for media use and communication. Supporting this, Wale & Mkuna [25] observed that farmers with access to social grants or off-farm income tend to prefer diverse information channels—including media, extension services, mobile phones, and NGO support—partly as they can afford associated costs such as transport and airtime.

Network connectivity significantly and positively influences the use of TV at the 1% significance level. This suggests that access to stable and quality network infrastructure enhances the use of television, likely as many modern TVs rely on digital signals, satellite and internet-based platforms. Mdoda & Mdiya [85] found a negative relationship between network coverage and interruption and ICT usage. Implying that network instability discourages technology use, a challenge increasingly evident among South African farmers who report connectivity issues as a major barrier to adopting ICT tools.

Electricity supply shows a significant negative influence on the computer usage at the 10% significance level. Farmers in areas with unstable electricity are less inclined to utilise computers for agricultural information. The results are consistent with Mdoda and Mdiya [85] and Otene et al. [89], who found that rising electricity disruptions directly limit ICT utilisation and damage electronic equipment. This challenge is particularly acute in regions experiencing persistent electricity failures, further discouraging technology adoption.

The study found the language used on ICT tools to positively affect the use of radio and basic cell phones at the 1% significance level. This suggests that individuals who are content with the language in which content is delivered are more likely to engage with these tools. On the contrary, Ireri et al. [90] found that many ICT tools still rely on national or global languages, such as English, which can limit access for users unfamiliar with those languages. Prior studies also highlight that ICT content is often unavailable in native languages, favouring individuals with formal education who can overcome language barriers [91].

ICT literacy positively influences smartphone adoption at the 5% significance level. This indicates that individuals with greater digital competence are more likely to adopt smartphones, which require skills such as installing apps, navigating internet features, and managing device settings. Prior research confirms that digital knowledge enhances mobile phone use among farmers [92]. However, many smallholder farmers struggle to filter relevant and reliable information from the vast amount of online content due to limited ICT literacy [75].

The study results found ICT awareness to positively and statistically influence the use of TV, computers and smartphones at a 1% significance level, respectively. This implies that those with greater knowledge about digital technologies are more likely to embrace and utilize them. Conversely, ICT awareness negatively influences the use of basic cell phones at the 10% significance level, indicating that individuals who are more aware of newer technologies may prefer more advanced communication tools. These findings align with previous research, highlighting a significant association between ICT awareness and the adoption of ICT tools among smallholder farmers [7,61].

5. Conclusions

This study examined the factors influencing smallholder farmers’ choices in using ICT tools to access agricultural information, focusing on demographic, economic, and contextual variables. The primary objective was to highlight the determinants shaping smallholder farmers’ preferences of ICT tools. The findings reveal that smallholder farmers utilise a range of ICT tools, including mobile phones, radio and TV, with preferences shaped by diverse and interrelated factors. For those not using ICTs, traditional sources such as extension services and fellow farmers remain vital, highlighting their continued relevance in rural areas even in the digital era. The key findings of the study show that smallholder farmers in the study area prefer mobile phones, radio and TV, respectively. Distinct factors seem to influence smallholder farmers’ preference of ICT tools. It is evident from the MVP model results that factors such as gender, age and language use positively influence the choice of radio. Farm size, off-farm income and language use positively influence the choice of a basic cell phone. Educational level has a negative impact on the choice of radio and basic cell phones. ICT literacy and ICT awareness positively influence the choice of smartphone, while age and farm size have a negative impact. Choice of a computer is positively influenced by educational level and ICT awareness and negatively influenced by gender, age, having off-farm income and electricity supply. The analysis also revealed that the choice of a TV is positively influenced by network connectivity and ICT awareness and negatively influenced by household size. The results indicate that the key factors influencing ICT choices vary between traditional ICT tools (radio, basic cell phone) and advanced ICT tools (smartphone, computer, TV). For traditional ICTs, language use is a consistently strong and positive factor influencing the choice of radio and basic cell phone. This suggests that individuals who engage more with local languages or communication practices are more likely to use these tools. For advanced ICTs, ICT awareness and educational level consistently emerge as the most influential positive factors. These tools require higher levels of digital literacy, which may explain their dependence on awareness and education. Additionally, age declines with the use of these tools, implying older individuals may not be as familiar with or inclined to adopt advanced digital communication technologies. The findings highlight a policy concern on the importance of these demographic, economic and contextual factors on the dissemination and access of agricultural information using ICT tools.

