Impact of Technical Training and Personalized Information Support on Farmers’ Fertilization Behavior: Evidence from China

Abstract

Improving farmers’ fertilization behavior through technical training is conducive to reducing agricultural non-point source pollution and maintaining soil health. The existing studies mainly discuss the impact of technical training on the adoption of soil testing formula fertilization (STFF) from the perspective of judging the training effect, but there are few further answers to how to improve the effect of technical training. Based on the theory of planned behavior and training transfer, this study adopted a logit model and OLS model to explore the role of training in the adoption of STFF, as well as the moderation effect of personalized information support. An empirical test was carried out using the survey data of farmers in Zhejiang Province, China. The results showed that participating in technical training is beneficial for farmers in adopting STFF, and receiving personalized information support can strengthen the promotion effect of technical training on the adoption of STFF. Further analysis reveals that there are certain differences in the above relationships among farmers with different operating scales. The findings could facilitate a deeper understanding of the improvement of technical training performance and accumulation of rural human capital and could also provide a useful reference for the policy formulation to optimize farmers’ fertilization behavior and promote sustainable agricultural development.

1. Introduction

The sustainable development of agriculture has attracted considerable attention around the world in recent decades [1]. Soil contamination of farmland is a growing problem worldwide, especially in countries where agricultural output is preferred. In 2014, the excess rate of farmland soil pollution points in China was 19.4%, up 14.8 percentage points from 4.6% in 1989 (Data source: National Soil Pollution Survey Bulletin issued by the Ministry of Environmental Protection and the Ministry of Land and Resources of China in April 2014). Excessive use of chemical fertilizer is the main source of agricultural non-point source pollution, which will aggravate the risk of environmental pollution of soil, water, and air of cultivated land [2]. China is one of the countries with the largest amount of fertilizer application [3]. The fertilization intensity in China increased from 265 kg/hm² in 2000 to 313 kg/hm² in 2020, which was significantly higher than the international safety standard of 225 kg/hm² [4]. Excessive application of nitrogen fertilizer is the main cause of soil acidification in farmland in China [5]. In addition, the increase in chemical fertilizer application volume in China did not significantly improve the efficiency of chemical fertilizer use [6]. In 2020, the utilization rate of chemical fertilizer for China’s three major grain crops was 40.2%, about 20 percentage points behind that of developed countries [7].

Farmers are the direct subject of applying fertilizer and protecting agricultural land. Optimizing the fertilization behavior of farmers or accelerating the adoption of precision fertilization technology by farmers is crucial for preventing agricultural land pollution and promoting sustainable agricultural development [8]. Soil testing and formula fertilization (hereafter abbreviated as STFF) are typical precision fertilization and environmentally friendly technologies promoted by the United Nations, which help to alleviate agricultural non-point source pollution and improve the economic and environmental benefits of agricultural production as a whole [9]. Soil testing and formula fertilization is based on a soil test and fertilizer field test, according to the law of the crop fertilizer demand, soil fertilizer supply performance and fertilizer effect, and on the basis of rational application of organic fertilizer, the application variety, quantity, application period and application method of nitrogen, phosphorus, potassium and medium and trace elements are proposed. After adopting STFF, farmers use the formula fertilizer, which is formulated with various straight fertilizers and complex fertilizers as raw materials. The formula fertilizer is suitable for specific farm regions and crop varieties (this explanation is from the Technical Specifications for Soil Testing and Formula Fertilization (2011 Revision) issued by the Ministry of Agriculture of China). Since 2005, China has implemented and promoted the STFF, and the government has also provided important support in terms of funding and other aspects [10]. However, it is not easy to popularize the application of STFF among farmers [11]. The phenomenon of farmers’ habitual and blind fertilization is still common [12].

