Perceptions of accountability for the transfer of training by leadership trainers

Schlenker’s model of responsibility was used in this study to investigate to what extent leadership trainers consider themselves accountable for the transfer of training. We conducted 15 semi- structured interviews and evaluated the answers using qualitative content analysis. With respect to the responsibility links of Schlenker’s model, the trainers described two out of three links as being rather strong. The interview data suggest that transfer- enhancing strategies were mostly clear to the trainers. They also reported feeling personally obliged to support trainees in their transfer efforts. Regarding the third link, the trainers perceived limited control over several transfer determinants. They explained that they could facilitate transfer but not produce it. The trainers identified the trainees, their supervisors, and the organisations as other responsible parties. The concept of trainers as transfer managers was scarcely reported among the data. Our findings suggest that client organisations could strengthen accountability by setting adequate and feasible training objectives and by monitoring their achievement. We discuss the theoretical and practical implications with regard to the promotion of transfer.


INTRODUCTION
Lifelong learning constitutes a vital aspect of society (Commission of the European Communities, 2001), and current developments such as digitalisation and demographic change particularly emphasise the need for effective and sustainable training and development to prepare professionals for present and future challenges.
Although there has been research investigating the determinants of effective training and development for many years, there are still some blind spots, and training does not always lead to the expected results (Burke & Saks, 2009). Researchers assess training effectiveness based on the degree to which trainees transfer the acquired knowledge, skills and attitudes to their workplaces (Baldwin & Ford, 1988). Even though transfer research is highly productive, the role of trainers in that context has been scarcely investigated (Bonnes et al., 2019). This is surprising considering that trainers are the individuals who plan and deliver training. Moreover, training research indicates that trainers can influence training quality and outcomes (Burke & Hutchins, 2008). Therefore, some researchers have raised the question of trainers' accountability for the (non-)success of transfer (Burke & Saks, 2009;Kopp, 2006). This understudied topic could be one of the missing links in the larger framework of training effectiveness and could help solve the "persistent but solvable problem" of insufficient transfer (Burke & Saks, 2009, p. 382). On the one hand, trainers' accountability is highly relevant for training practice as a lack of transfer results in high costs for organisations, in trainees being unable to realise their professional development potential and in some contexts such as medical training even in the harm or death of individuals (Kopp, 2006). On the other hand, the concept has theoretical relevance as previous training research is criticised for being atheoretical (Tews & Burke-Smalley, 2017) and trainers in particular are still severely underresearched.
Our study addresses this research desideratum by examining to what extent trainers consider themselves accountable for the transfer of training. This qualitative study focuses on leadership training, a common training and development objective, in Germany.
Within the following sections, we provide a brief overview of the literature, demonstrating that trainers are relevant on the one hand and understudied on the other. We then present the concept of accountability, Schlenker's (1997) triangle model of responsibility, and its application to trainers. We then propose our research questions and present our empirical study and findings. Finally, we discuss the potential for the optimisation of training based on the wider concept of accountability, and we describe the organisational implications of our study.

