Self-determination theory (SDT) represents one perspective for understanding motivational processes connected to learning in educational settings (Deci and Ryan 1985, 2000). Levesque-Bristol et al. (2006) applied key elements of SDT to develop the integrative model of learning and motivation (IMLM) to understand learning in higher education environments. Specifically, the model links together the classroom learning climate; the satisfaction of individuals’ basic psychological needs for autonomy, competence, and relatedness; self-regulated motivation; and learning processes, such as perceived knowledge transferability with the theoretical justification supplied through SDT. The purpose of this investigation was to understand and evaluate the basic IMLM tenets that link together the perceived learning climate, basic psychological needs satisfaction, self-regulated motivation, and perceived knowledge transferability. A conceptual model, grounded in SDT and the IMLM, was developed to outline hypothesized relationship among study variables (see Fig. 1) and then evaluated through structural equation modeling (SEM).

Fig. 1
figure 1

Conceptual model illustrating the hypothesized relationships among the learning climate, basic psychological needs satisfaction, self-determined motivation, and perceived knowledge transferability. Self-Determin Index = self-determination index, Knowledge Transfer = perceived knowledge transferability

Full size image

Grounding in Self-Determination Theory

SDT is a humanistic, person-centered motivational theory (Deci and Ryan 1985, 2000) which posits that humans thrive in environments that address the basic psychological needs of autonomy, competence, and relatedness (Deci and Ryan 1985) These three psychological needs have been found to support personal wellbeing across cultural contexts (Levesque et al. 2004). Autonomy suggests a sense of volition and ability to contribute to decision making in ways that reflects one’s own motives (Black and Deci 2000), whereas competence is the ability to master certain skills or achieve goals (Deci et al. 1996). Relatedness describes feelings of connection and a sense of belonging, which is viewed as important for addressing autonomy and competence (Deci and Ryan 2008). When these basic psychological needs are satisfied, individuals are more likely to engage in behavior that is more self-determined, or driven by internal factors rather than outside pressures (Deci and Ryan 2000).

Individuals have a variety of reasons or motives for engaging in behaviors, which can be differentiated based on their underlying levels of self-determination (Vansteenkiste and Ryan 2013). The prototype of self-determined motivation is intrinsic motivation, which refers to individuals’ tendencies to perform behaviors that they find to be self-satisfying (Deci and Ryan 2008). SDT distinguishes among forms of extrinsic motivations based on the relative degree of self-regulation (Deci and Ryan 1991). Integration is viewed as the most self-determined of the extrinsic motives (Deci and Ryan 2008). When a behavior is integrated, it is both valued and considered to be in full alignment or congruence with the other aspects of the self. Identification refers to a behavior found to be important to the individual and is therefore pursued voluntarily.

Introjection is a form of extrinsic motivation in which the regulation of the behavior is controlled by external incentives that have been partially internalized. Introjected behaviors are often regulated by guilt, pride, or other internal pressures that compel the individual and it is not considered to be self-determined (Vansteenkiste and Ryan 2013). Extrinsic regulation is the most externally regulated type of motivation and relates to situations in which motivation is limited to the pursuit of rewards of fear of punishment (Black and Deci 2000; Deci et al. 1996). Finally, amotivation represents an absence of motivation or regulation. It is a state in which one is not motivated to behave in a meaningful way.

The Integrative Model of Learning and Motivation

Based on SDT, Levesque-Bristol et al. (2006) developed the IMLM to illustrate how elements of the learning environment impact student motivation, which in turn predicts learning processes, including knowledge transferability, and then learning outcomes in a holistic or “integrative” way (see Levesque-Bristol et al. 2006). The learning environment is composed of a variety of elements including the students and instructors, the context in which learning occurs, the course content and instructional objectives, the goals that instructors and students bring to the learning experiences, and the strategies implemented by the instructor that impact student motivation. The combination of these elements contributes to the development of a certain type of learning climate within the classroom (Richards and Levesque-Bristol 2016). A teacher-centered learning climate is one in which the instructor’s perspectives and needs are prioritized over those of the students (De George-Walker and Keeffe 2010). Such an environment typically includes more traditional, didactic pedagogies in which the teacher is viewed as holding the knowledge that is transmitted to the students through lecture (Richards and Velasquez 2014).

