Transfer of Training - Human Resources/Psychology

art_5_grohmann.pdf

Home >Human Resource Management homework help >Transfer of Training - Human Resources/Psychology

Exploring the critical role of motivation to transfer in the training

transfer process

Anna Grohmann, Johannes Beller and Simone Kauffeld

The present study aims at exploring the critical role of motiva- tion to transfer within the training transfer process. In a sample of N = 252 employees of one industrial company, one peer rating and several self-ratings of transfer were used to investigate the mediating role of motivation to transfer in the relationship between training characteristics and training transfer. We furthermore used an online survey with N = 391 employees from different branches to corroborate the first analysis in a more diverse sample and to explore differential effects of motivation to transfer. The participants’ motivation to transfer was suc- cessfully identified as a mediational link between training char- acteristics and transfer. Moreover, quantile regression revealed that the positive effects of motivation to transfer on training transfer differ across quantiles. Understanding the role of moti- vation to transfer within the transfer process in greater detail helps to identify crucial parameters for successful transfer. Because motivation to transfer was found to be a linking mecha- nism between training characteristics and transfer, training professionals should enhance it by means of different instruc- tional methods. Quantile regression results point out that motivation to transfer should especially be emphasized and monitored depending on characteristics of the training par- ticipants. The present study answers the call for research on

❒ Anna Grohmann, Former Research Associate, Department of Industrial, Organizational and Social Psychology, Technische Universität Braunschweig, Germany. Email: a.grohmann@tu-braunschweig .de. Johannes Beller, Graduate Student, Technische Universität Braunschweig, Germany. Email: [email protected]. Simone Kauffeld, Professor, Department of Industrial, Organizational and Social Psychology, Technische Universität Braunschweig, Germany. Email: s.kauffeld@tu- braunschweig.de We would like to thank Renate Wirth for her efforts collecting data for Study 1. Earlier versions of this paper were presented at the 2012 congress of the German Psychological Society in Bielefeld, Germany, and at the 2012 conference of the European Association for Research on Learning and Instruction, Special Interest Group 14 in Antwerp, Belgium.

bs_bs_banner

International Journal of Training and Development 18:2 ISSN 1360-3736 doi: 10.1111/ijtd.12030

84 International Journal of Training and Development

mailto:[email protected]

the mediating role of motivation to transfer in the transfer process. This is one of the first studies that successfully applied quantile regression to training transfer research, suggesting new directions for future research.

Introduction Organizations invest considerable amounts of resources into enhancing their employees’ knowledge and skills (e.g. Aguinis & Kraiger, 2009). This effort pays off only when employees actually transfer the contents learned into practice (e.g. Hutchins et al., 2010). But what exactly can organizations do to promote high training transfer?

Previous research has identified training, work environment and trainee character- istics as major determinants for successful training transfer (Baldwin & Ford, 1988; Grossman & Salas, 2011).One important mediating factor that has been proposed in several theoretical training transfer models is motivation to transfer (Gegenfurtner et al., 2009, and references therein). However, to date, empirical papers examining the mediational relationship of the easily modifiable training characteristics, motivation to transfer and training transfer in the work setting are scarce (Gegenfurtner et al., 2009).

To explore the central role of motivation to transfer in the transfer process in more detail, we examine the relationship between training characteristics and different measures of training transfer via motivation to transfer in the work setting in two studies. We extend previous research by examining the mediational role of motivation to transfer between training characteristics and different measures of training transfer by means of path analysis, by using two measurement sources in Study 1 and by surveying diverse branches in Study 2. Because previous research has shown hetero- geneous results concerning the relationship between motivation to transfer and train- ing transfer (for an overview, see Gegenfurtner et al., 2009), we additionally use quantile regression (Koenker & Bassett, 1978) to examine the nature of these hetero- geneous results in our more diverse sample in Study 2.

Improving training transfer We refer to training transfer as changes in on-the-job behavior which describe if an individual applied what he or she learned in training to the actual work setting (Kraiger, 2002). According to Kraiger (2002), changes in on-the-job behavior address the question of whether behavior at work is different (e.g. occurs more often) after training. On-the-job behavior changes have to be distinguished from performance effectiveness, which refers to improved job performance resulting from participation in training (Kraiger, 2002). Because ‘evidence that trainees have applied the behaviors learned in training back on the job provides the most straightforward assessment of transfer’ (Kraiger, 2002, p. 361), we focus on transfer as on-the-job behavior change.

Drawing upon Baldwin and Ford’s (1988) seminal review of the transfer process, numerous studies have examined ways to facilitate the transfer of knowledge and skills from training programs to the job context (for a recent overview, see Grossman & Salas, 2011). Motivation to transfer, for example, has been shown to predict transfer inten- tions (Machin & Fogarty, 1997) and is generally assumed to be a central variable in the transfer process (e.g. Gegenfurtner et al., 2009; Grossman & Salas, 2011). According to a recent meta-analysis by Blume et al. (2010) that identified several important factors limiting and fostering training transfer, the empirically best-supported trainee charac- teristic which predicts transfer seems to be a trainee’s cognitive ability. However, cognitive abilities have been found to be more important in laboratory studies than in field contexts (Blume et al., 2010). Moreover, trainers cannot affect a trainee’s cognitive ability because there are, for example, organizational constraints in choosing employees with specific individual characteristics for participation in training (Blume et al., 2010). The same argument applies to factors related to the work environment (e.g. supervisor support), which have also been found to influence transfer (for an overview, see Grossman & Salas, 2011), but can only be modified with serious effort by training

professionals. By contrast, training-related characteristics, which demonstrably influ- ence training transfer as well (for an overview, see Bhatti & Kaur, 2010; Grossman & Salas, 2011), can directly be adapted by training professionals (e.g. Kauffeld & Lehmann-Willenbrock, 2010).

Two training-related characteristics that may impact training transfer are perceived content validity and transfer design (Holton, 2005). Both have been studied extensively in previous research (e.g. Holton et al., 2000; Kauffeld et al., 2008; for an overview, see Bhatti & Kaur, 2010). Perceived content validity is defined as ‘the extent to which trainees judge training content to reflect job requirements accurately’ (Holton et al., 2000, p. 345). Baldwin and Ford (1988) have underscored the importance of the prin- ciple of identical elements for training transfer, according to which a high similarity between training characteristics and practice improves transfer. Further studies show empirical evidence for this claim in organizational training settings. In a study on N = 569 employees from diverse organizations in Germany, perceived content validity was significantly related to training transfer (Bates et al., 2007). In a recent review, Bhatti and Kaur (2010) suggest that perceived content validity is central to the training transfer process in that it improves positive reactions towards the training and enhances the trainees’ self-efficacy. They point out that trainers should emphasize perceived content validity in order to enable high transfer (Bhatti & Kaur, 2010). That is, conceptually, the training should resemble the conditions which the participants face at work.

Transfer design refers to the extent ‘to which (1) training has been designed and delivered to give trainees the ability to transfer learning to the job, and (2) training instructions match job requirements’ (Holton et al., 2000, p. 345). Thus, transfer design focuses on the types of methods and procedures used by training professionals (e.g. training interventions) and investigates how helpful these are to the participants when they try to apply contents of the training on the job (Holton et al., 2000). By contrast, perceived content validity refers to the similarity between the training and the work- place, for example, with respect to the materials or tools used (see Holton et al., 2000). There are numerous studies exemplifying the positive effects of transfer design on transfer outcomes (for a recent review, see Grossman & Salas, 2011). For example, a study of 182 employees in a grocery company found that transfer design was signifi- cantly correlated with transfer (Velada et al., 2007). An additional stepwise regression showed that transfer design remained significantly predictive of training transfer for all steps (Velada et al., 2007). Regarding training interventions, a study by Lee and Kahnweiler (2000) examined the effect of a mastery learning intervention (e.g. utilizing self-directed feedback) in a total sample of N = 130 Navy recruits and showed positive effects on participants’ performance in a work-related task (for another example on this issue, see, e.g. Weissbein et al., 2011). In sum, perceived content validity and transfer design are among the most important training characteristics influencing transfer because these two factors facilitate training transfer and can most easily be changed by the trainer.

Training characteristics and training transfer: the mediational role of motivation to transfer

Motivation to transfer is referred to ‘as the trainees’ desire to use the knowledge and skills mastered in the training program on the job’ (Noe, 1986, p. 743). In a recent review, Gegenfurtner et al. (2009) proposed a comprehensive model of motivation to transfer, its determinants and its outcomes. In line with Baldwin and Ford’s (1988) transfer model, the determinants are grouped into factors related to the individual, organizational context and training (Gegenfurtner et al., 2009). On the training level, the characteristics, such as design and content validity, can substantially contribute to the participants’ motivation to transfer (Gegenfurtner et al., 2009; Kauffeld et al., 2008). An examination of 180 qualitative reports on trainees’ personal transfer experiences revealed that the participants balance their training experiences, such as explanations

given, against job requirements during training in order to develop their transfer intentions (Yelon et al., 2004). Furthermore, studies have shown that training interven- tions (e.g. reflecting on positive experiences with the training program) can have an impact on participants’ motivation to transfer (e.g. Kastenmüller et al., 2012).