DETERMINANTS OF THE TRANSFER OF TRAINING
Although there has been criticism of the term transfer (Hager & Hodkinson, 2009), the concept itself has received much attention within training and development research, and there is a rich body of literature investigating the determinants of the transfer of training. Most of the scholarly work is based on the framework by Baldwin and Ford (1988), which proposes three areas of impact factors.
Concerning the first factor, trainee characteristics, findings suggest that transfer is related to trainees' cognitive ability, personality, self-efficacy, and motivation to learn and to transfer (Blume et al., 2010;Colquitt et al., 2000;Hughes et al., 2020).
The second category of transfer determinants refers to the training design. For example, the sequencing of practice units and the concept of overlearning are related to transfer (Donovan & Radosevich, 1999;Driskell et al., 1992). Furthermore, specific training methods such as behaviour | 3 modelling training and error management training can promote transfer depending on the training objectives (Arthur et al., 2003;Keith & Frese, 2008;Lacerenza et al., 2017;Taylor et al., 2005).
Finally, successful transfer depends on the trainee work environment including factors such as the support of colleagues, supervisors and the organisation (Blume et al., 2010;Colquitt et al., 2000;Hughes et al., 2020). Ford et al. (2018) provide a review of the current state of transfer research. Most of the empirical findings blend into the three areas of impact factors suggested by Baldwin and Ford (1988). Nevertheless, suggestions for adjustments have arisen. Specifically, Burke and Hutchins (2008) suggested adding trainers to the model. In their qualitative study, training practitioners identified several trainer characteristics as being relevant for successful transfer. A survey by Donovan and Darcy (2011) supports these findings. Furthermore, substantial empirical evidence from education research stresses the importance of teachers for teaching and learning quality and outcomes (Hattie, 2009). These findings also support the relevance of trainers for training outcomes.
Apart from these empirical indicators, it is highly plausible that trainers play an important role in the transfer of training. They usually choose and deploy instructional methods and have at least some say in the timing and format of training. They can also affect trainee characteristics such as self-efficacy or motivation (Gegenfurtner et al., 2016;Hughes et al., 2020). Moreover, trainers can consult organisations regarding the appropriate support for trainees in transferring skills to their workplace (Broad & Newstrom, 1992).

THE CONCEPT OF ACCOUNTABILITY
The concept of accountability was introduced to the science of training by Kopp (2006), Saks (2009), Tews andBurke-Smalley (2017) and Grossman and Burke-Smalley (2018). While Kopp (2006), from a practical perspective, called for trainers to be held more accountable with regard to transfer, Burke and colleagues sought to incorporate the concept of accountability and its theoretical elements within the science of training. Building on Schlenker's (1997) theory, they argue that including the concept of accountability can provide guidance for scholars and practitioners.
Although little research on accountability for transfer exists thus far, some empirical findings corroborate its relevance. Findings by Saks and Burke (2012) suggest that the frequency of training evaluation positively relates to transfer if the evaluation assesses behavioural changes or organisational results. This might be an effect of a higher level of accountability resulting from the knowledge that training interventions are evaluated. In a meta-analysis by Taylor et al. (2005), behaviour modelling training was more effective when successful transfer was reinforced and lack of transfer was sanctioned. Additionally, in an exploratory study, Kontoghiorghes (2002) found that trainees who are expected to use newly acquired skills at their jobs are more motivated to transfer. There are a few other studies linking the accountability of trainees to learning (Cheramie & Simmering, 2010), to their intention to transfer (Baldwin & Magjuka, 1991) and to actual transfer (Saks & Belcourt, 2006).
While those few existing studies have focused on the accountability of trainees, Kopp (2006) pointed out that the accountability concept is valid for trainers as well, and he stated that trainers are rarely held accountable for the degree of transfer. He argued that a trainer's responsibility does not end with the training itself. In other words, trainers should become engaged not only in the training but also in the entire transfer process and take ownership of its success. Similarly, Broad and Newstrom (1992) suggested that a trainer plays the role of a transfer manager, implying that | 5 Finally, the audience could be parties within organisations, such as the human resources department or company management (Burke & Saks, 2009). As Weinert (2001) pointed out, competence consists not only of knowledge and skills but also of the motivational and volitional readiness to act accordingly. Regarding the transfer of training, Burke and Saks (2009) suggested including trainers' accountability in scientific discussions. The premise is that trainers need to perceive themselves as accountable for transfer to promote it. Hence, the main goal of this study was to gain a deeper understanding of how trainers perceive themselves as accountable for the transfer of training. Therefore, the study empirically employs Schlenker's (1997) responsibility triangle. Specifically, the study aims to determine how tight the links between the three elements of responsibility are. That is, to what extent do trainers perceive clarity, ownership, and personal control regarding the transfer of training?

