E-Learning in Organizations

Online courses for training purposes in organizations have been largely spread with accompanying cybernetic innovations, available interaction tools, and digital media (^{Raymond, Clarke, & Klein, 2014}). These instructional events are suitable for adults to attend in the work environment (^{Abbad et al., 2010}), provide continuous learning and professional qualification, and can reduce financial costs, cheapening training programs (^{Iglesias & Salgado, 2012}).

The science of training attempts to keep up with the advanced development, design, and delivery of online training programs in order to discover how to improve human performance through the definition of principles, guidelines, and criteria in monitoring e-learning in organizations (^{Garavan, Carbery, O'Malley, & O'Donnell, 2010}).

It is widely accepted that trainees' characteristics have a powerful role in the transfer of training and training outputs (^{Burke & Hutchins, 2007}): their personal features, life habits, expectations, strategies used to achieve learning, motivations, interests, and goals can influence the acquisition of skills during training and, subsequently, its application in work situations (^{Meneses, Abbad, Zerbini, & Lacerda, 2006}).

The skills and resources that a worker needs in an online training process are different from those needed for a face-to-face training, mainly because online training increases the demanded level of regulatory resources and motivation to persist in it, requiring the worker to plan and organize him/herself in a more relevant way (^{Ziljstra, Roe, Leonora, & Krediet, 1999}). For example, when employees are frequently interrupted while completing an online course (^{Sitzmann & Elly, 2010}), they must recover the pace of work and decide whether to continue in the training process, which, in turn, demands additional motivation.

In this sense, some research findings in e-learning point to the greater importance of cognitive and self-regulatory learning strategies to obtain better learning outcomes (^{Johnson, Gueutal, & Falbe, 2009}; ^{Martins & Zerbini, 2016}). In distance corporate environments, these learning strategies may meet the development of skills and performance of tasks that are mainly cognitive, apart from the adequacy of using self-regulatory processes to combine work activities (performance goals, excess workload) with studying (learning needs, development of new skills) during work-hours training.

When training sessions are effective, trainees develop new skills and intend to apply them at work, but limitations in work environment may interfere in the transfer process (^{Salas et al., 2012}). In that case, not only do the forms of interpersonal interaction during instruction and trainees' competencies change, but also the instructional design itself and the social support provided. Trainees must be accustomed to the technologies and, for the most part, do without help or feedback from a supervisor/instructor, which can be a challenge for some employees. Situational constraints might frustrate the trainees, by reducing their motivation to learn and their belief that additional effort will lead to better performance, hindering training results.

Figure 1 Conceptual Model Underlying the Hypothesized Relationships. 

Despite the differences between web-based and classroom instruction, the evaluation of training effectiveness requires measuring its effects on the individual employees' performance (^{Aguinis & Kraiger, 2009}; ^{Burke & Hutchins, 2007}). Thus, this paper includes a range of individual characteristics and work environment variables to explain the training outputs.

This study tests an effectiveness model (Figure 1) in a corporate online context to identify factors relating to learning strategies, reactions, barriers, and support of training transfer that can influence the transfer of training. The transfer of training was measured in terms of behavioral transfer (do trainees use their acquired knowledge and skills on the job?) and job performance (does training have an impact on the effectiveness of behavior?). Definitions and specific hypotheses for each of the relationships between variables that were tested are presented and supported below.

Learning Strategies

Learning strategies refer to cognitive, behavioral, and self-regulatory procedures; trainees use them to aid learning during a training program and achieve specific learning goals (^{Badia & Monereo, 2010}; ^{Warr & Allan, 1998}). E-learning programs require a range of different skills–knowing how to use a variety of web-based instructional procedures and resources, virtual tools, technological and pedagogical interactivity–tending to offer trainees increasingly larger amounts of control over their own learning process (^{DeRouin et al., 2005}), once content can be accessed anywhere and anytime. Because of that, those who have been trained in the most effective strategies can achieve better performance results (^{Bjork, Dunlosky, & Kornell, 2013}).