Many studies have explored the factors that affect farmers’ adoption of STFF. Among them, technical training is worthy of special attention because it is one potentially effective method to diffuse these new technologies [13,14]. Under the given conditions of farmers’ age and education level, technical training is a more operational measure for the human capital investment of farmers [15]. However, studies focused on the connection between technical training and STFF adoption in China are scant [16]. Relevant studies also have some disputes about the role of technical training in the adoption of STFF. Many studies have suggested that participating in technical training is beneficial for farmers to adopt STFF [17,18]. The main explanation is that technical training enables farmers to better understand the significance, function, principle and operation method of STFF so as to improve the probability of adoption [19,20]. On the contrary, some studies are skeptical about the effect of technical training. They found that the results of technical training in the empirical test are not significant. Possible reasons include that the training did not make farmers truly understand the essence of STFF [21], and farmers had a low evaluation of the previous training effect [22].

To sum up, the existing studies mainly explored the judgment of technical training performance, that is, whether technical training significantly affects farmers’ adoption of STFF, but they have rarely responded to a deeper question: how to improve the performance of technical training. In fact, if there is no additional effort after the training, the effectiveness of training may gradually weaken over time [23]. The adoption of STFF not only requires farmers to understand general information such as the basic principle of the technology but also requires farmers to master personalized information such as the soil nutrients of the land used. The advantage of technical training is to provide general information rather than personalized information. A reasonable prediction is that even though farmers have mastered relatively complete basic knowledge of STFF through technical training and are fully aware of the positive significance of STFF, without personalized information support, farmers will not be able to apply what they have learned or adopt STFF. Personalized information support provides important clues for deepening existing research, but few studies have conducted in-depth research on it [24].

The aim of this research is not only to judge the training effect on the adoption of STFF but also to explore the role of personalized information support in improving the training effect. Therefore, based on the theory of planned behavior, the theoretical framework of the impact of technical training on the adoption of STFF was constructed. The moderation effect of personalized information support in the relationship between technical training and the adoption of STFF was analyzed based on the theory of training transfer. To examine the research hypothesis, the empirical analysis was carried out through an econometric model using the survey data of farmers from Zhejiang Province in China.

2. Theoretical Analysis and Research Hypothesis

The logical relationship between technical training, personalized information support, and the adoption of STFF is shown in Figure 1.

(1)

Judge the training effect: the impact of technical training on the adoption of STFF technology

According to the planned behavior theory, there are three key factors influencing individual behavioral intentions: perceived behavioral control, attitudes, and subjective norms [25]. Technical training influences the adoption of STFF through the abovementioned three aspects.

• ① Perceived behavioral control. Compared with the traditional fertilization behavior based on experience or intuition, STFF is a knowledge-intensive technology. A lack of knowledge is one of the main constraints for farmers to adopt STFF [26]. The technical training will not only enhance the human capital of farmers in applying STFF but will also improve the self-efficacy (self-efficacy refers to an individual’s ability to undertake specific behaviors at different levels of ability [27]) of farmers in applying STFF by transmitting the scientific connotation, technical principles and operational processes of STFF and then affect the perceived behavioral control, which means an individual’s judgment about the difficulty of completing a task [28].
• ② Attitude. In practice, the technical training program on STFF is usually not only limited to technical knowledge, but also involves the significance and importance of STFF in environmental and ecological protection, sustainable development, and enhancing human well-being. For example, the most important farmers’ technical training project in Zhejiang Province is the “Quality Improvement Project for Ten Million Farmers” implemented by the government, which launched in 2012 and has continued to date. In this project, environmentally friendly technology such as STFF is one of the main training topics, and the training content usually involves technical principles, behavioral significance, relevant policies, and typical cases. This content helps to strengthen farmers’ sense of purpose, honor and responsibility of using STFF. The introduction to the technical advantages of STFF in cost saving and efficiency compared with traditional fertilization will also strengthen farmers’ cognition of the expected benefits of using STFF. Therefore, the training can strengthen farmers’ positive feelings towards STFF from the dimension of attitude.
• ③ Subjective norm. The technical training involving the promotion policy publicity of STFF can make the trainees more fully feel the determination and support of the government to promote STFF, to strengthen the trainees’ subjective norms. Subjective norms refer to the impact of social pressure on individual behavioral decision-making, including the extent to which the government promotes and supports the STFF [29]. Combining with the above analysis, the assumptions are proposed as follows:

Hypothesis 1:

Participating in technical training is beneficial for farmers to adopt STFF.

