Psychometric properties of Clinical Learning Environment Inventory and its association with Moroccan nursing students’ satisfaction: A PLS-SEM approach

Background

Nursing is a practical field, and as such, clinical practicum embodies a substantial part of the pre-registration nursing courses. Clinical experiences offer realistic prospects for students to gain skills in the actual context and help put theory into practice. Combining cognitive, psychomotor, and affective abilities with the capacity to come up with solutions in a real clinical context presents both achievements and challenges for students (Chan & Ip, 2007). Indeed, to shape nursing students’ skills and boost their self-confidence, a successful clinical education curriculum should focus on giving students a positive and practical learning experience.

There are significant differences between classroom learning and clinical setting because students’ full grasp of theoretical nursing principles does not guarantee their translation into practice (Yazdankhahfard et al., 2020). Unlike organized classroom learning, clinical practice occurs in a dynamic complex social context, which requires high levels of resilience in addition to cognitive, psychomotor, and emotional skills (Chan, 2001). The quality of placement sites has raised suspicions about students’ learning outcomes, and many studies suggest that not all CLE are conducive to learning (Chan et al., 2018). Clinical learning quality depends on many factors, such as applying theory to practice, excellent mentoring, constructive feedback, student integration through effective interpersonal relationships, task orientation, professional leadership, outstanding mentoring, and positive criticism (Tuomikoski et al., 2020). Moreover, impaired interpersonal relationships, a lack of assistance, and challenging learning prospects are some negative experiences that students may undergo (Ten Hoeve et al., 2018). These experiences can vary from one clinical context to another, as clinical education planning may change according to the location or country (Antohe et al., 2016).

The pre-registration nursing program in Morocco includes clinical instruction as a crucial component. Twelve public Higher Institutes of Nursing and Health Techniques (ISPITS) offer a 3-year Bachelor of Nursing. Students obtain the right to practice as registered Nurses after graduating from these institutes. However, as indeed other international nursing institutes, clinical education must be included in the curricula for at least 32 weeks (1,280 hours). Similarly, Moroccan nursing institutes consider the clinical practice a crucial part of students’ professional skills development. This recognition is justified by the overall proportion of credit hours allocated to the internship modules described in the nursing curricula. The amount of actual learning time varies according to nursing specialties (960 hours for family nurse practitioners to 1440 hours for polyvalent nurses). Students spend a significant amount of time in a clinical practicum before becoming staff nurses, where clinical educators help and gradually incorporate them into the nursing profession.

The measurement of the CLE using a validated tool is crucial. Different approaches were used to assess CLE, such as case studies, interaction analysis, and students’ and teachers’ perceptions of the environment’s psychosocial qualities. The CLE assessment from the student’s perspective was adopted because it is cost-effective. Moroccan nursing institutes need efficient tools to assess students’ perceptions of CLE quality. However, there is still a massive lack of knowledge about the Moroccan CLE setting (ElIdrissi et al., 2022; Guejdad et al., 2022).

The CLEI scale was developed from the CUCEI instrument (College and University Classroom Environment Inventory) after being modified for the unique hospital environment (Fraser, 2020). It consists of personalization, student involvement, task orientation, innovation, and individualization subscales. Each subscale is assessed by seven items. In addition, a sixth scale called “Student’s Satisfaction” was utilized by Chan (2002b) as a dependent variable. There are two versions of the CLEI scale, one actual and another preferred form. The two-form items are almost identical, with slight phrasing modifications. While the CLEI appears to be widely used, no previous study has reported a full structural validity of the CLEI scale combining both exploratory factor analysis and confirmatory factor analysis (Chan, 2001; Chan, 2002b, 2003; Chan & Ip, 2007; Ip & Chan, 2005; Midgley, 2006; Newton et al., 2010; Serena & Anna, 2009).

This study examined the psychometric properties of the CLEI scale and its subscales association with Moroccan nursing students’ satisfaction. Therefore, in reference to Chan’s study (Chan, 2001), a research model was conceptualized (Figure 1), and five hypotheses were formulated as follows:

H01: Personalization (PER) has a positive and significant impact on Students’ Satisfaction (SAT).

H02: Involvement (INV) has a positive and significant impact on Students’ Satisfaction (SAT).

H03: Task Orientation (ORI) has a positive and significant impact on Students’ Satisfaction (SAT).