According to ^{Warr and Allan's (1998)} classification of learning strategies, they can be divided into three components: cognitive (rehearsal, organization, elaboration), behavioral (interpersonal and written help-seeking, practical application), and self-regulatory (emotion and motivation control, comprehension monitoring).

Empirical studies have found that mainly cognitive and behavioral strategies during training are positively correlated with learning and transfer (^{Brandão & Borges-Andrade, 2011}; Crouse, Doyle, & Young, 2011; ^{Pantoja & Borges-Andrade, 2009}). Such strategies might be of great importance for achieving the best transfer results because trainees that are also workers mostly search the training content for utility and applicability: they look for connections between course material and their previous knowledge, and its implications at work, apart from the practical application of newly learned skills or behaviors.

However, self-regulatory strategies have shown positive and strong relationships with good academic results in e-learning environments (^{Aguinis & Kraiger, 2009}; ^{Johnson et al., 2009}; ^{Vovides, Sanchez-Alonso, Mitropoulou, & Nickmans, 2007}). In short, self-regulatory strategies express the learner's control of anxiety, concentration, attention, motivation, and the learning process itself. It appears to be more important in distance training than in face-to-face training, because trainees must more often force themselves to pay attention, maintain their interest in and concentration on the learning lesson, once online training is undertaken at the workplace, so that learning must compete with work activities and performance goals.

Although self-regulatory learning strategies seem relevant to explaining the transfer of training, providing trainees with skills that help them transfer successfully back to the workplace, most of the research outcomes have been limited to learning or have only indirectly tested relationships to transfer (^{Burke & Hutchins, 2007}). As the use of self-regulatory processes seems to be quite adequate in an online context, while studying in the workplace may require more effort, focused attention, and self-monitoring of learning, we predict self-regulatory learning strategies will influence the training outcomes and trainees' performance more than cognitive and behavioral strategies.

Hypothesis 1: Self-regulatory learning strategies will predict training transfer indicators.

Reactions to Training

Trainees' reactions to training reflect their satisfaction with the instructional characteristics in terms of pace, format, organization, content relevance, delivery quality, and amount learned (^{Alvarez, Salas, & Garofano, 2004}; ^{Klein, Noe, & Wang, 2006}). Reactions are an important emotional outcome in the area of training and development (^{Aguinis & Kraiger, 2009}) that influence the training transfer process (^{Mathieu, Tannenbaum, & Salas, 1992}), affecting subsequent performance.

Apart from learning and transferring skills, reactions are also effective training criteria. These favorable opinions about training are expected effects: first, because the more satisfied the participants with training, the more willing to transfer, increasing chances to achieve performance results at work; second, trainee reactions seem to play an important role along with other prior work experiences, and should be viewed as part of a larger training system (^{Bell et al., 2017}). Positive previous training experiences may contribute to participation in and motivation for future developmental programs, which can affect learning outcomes and job behavior.

Beyond testing effects of reactions as predictors, research might also test them as moderating (^{Gunawardena, Linder-VanBerschot, LaPointe, & Rao, 2010}) and mediating variables (^{Mathieu et al., 1992}). A supportive environment, pre-training motivation, and other trainee characteristics can influence reactions to training experiences, increasing employee participation and leading them to be more likely to demonstrate trained skills.

The instructional procedures matter to training success (^{Salas et al., 2012}) and should be chosen according to previous results that displayed positive relationships with the transfer process (^{Burke & Hutchins, 2007}). Reactions have been consistently demonstrated in studies to be related to training outcomes (^{Giangreco, Carugati, Sebastiano, & Della Bella, 2010}), as increases in the satisfaction of the participants (liking a training program and perceiving its utility) are concomitant with increases in willingness to transfer the new skills acquired when they return to the workplace.

Based on this, we suppose trainees' reactions to training will positively predict training transfer indicators.

Hypothesis 2: Trainees' reactions to training will be a positive predictor of training transfer indicators.