(2)

Improve the training effect: the role of personalized information support

As one of the most popular training performance evaluation frameworks, the Kirkpatrick framework [30] divides training performance into four levels: reaction, learning, behavior, and results. Reaction refers to the subjective evaluation of the training; learning refers to the degree to which the trainees acquire knowledge and skills from the training; behavior refers to professional behavior changes in the trainees due to participating in the training; results refer to the impact of the training on the benefits of the student’s organization. Among them, behavior is also known as “training transfer”, which includes the maintenance, use, and generalization of learned knowledge, skills, and attitudes (KSAs) in the workplace [31]. In the past thirty years, there has been a surge in research about training transfer in human resource development literature [32]. After farmers participate in the training, the adoption of STFF is a kind of training transfer. The key to improving the promotion effect of technical training on the adoption of STFF is to find the driving factors of training transfer.

There is a strand of literature on factors impacting transfer [33,34]. The main factors include three dimensions: individual, training design, and work environment. The framework has become the foundation of many studies on training transfer [35]. Among them, work environment factors are the contextual variables that support trainees to use their new skills and knowledge [36]. Providing information support is an important work support [37].

In information support, the effect of personalized information support on the training transfer of STFF deserves special attention because the application of STFF highly depends on personalized information support [38]. The personalized information support emphasized here mainly includes two aspects: the soil nutrient situation of agricultural land, and the corresponding reasonable fertilization standards and fertilization schemes. On the one hand, if farmers cannot fully grasp the soil nutrient information of their own agricultural land, it is difficult to carry out precise fertilization [39], even if they have mastered the technical principles of STFF through training. On the other hand, reasonable fertilization standards and fertilization scheme information based on soil nutrient information are also very important for the training transfer of STFF. In many cases, the main reason for farmers’ overuse of fertilizer is a lack of awareness of having overused it [40]. An important way to make farmers realize this point and carry out precise fertilization behavior is to inform them on the reasonable fertilization standard as well as the fertilization time, fertilizer formula, fertilization method, fertilization amount and other fertilization plan information of different growing periods of crops [26]. In conclusion, the following assumptions are proposed:

Hypothesis 2:

Personalized information support can strengthen the promotion effect of technical training on the adoption of STFF.

3. Empirical Research Design

3.1. Data Source

The data used in this study come from a survey of rice farmers in Zhejiang province in 2023. Rice is the grain crop with the highest grain proportion in China [41], and excessive fertilization in the rice production process is prominent [40]. Five cities were randomly selected for this survey (Jiaxing, Shaoxing, Jinhua, Quzhou and Taizhou). The map of the surveyed area is shown in Figure 2. In each city, two counties were randomly selected. Five villages were randomly selected in each county, and ten crop farmers were randomly selected in each village. A total of 500 questionnaires were distributed during the survey. After excluding questionnaires with incomplete or missing data, logical contradictions, and irregularities, we finally obtained 445 sample farmers, with a sampling efficiency of 89%.