Motivation to transfer was found to be the most important predictor of training transfer when compared with other individual, organizational and training-related variables (e.g. Bates et al., 2007; Kauffeld et al., 2008). If a person is not motivated, he or she might simply choose not to apply the newly learned skills into practice (Baldwin & Ford, 1988). Highly motivated individuals, however, will actively strive for possibilities to transfer what they have learned in training into practice (Gegenfurtner et al., 2009). In sum, it is plausible that motivation to transfer mediates the relationship between training characteristics and training transfer (e.g. Gegenfurtner et al., 2009; Kauffeld et al., 2008). This also implies that the effort put into creating stimulating training programs may be in vain if the trainers neglect the trainees’ motivation to transfer.

In line with these considerations, there are various theoretical frameworks which posit that motivation to transfer plays a central mediational role in the transfer process (e.g. Gegenfurtner et al., 2009; Holton, 2005; Noe, 1986). But despite this central role, papers empirically investigating the mediational model of training characteristics, motivation to transfer and transfer itself in the work setting are scarce (Gegenfurtner et al., 2009) and provide mixed support. As one of the few empirical studies, Liebermann and Hoffmann (2008) assume a linking role of motivation to transfer between training characteristics (i.e. opportunities for practice directly in the training and practical relevance) and training transfer but could not establish the hypothesized relationship between training characteristics and motivation to transfer. By contrast, a study by Van der Locht et al. (2013) supports motivation to transfer as a mediational link between the training characteristic identical elements (which corresponds to per- ceived content validity) and one transfer measure in a study that focused on a soft skill management program. Moreover, a study by Nijman et al. (2006) supports a mediational relationship of training characteristics (using one measure that covered different design aspects), motivation to transfer and training transfer for self-rated but not for supervisor-rated data, but only relies on separate regression analyses to examine this mediational model.

The present research Hypotheses

Therefore, the present paper explores whether motivation to transfer mediates the relationship between training characteristics (transfer design and perceived content validity) and different measures of transfer in the field by means of path analysis. We hypothesize that:

Hypothesis 1: The relationship between transfer design and different measures of training transfer is mediated by motivation to transfer.

Hypothesis 2: The relationship between perceived content validity and different measures of training transfer is mediated by motivation to transfer.

A comparison of the studies investigating the effects of motivation to transfer on training transfer shows a large discrepancy in the relationships reported: As Gegenfurtner et al. (2009) note, this ‘wide amplitude of findings, ranging from r = 0.04 (Tziner et al., 1991) to r = 0.63 (Machin & Fogarty, 1997), suggests that this relationship needs further elaboration’ (p. 414). We use quantile regression (Koenker & Bassett, 1978) as an explorary tool to obtain a better understanding of the heterogeneous effects of motivation to transfer on training transfer. Quantile regression is an analytic tech- nique which is extensively used in the field of econometrics and has also been applied to the investigation of training outcomes from an economic point of view (see, e.g., Arulampalam et al., 2010). By means of quantile regression, it is possible not only to calculate the effect of motivation to transfer on the mean of the transfer outcomes, but

also to study the effects of motivation to transfer on the whole distribution of the respective dependent variables. Thus, in the present paper, we explore whether the effects of motivation to transfer are heterogeneous across the distribution of transfer outcomes. Thereby we determine if, for example, the effect of motivation to transfer differs in case of highly transferring or lowly transferring individuals.

The use of quantile regression is beneficial to training transfer research for several reasons: (1) It is of interest for the practitioner to know how the effect of motivation to transfer varies for employees showing different amounts of transfer: Which employees benefit the most from motivation to transfer? Such in-depth knowledge helps trainers adapt their training programs to the trainees. (2) It is of theoretical interest to establish whether the effects of motivation to transfer differ across the response distribution of the transfer outcomes, i.e. if the regression slopes are unequal. This possible finding might, in part, explain the heterogeneity observed in the literature on motivation to transfer (see Gegenfurtner et al., 2009). (3) Transfer outcomes are likely not to be normally distributed with homogeneous variances (Beller & Baier, 2013; Micceri, 1989). Quantile regression, on the other hand, does not assume a normal distribution for transfer outcomes and is therefore reliable even in case of non-normality (Koenker, 2005; for more details on the advantages of quantile regression, see Beller & Baier, 2013). For the reasons outlined above (e.g. sub-sample specific effects, non-normal data, heterogeneous effects of motivation to transfer reported in the literature), we conse- quently presume:

Hypothesis 3: The effects of motivation to transfer will differ for the quantiles of the respective transfer outcomes. Therefore, it is assumed that the slopes are not equal across the distribution of the dependent variables.

The studies

In Study 1, we used three self-ratings and one peer rating of training transfer to investigate the mediating role of motivation to transfer between training characteristics and training transfer in a medium-sized company. In Study 2, we conducted an online survey to examine whether our results from Study 1 hold in a more diverse sample with employees from different branches. Additionally, we chose to exploratory conduct quantile regression for our more diverse sample in Study 2 to investigate the relation- ship between motivation to transfer and training transfer in more detail.

Method Sample and design – Study 1

A final sample of N = 252 employees of a medium-sized industrial company in Germany participated in Study 1. The respondents were asked to evaluate a specific professional training program they had attended (e.g. courses dealing with engine functionality or interpersonal skills). For our retrospective study, we considered only professional training courses that dated back at least one month and at most 24 months, the average being 7.3 months. This time frame allows employees to transfer training contents to practice while reducing potential memory bias (for a detailed explanation, see Grohmann & Kauffeld, 2013). Moreover, we excluded data from participants that did not respond on either of the study variables or participated more than once from our final sample. The average participants’ age was 37.7 years (SD = 10.7, 1.6% not specified) with a minimum of 16 years and a maximum of 63 years. Most participants were male (75.4% male, 23.4% female, 1.2% not specified), which is common for that industrial sector. Average tenure in terms of being employed in a specific function was 8.3 years (SD = 8.3, 1.6% not specified). Type of training content was rated by the test supervisor. 52.8% of the trainings dealt with closed skills (e.g. software training pro- grams) and 47.2% dealt with open skills (e.g. interpersonal skills). Additionally, we included a peer rating of the participants’ training transfer. Each training participant had to give one peer rating questionnaire to a colleague (for response rates, see Table 1).

Sample and design – Study 2

In Study 2, we conducted a cross-sectional online survey resulting in a final con- venience sample of N = 391 employees from different branches (e.g. service sector, manufacturing, administration, public health and education) in Germany. In our final sample, we included only employees that participated seriously in the online study according to their own statement, who reported on training that dated back between one and 24 months, the average being 7.4 months (see also Study 1) and excluded participants that rated more than one professional training course. The final data set contained a wide range of training contents (e.g. training programs on programming techniques or dealing with personality and charisma). We used a 3-point answering scale so respondents were able to rate the type of training content in a more differen- tiated manner (cf. Grohmann & Kauffeld, 2013). Fifty-seven per cent of the trainings dealt with closed skills, 20.2% focused on both open and closed skills and 21.8% dealt with open skills (1% not specified). The participants were 36.3 years of age on average (SD = 12.1, 0.8% not specified), ranging from 19 to 63 years. The gender ratio was well-balanced (47.1% female, 52.9% male). Organizational tenure at the time of the training was 8.3 years on average (SD = 9.0, 1.3% not specified).