Support of Training Transfer

Support of training transfer is a contextual variable that evaluates the support trainees receive to use their new skills and knowledge (^{Abbad et al., 2012}; ^{Burke & Hutchins, 2007}). Organizational and social (supervisor and peer) support are very consistent factors explaining the relationship between work environment characteristics (variables such as climate, job design, quality management, etc.) and transfer (^{Bell et al., 2017}; ^{Chiaburu, van Dam, & Hutchins, 2010}; ^{Grossman & Salas, 2011}; Joo, Lim, & Park, 2011).

Once new skills have been acquired, the work environment might facilitate their application. In this sense, organizational, peer, and supervisor support are of special relevance to training being applied at work (^{Chiaburu, 2010}; ^{Gunawardena et al., 2010}), or the person's own skills to transfer the knowledge.

We will distinguish social and organizational support (^{Abbad et al., 2012}; ^{Burke & Hutchins, 2007}). Social support means that the supervisors, coworkers, or peers set goals, give assistance (to discuss and to encourage the application of the new skills at work), and provide feedback. Organizational support refers to the available material and financial resources at the workplace and the workplace's physical conditions that could facilitate (or hinder) the skills transfer process.

Many studies show social support to be more important than organizational support for the transfer process (^{Bhatti, Battour, Sundram, & Othman, 2013}; ^{Dermol & ater, 2013}; ^{Homklin, Takahashi, & Techakanont, 2014}; ^{Pham, Segers, & Gijselaers, 2012}). One explanation is that the former is a closer and more concrete type of support, so learners are more likely to be engaged deeply in applying what they have learned in the presence of support from supervisors and peers, compared to the latter, which is more distal and diffuse, possibly leading to weaker influence on transfer (^{Chiaburu, 2010}). In some cases, support from peers and colleagues has proven to wield more consistent influence on trainee transfer than supervisory support (^{Chiaburu, 2010}; ^{Homklin et al., 2014}).

We predict that social support, that is, peer and supervisor support, will be more important than organizational support in an online context, especially when the training occurs during work hours. This type of support might present stronger relationships to transfer when the focus of training is on open skills (^{Blume, Ford, Baldwin, & Huang, 2010}) because managers can support their subordinates through “showing,” i.e., modeling appropriate behaviors (^{Bell et al., 2017}).

Considering the attitudinal nature of the training evaluated–its objectives consist of the employee's decision to turn off the light when s/he leaves work or a room that is not being used at the time, deciding not to print an e-mail, or to use only a single disposable cup during the whole day–the human aspect (role model) seems to play an important part.

When supervisors are fully aware of the training, they will be better prepared to encourage the trainees to transfer, influencing post-training behaviors; in addition, when a leader behaves in a manner that supports what was taught in training, trainees are more likely to apply what they learned (^{Bell et al., 2017}). For this reason, we predict that:

Hypothesis 3a: Social support will have a greater influence on training transfer indicators than organizational support.

Lastly, as an additional test of reaction relevance, its influence on training transfer should also be observed indirectly, since positive trainee reactions to a training program might influence the perception of support of transfer when they return to the workplace and, consequently, transfer. Trainees are more likely to respond positively to training when they perceive that learning or transfer is supported. When estimating the levels of support of transfer existing in their organizational environment, trainees might take into account their own reaction to training. The reported higher levels of reaction to training might lead to a positive perception of the support offered, influencing them to apply their knowledge and skills in the work setting.

Hypothesis 3b: Trainees' reactions to training will mediate the support of training transfer and training transfer indicators.

Barriers

Barriers can be defined as factors related to the learner, instructional, and contextual (social or organizational) characteristics that prevent, disrupt, or prematurely interrupt a learning process; likewise, they can hinder performance during training programs, impacting related results (^{Crouse et al., 2011}; ^{Hicks, Bagg, Doyle, & Young, 2007}).

In distance learning, new and specific aspects are present, for example, quality of the instructor's teaching performance, virtual instructional design, technical and social support provided by peers and organization to trainees (when and how it is provided to them), availability, and utility and ease of use of technology (^{Garavan et al., 2010}).