The selection of Zhejiang Province as the research site is mainly based on the following considerations: (1) Zhejiang, known as the “hometown of fish and rice”, has a long history of rice cultivation, as early as 7000 years ago. Rice is the most important food ration for Zhejiang residents. In 2022, the total output of early rice in Zhejiang was 702,000 tons and the yield per hectare was 27.59 kg, up 11.7% and 0.7% compared with the same period last year, respectively. The area increment and yield increment of early rice ranked first in China. (2) Non-point source pollution in Zhejiang province, especially the problem of excessive fertilization, is not optimistic. According to the second National Pollution Source Survey Bulletin of Zhejiang Province, the emission of agricultural source water in Zhejiang province is 5900 tons, the total nitrogen emission is 42,400 tons, and the total phosphorus emission is 6400 tons. The proportion of the three sources from the planting industry is 71.19%, 83.96% and 81.25%, respectively. Although the application rate of agricultural fertilizers and the application rate of fertilizers per unit area in Zhejiang Province have decreased in recent years, the application rate of fertilizers per unit area in 2022 (338.83 kg/hm²) is still higher than the international safety standard (225 kg/hm²). (3) Zhejiang has been endowed by the state with the positioning of the demonstration of agricultural green development. Zhejiang is the first pilot zone in China to promote the pilot construction of green agricultural development. To this end, Zhejiang has introduced relevant policies to reduce quantitative input and reduce unreasonable fertilization. (4) Zhejiang took the lead in exploring the personalized information support of STFF. In addition to the traditional method represented by the field guidance or suggestion card, Zhejiang is vigorously promoting PIS based on digital technology. This novel PIS has more advantages in cost and efficiency than the traditional method. For example, a smartphone-based application called “Zhe’yang’shi” (see Li et al. (2022) [38] for a detailed introduction to “Zhe’yang’shi” app V1.0) has been gradually promoted in Zhejiang province in recent years. The app provides information about soil nutrients and fertilization plans, aiming to promote precision fertilization. Farmers can adopt the advice for STFF simply by inputting their location and crop variety in the application.

3.2. Key Variables and Descriptive Statistics

We selected STFF_whether and STFF_degree as dependent variables. Both variables depict farmers’ adoption behavior of STFF. The difference is that the former variable is a binary variable, indicating whether the household applied STFF. The latter variable is a continuous variable to show the acreage of land that applied STFF, which further presents the adoption behavior of STFF.

As for independent variables, Training is the number of times the household head gained technical training on agricultural green technology annually. PIS is whether the household received personalized information support about STFF. Referring to the literature, we included the household head’s age, education, household size, deposits, land quality, off-farm work and support level of government as control variables. To investigate the heterogeneity by different farm scales, we divided the samples into two categories. One is large scale (≥100 mu) and the other is small scale (<100 mu).

Table 1. Definitions of the variables and summary statistics.

Variables	Definition	Mean	Std. Dev.	Min	Max
Dependent variables
STFF_whether	Whether the household applied soil testing and formula fertilization technology: 1 = yes, 0 = no	0.81	0.39	0	1
STFF_degree	Acreage of land that applied soil testing and formula fertilization technology (log)	3.10	2.59	0	9.23
Independent variables
Training	The number of times household head gained technical training	1.90	1.55	0	4
PIS	Whether received personalized information support about soil testing and formula fertilization technology	0.87	0.34	0	1
Age	Age of household head (years)	42.37	11.59	20	72
Education	The schooling years of household head (years)	12.18	3.41	6	19
Household size	Number of people residing in a household	5.06	1.16	3	8
Deposits	Amount of household deposits per capita (thousand yuan)	81.95	129.33	30	1050
Land quality	1 if farmland quality is good, 0 otherwise	0.49	0.50	0	1
Off-farm work	The off-farm work time (months)	4.64	3.85	0	12
Support	The STFF support level of government: 1 to 5 means low to high	3.75	1.02	1	5
Large scale	1 if the farm scale is above 100 mu, 0 otherwise	0.52	0.50	0	1

Note: Sample size is 445. 1 mu = 1/15 hectare.

lists the definitions and summary statistics of the variables.

3.3. Econometric Model

In this paper, we applied two models to conduct the empirical analysis. First, because STFF_whether is a dichotomous decision, we estimate logit models as follows:

$$\ln \left(\frac{P_i}{1-P_i}\right)={\alpha }_0+{\alpha }_{1}Trainin{g}_i+X_i{\gamma }_i+{\epsilon }_i$$

(1)

$$\ln \left(\frac{P_i}{1-P_i}\right)={\alpha }_0+{\alpha }_{1}Trainin{g}_i+{\alpha }_{2}Trainin{g}_i\ast PI{S}_i+X_i{\gamma }_i+{\epsilon }_i$$

(2)

where $i=1, 2,\dots , n$ refers to different individual farmers, and $P_i$ indicates the probability that a farmer $i$ chooses to adopt STFF. ${\alpha }_0$ is a constant. $Trainin{g}_i$ is the number of times that a farmer $i$ gained about agricultural green technology. $X_i$ is a vector of explanatory variables, and ${\gamma }_i$ denotes the parameters to be estimated. ${\epsilon }_i$ is a random disturbance term. More specifically, $X_i$ includes variables representing the household head’s age, education, household size, deposits, land quality, off-farm work and support level of government. $PI{S}_i$ denotes whether a farmer $i$ received personalized information about STFF.