Measures – Study 1

In Study 1, transfer design, perceived content validity and motivation to transfer were measured with the respective scales of the German version (GLTSI; Kauffeld et al., 2008) of the Learning Transfer System Inventory (LTSI; Holton et al., 2000). For all scales, a five-point answering format was used, which ranged from 1 = strongly disagree to 5 = strongly agree (Holton et al., 2000). Moreover, we stated some of the GLTSI items in past tense which was necessary because we conducted a retrospective study. We chose a more general and perception-based measure of transfer design to adequately cover the diverse training courses in our study. A sample item from the transfer design scale, comprising altogether four items, is: ‘The activities and exercises the trainers used helped me know how to apply my learning on the job’ (Holton et al., 2000; Kauffeld et al., 2008). The perceived content validity scale com- prised five items. A sample item is: ‘What is taught in training closely matches my job requirements’ (Holton et al., 2000; Kauffeld et al., 2008). A sample item of the moti- vation to transfer scale, comprising four items, is: ‘I get excited when I think about

Table 1: Valid responses, missing data, means and standard deviations in Studies 1 and 2

Scale Study Valid responses

Percentage of missing

data

M SD

Transfer design 1 238 5.6 3.19 1.03 2 390 0.3 3.69 0.80

Perceived content validity 1 239 5.2 2.93 0.97 2 390 0.3 3.56 0.87

Motivation to transfer 1 241 4.4 2.99 1.11 2 391 0 3.43 0.88

Perceived application to practice 1 250 0.8 4.82 2.84 2 391 0 6.15 2.65

Transfer quantity 1 186 26.2 1.75 1.85 2 374 4.3 2.23 2.24

Transfer quality 1 182 27.8 4.46 3.86 2 373 4.6 5.71 3.65

Peer rating: perceived application to practice

1 128 49.2 4.18 2.98

trying to use my new learning on my job’ (Holton et al., 2000; Kauffeld et al., 2008). Investigation of item wording of the motivation to transfer scale revealed that one item did not match the definition by Holton et al. (2000), as it was adapted to past tense. Therefore, we excluded this item from the motivation to transfer scale in our analyses. Perceived application to practice was measured with a scale from an initial version (Kauffeld et al., 2009) of the Questionnaire for Professional Training Evalu- ation (Q4TE, Grohmann & Kauffeld, 2013). This scale consists of two items and uses an 11-point answering format from 0% = completely disagree to 100% = completely agree. We recoded the perceived application to practice scale in values ranging from 0 to 10 (e.g. 0% in 0, and 100% in 10) in order to get the same metric of each depend- ent variable. The items are: ‘I use the knowledge and skills learned in the training in my everyday work’ and ‘Due to the training, my behavior at work has changed’ (Kauffeld et al., 2009). To be more precise in measuring transfer, we included the more behavior-oriented transfer measures transfer quantity and transfer quality (Kauffeld et al., 2008). According to Kauffeld et al. (2008), transfer quantity describes the number of steps implemented on the job after the training. Respondents could report up to ten steps in an open response format. Examples of these steps for a presentation training are: designing power point slides, holding a presentation in front of an audience and dealing with disruptive persons. We chose a maximum of ten steps for our study because previous studies had proved this a feasible number (e.g. Kauffeld et al., 2009). Transfer quality, on the other hand, was measured on an 11-point scale ranging from 0% = unsuccessful to 100% = very successful as the degree of implementation reached for each individual step (Kauffeld et al., 2008). Respondents were asked to evaluate how successfully they managed to apply each of the training contents to the job (Kauffeld et al., 2008). To scale this variable in the same way as all dependent variables, we recoded each value (e.g. 0% in 0, and 100% in 10); thus, values ranged between 0 and 10. According to Kauffeld et al. (2008), the mean value of all of these ratings is referred to as transfer quality and addresses the degree of success in applying the training contents into practice. In addition to self- ratings, we included one peer rating of the perceived application to practice scale (Kauffeld et al., 2009). The peer ratings of transfer quantity and transfer quality were not included in our analyses because they appeared to be too specific to provide a valid assessment as not all peer raters had necessarily participated in the same train- ing course as the participants.

Cronbach’s alpha values and intercorrelations were calculated with SPSS and are shown in Table 2. All scales had good reliabilities (α ≥ 0.80) or at least a satisfactory value as in the case of the self-rating of the perceived application to practice scale (α = 0.72, see Table 2).

Measures – Study 2

Just as in Study 1, we measured transfer design, perceived content validity and moti- vation to transfer with the respective scales from the GLTSI (Kauffeld et al., 2008). Perceived application to practice was measured with a scale from the final version of the Q4TE (Grohmann & Kauffeld, 2013). The respective items are ‘In my everyday work, I often use the knowledge I gained in the training’ and ‘I successfully manage to apply the training contents in my everyday work’ (Grohmann & Kauffeld, 2013). Furthermore, we measured transfer quantity and transfer quality (Kauffeld et al., 2008; see also Study 1). Cronbach’s alpha values and intercorrelations of all scales are presented in Table 3. All scales demonstrated good reliabilities (α ≥ 0.80).

Data analysis – Study 1

Path analysis was used to examine the hypothesized mediating effects of motivation to transfer on the relationship between training characteristics and training transfer (see Figure 1). Analyses were run with Mplus 6.1 (Muthén & Muthén, 1998–2010), and the

Table 2: Intercorrelations and Cronbach’s α values for all study variables as well as age, gender, tenure, course duration and type of training content in Study 1

Scale 1 2 3 4 5 6 7

(1) Transfer design (0.90) (2) Perceived content

validity 0.75** (0.88)

(3) Motivation to transfer

0.73** 0.63** (0.88)

(4) Perceived application to practice

0.58** 0.50** 0.70** (0.72)

(5) Transfer quantity 0.43** 0.28** 0.42** 0.36** (a) (6) Transfer quality 0.47** 0.37** 0.49** 0.41** 0.57** (a) (7) Peer rating:

perceived application to practice

0.38** 0.45** 0.48** 0.62** 0.22* 0.25* (0.80)

(8) Age −0.02 −0.01 0.02 0.08 −0.07 0.01 −0.01 (9) Genderb 0.10 0.13* 0.00 0.00 0.04 −0.10 −0.06

(10) Tenure −0.12 −0.13* −0.08 −0.05 −0.19* −0.10 −0.09 (11) Course duration 0.35** 0.20** 0.33** 0.23** 0.48** 0.36** −0.00 (12) Type of trainingc 0.16* 0.07 0.18** 0.14* 0.44** 0.27** 0.12

Table 3: Intercorrelations and Cronbach’s α values for all study variables as well as age, gender, tenure, course duration and type of training content in Study 2

Scale 1 2 3 4 5 6

(1) Transfer design (0.85) (2) Perceived content validity 0.76** (0.89) (3) Motivation to transfer 0.69** 0.65** (0.80) (4) Perceived application to

practice 0.62** 0.64** 0.67** (0.87)

(5) Transfer quantity 0.33** 0.27** 0.38** 0.30** (a) (6) Transfer quality 0.48** 0.48** 0.50** 0.55** 0.49** (a) (7) Age −0.04 −0.02 −0.04 0.03 0.21** 0.09 (8) Genderb 0.03 0.02 −0.05 −0.01 −0.07 −0.02 (9) Tenure −0.01 −0.02 −0.06 0.02 0.18** 0.06

(10) Course duration 0.09 0.04 0.11* 0.11* 0.12* 0.06 (11) Type of trainingc,d −0.09* −0.19** −0.06 −0.16** 0.00 −0.03

* p < 0.05, ** p < 0.01 (two-tailed). Cronbach’s α values are included in parentheses. a No Cronbach’s α values were calculated for transfer quantity and quality (one item each). b 1 = female and 2 = male. c 1 = closed skills, 2 = both (open as well as closed skills) and 3 = open skills. d Kendall’s τ is reported because type of training content is ordinal.

maximum likelihood (ML) estimator was applied. Due to the small sample size, we preferred path analysis over structural equation modeling to reduce model complexity while investigating all relationships between study variables simultaneously. Age, gender, tenure, course duration and type of training content were taken into account as covariates for the following reasons: Older employees have been found to be less motivated to learn (Colquitt et al., 2000), which might subsequently influence their motivation to transfer as well as the training transfer itself. Some previous studies have found support for gender effects on knowledge acquisition (for examples, see Colquitt et al., 2000), which may indicate potential gender effects in training transfer as well. Moreover, employees with more job experience (i.e. a higher tenure) were found not to improve as much after training as employees with less job experience (Warr & Bunce, 1995). Thus, tenure may influence successful training transfer. Previous studies with varying course duration included this variable as a covariate (e.g. Bell et al., 2011). In line with these studies, we assume that course duration (e.g. one-day training versus four-week training) may influence the amount of what can be transferred to practice. Finally, Blume et al. (2010) have identified differences in predictor-transfer relationships between training programs that deal with open skills (e.g. communication training) and those that deal with closed skills (e.g. software training).

Our empirical analyses showed no significant relationship between gender or age and the dependent study variables (see Table 2). In Study 1, there were some significant correlations between the remaining covariates and our dependent variables (see Table 2). We therefore included tenure, course duration and type of training content as covariates in our path model in Study 1 (results not depicted). The specified path model (see Figure 1) comprised two independent variables (transfer design and perceived content validity), one mediating variable (motivation to transfer), four dependent vari- ables (self-ratings of perceived application to practice, transfer quantity and transfer quality as well as a peer rating of perceived application to practice) and three covariates (tenure, course duration and type of training content). To account for the theoretical overlap of the dependent variables, we allowed for residual covariances among our dependent variables, which is the Mplus default for the present path model (Muthén & Muthén, 1998–2010; see Figure 1). Moreover, we used the Mplus default of full information maximum likelihood (FIML) to deal with missing data (see also Table 1; Muthén & Muthén, 1998–2010). As we did not include any model constraints, the path model is saturated (e.g. Preacher et al., 2010).