A student-centered learning climate, on the other hand, “is characterized by high levels of student engagement and empowerment so that students become central to the learning process” (Richards and Levesque-Bristol 2014, p. 44), and promotes basic psychological needs satisfaction. There are a variety of ways to foster the creation of student-centered learning environment. For example, an instructor could offer choices in the type of work students can submit to demonstrate mastery of a skill (Vansteenkiste and Ryan 2013). Alternatively, when choice is not possible, the provision of a rationale for doing the work can also contribute to a student-centered environment. Further, instructors taking a student-centered approach intentionally account for students’ perspectives, empower students to take responsibility for their own learning, and seek to align instruction with students needs and interests as learners (Jang et al. 2010). Recognizing that the type of students in a class (e.g., non-science majors), their experience with the content (e.g., chemistry), and the course format (e.g., blended, online, traditional) may call for different teaching strategies, is also more likely to foster competence and autonomy (Deci and Ryan 2000; De George-Walker and Keeffe 2010).

Levesque-Bristol et al. (2006) explained that establishing an autonomy supportive, student-centered learning environment is critical to basic psychological need satisfaction, an assertion that is supported by SDT research in education settings (Ryan and Deci 2013, 2017; Black and Deci 2000). SDT researchers have also argued that social contexts that support one need also tend to support other needs (Vansteenkiste and Ryan 2013). While it is now recognized that basic psychological needs satisfaction is important with regard to internalization, development, and overall wellbeing, there is still much to learn about the relationships and interplay among the three needs (Ryan and Deci 2017; Van den Broeck and Rosen 2016). Recent research in educational settings has found evidence for the predominance of competence on motivation in the classroom (Bettencourt and Sheldon 2001; Yu 2019; Ntoumanis 2001; Talley et al. 2012). This perspective views competence as the most proximal of the three needs for developing self-regulated motivation, but recognizes that autonomy and relatedness are important mediating variable om considering the relationship between the learning environment and competence development.

As motivation becomes more self-regulated, the IMLM predicts that the a course environment that satisfies students basic psychological needs will facilitate psychological outcomes such as adaptation, adjustment, and growth as well as learning processes and strategies, such as knowledge transferability and ultimately student learning (Levesque-Bristol et al. 2006; Deci and Ryan 2008). In these environments, individuals have more responsibility for and control over their own learning, which helps them identify more with the course (Levesque-Bristol et al. 2006). Accordingly, the instructor’s role is to present course content in a learning environment that is perceived as student-centered, addresses the basic psychological needs, and encourages more self-regulated forms of motivation. These conditions have been linked to student learning (Levesque-Bristol et al. 2010; Black and Deci 2000; Williams and Deci 1996). Specifically, students whose motivation is more self-regulated are likely to experience deeper forms of learning than opposed to surface-level learning (Kyndt et al. 2011; Hooper 2009). Svinicki (2004) explains this relationship by noting that increased motivation helps the learner focus on relevant environmental stimuli and persevere even when learning become challenging. Further, applying what has been learned relates to success not only during college, but also in the future workplace and in everyday life (Haskell 2000).

The value of learning with deep understanding is that it resides and travels with the learner and facilitates application, or transfer, to other environments (Everett 2010). Toward this end, Levesque-Bristol et al. (2006) explained that the motivation-learning relationship is mediated by greater perceived knowledge transferability, alongside metacognition and engagement. Perceived knowledge transferability is students’ belief that what has been learned is important and can transfer beyond the immediate learning environment (Rhodes et al. 2006). The information is viewed as relevant to the learner, and therefore important to know, which enhances the likelihood of learning (Martin and Dowson 2009). Previous knowledge and experience, characteristics of the present task, and motivation interact to determine knowledge transfer (Nokes-Malach and Belenky 2011). Accordingly, perceived knowledge transferability is more likely if students are motivated to use what they learn in class and perceive coursework to be relevant for their future (Axtell et al. 1997).