THE GOALS OF THE STUDY
Another goal was to identify other parties that the trainers regarded as accountable for transfer. Finally, the study investigates whether trainers perceive themselves as transfer managers, coordinating with other stakeholders, as described by Broad and Newstrom (1992 (Burke & Saks, 2009;Schlenker et al., 1994) | 3 213

Research design
The idea to investigate trainers' accountability in the transfer process is new, and the research objectives are rather exploratory. Because of the need to gain an in-depth understanding of trainers' perceived accountability for the transfer of training, we chose a qualitative approach. As our study represents a first attempt to investigate this subject, we wanted to give trainers the opportunity to express their thoughts in their own voice and from their own perspective without presenting them with predefined thought patterns (Alshenqeeti, 2014). Insch et al. (1997) stress that qualitative research reveals subjects' immediate interpretation and delineation of the topic of interest and thus often opens unconsidered avenues free of preconceived biases. These advantages are ensured by the use of open questions during data collection and the fact that new, inductive categories can be formed during data analysis.
Our choice of leadership trainers as constituting an initial sample to investigate trainers' accountability for transfer is based on research that suggests fostering transfer might be especially important for soft-skill training (Blume et al., 2010;Wisshak & Hochholdinger, 2020). Leadership training represents a typical soft-skill training objective, and its transfer success is highly relevant for individuals, teams and organisations.
We conducted 15 semi-structured interviews in which we asked leadership trainers about several aspects of the transfer of training (Barth & Hochholdinger, 2018). One question tapped trainers' perceived accountability for the transfer of training. For the present study, we analysed the answers to this question via qualitative content analysis (Mayring, 2014). Mayring's approach is widely used in German-speaking countries and is appropriate for text interpretation and analysis. Adopting this approach, the researcher follows clear rules and stepwise procedures to obtain final thematic categories that can consist-as is the case in our study-of a combination of a priori categories and inductive categories (Mayring, 2019).

Study participants
Regarding the demographics of the 15 study participants, 7 trainers were women, and 8 were men. On average, they were 50 years of age (minimum = 22; maximum = 76), and they had 18 years of experience as trainers (minimum = 7; maximum = 50).
The majority (11 trainers) had a higher education degree. Furthermore, all the trainers possessed additional non-formal qualifications such as train-the-trainer certificates (9) and certificates for coaching (10) and organisational development or management (5). Characteristically for trainers in Germany, the majority (12) of the trainers worked as freelancers while only 3 trainers were employed by training institutes.

Participant selection and recruitment
The participants were selected to obtain a purposive sample. As stated above, we chose leadership trainers to rule out an effect of different training topics. Within the population of leadership trainers, we intended to obtain a sample representing the maximum variation in terms of age, sex, and professional experience and to include both trainers who worked as freelancers and employed | 7 trainers (Patton, 1990). We halted data collection when we reached a point of saturation, meaning that no new aspects or further information emerged in the interviews (Glaser & Strauss, 1967).
The participants were recruited via their web pages, through a training institute, and via snowball selection. The first contact was via email or telephone, and the trainers were asked to participate in an interview study with no compensation given. They received elaborate information on the aims of the study, why we had chosen them as participants, what they could expect from the interview and how we would use the data.

Data collection
Data were collected via semi-structured interviews conducted face-to-face (11 interviews) in the offices of a German university or via telephone (4 interviews). A single person from the research team conducted the interviews, and other than the interviewee and the interviewer, no other persons were present.
The interviews had an average duration of 77 min (Min = 49, Max = 119) and were recorded as audio files and transcribed and anonymised for analysis.
The interviewer used a protocol that contained the interview questions. At the beginning of the interviews, the trainers were asked a few sociodemographic questions concerning their age, degree, biography, training content, target group and training formats. Then, we provided the trainers with the following definition of transfer: Transfer of training means that trainees permanently apply the knowledge and skills that they acquired during training at their workplace.
Then, the trainers were asked five open questions about transfer. One of these questions referred to their perceived accountability and was phrased as follows: In what way do you feel responsible regarding the transfer of training? Or, put differently, what role do you perceive for yourself regarding the transfer of training?
We based the question on the proposition to investigate trainers' perceived accountability by adopting Schlenker's (1997) triangle model, as described above. Nevertheless, the interviewer asked the open question without referring to the model and then engaged further only by actively listening, asking for examples or clarifications or leading the interviewee back to the topic; e.g., "Okay. Is there anything else regarding the topic of accountability…?" (A, 359).
The participants agreed in writing to the publication of the data and the results of the study. The anonymised transcripts are available at a repository of the Verbund Forschungsdaten Bildung (Barth & Wisshak, 2021).