Transfer design

Perceived content validity

Transfer quantity

Transfer quality

Perceived application to practice

Peer rating: Perceived application to practice

Motivation to transfer

All direct effects from transfer design and perceived content validity on the dependent variables were specified, but are not depicted for sake of clarity.

These covariates were included in both studies in each regression equation of the mediational model (results not depicted for clarity): tenure , course duration and type of training content. In Study 2, we additionally included age as a covariate.

Figure 1: Path model for the mediational relationship of training characteristics, motivation to transfer and transfer outcomes. In Study 1, the entire model was tested. In Study 2, the

peer rating was not included in addition to self-ratings of transfer outcomes.

Bootstrapping offers multiple advantages over traditional mediation analysis approaches (for an overview, see Preacher & Hayes, 2008). Therefore, all path analyses were based on N = 1000 bias-corrected bootstrap samples in Mplus (MacKinnon et al., 2004). In line with Preacher et al. (2010), we estimated 90% confidence intervals (CI), which correspond to a one-tailed test with α = 5%. In our mediation analysis section, we report on the total effect c, the indirect effect ab and the direct effect c’ (e.g. Preacher & Hayes, 2008).

Data analysis – Study 2

Path analysis As in Study 1 we used Mplus 6.1 (Muthén & Muthén, 1998–2010) to analyze the hypothesized mediating effects of motivation to transfer. Moreover, age, gender, tenure, course duration and type of training content were taken into account as covariates. We found no gender effects. However, age, tenure, course duration and type of training content showed some significant relationships with our dependent variables (see Table 3) and were therefore included as covariates (results not depicted). We ran a saturated path model (see Figure 1) with two independent variables (transfer design and perceived content validity), motivation to transfer as a mediating variable, three depend- ent variables (perceived application to practice, transfer quantity and transfer quality) and four covariates (age, tenure, course duration and type of training content).

Quantile regression Furthermore, we used quantile regression (Koenker & Bassett, 1978) to explore the effects of motivation to transfer on the whole distribution of the dependent variables. In the present paper, the quantreg package version 4.76 (Koenker, 2011) was applied in R (R Development Core Team, 2011) to fit quantile regressions. Three quantile regres- sions were calculated with motivation to transfer as the predictor and perceived appli- cation to practice, transfer quantity and transfer quality, respectively, as the outcome variables. In each regression, we controlled for transfer design, perceived content validity, age, tenure, course duration and type of training content (results not depicted). Bootstrapped standard errors were used, being the recommended statistical inference method for quantile regression and having been shown to be especially robust in most situations (Koenker, 2005). To test whether the differences in slopes between quantiles could be attributed to sampling error, we conducted the joint test for equality of slopes (Koenker & Bassett, 1982).

Results and discussion – Study 1 Path analysis

Means, standard deviations, valid responses and percentage of missing data of all study variables were calculated with SPSS and are presented in Table 1. Path analysis results concerning the mediating effect of motivation to transfer on the relationship between training characteristics and training transfer are shown in Table 4. In line with common practice in mediational research (Rucker et al., 2011), we report unstandardized values in the present paper. For completeness, we also included standardized BC bootstrap results in Table 4, which are overall similar to the unstandardized values.

Self-ratings

According to unstandardized BC bootstrap results (see Table 4), the total effects of transfer design on perceived application to practice (c = 1.21; SE = 0.24; 90% CI: 0.81, 1.59), transfer quantity (c = 0.47; SE = 0.16; 90% CI: 0.17, 0.72) and transfer quality (c = 1.14; SE = 0.40; 90% CI: 0.40, 1.72) were significant. The indirect effects of transfer design on perceived application to practice (ab = 0.85, SE = 0.16; 90% CI: 0.62, 1.14), transfer quantity (ab = 0.19; SE = 0.10; 90% CI: 0.06, 0.38) and

T ab

le 4:

U n

st an

da rd

iz ed

as w

el l

as st

an da

rd iz

ed es

ti m

at es

an d

bi as

-c or

re ct

ed -b

oo ts

tr ap

re su

lt s

fo r

m ed

ia ti

n g

ef fe

ct s

of m

ot iv

at io

n to

tr an

sf er

on th

e re

la ti

on sh

ip be

tw ee

n tr

ai n

in g

ch ar

ac te

ri st

ic s

(t ra

n sf

er de

si gn

, pe

rc ei

ve d

co n

te n

t va

li di

ty )

an d

tr ai

n in

g tr

an sf

er (p

ee r

ra ti

n g

an d

se lf

-r at

in gs

of pe

rc ei

ve d

ap pl

ic at

io n

to pr

ac ti

ce ,

tr an

sf er

qu an

ti ty

an d

tr an

sf er

qu al

it y)

in S

tu di

es 1

an d

M ed

ia ti

n g

ef fe

ct o

f m

o ti

va ti

o n

to tr

an sf

er o

n th

e re

la ti

o n

sh ip

b et

w ee

n ..

. To

ta l

ef fe

ct c

In d

ir ec

t ef

fe ct

ab D

ir ec

t ef

fe ct

c’

U n

st d

. St

d .

U n

st d

. St

d .

U n

st d

. St

d .

St u

d y

1 Tr

an sf

er d

es ig

n P

er ce

iv ed

ap p

li ca

ti o

n to

p ra

ct ic

e 1.

21 a

0. 45

a 0.

85 a

0. 31

a 0.

36 a

0. 13

Tr an

sf er

q u

an ti

ty 0.

47 a

0. 26

a 0.

19 a

0. 11

a 0.

28 0.

16 Tr

an sf

er q

u al

it y

1. 14

a 0.

31 a

0. 56

a 0.

15 a

0. 58

0. 16

P ee

r ra

ti n

g :

P er

c. ap

p l.

to p

ra ct

ic e

0. 54

0. 19

0. 50

a 0.

18 a

0. 04

0. 01

P er

ce iv

ed co

n te

n t

va li

d it

y P

er ce

iv ed

ap p

li ca

ti o

n to

p ra

ct ic

e 0.

48 a

0. 17

a 0.

36 a

0. 13

a 0.

12 0.

04 Tr

an sf

er q

u an

ti ty

0. 04

0. 02

0. 08

a 0.

04 a

−0 .0

4 −0

.0 2

Tr an

sf er

q u

al it

y 0.

51 0.

13 0.

24 a

0. 06

a 0.

28 0.

07 P

ee r

ra ti

n g

: P

er c.

ap p

l. to

p ra

ct ic

e 0.

97 a

0. 33

a 0.

21 a

0. 07

a 0.

76 a

0. 25

St u

d y

2 Tr

an sf

er d

es ig

n P

er ce

iv ed

ap p

li ca

ti o

n to

p ra

ct ic

e 1.

05 a

0. 32

a 0.

56 a

0. 17

a 0.

49 a

0. 15

Tr an

sf er

q u

an ti

ty 0.

68 a

0. 24

a 0.

36 a

0. 13

a 0.

33 a

0. 12

Tr an

sf er

q u

al it

y 1.

16 a

0. 26

a 0.

51 a

0. 11

a 0.

66 a

0. 14

P er

ce iv

ed co

n te

n t

va li

d it

y P

er ce

iv ed

ap p

li ca

ti o

n to

p ra

ct ic

e 1.

11 a

0. 37

a 0.

37 a

0. 12

a 0.

74 a

0. 25

Tr an

sf er

q u

an ti

ty 0.

25 0.

10 0.

23 a

0. 09

a 0.

02 0.

01 Tr

an sf

er q

u al

it y

1. 23

a 0.

30 a

0. 33

a 0.

08 a

0. 91

a 0.

22 a

N o

te :

W e

es ti

m at

ed 90

% co

n fi

d en

ce in

te rv

al s

(C I)

b as

ed o

n N

= 10

00 b

o o

ts tr

ap sa

m p

le s

u si

n g

F IM

L in

M p

lu s.

In St

u d

y 1,

te n

u re

,c o

u rs

e d

u ra

ti o

n an

d ty

p e

o f

tr ai

n in

g co

n te

n t

w er

e in

cl u

d ed

as co

va ri

at es

.I n

St u

d y

2, w

e ad

d it

io n

al ly

in cl

u d

ed ag

e as

a co

va ri

at e.

A s

ty p

e o

f tr

ai n

in g

co n

te n

t w

as o

rd in

al in

St u

d y

2, it

w as

d u

m m

y -c

o d

ed b

ef o

re an

al y

si s.

In b

o th

st u

d ie

s so

m e

o f

th e

x- va

ri ab

le s

co n

ta in

ed m

is si

n g

va lu

es ,

re su

lt in

g in

N =

23 4

fo r

St u

d y

1, an

d N

= 37

7 fo

r St

u d

y 2.