In addition to being a key component of the IMLM, evidence suggests that knowledge transfer is more likely when individuals feel competent in the subject matter because they are better able to cope with constraints in the new environment where they are trying to apply previously learned skills (Billing 2007; Wang and Haggerty 2009). As such, competence has been identified as a key variable in the learning process and a major antecedent of knowledge transfer and future performance (Billing 2007; Kraiger et al. 1993; Wang and Haggerty 2009). Based on the preceding discussion of prior research and key tenets of SDT and the IMLM, we have developed a research-based, theoretically grounded, conceptual model to outline selected anticipated relationships among the learning climate, basic psychological needs satisfaction, self-regulated motivation, and perceived knowledge transferability (see Fig. 1). This conceptual model is supported with the following specific hypotheses:

  • H1: perceptions of a student-centered learning environment will enhance autonomy and relatedness directly

  • H2: Autonomy and relatedness will mediate the relationship between perceptions of a student-centered learning environment and competence

  • H3a: Competence will directly affect self-regulated motivation

  • H3b: Competence will serve as a mediator in understanding the predictive capacity of autonomy and competence on self-regulated motivation

  • H4: Self-regulated motivation will be a partial mediator in the relationship between competence and perceived knowledge transferability

  • H5: Basic psychological needs satisfaction and self-regulated motivation will serve a mediating capacity in understanding the association between the perceptions of student-centered learning and knowledge transferability

Method

Participants and Setting

The participants in the current investigation included 4385 (2200 males, 2185 females) undergraduate students at a public, research-focused university in the U.S. Midwest. The participants were on average 20.20 years old (SD = 2.64), and 322 (7.34%) were underrepresented minority students. With reference to ethnicity and nationality, the students were primarily European American (n = 2965; 67.50%), with Asian American (n = 178; 4.10%), Hispanic American (n = 158; 3.60%), and African American (n = 104; 2.40%) reflecting other common ethnicities. A total of 821 (18.70%) participants were international students. The composition of the sample included first year (n = 959; 21.90%), second year (n = 1357; 30.90%), third year (n = 1161; 26.50%), and fourth year students (n = 908; 20.70%). The students represented a broad range of majors, with the most common including health and human sciences (n = 791; 18.00%), technology (n = 655; 14.90%), engineering (n = 645; 14.70%), and science (n = 545; 12.40%). The average student had an overall GPA of 3.13 (SD = .56) and earned a GPA 3.13 (SD = .73) while being registered for 15.18 (SD = 2.70) credit hours during the semester in which data were collected. Complete participant demographic information is provided in Table 1.

Table 1 Demographic information for the participants
Full size table

Research Procedures

This investigation was one part of a larger study examining a campus-wide course redesign initiative at a large, research-focused university in the U.S. Midwest, and all of the relevant research procedures were approved by the lead author’s institutional review board. SDT and the IMLM framed the redesign so as to help faculty members deliver instruction in a manner that facilitated perceptions of student-centered learning. Participating faculty members were encouraged to consider how creating a student-centered learning climate could promote psychological need satisfaction and increase self-regulated motivation as they redesigned their courses. This occurred through a semester-long faculty learning community facilitated by the university’s faculty development center and included ongoing support for implementation of redesigned courses. Although all the students were taking courses that were part of the course redesign program, there was great variability in the types of courses, instructors, and implementation strategies are built into this program (see Levesque-Bristol et al. 2019 for a complete program overview). Since the IMLM is believed to apply to all college-level learning situations (Levesque-Bristol et al. 2006), and we were concerned with testing the relationships among the variables generally, all 11,656 students enrolled in at least one of the 186 courses received an invitation to participate in the study during the final two weeks of the semester. In total, 4450 provided at least partial responses, resulting in a response rate of 38.18%. After removing cases with a large amount of missing data, the final sample included 4385 participants; an adjusted response rate of 37.62%.

Study Instrumentation

The current investigation drew upon several previously validated instruments informed by SDT and linked to the IMLM. The final set of questions related to perceived knowledge transferability were developed for the purpose of this study. All instruments are described below and participants responded to all psychometric questions using a seven-point, Likert-type scale anchored by strongly disagree (1) and strongly agree (7).