The findings of this study carry important implications for agricultural policy and rural development in Mpumalanga Province. To promote agricultural transformation through inclusive access to information, ICT initiatives should be tailored to the specific socioeconomic realities of smallholder farmers in the province. In particular, policymakers and agricultural stakeholders should prioritise the dissemination of agricultural content through ICT tools that are already widely used, such as radio, smartphone, and basic cell phone, and ensure that messages are delivered in local languages such as siSwati, Xitsonga and Sepulana to match the literacy and linguistic diversity of rural communities. Given that ICT awareness and education are strong predictors of advanced ICT adoption, the Mpumalanga Department of Agriculture, Rural Development, Land and Environmental Affairs (DARDLEA), in collaboration with local extension officers should implement community-based digital literacy campaigns. These could include farmer field schools integrating ICT demonstrations, partnerships with local schools and training centres to host ICT workshops, and awareness campaigns in rural municipalities such as Mbombela, Nkomazi and Bushbuckridge where digital exclusion is most pronounced. To address affordability and infrastructural constraints, provincial policies should support financial mechanisms such as subsidised devices, low-interest credit schemes, and cooperative purchasing models that enable low-income farmers to acquire computers and smartphones. In areas with weak connectivity, particularly remote wards of Ehlanzeni District, partnerships with mobile network providers could expand internet coverage and reduce data costs through tailored rural bundles for farmers. Recognising that older farmers face challenges in adopting advanced technologies, digital inclusion programmes should be specifically designed for elderly farmers. Practical measures may include peer learning schemes where younger, digitally literate farmers mentor older counterparts, ICT hubs in community centres where elderly farmers can practice using computers in low-pressure environments, and training sessions delivered in local languages using simplified, step-by-step methods. Lastly, gender-inclusive ICT policies are critical. Women farmers in Mpumalanga often face greater structural and cultural barriers to digital adoption. Targeted interventions should therefore provide women with equal access to training, resources and credit facilities. Examples include organising women-only ICT workshops within cooperatives, supporting mobile-based platforms that enable flexible participation by women with household responsibilities, and incorporating gender-sensitive training modules into extension programmes. Ensuring equitable ICT adoption will foster a more inclusive and resilient agricultural sector in Mpumalanga.

This research acknowledges limitations such as the focus on a sample of only 308 smallholder farmers within Mbombela Local Municipality, which does not provide a representative overview of smallholder farmers across Mpumalanga or South Africa as a whole. Therefore, the findings cannot be generalised to smallholder farmers in other regions or municipalities within the country. Future research directions should focus on expanding the geographical scope of the study to include a more diverse and representative sample of smallholder farmers across various provinces. Additionally, longitudinal studies could offer a more comprehensive analysis of the long-term effects of agricultural policies on the dynamics of ICT utilisation among smallholder farmers. Moreover, the sample was confined to farmers registered with the Department of Agriculture, potentially excluding informal or unregistered farmers who may experience greater digital barriers, thus limiting the generalisability of the findings to more marginalised populations. The study employed a quantitative, cross-sectional design. This is valuable for identifying statistical associations at a single point in time; however, it does not allow for establishing causal relationships and capturing changes over time. As a result, while the study highlights important linkages between farmers’ characteristics and ICT use, it cannot determine the direction of influence and account for evolving dynamics. Furthermore, the data relied on farmers’ self-reported ICT use, which may be prone to recall errors and social desirability bias. These methodological constraints limit the precision and depth of the evidence produced. Gender-related findings also presented inconsistencies with existing literature, particularly regarding male farmers’ lower likelihood of computer use, which could not be fully explored due to the absence of qualitative data on socio-cultural and intra-household dynamics. Similarly, the observed complex relationship between off-farm income and ICT tool usage suggests deeper economic behaviours not captured by the current income categorisation. Moreover, the study did not assess how effectively farmers utilise agronomic applications on ICT tools, which is a critical component of digital adoption. Future research would benefit from mixed-methods approaches, more inclusive sampling strategies, disaggregated analysis of key variables such as income type and household roles, and the inclusion of application-specific and qualitative data, such as interviews or field observations, to deepen understanding of the barriers, facilitators, and real-world utility of ICTs in agriculture.

In conclusion, agricultural information needs demand immediate global attention, as their unmet expectations pose economic risk to GDP, the local economy and the livelihoods of smallholder farmers in rural communities. In this fourth industrial revolution era, it is important to implement proactive strategies that will reduce the digital divide among farming communities and ensure inclusive, efficient access to agricultural information. These are essential to safeguarding local economic stability and sustainable attainment of livelihood outcomes for smallholder farmers.