Previous studies have found that barriers were mainly related to resource constraints like lack of time and access to technology (Internet, computers, etc.) or to personal features: a learner's lack of interest in the course content, misperceptions about its nature and complexity, conflict with work schedules and activities, self-regulatory learning processes (^{Hicks et al., 2007}; ^{Klein et al., 2006}). These difficulties may interfere in the choice to persist in training or even cause course evasion (^{Abbad et al., 2010}).

Environmental conditions or events can be perceived as facilitators or obstacles by learners, depending on their individual characteristics and the instruction itself (^{Klein et al., 2006}): whether they are able to combine work activities with studying and master the required technologies, whether the course meets their personal and professional expectations, the amount of resources in terms of time, finances, or knowledge available for study.

When perceived as facilitators, those factors are called enablers and might express possibilities for trainees during training programs, because they are able to motivate, support, and enhance learning, helping them acquire new knowledge and skills. Findings suggest that the enablers would be related to organizational support, job or task, and/or family, and access to technology and personal attributes (motivation to learn, learner's own needs and interest).

Therefore, we expect the extent to which personal and environmental features as a whole are perceived to be barriers will influence the transfer of training. If trainees perceive features as barriers, these obstacles might hinder their performance, leading to lower levels of transfer.

Hypothesis 4a: Barriers will be negatively related to training transfer indicators.

Furthermore, we examine the moderating role of support in transferring, buffering the negative influence of perceived barriers over training transfer indicators, i.e., the effects of barriers on transfer will decrease when trainees perceive support for transfer.

Hypothesis 4b: The more trainees perceive support to transfer, the less negative the influence of barriers to training effectiveness.

Training Transfer

Training transfer refers to the effective application at the workplace of new knowledge, skills, and abilities (KSAs) acquired during a training program, and has been the main indicator of the effectiveness of training at an individual level (^{Bell et al., 2017}; ^{Burke & Hutchins, 2007}; ^{Grossman & Salas, 2011}). In this paper, we measure training transfer as both behavioral transfer and job performance. Behavioral transfer encompasses the direct and specific effects of training, related to learning goals, on the trainees' subsequent performance in their job positions (^{Zerbini et al., 2012}). Job performance refers to the indirect influence of training on broader trainee performance, attitudes, and motivation (^{Abbad, Pilati, Borges-Andrade, & Sallorenzo, 2012}). This definition extends the performance concept provided by ^{Campbell (1990)} by examining the effects of training on broader job performance, specifically performance that was not directly related to the training itself (^{Bell et al., 2017}).

Obviously, the second measure has been more frequently used in empirical studies, once it standardizes evaluation indicators, making them generic and objectives, becoming very useful in comparing the effects of different training regimes on trainees' performance (^{Pilati & Abbad, 2005}). The former requires the analysis of learning goals, which are not always available and vary from one course to another, requiring the elaboration and validation of a specific instrument for each training program evaluated (^{Zerbini et al., 2012}). Consequently, the use of two different measures has some clear advantages, allowing the evaluation of the training outcomes on job positions and, additionally, their effects on the trainees' broader performance, including their motivation, self-confidence, and openness to changes in work processes.

We assumed that the direct effects of training on the specific developed skills would influence the participants' broader performance since work behavior occurs in a combined way. In other words, a specific and effective change caused by training (i.e., trainees apply the new KSAs directly to their job positions) would lead to a change in the way they perform indirectly at work in other activities, not necessarily related to the learning goals of training.

Hypothesis 5: Training effectiveness related to the job (behavioral transfer) will be positively related to the worker's broader performance (job performance).

Participants

Participants were employees of a large public Brazilian bank that, in 2016 and at the workplace, participated in the Operational Efficiency online training, with the objective of identifying ways to promote operational efficiency in work activities at the company. It is a 2-hour self-instructional course that approached, in six units, the differences between efficiency and effectiveness, exposed some trends in the banking scenario, taught how to measure operational efficiency, and proposed ways to improve and promote it at work.

This training is strategic training available to all bank employees and not specific to a single functional category. It was developed more to enhance the individual performance than group or unit performance; so, even if the employees were in different work units (support, business, tactics, or strategic), there would hardly be differences between those units. Nevertheless, due to training characteristics and the individual analysis of performance by the organization, we could not analyze data using multi-level analysis.