Learning Climate

Students’ perception of the course learning climate was evaluated using the shortened, six-item version of the Learning Climate Questionnaire (LCQ; Williams and Deci 1996). Higher scores on the LCQ are associated with the perception of more student-centeredness or autonomy-support. Example items include: “I feel my instructor provides me with choices and options” and “my instructor conveyed confidence in my ability to do well in the course.” The internal consistency reliability of the LCQ has been illustrated previously (Williams and Deci 1996; Levesque-Bristol et al. 2010), and was appropriate in this study (Cronbach’s α = .95).

Basic Psychological Needs Satisfaction

The Basic Psychological Needs at Work Scale (BPNS; Kasser et al. 1992) was adapted to examine to autonomy, competence, and relatedness in classroom environments (Levesque-Bristol et al. 2010). In this study, only the positively worded items were used to understand students’ perceptions of autonomy (4 items), competence (3 items), and relatedness (4 items) satisfaction. Example items included “I am free to express my ideas and opinions in this course” (autonomy) and “most days I feel a sense of accomplishment from this course” (competence). The internal consistency for the BPNS has been illustrated previously (Levesque-Bristol et al. 2010), and was appropriate in this study (Cronbach’s α ranged from .80 to .87).

Motivation for Learning

The situational motivation scale (SIMS; Guay et al. 2000) was used to measure motivation for learning. This scale includes a total of 18 items that measure intrinsic motivation (IM), integration (INTEG), identification (IDEN), introjection (INTRO), external motivation (EM), and amotivation (AM) using three items each. Example items included “because acquiring all kinds of knowledge is fundamental for me” (INTEG) and “because it’s really fun” (IM). Previous research has demonstrated adequate internal consistency reliability for the SIMS (Guay et al. 2000), and the instrument met minimum criteria for acceptability in this study (Cronbach’s α ranged from .84 to .96). In accordance with procedures outlined by Levesque-Bristol and collogues (Levesque-Bristol et al. 2010), self-determination indices (SDIs) were calculated to convert the SIMS items into scores reflecting overall motivation. Three scores were constructed by combining one item from each type of motivation using the formula outlined in Eq. 1.

$$ {\mathrm{SDI}}_{\mathrm{i}}=3\ast \left({\mathrm{IM}}_{\mathrm{i}}\right)+2\ast \left({\mathrm{INTEG}}_{\mathrm{i}}\right)+1\ast \left({\mathrm{IDEN}}_{\mathrm{i}}\right)-1\ast \left({\mathrm{INTRO}}_{\mathrm{i}}\right)-2\ast \left({\mathrm{ER}}_{\mathrm{i}}\right)-3\ast \left({\mathrm{AM}}_{\mathrm{i}}\right) $$
(1)

Perceived Knowledge Transferability

Eight items were created for the purpose of this study to examine the extent to which students believed that the information they were learning was transferrable to their lives, careers, and future courses. To craft the questions, we explored the literature on knowledge transfer in various fields, such as education and business (Rhodes et al. 2006; Martin and Dowson 2009; Nokes-Malach and Belenky 2011; Axtell et al. 1997). We discussed the items internally, and received feedback from a panel of five external experts. Modifications to improve the structure of the items were made following feedback. Higher scores reflect the perception of knowledge transferability, which indicates that material was viewed as relevant beyond the immediate course. The final items included: (a) “I feel as if the material covered in this course is relevant to my future career,” (b) “information learned in this course will inform my future learning experiences,” (c) “I understand how I will use the information learned in this class in my professional life,” (d) “I believe that it is important for me to learn the information included in this course,” (e) “given the future career that I have chosen, it is important for me to learn the information covered in this class,” (f) “I feel confident in my ability to apply the course material in my professional life,” (g) “I feel confident in my ability to apply the course material in other classes that I have,” and (h) “the information learned in this course will help me become a more well-rounded individual.” The internal consistency reliability for the unidimensional perceived knowledge transferability construct was good (Cronbach’s α = .97).