Second, for the dependent variable STFF_degree, we apply an OLS estimation based on the following models:

$$STFF\_degre{e}_i={\beta }_0+{\beta }_{1}Trainin{g}_i+{\beta }_{2}Trainin{g}_i+X_i{\gamma }_i+{\mu }_i$$

(3)

$$STFF\_degre{e}_i={\beta }_0+{\beta }_{1}Trainin{g}_i\times PI{S}_i+{\beta }_{2}Trainin{g}_i+X_i{\gamma }_i+{\mu }_i$$

(4)

where $STFF\_degre{e}_i$ is in logarithmic form and denotes the acreage of land that applied STFF technology. ${\beta }_0$ is a constant. $X_i$ is a vector of explanatory variables same as it in Equation (1) and ${\gamma }_i$ denotes the parameters to be estimated. ${\mu }_i$ is a random disturbance term. $PI{S}_i$ denotes whether a farmer $i$ received personalized information about STFF technology. In Equations (2) and (4), we introduce an interaction term, $Trainin{g}_i\times PI{S}_i$, into models to capture the moderation effects of personalized information support.

4. Empirical Results

(1)

The impact of technical training on the adoption of STFF

Table 2. Impacts of training on the adoption of STFF.

	STFF_Whether			STFF_Degree
	Total Sample	Small Scale	Large Scale	Total Sample	Small Scale	Large Scale
	(1)	(2)	(3)	(4)	(5)	(6)
Training	0.0290 ***	0.0127 *	0.0753 ***	0.7410 ***	0.2590 ***	1.0760 ***
	(0.0106)	(0.0070)	(0.0254)	(0.0724)	(0.0582)	(0.1320)
Age	−0.0084 ***	0.0009	−0.0147 ***	−0.0333 ***	0.0001	−0.0547 **
	(0.0019)	(0.0012)	(0.0046)	(0.0122)	(0.0103)	(0.0227)
Education	0.0115 *	0.0034	0.0532 ***	−0.0098	−0.0294	0.0809
	(0.0066)	(0.00223)	(0.0201)	(0.0406)	(0.0197)	(0.0984)
Household size	0.0119	−0.0130	0.0007	0.3010 ***	0.1790 **	0.2610 *
	(0.0126)	(0.0086)	(0.0235)	(0.0918)	(0.0766)	(0.1490)
Deposits	0.0000	0.0001 **	0.0000	0.0002 **	0.0025 ***	0.0000
	(0.0000)	(0.0001)	(0.0000)	(0.0001)	(0.0003)	(0.0001)
Land quality	0.1410 ***	0.0540 **	0.2340 ***	0.4690 **	−0.1402	1.0280 ***
	(0.0322)	(0.0229)	(0.0712)	(0.2200)	(0.1700)	(0.3790)
Non-farm work	−0.0012	−0.0021	−0.0056	−0.0996 ***	−0.1210 ***	−0.1060 **
	(0.0033)	(0.0023)	(0.0067)	(0.0275)	(0.0226)	(0.0470)
Support	−0.0006	−0.0175	0.0260	0.4430 ***	−0.0277	0.5450 ***
	(0.0156)	(0.0119)	(0.0299)	(0.1080)	(0.0938)	(0.1720)
Constant	0.2320	0.0811	−0.0096	0.2320	1.7900 ***	−0.4120
	(0.172)	(0.0912)	(0.405)	(1.1210)	(0.6510)	(2.3600)
Adj. R-sq	-	-	-	0.306	0.485	0.315
Observations	445	215	230	445	215	230

Note: The values in parentheses are standard errors. * p < 0.1, ** p < 0.05, *** p < 0.01.