P ee

r ra

ti n

g :

P er

c. ap

p l.

to p

ra ct

ic e

= P

ee r

ra ti

n g

: P

er ce

iv ed

ap p

li ca

ti o

n to

p ra

ct ic

e. a

E m

p ir

ic al

90 %

co n

fi d

en ce

in te

rv al

(l o

w er

an d

u p

p er

5% li

m it

) d

o es

n o

t in

cl u

d e

ze ro

transfer quality (ab = 0.56; SE = 0.22; 90% CI: 0.24, 0.98) via motivation to transfer were also significant. A significant direct effect of transfer design on perceived appli- cation to practice remained (c’ = 0.36; SE = 0.22; 90% CI: 0.03, 0.74); however, there were no significant direct effects on transfer quantity (c’ = 0.28; SE = 0.18; 90% CI: −0.04, 0.55) and transfer quality (c’ = 0.58; SE = 0.44; 90% CI: −0.18, 1.28).

For perceived content validity, we found a significant total effect on perceived appli- cation to practice (c = 0.48; SE = 0.24; 90% CI: 0.07, 0.84), but no significant total effects on transfer quantity (c = 0.04; SE = 0.17; 90% CI: −0.23, 0.31) and transfer quality (c = 0.51; SE = 0.38; 90% CI: −0.11, 1.14). Furthermore, we identified significant indirect effects of perceived content validity on perceived application to practice (ab = 0.36; SE = 0.14; 90% CI: 0.15, 0.61), transfer quantity (ab = 0.08; SE = 0.05; 90% CI: 0.02, 0.19) and transfer quality (ab = 0.24; SE = 0.14; 90% CI: 0.07, 0.52) via motivation to transfer. None of the direct effects between perceived content validity and perceived application to practice (c’ = 0.12; SE = 0.20; 90% CI: −0.25, 0.43), transfer quantity (c’ = −0.04; SE = 0.18; 90% CI: −0.32, 0.25) and transfer quality (c’ = 0.28; SE = 0.39; 90% CI: −0.37, 0.89) were significant.

In sum, motivation to transfer was found to mediate the relationship between transfer design and all training transfer scales. For perceived content validity, we found significant indirect effects on all training transfer scales via motivation to transfer, but mostly no total or direct effects. However, recent literature on mediation assumes that a significant total effect is not necessarily a prerequisite for investigating mediation relationships (Rucker et al., 2011, and references therein). Therefore, we can conclude that motivation to transfer acts as a linking mechanism between per- ceived content validity and all transfer measures as well. In line with Hypothesis 1 and 2, motivation to transfer does mediate the relationship between training charac- teristics (transfer design and perceived content validity) and the different self-rated measures of training transfer (perceived application to practice, transfer quantity and transfer quality).

Peer rating

According to unstandardized BC bootstrap results (see Table 4), the total effect of transfer design on the peer rating of perceived application to practice was not signifi- cant (c = 0.54; SE = 0.39; 90% CI: −0.15, 1.16), whereas the indirect effect via motivation to transfer was significant (ab = 0.50; SE = 0.26; 90% CI: 0.09, 0.94). The direct effect of transfer design on the peer rating was not significant either (c’ = 0.04; SE = 0.52; 90% CI: −0.88, 0.80).

For perceived content validity, the total effect (c = 0.97; SE = 0.35; 90% CI: 0.36, 1.50), the indirect effect via motivation to transfer (ab = 0.21; SE = 0.12; 90% CI: 0.05, 0.44) and the direct effect on the peer rating of perceived application to practice (c’ = 0.76; SE = 0.35; 90% CI: 0.15, 1.31) were all significant. Even when all covariates are excluded from our path model in Study 1 (see Figure 1), all indirect effects via motivation to transfer remain significant (results not depicted).

In sum, motivation to transfer was found to be a linking mechanism between transfer design and the peer rating of perceived application to practice as indicated by a significant indirect effect via motivation to transfer. Furthermore, we found motivation to transfer to mediate the relationship between perceived content validity and the peer rating of perceived application to practice. Thus, the results lend further support to Hypothesis 1 and 2.

Results and discussion – Study 2 Path analysis

Means, standard deviations, valid responses and percentage of missing data of all of the study’s variables are shown in Table 1. Unstandardized as well as standardized BC bootstrap results are included in Table 4. Both reveal overall similar findings.

The examination of unstandardized BC bootstrap results showed that the total effects of transfer design on perceived application to practice (c = 1.05; SE = 0.22; 90% CI: 0.68, 1.40), transfer quantity (c = 0.68; SE = 0.21; 90% CI: 0.37, 1.03) and transfer quality (c = 1.16; SE = 0.34; 90% CI: 0.60, 1.74) were significant (see Table 4). The indirect effects of transfer design on perceived application to practice (ab = 0.56; SE = 0.12; 90% CI: 0.38, 0.78), transfer quantity (ab = 0.36; SE = 0.09; 90% CI: 0.24, 0.52) and transfer quality (ab = 0.51; SE = 0.17; 90% CI: 0.25, 0.81) via motivation to transfer were also significant. Moreover, transfer design had significant direct effects on perceived application to practice (c’ = 0.49; SE = 0.21; 90% CI: 0.10, 0.80), transfer quantity (c’ = 0.33; SE = 0.20; 90% CI: 0.01, 0.68) and transfer quality (c’ = 0.66; SE = 0.35; 90% CI: 0.09, 1.26).

For perceived content validity, we identified significant total effects on perceived application to practice (c = 1.11; SE = 0.20; 90% CI: 0.78, 1.45) and transfer quality (c = 1.23; SE = 0.34; 90% CI: 0.65, 1.79), but a non-significant total effect on transfer quantity (c = 0.25; SE = 0.19; 90% CI: −0.10, 0.53). We found significant indirect effects of perceived content validity on perceived application to practice (ab = 0.37; SE = 0.10; 90% CI: 0.23, 0.56), transfer quantity (ab = 0.23; SE = 0.07; 90% CI: 0.14, 0.38) and transfer quality (ab = 0.33; SE = 0.12; 90% CI: 0.17, 0.55) via motivation to transfer. There were also significant direct effects between perceived content validity and perceived appli- cation to practice (c’ = 0.74; SE = 0.21; 90% CI: 0.40, 1.10) as well as transfer quality (c’ = 0.91; SE = 0.34; 90% CI: 0.35, 1.46). However, there was no significant direct effect between perceived content validity and transfer quantity (c’ = 0.02; SE = 0.19; 90% CI: −0.31, 0.32). When all covariates were excluded from the path model (see Figure 1), all indirect effects via motivation to transfer were still significant (results not depicted).

In sum, the participants’ motivation to transfer mediated the relationship between transfer design and all training transfer scales. For perceived content validity, the effect on perceived application to practice and on transfer quality was mediated by motiva- tion to transfer. We found a significant indirect effect of perceived content validity on transfer quantity via motivation to transfer, but no such total or direct effect. Our results underscore the linking role of motivation to transfer in the relationship between per- ceived content validity and transfer quantity. In further support of Hypothesis 1 and 2, we identified motivation to transfer as a linking mechanism between training charac- teristics (transfer design and perceived content validity) and training transfer in a more diverse sample, which also included employees from non-industrial branches.

Quantile regression

Table 5 contains the bootstrapped quantile regression results. We report unstan- dardized values in the text; however, standardized bootstrap results point in the same direction (results not depicted). For comparison purposes, the usual mean-based OLS regression coefficients are also included in Table 5.

The effect of motivation to transfer on perceived application to practice is significant for every quantile of perceived application to practice, with the largest quantile regres- sion coefficient occurring at the 0.15 quantile, b = 1.54, p < 0.001, and then steadily dropping until the 0.85 quantile, b = 0.61, p = 0.022. The equality of slopes hypothesis was significantly rejected, F(32, 1853) = 3.03, p < 0.001. Regarding transfer quantity, every effect of motivation to transfer was significant, with the largest regression coef- ficient occurring in the upper part (0.75, b = 1.11, p < 0.001) of the distribution. The corresponding test for equality of slopes turned out to be significant, F(32, 1773) = 3.18, p < 0.001. For transfer quality, the quantile regression estimates are significant for the 0.15, b = 2.00, p = 0.004, 0.25, b = 1.63, p < 0.001, and 0.50, b = 0.86, p = 0.006, part of the distribution, with the largest effect occurring in the lower part (0.15) of transfer quality. The joint test for equality of slopes rejected the equal slopes hypothesis, F(32, 1768) = 7.61, p < 0.001. Notably, even if all covariates were excluded from the analyses, we obtained overall similar results (not depicted), which correspond to the general trends found before.