Analysis
To answer the aforementioned research questions, we used qualitative content analysis by Mayring (2014). More specifically, we used the procedure that Mayring calls content structuring/theme analysis. This procedure aims to structure the material into an a priori coding scheme while still being able to add new categories that appear during analysis. It is guided by clear rules within a stepwise model with iterative loops during data analysis. The procedure can be described as a combination of conventional content analysis (deductive part) and directed content analysis (inductive part) (Hsieh & Shannon, 2005). Eventually, this approach allows for an additional quantitative step by analysing the frequencies of the coded segments within each category (Mayring, 2014).
We used Schlenker's (1997) triangle model as a coding scheme to differentiate between weak and strong linkages. One category was added inductively, namely, the moderate identity-event link. This category emerged during the analysis and is semantically located on the continuum between high and low control of the transfer process. The two categories for research questions 5 and 6 were also developed a priori based on the literature. See Table 1 for the categories and their definitions.
We analysed the interviews using MAXQDA 2020 software (VERBI Software, 2019), which allows the assignment of text passages to the categories of the coding scheme and the ability to comment on the coding procedure via memos. We coded the trainers' answers to the question about accountability, and we coded additional sections from the interviews if they referred to accountability.
Both of the study's authors coded the material. We agreed on the coding scheme, the category definitions, anchor examples for each category, and a set of rules for the coding process (Mayring, 2014). The smallest analytical unit was one sentence, and each unit could be coded only once. We optimised the coding scheme in several loops.
To calculate the interrater reliability, we coded 5 of the 15 interviews independently (Mayring, 2014). The agreement across all categories was 87.67 per cent. See Table 2 for the agreement rates for the single categories. The coefficient kappa, which is considered acceptable at a value greater than 0.60, was adjusted for random matches and was found to be 0.86 (Brennan & Prediger, 1981;Cohen, 1960;Landis & Koch, 1977). Due to the high level of agreement, it could be assumed that the coding scheme enabled a relatively clear assignment of the trainers' statements. Therefore, the remaining 10 transcripts were coded by one person. In total, 175 segments were coded.

FINDINGS
In this section, we describe the leadership trainers' perceptions of their accountability for the transfer of training as expressed within the interviews. For their anonymity, the trainers are referred to as trainers A, B, C, etc.

RQ 1: Transferability of the triangle model
In general, none of the trainers expressed that they considered the concept of accountability inappropriate or unimportant in the context of training transfer.
Good interrater agreement is a first indicator that the linkages from the model can be transferred to the trainers' responses and that they are mutually exclusive. As a second indicator, we examined whether each trainer addressed each link of the triangle model. The codings per trainer and category are presented in Table 3. Of the 15 trainers, 11 trainers addressed every link of Schlenker's (1997) model. Trainers H, K and N did not refer to the identity-event link, and Trainers M and N did not comment on the prescription-event link. It should be noted that within these interviews, very few units were coded (3 each in K and M and only 1 in N).

RQ 2: Prescription-event link
Overall, the trainers reported that goals and procedures were clear (40 codings) more often than they reported them as being unclear (11). They identified several transfer-enhancing measures to be taken before, during and after training.

| 9
Before training, they mentioned needs analysis (J), clarification of the terms of the training with the respective company (B), consulting clients on realistic training outcomes (F), and deciding what kind of transfer is intended (I). In this context, one trainer mentioned that he prefers to clarify responsibilities in this early stage (O). Similarly, trainer I emphasised that the promotion of transfer is already initiated in advance of the training. The trainer believes that he or she is capable of promoting transfer due to his or her abilities, skills, autonomy, or resources The trainer has control over the transfer process The trainer has high self-efficacy regarding transfer

Moderate identity-event link (Trainer as facilitator)
The trainer perceives himself as the facilitator, supporter, and initiator of transfer 7. Weak identity-event link (low control) The trainer does not believe that he or she is capable of promoting transfer due to his or her abilities, skills, autonomy, or resources The trainer has no or little control over the transfer process The trainer has low self-efficacy regarding transfer