Analytic Procedure

Confirmatory Factor Analysis

The IBM SPSS 23.0 program was used for preliminary data cleaning and screening in line with recommendations in the literature (Tabachnick and Fidell 2013) and LISREL 9.1 was used for all latent variable modeling. Prior to examining the relationships specified in the conceptual framework (Fig. 1), a concurrent CFA was used to examine the factorial, convergent, and discriminant validity of the measurement model. A concurrent CFA was used rather than individual CFAs because it is more parsimonious allows for the examination of discriminant validity through the latent correlations among study constructs (Teo et al. 2009). Factor (λ) loadings, composite reliability (ρc), and average variance extracted (AVE) scores were used to examine convergent validity. A λ loading ≥ .40 is statistically significant and an adequate indicator of the underlying construct. Similar to Cronbach’s α, the ρc values estimate internal consistency reliability (Diamantopoulos and Siguaw 2000), whereas AVE signifies the proportion of variance attributable to the underlying construct in relation to measurement error (Fornell and Larcker 1981). To be considered adequate, ρc for a latent construct should be ≥ .70 whereas AVE should be ≥ .50. The correlations among study constructs are often examined along with a comparison of \( \sqrt{\mathrm{AVE}} \) for a construct to the correlation between that construct and others in the model to determine discriminant validity (Teo et al. 2009). The construct is independent if all correlations in the model are ≤ .85 and the \( \sqrt{\mathrm{AVE}} \) values are > the relevant correlations.

Structural Equation Modeling

Following preliminary CFA analyses, maximum likelihood SEM was used to examine the hypothesized relationships outlined in Fig. 1 (Kline 2011). As an extension of CFA, SEM is a latent variable modeling approach that includes both the measurement model evaluated previously through CFA and a structural model that reflects hypothesized relationships among the study latent constructs. One key advantage of SEM is that all of these relationships are examined simultaneously through the application of variance and covariance matrix estimation. Following an initial model run, the significant tests for the structural pathways are examined and non-significant pathways are removed before the model is rerun (Hatcher 1994). Given that a maximum likelihood estimation was used, the normality of indicators variables was examined extensively (Byrne 1998). All indicators met Kline’s (2011) recommendations related to univariate normality, including skewness < |3.00| and kurtosis < |10.00|. Further analyses indicated. However, that the assumption of multivariate normality was not met, which led to the adoption of robust maximum likelihood estimation procedures. The robust maximum likelihood approach reduces dependence on normality by correcting the standard errors and χ2 statistics used in latent variable modeling (Li 2015). Toward this end, the Satorra-Bentler scaled χ2 (C3) is presented rather than the traditional χ2 for all model assessments (Satorra and Bentler 1994).

Goodness-of-Fit Estimates

Goodness-of-fit estimates are used in latent variable modeling to examine the extent to which the data fit the hypothesized model (Hatcher 1994). When a model fit is good it can be taken to mean that the data are a good fit for the model, whereas a poor fitting model likely indicates a problem with misspecification (Schreiber et al. 2006). Fit indices adopted in this study reflected recommendations in the literature (Brown 2006; Schreiber et al. 2006) and included χ2, the Tucker-Lewis index (TLI), the comparative fit index (CFI), the standardized root mean square residual (SRMR), and the root mean square error of approximation (RMSEA). While it is customary to provide the χ2 test statistic, the test is very sensitive to the size of the sample and is no longer considered an accurate model fit indicator (Teo et al. 2009; Schreiber et al. 2006). Therefore, it is provided, but not interpreted in this study. The RMSEA and SRMR ≤ .08, whereas TLI and CFI values should ≥ .90 (Brown 2006).

Results

Preliminary Analyses

Table 2 provides descriptive statistics for all indicator variables as well as the composite scores. Participants perceived a relatively student-centered learning climate (M = 5.37, SD = 1.32); high perceived knowledge transferability (M = 5.18, SD = 1.39); and moderate-to-high autonomy (M = 4.89, SD = 1.19), competence (M = 4.69, SD = 1.27), and relatedness (M = 4.98, SD = 1.12). On a scale ranging from −33.00 to 36.00, the average SDI score was 7.23 (SD = 11.37), which indicates that students felt a moderate level of self-determination in their course.