shows the estimated results of the impact of training on STFF adoption behavior. The marginal effects of logit model equations are reported in columns (1) to (3). The dependent variables in columns (1) to (3) are whether to adopt STFF, and the dependent variables in columns (4) to (6) are the degree of STFF adoption. The positive and significant results for the training variable in the models indicate that the more frequent the participation of training, the greater the probability that those farmers will adopt STFF by 2.9% for the total sample. We also find that participating in technical training brings about a significant increase in the STFF degree (by 74.1%). In conclusion, participating in technical training is beneficial for farmers’ STFF behaviors. Hypothesis 1 was verified. In terms of farm scale, the results indicate that the farmers with large-scale farms experience a higher rate of STFF adoption after receiving technical training than those with small-scale farms.

(2)

The role of personalized information support

Table 3. The moderation effects of personalized information support.

	STFF_Whether			STFF_Degree
	Total Sample	Small Scale	Large Scale	Total Sample	Small Scale	Large Scale
	(1)	(2)	(3)	(4)	(5)	(6)
Training × PIS	0.0617 ***	0.0028	0.0665 ***	0.7100 ***	0.1750	0.6100 **
	(0.0192)	(0.0112)	(0.0244)	(0.146)	(0.112)	(0.284)
Training	0.0289 ***	0.0128*	−0.0145 ***	0.7370 ***	0.2810 ***	1.0040 ***
	(0.0102)	(0.0071)	(0.0048)	(0.0706)	(0.0597)	(0.135)
Age	−0.0084 ***	0.0008	0.0465 **	−0.0257**	−0.0025	−0.0469 **
	(0.0018)	(0.0012)	(0.0207)	(0.0120)	(0.0104)	(0.0228)
Education	0.0053	0.0033	0.0138	−0.0136	−0.0330 *	0.0420
	(0.0065)	(0.0022)	(0.0251)	(0.0396)	(0.0198)	(0.0993)
Household size	0.0140	−0.0132	0.0000	0.2760 ***	0.1780 **	0.2580 *
	(0.0120)	(0.0087)	(0.0000)	(0.0896)	(0.0764)	(0.1408)
Deposits	0.0000	0.0001 **	0.2420 ***	0.0003 ***	0.0025 ***	0.0001
	(0.0000)	(0.0001)	(0.0715)	(0.0001)	(0.0003)	(0.0001)
Land quality	0.1360 ***	0.0521 **	−0.0094	0.3110	−0.2220	0.9680 **
	(0.0307)	(0.0238)	(0.0074)	(0.2170)	(0.1770)	(0.3770)
Non-farm work	−0.0050	−0.0022	0.0018	−0.1360 ***	−0.1280 ***	−0.1370 ***
	(0.0035)	(0.0024)	(0.0338)	(0.0278)	(0.0229)	(0.0488)
Support	−0.0108	−0.0178	0.0665 ***	0.3310 ***	−0.0191	0.3710 *
	(0.0152)	(0.0121)	(0.0244)	(0.1080)	(0.0936)	(0.1890)
Constant	0.3450 **	0.0896	0.1100	0.6600	1.9400 ***	0.5800
	(0.1690)	(0.0980)	(0.4340)	(1.0960)	(0.6560)	(2.3860)
Adj. R-sq	-	-	-	0.340	0.488	0.425
Observations	445	215	230	445	215	230

Note: The values in parentheses are standard errors. * p < 0.1, ** p < 0.05, *** p < 0.01.

reports the estimated results of the moderation effect of personalized information support, including the interaction term of training and personalized information support. The marginal effects of logit model equations are reported in columns (1) to (3). The dependent variables in columns (1) to (3) are whether to adopt STFF, and the dependent variables in columns (4) to (6) are the degree of STFF adoption. The results show that the coefficients of the interaction term are significantly positive for both STFF adoption and the STFF degree. That indicates that personalized information support can strengthen the promotion effect of technical training on the farmers’ soil testing and formula fertilization technology behaviors. Hypothesis 2 was verified.