Other responsible stakeholders
Other parties that trainers identify as being accountable for the transfer of training include the trainees, the organisation or the human resources department 9. Trainer as transfer manager The trainer works towards an organisation-wide focus on transfer and transfer strategies The trainer initiates a transfer partnership with managers and trainees and coordinates with those parties (Broad & Newstrom, 1992) Note: The coding scheme is based on the triangle model of responsibility from Schlenker et al. (1994). The inductive category (category 6) is displayed in italics.
Many of the reported transfer-enhancing strategies associated with accountability concerned the training itself including letting the participants define their transfer goals (O); building relationships with trainees (E); and using role-play (C), (peer) feedback (C), and learning journals (O). Furthermore, the trainers reported providing practice-oriented training (O) and working on realistic problems (C). They also made training success visible to promote transfer to the workplace (C).  Brennan and Prediger (1981).

| 11
Trainers also reported a few post-training interventions such as coaching, follow-ups and reminders (E; I).
The category weak goal and process clarity was coded 11 times. Two trainers admitted that their knowledge about transfer-enhancing strategies was limited (F; G). The other codes refer to questionable or conflicting goals.
For instance, one trainer reported that while the official company objective was to ensure that more women attain management positions, she was assigned to train women who had already obtained management positions. She thought that other measures would be more suitable for reaching the company goals using the metaphor of a " fig leaf". "[…] There I had the feeling that other measures could have been taken to get more women into leadership positions rather than training the women leaders" (D, 130).
Similarly, another trainer recounted that his assignment was to train participants on a more participative leadership style, although the organisation was strictly hierarchical. He reported that in such cases, he sometimes focused on alternative, hidden training goals such as how to successfully cope with the existing organisational culture. He referred to this strategy as "submarine" because the actual training goals are out of sight of the organisation (A, 226-234).
Despite being part of the a priori definition of the prescription-event link, consequences of (non-)successful transfer were not reported by the trainers.

RQ 3: Prescription-identity link
Most trainers who commented on this topic expressed a clear sense of ownership (44 codings).
Many trainers revealed that they feel a strong personal and sometimes also a moral obligation towards the training goals, especially towards the trainees: "To have responsibility for the person, to have responsibility for the goal and to work here with people from an ethical and moral point of view and to work on personalities when it comes to leadership issues" (B, 96). "I want people to be able to implement it. I have a personal ambition in there" (C, 370). The trainers also stated that their responsibility refers to the whole transfer process, not only the training itself: "So my responsibility as a trainer does not stop at the end of the training" (I, 342).
Trainer N explained that she would cut the branch she is sitting on if she did not ensure that transfer occurs. Trainer G mentioned that she once suggested to a company that she regularly works for that it should pay her depending on the success of her training. She reported feeling an especially strong obligation towards this company because it was her main client. With this client, she did not feel like an external trainer; instead, she almost felt as if she were part of the organisation. Therefore, she had a stronger sense of accountability in this setting than in other settings: "So I think sometimes it changes for me with one of my main clients in that I am no longer one hundred percent external but see myself as part of the company" (G, 222).
Only 11 statements from the trainers referred to a weak prescription-identity link, meaning that they felt a low obligation towards promoting transfer. Trainer G mentioned a tendency to feel less responsible if she does not have a relationship with a company: "If you are only external and the clients say 'Do [this or that] quickly for once…', then you can rather keep the attitude that 'I have tried, but…that was just a job for three or four days'" (234).
The trainers did not report being answerable to or accountable for transfer by the companies. Trainer O remembered being asked once 12 years ago what he did to promote transfer. Although this finding concerns the audience in Schlenker's (1997) accountability pyramid, we report it here because it is the only statement in this direction.