Table 2 Mean, standard deviation, skewness, and kurtosis of indicator variables and composite scores
Full size table

Examination of the Measurement Model

Following preliminary analyses, the analytic approach transitioned to concurrent CFA to examine the psychometric properties of the measurement model (Teo et al. 2009). The initial CFA analysis indicated that the data were an appropriate fit for the hypothesized model, C3(335) = 6506.51, p < .001; RMSEA = .08; SRMR = .05; TLI = .94; CFI = .95. Relative to construct reliability, the top panel of Table 3 includes the λ loadings and ρc and AVE values, all of which were strong and above the associated cut points. The bottom panel of Table 3 includes the latent correlation matrix. In line with expectations, all variables in the model correlated positively and significantly at the α = .01 level. The strongest correlations were among competence and LCQ (r = .85; p < .001), autonomy and competence (r = .85; p < .001), and competence and knowledge transferability (r = .83; p < .001). In relation to discriminant validity, all of the correlations were ≤ .85, and most of the \( \sqrt{\mathrm{AVE}} \) were larger than the associated correlations (see bottom panel of Table 3). Collectively, these CFA analyses provide support for the psychometric quality of the measurement model used in this study, which makes it appropriate to advance into SEM.

Table 3 Convergent and discriminant validity for all constructs included in the model
Full size table

Evaluation of the Conceptual Model

After verifying the psychometric quality of the measurement model, SEM was used to evaluate the hypothesized relationships in the structural model as outlined in Fig. 1. The latent disturbances among autonomy and relatedness were correlated in the model since the variables are conceptually connected. The initial SEM run indicated good model fit, C3(342) = 6691.75, p < .001, RMSEA = .08, SRMR = .05, TLI = .94, CFI = .95. Examination of the significance tests for the pathways in the structural model indicated that all were significant, eliminating the need to remove any and rerun the model. Figure 2 presents the final structural model with completely standardized regression coefficients. The learning climate directly predicted autonomy (β = .84, p < .01) and relatedness (β = .59, p < .01), and was indirectly related to increased competence (β = .70, p < .01) and perceived knowledge transferability (β = .59, p < .01). Autonomy directly predicted competence (β = .73, p < .01), and was indirectly related to perceived knowledge transferability (β = .60, p < .01) and self-regulated motivation (β = .52, p < .01).

Fig. 2
figure 2

Final structural model with completely standardized regression path coefficients, values in parentheses and dashed lines represent indirect relationships, error terms are omitted for simplicity in representation, C3(342) = 6691.75, p < .001, RMSEA = .08, SRMR = .05, TLI = .94, CFI = .95, Self-Determin Index = self-determination index, Knowledge Transfer = perceived knowledge transferability, *p < .05, **p < .01

Full size image

Relatedness significantly predicted competence, but the β was small compared to some of the other pathways in the model (β = .13, p < .01). Similar findings were true relative to the role of relatedness in predicting self-regulated motivation (β = .09, p < .01) and knowledge transferability (β = .11, p < .01). Competence directly predicted both self-regulated motivation (β = .71, p < .01) and perceived knowledge transferability (β = .72, p < .01), and had an indirect effect on perceived knowledge transferability (β = .11). Finally, self-regulated motivation directly predicted perceived knowledge transferability (β = .15, p < .01). Collectively, these results confirm that the learning climate, basic psychological needs satisfaction, and self-regulated motivation directly and indirectly predicted perceptions of knowledge transferability.

Discussion

The purpose of this investigation was to understand and evaluate the basic IMLM tenets that link together the perceived learning climate, basic psychological needs satisfaction, self-regulated motivation, and perceived knowledge transferability. We focused on these variables in an effort to test an extended motivational sequence specified through SDT and the applied within the IMLM using latent variable modeling applications. Generally, the results support the basic assumptions of the IMLM and indicate that a more student-centered learning climate is associated with autonomy, competence, and relatedness the development of self-regulated motivation; and increased perceived knowledge transferability. Based in SDT and the IMLM, the learning climate was modeled as an antecedent for the development of motivational outcomes in educational environments (Ryan and Deci 2013), and perceived knowledge transferability, through its association with basic psychological needs satisfaction. In the proposed model, the learning climate was tested as the antecedent, knowledge transferability as the learning process outcome, and the motivation variables served as mediating variables.