In sum, our findings corroborate that motivation to transfer matters because it shows significant effects on almost all quantiles of the dependent variables. Moreover,

T ab

le 5:

U n

st an

da rd

iz ed

qu an

ti le

re gr

es si

on es

ti m

at es

fo r

m ot

iv at

io n

to tr

an sf

er on

th e

de pe

n de

n t

va ri

ab le

s (p

er ce

iv ed

ap pl

ic at

io n

to pr

ac ti

ce ,

tr an

sf er

qu an

ti ty

an d

tr an

sf er

qu al

it y)

in S

tu dy

15 th

(S E

) 25

th (S

E )

50 th

(S E

) 75

th (S

E )

85 th

(S E

) O

L S

(S E

)

A P

(N =

37 7)

1. 54

** *

(0 .2

7) 1.

17 **

* (0

.2 9)

1. 03

** *

(0 .2

8) 0.

96 **

* (0

.2 4)

0. 61

* (0

.2 7)

1. 17

** *

(0 .1

5) Q

U A

N T

(N =

36 1)

0. 34

** (0

.1 3)

0. 48

** *

(0 .1

3) 0.

75 **

* (0

.1 4)

1. 11

** *

(0 .2

8) 0.

92 **

(0 .2

8) 0.

74 **

* (0

.1 7)

Q U

A L

(N =

36 0)

2. 00

** (0

.6 9)

1. 63

** *

(0 .4

8) 0.

86 **

(0 .3

1) 0.

35 (0

.3 0)

0. 22

(0 .1

9) 1.

05 **

* (0

.2 7)

* p

< 0.

05 ,

** p

< 0.

01 ,

** *

p <

0. 00

1 (t

w o

-t ai

le d

). W

e re

p o

rt th

e ef

fe ct

s o

f m

o ti

va ti

o n

to tr

an sf

er o

n th

e d

if fe

re n

t q

u an

ti le

s (0

.1 5,

0. 25

, 0.

50 ,

0. 75

an d

0. 85

) o

f th

e d

ep en

d en

t va

ri ab

le s.

T h

e es

ti m

at es

h av

e b

ee n

co rr

ec te

d fo

r tr

an sf

er d

es ig

n ,p

er ce

iv ed

co n

te n

tv al

id it

y, te

n u

re ,c

o u

rs e

d u

ra ti

o n

,a g

e an

d d

u m

m y

-c o

d ed

ty p

e o

ft ra

in in

g co

n te

n t.

St an

d ar

d er

ro rs

S E

ar e

d ep

ic te

d in

p ar

en th

es es

. C

as es

w it

h in

co m

p le

te d

at a

w er

e re

m ov

ed li

st w

is e

fr o

m th

e an

al y

se s.

A P

= se

lf -r

at ed

p er

ce iv

ed ap

p li

ca ti

o n

to p

ra ct

ic e,

Q U

A N

T =

tr an

sf er

q u

an ti

ty ,

Q U

A L

= tr

an sf

er q

u al

it y,

O L

S =

o rd

in ar

y le

as t

sq u

ar es

medium- and low-transferers generally benefit from an increased motivation to trans- fer. In line with Hypothesis 3, the quantile regression analyses add that the effect of motivation to transfer is significantly heterogeneous across the distributions of all the dependent variables. Regarding transfer quantity, we found an increasing impact of motivation to transfer with higher quantiles. In case of perceived application to practice and transfer quality, however, the regression coefficients decrease with higher quantiles.

General discussion The purpose of the present study was to examine the critical role of motivation to transfer in the transfer process. In Study 1, one peer rating was used in addition to three self-ratings of transfer to examine the role of motivation to transfer in mediating the relationship between training characteristics and training transfer. In Study 2, we investigated the mediating role of motivation to transfer in a more diverse sample of employees from different branches. Moreover, we identified differential effects of motivation to transfer on training transfer by means of quantile regression.

As hypothesized, we found motivation to transfer to be a linking mechanism between training characteristics (transfer design and perceived content validity) and training transfer in both studies. In all our analyses, the indirect effects between training characteristics (transfer design and perceived content validity) and all transfer outcomes via motivation to transfer were significant regardless of the measurement source (peer rating or self-rating of the transfer outcomes). As theoretically assumed (e.g. Gegenfurtner et al., 2009), our results highlight the importance of motivation to transfer as a mediational link between training characteristics and training transfer.

Although our findings support the linking role of motivation to transfer, significant direct effects of training characteristics (transfer design and perceived content validity) on transfer outcomes were also found in half of the analyses. This finding is in line with previous theorizing that assumes a direct link between training characteristics and training transfer (e.g. Holton, 2005). In addition to motivation to transfer, other medi- ating variables might be included in future studies (e.g. self-efficacy; for an overview on possible mediators, see Grossman & Salas, 2011). However, our results underscore the empirical relevance of motivation to transfer as an important linking mechanism in the training transfer process.

Previous empirical studies show a wide range of relationships between motivation to transfer and training transfer, from close to zero to rather high values (Gegenfurtner et al., 2009). By means of quantile regression, it was possible empirically to explore the nature of the heterogeneity. (1) In general, motivation to transfer had a significant effect on transfer outcomes in all quantile regression analyses for most quantiles. This is in line with studies highlighting the importance of motivation to transfer for training transfer (e.g. Bates et al., 2007; Kontoghiorghes, 2004). (2) As hypothesized, motivation to transfer also had heterogeneous effects on training transfer. For transfer quantity, motivation to transfer tends to benefit most the participants who transferred a large amount of what they learned (i.e. high transferers). In case of transfer quality and perceived application to practice, the results showed higher benefits for participants who transferred a low amount of training contents (i.e. low transferers).

Although our findings need to be further substantiated in future research, they do suggest that the effect of motivation to transfer on transfer outcomes is dependent on the standing of the person in the distribution. This could in part explain the heterogeneity reported between motivation to transfer and transfer outcomes (Gegenfurtner et al., 2009) and extends previous theorizing: The effect of motivation to transfer depends not only on the operationalization of the dependent variable (Gegenfurtner et al., 2009), but also on the sample considered because the effect of motivation to transfer differs between quantiles of transfer outcomes. In case of trans- fer quantity, motivation to transfer tends to have the highest effects on individuals who are high transferers. Consequently, to make good employees even better in terms of transfer quantity, one essential impact factor is to enhance their motivation

to transfer. Transfer quality seems to function differently according to our analysis: Motivation to transfer might not have a large effect on the transfer quality of high transferers (i.e. employees who belong to the higher quantiles of the transfer outcome distributions). Yet, the finding that positive effects of psychological variables may reach a limit and subsequently decline is not new. There is extensive literature in psychology on theories dealing with differing effects of predictors or differing effects across samples, often positing inverted U-relationships (for a recent review, see Grant & Schwartz, 2011).

Implications, limitations and suggestions for future research Implications

According to the present study results, transfer design and perceived content validity act as a starting point for successful training transfer, and companies would do well to mind these factors. Trainers can improve the transfer design of a training program, for example, if they include transfer-related case studies that show how parti- cipants can effectively apply training contents and skills back at work (Holton et al., 2000; Kauffeld, 2010). In addition, exercises that enable reflection on solutions for potential barriers for successful transfer may be included directly in the training program (Kauffeld, 2010). Perceived content validity can be enhanced if training con- tents correspond to job requirements (Holton et al., 2000). Accordingly, to enable high training transfer, training courses should deal with work-related topics which ideally showcase the importance of the training contents for the job (Kauffeld, 2010).

Moreover, motivation to transfer has been shown to serve as a linking mechanism between training characteristics and training transfer. Thus, companies have to focus on enhancing the employees’ motivation to transfer, for example by planning and monitoring specific steps to apply training contents to the job, by selecting tandem partners (e.g. other training participants) to help employees reflect upon the possibil- ities for transferring what they have learned, and by integrating transfer coaching via telephone or face-to-face (Kauffeld, 2010).

Although there have been numerous studies on motivation to transfer and transfer outcomes (for a recent review, see Gegenfurtner et al., 2009), it has not yet been exam- ined whether the distribution of transfer outcomes in itself is important in determining the effect of motivation to transfer: Do high transferers respond differently to improved motivation to transfer than those employees who transfer less? The results of the present study point in this direction. Thus, trainers should make sure that interventions enhancing employees’ motivation to transfer are individually tailored for participants. This could, for example, be based on previous experiences with the respective individ- ual transfer amounts. Furthermore, trainers may differentiate based on training objec- tives. In trainings that aim at individuals to transfer many skills into practice (i.e. which aim at a quantitatively high transfer), motivation to transfer will benefit the high transferers most. For example, in trainings on selling techniques, from which partici- pants should transfer as many techniques as possible in order to convince the custom- ers, trainers should focus on enhancing participants’ motivation to transfer to facilitate high training transfer. By contrast, in trainings that require a profound understanding of training contents (i.e. which aim at a qualitatively high transfer), motivation to transfer will benefit the low or medium transferers, but not the high transferers most. For example, in trainings on mathematical software that are targeted at understanding one specific computer program in depth and applying this knowledge, trainers should rather elaborate on further techniques (e.g. concerning the work environment) to promote high training transfer.