RQ 4: Identity-event link
Concerning the trainers' perceived control over the transfer of training, we differentiated three levels: strong (7 codings); moderate (the trainer as facilitator, 14 codings); and weak control (15 codings).
Only 7 statements expressed strong control: "Because I believe and I am convinced of myself, that I can change something there" (G, 223).
More trainers (8 trainers, 14 statements) expressed that trainers can facilitate transfer. We added this category inductively to the coding scheme. The expression stems from Trainer C: "As a facilitator, as they say today" (342). The trainers used keywords such as giving an "impetus" (I, 355) and creating the appropriate "framework" (J, 223) for transfer. One trainer emphasised that working with adults means that trainees must decide for themselves whether they want to use the acquired knowledge and skills (F, 304). He said that he perceived himself as being responsible for encouraging participants and for reflecting on transfer. Controlling transfer, on the other hand, was not his responsibility. Another trainer compared leadership training to a fitness centre (O). While it provides the necessary equipment, customers have to use that equipment and practice to build and maintain their muscles. The same trainer emphasised that in leadership training, there is no such thing as a one and final truth. Instead, he offers several alternatives, accompanies the trainees through a process, and reflects on the process with them.
In addition to their role as facilitators, the trainers expressed weak control over the transfer of training (15 codings). They highlighted their inability to produce transfer: "Well, so if I had the responsibility, I'd have a problem. Because […] this linkage is not what you might think of in engineering terms" (A, 215-216). "But I can't produce transfer; everyone has to do it themselves" (C, 347). Some trainers mentioned financial or organisational restrictions on the part of the companies (A; D; I). "But you are just not free in some settings. Sometimes customers […] say, for reasons of cost or organisation, you can only have what you have" (A, 219-220). Furthermore, the trainers described how they were powerless if trainees were unwilling to engage in the training (E) or if trainees decided to check their email during the training or leave the room for telephone calls (J).
[…] but basically-if you are quite honest-if someone doesn't feel like it, was sent to a training and the organisation doesn't call up the acquired knowledge, I can dance on the table and do whatever, but nothing happens at all. (D, 125)

RQ 5: Other responsible parties
Within the 26 coded segments that refer to other stakeholders, trainees were named 22 times, organisations or human resources departments 3 times, and trainees' supervisors once.
And of course, the responsibility for a transfer is ultimately with the participant, where else. So maybe a touch of the organisation: if it does little or nothing to ensure that the things that people experience there really do fall on fertile ground. (A, 221, 222) The trainers argued that the trainees were adults and that in adult education, learners are responsible for their own learning and transfer process (H; J). The trainers emphasised that | 13 their role differs from that of teachers (H, 113). One trainer mentioned that sometimes she reminded the trainees of how expensive training was and how high their companies' loss would be if it did not pay off (L).

RQ 6: Trainers as transfer managers
Only 7 segments were coded with the category trainer as transfer manager. Trainer A reported that he sometimes has to convince the organisation to provide the appropriate infrastructure for the transfer of training. Trainers D, I and J described how they try to include organisations' human resources professionals and trainees' supervisors in transfer-related considerations.