The association between learning climate and autonomy and relatedness is proposed to be a direct one (H1), whereas the association with competence is indirect, through autonomy and relatedness (H2). The second hypothesis was based on evidence indicating that autonomy and relatedness are required for students to develop feelings of competence (e.g., Talley et al. 2012). Results support these hypotheses indicating that perceptions of the learning environment directly predict autonomy and competence and have an indirect relationship with competence. These findings support previous research suggesting that the development of autonomy and relatedness is required for individuals to feel competence (Bettencourt and Sheldon 2001; Yu 2019), and highlight the role of the instructor in developing an environment in which students have the opportunity to build relationships, and have opportunities to exercise their voices (Hooper 2009). Considering that prior research has identified the importance of basic psychological needs satisfaction in fostering autonomous motivation (Deci and Ryan 2008), the SDI was entered into the model next as a measure of self-regulated motivation. It was further hypothesized that competence would positively predict self-regulated motivation (H3a), and that the association of autonomy and relatedness would be mediated through competence (H3b). Results support these two sub-hypotheses, thus affirming the importance of basic psychological needs satisfaction, and competence in particular, in fostering self-regulated motivation (Talley et al. 2012; Yu 2019).

Perceived knowledge transferability was positioned at the end of the model based on relationships hypothesized in the IMLM (Levesque-Bristol et al. 2006), prior research highlighting the relationship between intrinsic motivation and deeper forms of learning (Kyndt et al. 2011), and linkages between competence and perceived knowledge transfer in prior research (Billing 2007; Wang and Haggerty 2009). It was hypothesized that relationship between competence on perceived knowledge transferability would be partially mediated through self-regulated motivation (H4). Support was found for this hypothesis, although competence was a stronger direct predictor of perceived knowledge transferability than it was in the indirect sense. In addition to the hypothesized direct pathways in the model, there were several notable indirect relationship between the motivational variables and perceived knowledge transferability (H5). Importantly, results indicated that students’ perception of the learning climate, along with autonomy and relatedness, exhibited significant indirect relationships with both self-regulated motivation and knowledge transferability. This reaffirms the importance of the full motivational sequence outlined in the IMLM and the position of the learning climate as an antecedent, the motivation variables as mediators, and perceived knowledge transferability as a learning process and outcome (Ryan and Deci 2013; Levesque-Bristol et al. 2006).

Implications for Practice

In addition to providing evidence that supports key tenets of SDT and the IMLM, this study highlights implications for teaching practice in college environments. Among the priorities of college education is the need to help individuals develop the knowledge and skills needed for success in their future careers (Thomas 2007; Haskell 2000). It is critical, therefore, to develop environment conducive to the perception of knowledge transferability so to help increase the relevance of information students are learning. The ability of instructors to foster perceptions of student-centeredness is essential to this perception, which reaffirms the importance of actively engaging students in the learning process (Hooper 2009). In particular, it appears that learning needs to take place in an environment that fosters autonomy support and, to a lesser extent, promotes relatedness. Although the direct relationship between relatedness and competence was significant, it was small compared to the relationship between autonomy and competence.

Importantly, when asked about the biggest mistakes that instructors make when teaching, college students highlight the predominance of a teacher-centered learning environment that lacks engagement and autonomy support (Richards and Velasquez 2014). Questions about constructing an appropriate learning environment connect back to discussions and debates about behaviorist and constructivist-oriented pedagogies (e.g., Benware and Deci 1984). While behaviorist approaches tend to disenfranchise students by positioning the instructor as the all-knowing “sage of the stage,” constructivist-oriented approaches integrate active learning strategies that help students feel as if they are integral components of the course experience (National Research Council 2000; Wang et al. 2017; Alessio 2004). Active learning strategies appear central to the development of course environments that foster motivational outcomes that lead to deeper learning (Ryan and Deci 2013; Kyndt et al. 2011).