Additionally, managers have evaluated the influences of that course on their subordinates' work behaviors. Managers can easily observe the trained behaviors (e.g., put into practice actions that can reduce administrative expenses, decrease the consumption of the bank's resources, and increase its results) on a daily basis at the workplace. In addition, the bank commonly performs this type of evaluation, including both workers' and supervisors' reports.

The answers obtained as to the demographic and professional characterization of 1,639 employees (participants in training) and 2,261 managers (raters of the employees who participated in training), respectively, show that the majority are males (56.8%/67.7%), aged between 46 and 55 years (26.1%/41.7%), have 1 to 3 years of experience on the job (20.3%/27.5%), work in the Business and Management Support Unit (37.8%/59.1%), and hold an undergraduate degree or higher (63.3%/86.3%).

Instruments

We used an adapted version of the following instruments:

Learning strategies. An 18-item questionnaire, using a 5-point rating scale from 1 (never) to 5 (always), that measures the frequency with which the participants employed learning strategies (^{Martins, Zerbini, & Medina, 2018a}). It was represented by items from its 4-factor structure, for instance: “I increased my effort when I began to lose interest” (motivation control); “I identified daily situations in which I could try the material out in practice” (elaboration/practical application); “I tried to persuade myself not to worry about mistakes I made” (emotion control); and “I asked other course members for help when I did not fully understand the material” (cognitive/help-seeking). The internal consistency reliability estimate of this scale in this study was .89.

Reactions to training. Participants' reactions to training were assessed using a 9-item scale (sample items included “language used in course material” and “contents available in each unit”), with response alternatives scored from 1 (very bad) to 5 (excellent) (^{Martins, Zerbini, & Medina, 2018b}). The internal consistency reliability estimate of this scale in this study was .92.

Support of training transfer. Supervisor, peer, and organizational support were assessed using a 14-item scale (e.g., “the necessary information for the effective use of new skills learned in training is provided”, “difficulties in the effective use of new skills learned in training are removed”, and “the necessary material resources are provided for the effective use of the new skills learned in training”), with response alternatives scored from 1 (never) to 5 (always) (^{Abbad & Sallorenzo, 2001}). The internal consistency reliability estimate of this scale in this study was .94.

Barriers. To measure perceived barriers, a list of 13 items were presented to the participants, reflecting different personal and environmental features that could be viewed as hindering their course performance (sample items included: “time available to study”, “Internet connection”, and “reconciliation of the course with professional activities”), with response alternatives scored from 1 (did not hinder my performance) to 5 (definitely hindered my performance) (^{Umekawa & Zerbini, 2015}). The internal consistency reliability estimate of this scale in this study was .98.

Training transfer. To measure training transfer, two instruments were used (^{Pilati & Abbad, 2005}), one for self-evaluation (α = .89) and another for hetero-evaluation (α = .94), both with 7 items that evaluate the indirect influence on the broader performance, attitudes, and motivation of trainees (job performance–see Appendix). Participants rated their agreement with each item using a 5-point Likert scale, from 1 (do not agree) to 5 (totally agree).

Additionally, an 8-item instrument, with alternative answers scored from 1 (never) to 5 (always), was developed to measure the direct and specific effects of training Operational Efficiency on job position (behavioral transfer) related to its learning goals (e.g., minimize resource consumption, reduce administrative expenses, contain costs, apply resources, and share actions with colleagues to promote operational efficiency in daily activities at work). The internal consistency reliability estimated for the scale was .85 and .89, respectively, in self-evaluation and hetero-evaluation.

Procedure

This paper uses multi-source data collected from workers that had participated in the online training and from their supervisors, who had evaluated course influence on their subordinates' work behaviors. An online application of the instruments to a potential population of 3,600 employees was carried out; their supervisors responded to the training transfer measures (behavioral transfer and job performance) related to subordinates who were enrolled in the course. Due to the high number of employees in the bank (about 110,000) and the number of bank offices (about 15,000), each subordinate was evaluated by one supervisor, with some exceptions.