DISCUSSION
Effective training and development can help individuals, organisations, and society thrive (Aguinis & Kraiger, 2009). Nevertheless, training is not always successful, and scholars are still working to identify relevant levers for optimising transfer. As prior research has suggested that the interdisciplinary concept of accountability might be one of those levers, we took the next step in this direction by deploying the concept of accountability to trainers. Therefore, one of the central contributions of this study lies in the requested application of Schlenker's (1997) triangle model of responsibility to the transfer of training (Burke & Saks, 2009;Kopp, 2006). The current study indicates that the model is indeed transferable to this context. The differentiation among the three links can be applied to empirical data and can help dissect the complex construct of responsibility and support a better understanding of its premises. More specifically, we used the model to analyse trainers' perceptions of their accountability for the transfer of training. While some scholarly work has investigated the accountability of trainees, this is the first study to focus on trainers, which represents the second major contribution.
The third central contribution of the study lies in the investigation of the specific qualities and dimensions of the responsibility links. According to the 15 leadership trainers, the first link, namely, goal and procedural clarity, is rather strong, although the trainers mentioned a few constraints. The trainers reported using a variety of transfer-enhancing strategies, most of which are recommended by prior research such as providing (peer) feedback and promoting trainees' self-efficacy (Colquitt et al., 2000;Lacerenza et al., 2017). Nevertheless, two trainers expressed a lack of knowledge regarding the promotion of transfer. Furthermore, the trainers reported some goal ambiguity. They used metaphors such as a fig leaf or submarine to depict training when they considered the official objectives not to be useful or realistic. These are examples of organisational ambiguities and restrictions that, according to Schlenker (1997), can reduce personal responsibility and lead to inferior performance. The trainers in our study describe coping with such situations by prioritising their accountability towards their trainees over their accountability towards the client organisation.
Concerning the second link of Schlenker's triangle model, ownership, the trainers seemed to feel a strong obligation-especially towards the trainees-for promoting transfer.
Regarding the third link, namely, control over the transfer of training, we added the category trainer as facilitator, located between strong and weak control. The trainers emphasised that they could facilitate transfer only by making offers to the trainees such as reflecting on what they have learned and making transfer opportunities visible (C, O). In naming various transfer barriers on the side of the trainees and the workplace, such as involuntary training participation, lack of motivation to transfer, and insufficient time and opportunities to transfer, the trainers confirmed prior research findings (Gegenfurtner et al., 2016;Hughes et al., 2020). At the same time, the trainers' description of themselves as facilitators points to a constructivist understanding of training, which is well established in adult education theory (Knowles, 1975). Ultimately, the qualitative data do not allow for any conclusions as to why some trainers perceived their control as lower than others. One possible explanation could be their different learning biographies that seem to result in differing training approaches. Trainer A, who expressed doubts about his control, argued from a systemic point of view that trainers can give an impulse into a system, but they never know what it will lead to in the end. Trainer J, who similarly showed a weak identity-event link, never attended a train-the-trainer program. As a trained psychotherapist, his main approach to training was transactional analysis, an approach that stems from a form of psychotherapy and is related to psychoanalysis (Berne, 2016). Trainer B on the other hand, who has a background in banking and sales, showed a rather strong identity-event link. She seemed to pursue a more behaviouristic approach, where she manipulates trainees by breaking into their comfort zones (298). She stressed that raising awareness was not enough for her, but that she was determined to bring about lasting change.
Summarising the contributions of this paper to the disciplinary discussion, our study elaborates on prior research in several ways. It complements theoretical and empirical research regarding the accountability of trainees by adding trainers' perspectives on their accountability (Saks & Belcourt, 2006;Taylor et al., 2005). The study generally supports the proposition by Grossman and Burke-Smalley (2018) to include an accountability framework in training research while elaborating on that proposition by focusing on trainers.

Limitations
The study provides new and exploratory empirical insights while being embedded in an established theoretical framework. Nevertheless, it has some limitations. First, the qualitative approach resulted in a small and non-representative sample.
Furthermore, we focused on leadership training, where the promotion of transfer is particularly important. Readers should keep this in mind when generalising the findings to other training objectives since the trainers' perceived accountability could be different. After all, prior research suggests that the skill type taught during training functions as a moderator between certain input factors and the transfer of training (Blume et al., 2010;Grossman & Burke-Smalley, 2018).
Moreover, the sample did not include in-house trainers, meaning that accountability might be underrated in our study. Whether a trainer is employed by a company might be related to her accountability and her role as a transfer manager. The finding by Freitas et al. (2017) that job resources such as autonomy and support within an organisation can affect in-house trainers' felt responsibility supports this assumption. Therefore, future studies should investigate in-house and external trainers and potential differences in their accountability regarding transfer.
Finally, our study was conducted in only one country. As Ashton (2004) is pointing out, workplace learning is organised very differently, depending on work organisation and institutional frameworks in the countries in question. In Germany, initial vocational education and training is covered comprehensively by the successful dual system and full-time vocational schools. The

| 15
continuing vocational training sector however, is very diverse and barely regulated, as are the qualifications of the trainers working within it. This might limit the generalizability of our findings to other national contexts. Thus, future research should take such national circumstances into account.
For the aforementioned reasons, the scope of the transferability of our findings is limited.