The focus on the motivational variables in higher education was supported in the current investigation and aligns with key tenets specified through SDT and adopted within the IMLM (Ryan and Deci 2013; Levesque-Bristol et al. 2006). Even though the strongest direct association with knowledge transferability was with the satisfaciton of the need for competence, we want to highlight the importance of the other needs and the creation of a student-centered environment. It may make the most sense to emphasize the satisfaction of the need for competence in higher education in order to foster relevant outcomes. However, instructors and practitioners should also emphasize the creation of environments that are student-centered, which meet the needs of autonomy and relatedness, because these variables will foster self-regulated motivation and knowledge transferability through perceived competence. Although the need for competence appears predominant, it needs to be fostered in an envrionment that is autonomy supportive.

Limitations of the Research Design

While this study offers several implications for practice, these are tempered by limitations that should be acknowledged. First, participants were primarily White and all were recruited from one, research university in the U.S. Midwest. The sample may, therefore, threaten generalizability of these findings beyond similar universities environments. For example, caution should be taken when generalizing findings as smaller colleges and universities with more teaching-focused missions, and more diverse contexts, such as historically black colleges and universities. The cross-sectional design used in this study should also be acknowledged as a limitation because students’ perceptions of learning experiences are likely to change over time and in response to course experiences. Accordingly, more robust analyses would be possible through a repeated measures design. The cross-sectional design also precluded longitudinal applications of SEM, which could produce more reliable results (Tabachnick and Fidell 2013).

Further, while response rates ranging from 30 to 40% are relatively desirable when using online surveys, and the response rates do not always correlate with the representativeness of a sample (Lambert and Miller 2014), response bias cannot be ruled out and should be viewed as a potential limitation. Further, while sampling a variety of courses from several disciplines helps in the generalizability of the results, it also precluded an examination of differences across student demographics and individual courses. A more in-depth analysis of course learning climates, or experimental manipulation that compares student-centered and teacher-centered climates for certain courses, may have yielded more concrete recommendations for the ways in which college instructors can promote perceptions of knowledge transferability.

In terms of the measures used, although most of the surveys used were previously validated, the measure of perceived knowledge transferability was created for the purpose of this study. Preliminary CFA analyses provided initial support for the factorial validity of this measure, but additional research will be necessary further evaluate validity and reliability. In addition, even though the creation of self-determination indices (SDIs) as a measurement of self-regulated motivation is a popular scoring protocol used in several studies (Grolnick and Ryan 1989; Levesque et al. 2004; Richards and Levesque-Bristol 2016; Vallerand 1997), they should also be interpreted with caution. This is particularly given the weights assigned to the different motivational constructs and the practice of multiplying these weights in the scoring protocol (Chemolli and Gagne 2014). Accordingly, future researchers may consider other options for measuring student motivation.

Future Directions

One key contribution of this study is linking motivation and perceived knowledge transferability, which is an association that has not been well understood to date (Nokes-Malach and Belenky 2011). Accordingly, if college instructors want to enhance student motivation along with the perception that learning will transfer beyond the current course, the learning climate seems paramount (Levesque-Bristol et al. 2006). While learning outcomes were not measured as part of the current investigation, the IMLM and previous research support a link between perceptions of knowledge transferability and actual learning (Martin and Dowson 2009). Nevertheless, there is a need for more direct evaluations of student performance to fully examine connections between motivation, perceived knowledge transferability, and learning (Brooks et al. 2011). Similarly, longitudinal research is necessary to understand how student-centered learning environments develop over the course of a semester or across courses within the same program. Scholars should also consider how student demographic variables, nationality, ethnicity, gender, class rank, and college major relate to perceptions of the learning environment. While evidence suggests that the relationships among variables related to SDT and specified through the IMLM are invariant across cultural and geographic boundaries (Levesque et al. 2004), further investigation into these relationships may help scholars and college-level instructors better understand how students experience the learning environment. Finally, other elements of the IMLM, such as the course content and institutional and extended community, that are believed to frame the teaching-learning exchange need to be evaluated in more detail (Levesque-Bristol et al. 2006). While the learning climate is important, college education does not take place in a cultural bubble and an abundance of external and internal factors need to be considered in greater detail.