The evaluation was completed after approximately 6 months from the end of training, so its effect could be observed at an individual level. The self- and hetero-evaluation obtained, respectively, an overall response rate of 61.1% (n = 2,201) and 66.9% (n = 2,411). To link managers' performance ratings to workers' reports of training transfer, we analyzed the relationships between overall average levels of training transfer reported by workers and their supervisors.

Data Analysis

To run the analyses, the SPSS/AMOS 22.0 was used. Preliminary analyses were done to check the existence of missing values, univariate, and multivariate outliers. Normality was assessed by the skewness and kurtosis of item values, which should range from -2.0 to 2.0, although the larger the sample size, the less concern about normality.

To test the proposed research model and identify different relationships between variables, a structural equation modeling (SEM) was conducted. The estimation method used was maximum likelihood, which is very reliable in cases where distributions of the variables are normal. To judge model fit, the following goodness-of-fit indices were considered acceptable: when CMIN/DF (χ²/df) is less than 5, incremental indexes (CFI and TLI) are higher than .90 (ideally, above .95), and error rate (RMSEA) is less than .08 (ideally, below .05). To estimate mediators and moderators of training effectiveness at work, PROCESS/SPSS procedure (bootstrap confidence interval [CI] method) was used.

To take into account unavoidable clustering effects that occur when managers rate multiple workers, we determined intra-rater reliability by calculating intraclass correlation coefficient (ICC), using a two-way fixed model (alpha) and absolute agreement type, (σ_error)²= (σ_o)²+ (σ_residual)². In this way, we could establish the reliability or consistency of repeated measures (for example, two individual raters testing the same characteristic or measure), and determine how consistent those measures are.

Initial Analyses

Means, standard deviations, correlations, and alpha reliabilities are presented in Table 1.

Table 1 Means, Standard Deviations, Correlations, and Reliabilities 

Note. Parenthetical numbers on the diagonal are coefficient alpha reliability estimates.*p < .01, **p < .05.

Descriptive statistics indicated that workers transferred the new competencies learned to their job positions (behavioral transfer: M = 4.16, SD = .84) and training affected their broader performance as well (job performance: M = 4.31, SD = .70). Supervisors confirmed those reports (behavioral transfer: M = 4.04, SD = .89; job performance: M = 4.07, SD = .83), showing a high convergence between the evaluations of both sources and their agreement on training contributions. Among supervisors, regarding ratings of behavioral transfer and job performance, results present very good reliability, corroborated by ICC = .85 (95% CI [.832, .865], F = 6.722). Raters are consistent, i.e., supervisors agree absolutely on the measurement they are getting, so that we have consistent results and reliable data.

Hypotheses Testing

With respect to distributional assumptions, the absolute values of skewness (range from .08 to 1.60) and kurtosis (range from .002 to 1.94), with only a few exceptions (6 out of 69 items: 2.34 to 3.94), which is still within the recommended bounds by ^{Hancock and Mueller (2010)}, did not exceed the reference interval, indicating the normal distribution of data.

In order to test hypotheses, direct effects of learning strategies (cognitive/help-seeking, emotion control, elaboration/practical application, and motivation control), reactions to training, support (supervisor, peer, and organizational support), and barriers to training transfer (behavioral transfer and job performance) were examined by reviewing the standardized regression coefficient (β) weights at alpha level of p < .0001 and p < .05.

Self-evaluation. The structural model provided a good fit to the data (CMIN/DF = 4.87, CFI = .91, TLI = .90, RMSEA = .04). First, the effects of elaboration/practical application learning strategies (β = .477, p < .0001), trainees' reactions to training (β = .204, p < .0001), organizational support (β = .159, p < .0001), and peer support (β = .081, p < .05) of the behavioral transfer were statistically significant. Second, the effects of motivation control (β = .129, p < .0001), cognitive/help-seeking (β = .110, p < .05), and elaboration/practical application learning strategies (β = .093, p < .05), besides trainees' reactions to training (β = .247, p < .0001) on job performance were statistically significant. The direct effect of the behavioral transfer on job performance was of β = .334 (p < .0001).