Implications for future research and practice
Despite its limitations, the study has significant implications for future research and practice. One implication for future research is that Schlenker's triangle model of responsibility can be successfully adapted to trainers. Building on this, future studies can measure the perceived accountability for a larger sample of trainers and relate it to external criteria such as transfer success. Burke and Saks (2009) proposed some example items for measuring the accountability of trainees, trainers and supervisors. Our findings can be further utilised by future research focusing more strongly on trainers as relevant transfer facilitators. In the literature review of this article, we argued that trainers play an important role in the quality and effectiveness of training but are often overlooked within the transfer literature. The trainers in our study confirmed that they are relevant agents in the promotion of transfer. Several trainers perceive themselves as facilitators or catalysts of transfer, implying that their transfer-enhancing actions are an important prerequisite. The question why some trainers are more convinced about their control over the transfer process than others could be an interesting starting point for future research. The answers could provide us with ways of helping trainers to facilitate transfer in the future.
In addition to these suggestions for future research, the study has several implications for training practice. Although the trainers reported feeling accountable for promoting transfer regarding the design and delivery of training, there seem to be some organisational aspects that could be optimised. Lack of alignment of training objectives with the organisational situation and trainees' needs seems to result in role and goal ambiguity and diminished accountability. The trainers seem to switch to alternative training objectives in such cases without coordinating with the organisations. Regarding the empirical evidence on transfer success, we recommend that training needs be carefully assessed (Burke & Hutchins, 2007;Colquitt et al., 2000). Training programmes should not only be thoroughly aligned with organisational goals but should also consider trainees' current and future work environments. Some might argue that this is especially the case for e.g., personal or professional development measures where training goals are not necessarily derived from organisational objectives (van der Sluis, 2007).
Trainers, for their part, could communicate problems back to organisations, consulting them on transfer-enhancing strategies and acting as transfer managers (Broad & Newstrom, 1992). Both trainers and responsible parties in companies could expedite the involvement of trainers in strategic decisions and the building of partnerships among trainers, executives and trainees. Based on such collaboration, it should be possible to conduct the aforementioned needs analyses and joint strategic training planning. The recently accelerated digitalisation process might facilitate a change in this direction. Web-based training bears the opportunity to provide more flexible, demand-oriented learning opportunities that can be closely linked to individual and organisational development. For example, it is easier to provide follow-up and additional coaching services. If trainers and organisations seize this opportunity to enter into closer coordination, it might lead to better training outcomes.
Another noteworthy finding of our study is that the client organisations, representing the audience in Schlenkers' model, did not seem to have any transfer expectations of the trainers. Consequently, the trainers did not report any consequences for (non-)successful transfer. Therefore, the audience's potential influence on the trainers' accountability as it is intended in the model did not show in the data. This finding is in line with Kopp's (2006) observation that trainers are usually not held accountable for the transfer of training in organisations, and with the fact that transfer is rarely evaluated (Saks & Burke, 2012). However, according to the review by Hall et al. (2017), accountability requires at least the possibility of evaluation and associated consequences. Similarly, Burke and Saks (2009) stated that for the training and development field, a prerequisite for establishing accountability mechanisms across all stakeholders is the evaluation of transfer criteria. Consequently, organisations should collect reliable data about the transfer of training to connect the responsible parties to transfer outcomes. At the same time, regular feedback on their (non-)successes could be a valuable way for trainers to develop their transfer-promoting strategies, professional competence, and self-efficacy.

CONCLUSION
In summary, this article provides an empirical contribution to the long-requested research on accountability for the transfer of training. As Burke and Saks (2009) and Kopp (2006) expected, our qualitative data suggest that the accountability of trainers might have some potential as an explanatory variable for the (non-)success of transfer and for the optimisation of training success in the future. With regard to leadership trainers, we can conclude that they are aware of transferenhancing measures and are committed to using them. However, cooperation between trainers and organisations seems to be improvable in promoting accountability in all parties and-as a consequence-the transfer of training.
We hope that our study will inspire scholars and practitioners to increasingly incorporate both the concept of accountability and the role of trainers in further discussions about the transfer of training.