The empirical results also indicate that the moderation effect of personalized information support is significant on a large scale but not on a small scale. The main reasons are as follows: (1) owing to smaller areas and longer self-management, small-scale farmers tend to know more about the nutrients of their land, while large-scale farmers usually transfer land from others and operate for a shorter time. Therefore, small-scale farmers have relatively weak dependence on personalized information support when making decisions regarding STFF adoption, or there is a less marginal role of PIS in promoting the migration of STFF training; (2) many small-scale farmers (especially the aging ones) believe more in traditional habits and subjective experiences in fertilization decisions than in what others tell them; (3) some small-scale farmers will not change behavior through PIS even if they know that fertilizer has been overused. That is because they see excessive fertilization as a “yield insurance” to resist uncertainty in agricultural production and believe the cost of excessive fertilization is acceptable because of the small size of the land operated.

5. Conclusions and Implication

This study investigated the role of training and personalized information support in the adoption of STFF through theoretical and empirical analysis. After the expatiation of the impact of training on the adoption of STFF based on plan behavior theory, the moderation effects of personalized information support in the relationship between training and STFF is analyzed based on the theory of training transfer. Using the survey data of 445 farmers in Zhejiang province, empirical results show that attending training increases the adoption of STFF; a higher frequency of participation of farm households in training increases the likelihood of these farmers to adopt STFF, increasing by 2.9%, and the degree of STFF will increase by 74.1%. The results also show that personalized information support can strengthen the positive effect of training. Furthermore, there are certain differences in the above relationship between different operation scales. In terms of farm scale, the results indicate that the farmers with large-scale farms experience more promotion on STFF adoption behavior from receiving the technical training than those with small-scale farms. Moreover, the moderation effect of personalized information support is significant on a large scale but not on a small scale. Therefore, it is important to provide training and personalized information support to promote STFF. The findings could facilitate a deeper understanding of the improvement of technical training performance and accumulation of rural human capital and also provides a useful reference for the policy formulation to optimize the farmers’ fertilization behavior and promote sustainable agricultural development.

There are some implications of the training effect. The transformation from the traditional fertilization behavior relying on empirical intuition and subjective judgment to the STFF relying on scientific data put forward higher requirements for farmers’ human capital. Technical training is an important method of human capital investment, which has the advantages of pertinence, closeness and operability compared with academic education. The findings presented in this paper provide new empirical evidence to prove the effectiveness of technical training in promoting the popularization of STFF. In view of the positive externalities of both ecological and environmental protection and new technology demonstration, it is legitimate and reasonable that government organizes and supports the education of farmers on STFF. Therefore, developing countries should pay more attention to strengthening the technical training of STFF, which can not only improve the objective ability of STFF but can also strengthen the subjective cognition of STFF. In addition, this paper found that the facilitation effect of training is not ideal for small-scale farmers. This means that it is difficult to rely solely on training to promote the adoption of STFF by small farmers, and concerted efforts need to be made in rural education, agricultural management systems, policy incentives and other aspects, especially for countries where small-scale farmers are widely present.

A deeper discussion is needed on how to improve the training effect. The theory of training transfer provides an analytical framework and indicates the importance of information support for the application of knowledge. After judging the training performance, the findings further show that personalized information support can enhance the positive effect of technical training for the adoption of STFF. The technical characteristics of STFF determine that the training transfer of STFF is highly dependent on personalized information support. The generalized information about the technical principles and the personalized information about the soil quality together constitute the “secret key” to applying STFF. To some extent, training is a kind of generalized information support. Therefore, in the process of promoting STFF with information support, both internal and external perspectives of training should be considered. It is necessary not only to focus on setting-up the training course but also to strengthen the corresponding personalized information support after training. Moreover, this study found that the effect of personalized information support was significant for large-scale farmers but not for small-scale farmers. This means that it is necessary not only to continue to increase the personalized information support for large farmers but also to further play the leading and union role of large farmers to small farmers. In this regard, China is committed to innovating and optimizing the agricultural management system through the connection of interests, standard unification and socialized services, so as to change the incentive and restraint of small farmers to adopt STFF, and then optimize their fertilization behavior. This is also a useful reference for other developing countries around the world.