Hetero-evaluation. The fit statistics indicated that the measurement model exhibits a good fit (CMIN/DF = 4.67, CFI = .92, TLI = .91, RMSEA = .03). For behavioral transfer, supervisor support (β = .074, p < .05), and job performance, cognitive/help-seeking learning strategies (β = .052, p < .05) were statistically significant. The direct effect of behavioral transfer on job performance presented a very high value (β = .878, p < .0001).

In sum, the results have completely supported Hypotheses 2 and 5, which predicted that trainees' reactions to training would be a positive predictor of training transfer indicators, and that behavioral transfer would be positively related to job performance. On the other hand, since the effects of barriers on the effectiveness of training indicators were not statistically significant, Hypothesis 4a was not confirmed.

Regarding Hypothesis 1, which predicted that self-regulatory learning strategies (motivation and emotion control) would explain training transfer indicators, findings showed that it was partially confirmed. Emotion control did not significantly predict performance results; however, motivation control not only predicted job performance, but also exhibited a greater β weight than the other strategies of cognitive/help-seeking and elaboration/practical application. Nevertheless, the combination of cognitive and behavioral strategies (elaboration/practical application) also predicted both behavioral transfer and job performance.

Concerning the different types of support (organizational, peer, and supervisor), all revealed a significant relationship with training transfer indicators. Although, contrary to what Hypothesis 3a stated, which predicted that social support (supervisor and peer) would have a greater influence on training transfer indicators than organizational support, organizational support exhibited a greater β weight than peer or supervisor support.

These findings elucidate the significant contribution of individual and contextual variables in predicting training transfer, which means that there are some learning strategies that are more important than others and, consequently, should be chosen while studying. Then, after training is ended, trainees' levels of satisfaction with instructional proceedings and their perceived support when they return to work also matter. Moreover, job-related training effectiveness (behavioral transfer) is positively associated to a worker's broader performance (job performance).

Reactions to training as mediators. Trainees' reactions to training were specified as a mediating variable between support of training transfer and training transfer indicators in Hypothesis 3b. Mediation analysis corroborated this hypothesis, according to the results shown in Table 2.

Table 2 Simple Mediation of Reactions to Training between Support and Training Transfer 

Note. M = RP; XM = a; MY = b; XY = c'; Model 4, bootstrap CI method (bias corrected), number of samples bootstrap = 5,000; Boot LL CI = lower limit; Boot UL CI = upper limit; κ² = kappa-squared (^{Preacher & Kelly, 2011}).*p < .0001.

“Support of training transfer” (supervisor, peer, and organizational support) had a positive indirect effect on training transfer indicators (behavioral transfer and job performance) through increasing “reactions to training”, confirming its mediator role–the ab effect was statistically significant (value 0 is not within the 95% confidence interval, which can be observed by positive signals of LL and UL). “Support of training transfer” had a positive effect on “reactions to training,” and “reactions to training” had a positive effect on “training transfer,” leading to an overall positive indirect effect of “support of training transfer” on “training transfer.”

Support of training transfer as moderators. Results indicate (Model 1, bootstrap CI method, bias corrected, number of samples bootstrap = 5,000) that “organizational support” marginally moderates the effect of perceived “barriers” on “behavioral transfer” (β_interaction = .03, p < .10), such that for participants who judged the support offered by the organization to be sufficient “barriers” effects on “behavioral transfer” were attenuated.

For trainees with high scores for “organizational support” (M = 4.64, 5-point scale), “barriers” had a positive effect on “behavioral transfer” (95% CI = [.0120, .0874]). However, for those with low (M = 3.00, 95% CI = [-.0558, .0466]) to moderate averages (M = 3.82, 95% CI = [-.0090, .0541]) on perceived “organizational support”, the effect became insignificant. Thus, the “behavioral transfer” prediction of “barriers” through levels of “organizational support” is almost inexpressive, according to its small magnitude, marginally supporting Hypothesis 4b.