Limitations and suggestions for future research

Several limitations of the present paper need to be considered. First, causal interpreta- tions of the relationships observed are not warranted because we implemented a

cross-sectional, retrospective design in both studies. Although cross-sectional retro- spective study designs are rather common in training transfer studies (Blau et al., 2012), we suggest caution with respect to any causal conclusions from the present results. Ideally, mediation relationships should be examined using a double randomized experimental design (MacKinnon et al., 2007). Due to ethical issues, however, we could not expose the study participants to experimental manipulations. However, previous theorizing (e.g. Gegenfurtner et al., 2009) and empirical studies (e.g. Nijman et al., 2006) hint at the hypothesized direction of the relationships.

Second, measuring all study variables with questionnaires can increase the potential for common method bias (e.g. Podsakoff et al., 2012) and may produce positively biased relationships between variables (e.g. Blume et al., 2010; Podsakoff et al., 2012). More- over, in Study 2, self-selection bias may occur as we collected a convenience sample via online survey (e.g. Reips, 2000). To ensure data quality, we followed recommendations by Podsakoff et al. (2012) concerning the study procedure (e.g. by implementing a short questionnaire). In addition, we separated the source of the dependent variable in Study 1 (peer versus self-rating; see Blume et al., 2010). Moreover, we used validated scales (Grohmann & Kauffeld, 2013; Holton et al., 2000; Kauffeld et al., 2008, 2009). As some of our study variables assess the individual participants’ perceptions (e.g. motivation to transfer) rather than observable behavior, self-reports present an appropriate data source (Podsakoff et al., 2012). Furthermore, questionnaires offer an effective and reli- able way for assessing the psychological constructs under study here because the training courses covered were rather heterogeneous, especially in Study 2.

Third, in the present paper, motivation to transfer was measured as a unidimensional construct at one point in time. According to a recent review by Gegenfurtner et al. (2009), it is common practice in HRD research to conceptualize motivation to transfer unidimensionally. By contrast, motivation to learn, a closely related construct, has been proposed multidimensionally and tested empirically in several studies (e.g. Kim et al., 2012; Weissbein et al., 2011). To date, only few studies have investigated motivation to transfer as a multidimensional construct (e.g. Gegenfurtner, 2013). However, there is no consensus about which dimensions of motivation to transfer should be distin- guished in HRD research. Thus, future studies should further elaborate on this. More- over, to account for a dynamic view of motivation (e.g. Beier & Kanfer, 2010), motivation to transfer could be assessed as a multidimensional construct at different points in time (e.g. at the end of a training and some months after the training took place). This would help to identify when exactly which kind of motivation to transfer results in high or low performers (in terms of training transfer).

Fourth, transfer design and perceived content validity were highly interrelated in our analyses, which might be due to a theoretical overlap because both are training characteristics. Moreover, both were measured with self-report items. Future studies should assess transfer design with more objective measures (e.g. evaluate whether specific training interventions were used) to examine the different effects of transfer design and perceived content validity in more depth. However, factor analyses results underscore the distinction between perceived content validity and transfer design (e.g. Kauffeld et al., 2008). Moreover, because we surveyed diverse training courses in both studies, our (more subjective) self-report measure allowed us to assess transfer design more accurately because it covered all kinds of different training programs with all the different possibilities of transfer design aspects. Thereby we were able to address the limitations of previous transfer studies that surveyed only one company (e.g. Chiaburu & Tekleab, 2005) in Study 2, which included participants from different branches.

Fifth, this paper investigated only transfer design and perceived content validity as independent variables because these were identified as two meaningful training char- acteristics in previous HRD studies (e.g. Bates et al., 2007; Kauffeld et al., 2008; Velada et al., 2007). Other training characteristics, for example, trainer competencies (Hutchins et al., 2010), training reputation and voluntary training participation (Aziz & Ahmad, 2011) may also play a central role in the transfer process and should additionally be investigated in future studies. As Holton et al. (2000) suggest, researchers should measure the entire system of variables (work environment, training and trainee

characteristics) that influence training transfer. Hence, it becomes possible to investi- gate the unique influence of different training characteristics, for example, by empiri- cally examining the model by Gegenfurtner et al. (2009) in which motivation to transfer is proposed to be the central mediational variable in the transfer process.

Finally, our exploratory quantile regression results need to be substantiated in future research. Because our two samples were markedly different, we chose to apply quantile regression only to the more diverse sample 2. In general it might not be problematic to analyze whether certain variables have an effect across different samples on average. However, these generalizations might become more problematic when analyzing how effects differ via quantile regression, because the heterogeneous effects themselves might differ across sub-populations. One possible solution to this problem is the use of representative samples. As we are, to the best of our knowledge, the first to explore the effects of motivation to transfer on different quantiles of the training transfer distribu- tions, it is difficult to speculate on the representativeness of our results. Although the present paper underscores the importance of examining the differential role of moti- vation to transfer, open questions can only be answered by further empirical studies. Future research should focus on investigating the differential effects obtained with quantile regression in more detail and use more diverse transfer measures (for exam- ples, see, e.g., Kraiger, 2002). Objective transfer measures can strengthen our findings by helping to examine if the differential effects of motivation to transfer on transfer outcomes are not a result of a possible confounding variable inherent in self-ratings of training transfer. Based on present research dealing with, for example, the inverted U-shaped function of learning (Salehi et al., 2010), potential non-linear effects of moti- vation to transfer regarding the distribution of motivation to transfer itself might also be analyzed in future studies.

Conclusion Participants’ motivation to transfer is crucial to the transfer process. The present paper highlights the importance of motivation to transfer as an influential linking mechanism between training characteristics and training transfer. Application of quantile regres- sion in training research revealed differential effects of motivation to transfer on trans- fer outcomes. These results underscore the importance of considering personal characteristics (e.g. motivation to transfer) to further increase specific transfer criteria (e.g. high transfer quantity) and offer a starting point for understanding the relation- ship between motivation to transfer and training transfer in more detail.

References Aguinis, H. and Kraiger, K. (2009), ‘Benefits of training and development for individuals and

teams, organizations, and society’, Annual Review of Psychology, 60, 451–74. Arulampalam, W., Booth, A. L. and Bryan, M. L. (2010), ‘Are there asymmetries in the effects of

training on the conditional male wage distribution?’, Journal of Population Economics, 23, 251–72. Aziz, S. F. A. and Ahmad, S. (2011), ‘Stimulating training motivation using the right training

characteristic’, Industrial and Commercial Training, 43, 53–61. Baldwin, T. T. and Ford, J. K. (1988), ‘Transfer of training: a review and directions for future

research’, Personnel Psychology, 41, 63–105. Bates, R., Kauffeld, S. and Holton, E. F. III (2007), ‘Examining the factor structure and predictive

ability of the German-version of the Learning Transfer Systems Inventory’, Journal of European Industrial Training, 31, 195–211.

Beier, M. E. and Kanfer, R. (2010), ‘Motivation in Training and Development: A Phase Perspec- tive’, in S. W. J. Kozlowski and E. Salas (eds), Learning, Training, and Development in Organizations (New York: Routledge), pp. 65–97.

Bell, S. T., Towler, A. J. and Fisher, D. M. (2011), ‘A multilevel examination of the influence of trainee-trainer gender dissimilarity and trainee-classroom gender composition dissimilarity on trainee knowledge acquisition’, Human Resource Development Quarterly, 22, 343–72.

Beller, J. and Baier, D. (2013), ‘Differential effects: are the effects studied by psychologists really linear and homogeneous?’, Europe’s Journal of Psychology, 9, 378–84.

Bhatti, M. A. and Kaur, S. (2010), ‘The role of individual and training design factors on training transfer’, Journal of European Industrial Training, 34, 656–72.

Blau, G., Gibson, G., Bentley, M. and Chapman, S. (2012), ‘Testing the impact of job-related variables on a utility judgment training criterion beyond background and affective reaction variables’, International Journal of Training and Development, 16, 54–66.

Blume, B. D., Ford, J. K., Baldwin, T. T. and Huang, J. L. (2010), ‘Transfer of training: a meta- analytic review’, Journal of Management, 36, 1065–105.

Chiaburu, D. S. and Tekleab, A. G. (2005), ‘Individual and contextual influences on multiple dimensions of training effectiveness’, Journal of European Industrial Training, 29, 604–26.

Colquitt, J. A., LePine, J. A. and Noe, R. A. (2000), ‘Toward an integrative theory of training motivation: a meta-analytic path analysis of 20 years of research’, Journal of Applied Psychology, 85, 678–707.

Gegenfurtner, A. (2013), ‘Dimensions of motivation to transfer: a longitudinal analysis of their influence on retention, transfer, and attitude change’, Vocations and Learning, 6, 187–205.

Gegenfurtner, A., Veermans, K., Festner, D. and Gruber, H. (2009), ‘Motivation to transfer train- ing: an integrative literature review’, Human Resource Development Review, 8, 403–23.

Grant, A. M. and Schwartz, B. (2011), ‘Too much of a good thing: the challenge and opportunity of the inverted U’, Perspectives on Psychological Science, 6, 61–76.

Grohmann, A. and Kauffeld, S. (2013), ‘Evaluating training programs: development and corre- lates of the questionnaire for professional training evaluation’, International Journal of Training and Development, 17, 135–55.

Grossman, R. and Salas, E. (2011), ‘The transfer of training: what really matters’, International Journal of Training and Development, 15, 103–20.

Holton, E. F. III (2005), ‘Holton’s evaluation model: new evidence and construct elaborations’, Advances in Developing Human Resources, 7, 37–54.

Holton, E. F. III, Bates, R. and Ruona, W. E. A. (2000), ‘Development of a generalized learning transfer system inventory’, Human Resource Development Quarterly, 11, 333–60.

Hutchins, H. M., Burke, L. A. and Berthelsen, A. M. (2010), ‘A missing link in the transfer problem? Examining how trainers learn about training transfer’, Human Resource Management, 49, 599–618.

Kastenmüller, A., Frey, D., Kerschreiter, R., Tattersall, A. J., Traut-Mattausch, E. and Fischer, P. (2012), ‘Perceived openness of climate during training and transfer motivation: testing two short and simple interventions’, Journal of Vocational Education and Training, 64, 211–25.

Kauffeld, S. (2010), Nachhaltige Weiterbildung. Betriebliche Seminare und Trainings entwickeln, Erfolge messen, Transfer sichern [Sustainable Human Resource Development: Developing, Evaluating, and Trans- ferring Professional Training Courses] (Berlin: Springer).

Kauffeld, S. and Lehmann-Willenbrock, N. (2010), ‘Sales training: effects of spaced practice on training transfer’, Journal of European Industrial Training, 34, 23–37.

Kauffeld, S., Bates, R., Holton, E. F. III and Müller, A. C. (2008), ‘Das deutsche Lerntransfer- System-Inventar (GLTSI): Psychometrische Überprüfung der deutschsprachigen Version [the German version of the Learning Transfer System Inventory (GLTSI): psychometric validation]’, Zeitschrift für Personalpsychologie, 7, 50–69.

Kauffeld, S., Brennecke, J. and Strack, M. (2009), ‘Erfolge sichtbar machen: Das Maßnahmen- Erfolgs-Inventar (MEI) zur Bewertung von Trainings [Visualizing Training Outcomes: The MEI for Training Evaluations]’, in S. Kauffeld, S. Grote and E. Frieling (eds), Handbuch Kompetenzentwicklung (Stuttgart: Schäffer-Poeschel), pp. 55–78.

Kim, K., Oh, I.-S., Chiaburu, D. S. and Brown, K. G. (2012), ‘Does positive perception of oneself boost learning motivation and performance?’, International Journal of Selection and Assessment, 20, 257–71.

Koenker, R. (2005), Quantile Regression (New York: Cambridge University Press). Koenker, R. (2011), Quantreg: Quantile Regression. Available at: http://cran.r-project.org/web/

packages/quantreg/index.html (accessed 15 December 2011). Koenker, R. and Bassett, G. B. Jr. (1978), ‘Regression quantiles’, Econometrica: Journal of the

Econometric Society, 46, 33–50. Koenker, R. and Bassett, G. B. Jr. (1982), ‘Robust tests for heteroscedasticity based on regression

quantiles’, Econometrica: Journal of the Econometric Society, 50, 43–61. Kontoghiorghes, C. (2004), ‘Reconceptualizing the learning transfer conceptual framework:

empirical validation of a new systemic model’, International Journal of Training and Development, 8, 210–21.

Kraiger, K. (2002), ‘Decision-Based Evaluation’, in K. Kraiger (ed.), Creating, Implementing, and Managing Effective Training and Development: State-of-the-Art Lessons for Practice (San Francisco, CA: Jossey-Bass), pp. 331–75.

http://cran.r-project.org/web/packages/quantreg/index.html

Lee, C. D. and Kahnweiler, W. M. (2000), ‘The effect of a mastery learning technique on the performance of a transfer of training task’, Performance Improvement Quarterly, 13, 3, 125–39.

Liebermann, S. and Hoffmann, S. (2008), ‘The impact of practical relevance on training transfer: evidence from a service quality training program for German bank clerks’, International Journal of Training and Development, 12, 74–86.

Machin, M. A. and Fogarty, G. J. (1997), ‘The effects of self-efficacy, motivation to transfer, and situational constraints on transfer intentions and transfer of training’, Performance Improvement Quarterly, 10, 2, 98–115.

MacKinnon, D. P., Lockwood, C. M. and Williams, J. (2004), ‘Confidence limits for the indirect effect: distribution of the product and resampling methods’, Multivariate Behavioral Research, 39, 99–128.

MacKinnon, D. P., Fairchild, A. J. and Fritz, M. S. (2007), ‘Mediation analysis’, Annual Review of Psychology, 58, 593–614.

Micceri, T. (1989), ‘The unicorn, the normal curve, and other improbable creatures’, Psychological Bulletin, 105, 156–66.

Muthén, L. K. and Muthén, B. O. (1998–2010), Mplus User’s Guide, 6th edn (Los Angeles, CA: Muthén & Muthén).

Nijman, D.-J. J. M., Nijhof, W. J., Wognum, A. A. M. and Veldkamp, B. P. (2006), ‘Exploring differential effects of supervisor support on transfer of training’, Journal of European Industrial Training, 30, 529–49.

Noe, R. A. (1986), ‘Trainees’ attributes and attitudes: neglected influences on training effective- ness’, Academy of Management Review, 11, 736–49.

Podsakoff, P. M., MacKenzie, S. B. and Podsakoff, N. P. (2012), ‘Sources of method bias in social science research and recommendations on how to control it’, Annual Review of Psychology, 63, 539–69.

Preacher, K. J. and Hayes, A. F. (2008) ‘Contemporary Approaches to Assessing Mediation in Communication Research’, in A. F. Hayes, M. D. Slater and L. B. Snyder (eds), The SAGE Sourcebook of Advanced Data Analysis Methods for Communication Research (Thousand Oaks, CA: Sage), pp. 13–54.

Preacher, K. J., Zyphur, M. J. and Zhang, Z. (2010), ‘A general multilevel SEM framework for assessing multilevel mediation’, Psychological Methods, 15, 209–33.

R Development Core Team (2011), R: A Language and Environment for Statistical Computing (Vienna, Austria: R Foundation for Statistical Computing).

Reips, U.-D. (2000), ‘The Web Experiment Method: Advantages, Disadvantages, and Solutions’, in M. H. Birnbaum (ed.), Psychological Experiments on the Internet (San Diego, CA: Academic Press), pp. 89–118.

Rucker, D. D., Preacher, K. J., Tormala, Z. L. and Petty, R. E. (2011), ‘Mediation analysis in social psychology: current practices and new recommendations’, Social and Personality Psychology Compass, 5, 6, 359–71.

Salehi, B., Cordero, M. I. and Sandi, C. (2010), ‘Learning under stress: the inverted-U-shape function revisited’, Learning & Memory, 17, 522–30.

Tziner, A., Haccoun, R. R. and Kadish, A. (1991), ‘Personal and situational characteristics influ- encing the effectiveness of transfer of training improvement strategies’, Journal of Occupational Psychology, 64, 167–77.

Van der Locht, M., Van Dam, K. and Chiaburu, D. S. (2013), ‘Getting the most of management training: the role of identical elements for training transfer’, Personnel Review, 42, 422–39.

Velada, R., Caetano, A., Michel, J. W., Lyons, B. D. and Kavanagh, M. J. (2007), ‘The effects of training design, individual characteristics and work environment on transfer of training’, International Journal of Training and Development, 11, 282–94.

Warr, P. and Bunce, D. (1995), ‘Trainee characteristics and the outcomes of open learning’, Personnel Psychology, 48, 347–75.

Weissbein, D. A., Huang, J. L., Ford, J. K. and Schmidt, A. M. (2011), ‘Influencing learning states to enhance trainee motivation and improve training transfer’, Journal of Business and Psychology, 26, 423–35.

Yelon, S., Sheppard, L., Sleight, D. and Ford, J. K. (2004), ‘Intention to transfer: how do autono- mous professionals become motivated to use new ideas?’, Performance Improvement Quarterly, 17, 2, 82–103.

Copyright of International Journal of Training & Development is the property of Wiley- Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.