H04: Innovation (INO) has a positive and significant impact on Students’ Satisfaction (SAT).

H05: Individualization (IND) has a positive and significant impact on Students’ Satisfaction (SAT).

Methods

Results

Partial Least-Squares Structural Equation Modeling

Measurement model: Construct reliability and convergent validity

The construct reliability, convergent validity, and discriminant validity were computed from the measurement model (Hair et al., 2019). We, therefore, analyzed the individual loads (λ) with their respective latent variables (λ ≥ 0.7 is accepted). CCA maintained five factors and discarded two items with lower loading values (INV20 = 0.34 and INO41 = 0.52). The construct reliability was estimated by the outer loading and CR values, while the convergent validity was assessed by the AVE values (Table 3 and Figure 2).

Every indicator below 0.7 crossed the 0.4 level, with ORI22 having the lowest value at 0.61. Whether to keep or delete an item depends on the high outer loadings of the other items and their impact on CR and AVE values when the outer loading is between 0.4 and 0.7 (Manfrin et al., 2019). The reliability coefficient for each of the constructs exhibits sufficient internal consistency of the measures since the CR coefficients for each construct were above 0.70 cutoffs (rhoC= 0.83-0.92). Concerning the convergent validity, the AVE values (0.56 - 0.71) demonstrated that the five constructs had reached the least benchmark of 0.50.

As shown in Table 3, √AVE for each latent variable was higher than all factors matrix correlations, and the highest HTMT value was 0.27 (IND, SAT), below the maximum level of acceptable for HTMT levels. Moreover, the value “1” was not present in the HTMT confidence interval, represented by the values in brackets (Table 4). This demonstrated that the scale being employed had strong discriminant validity (Hair Jr et al., 2020).

Table 3: Construct validity and reliability of PLS-SEM of CLEI

Factor/item	Factors loading	Cronbach’s alpha	Composite reliability (rhoC)	AVE
Personalization
PER1	0.87	0.93	0.92	0.71
PER7	0.78
PER13	0.77
PER19	0.89
PER25	0.76
PER31	0.90
PER37	0.90
Involvement
INV2	0.80	0.89	0.91	0.62
INV8	0.82
INV20	0.89
INV26	0.77
INV32	0.86
Task orientation
ORI4	0.87	0.89	0.90	0.60
ORI10	0.77
ORI16	0.74
ORI22	0.61
ORI28	0.80
ORI34	0.78
ORI40	0.79
Innovation
INO5	0.83	0.91	0.88	0.63
INO11	0.86
INO17	0.68
INO23	0.83
INO29	0.82
INO35	0.72
Individualization
IND12	0.79	0.75	0.83	0.56
IND24	0.70
IND30	0.81
IND42	0.69
Satisfaction
SAT3	0.89	0.92	0.90	0.70
SAT9	0.87
SAT15	0.77
SAT21	0.84
SAT27	0.67
SAT33	0.88
SAT39	0.85

Table 4: Discriminant validity coefficients

Factor	Fornell–Larcker test						Heterotrait–monotrait ratio (HTMT) [2.5% - 97.5% CI]
	F1	F2	F3	F4	F5	F6	F1	F2	F3	F4	F5
PER	0.84
INV	0.16	0.82					0.22 [0.11, 0.33]
ORI	0.12	-0.05	0.77				0.16 [0.12, 0.23]	0.14 [0.08, 0.22]
INO	0.22	-0.17	-0.15	0.79			0.25 [0.12, 0.40]	0.22 [0.11, 0.35]	0.16 [0.11, 0.30]
IND	0.06	0.02	0.18	-0.06	0.75		0.11 [0.10, 0.23]	0.17 [ 0.10, 0.25]	0.23 [0.13, 0.38]	0.19 [0.13, 0.27]
SAT	-0.10	-0.16	0.28	-0.12	-0.04	0.83	0.10 [ 0.08, 0.22]	0.17 [ 0.10, 0.30]	0.28 [0.17, 0.40]	0.13 [0.08, 0.24]	0.12 [0.07, 0.22]

Structural Model Assessment

All VIF values are lower than five, which ascertained the absence of a multicollinearity effect among CLEI factors (Table 5). The structural model was tested using a resample of 1000 bootstrap samples (Streukens & Leroi-Werelds, 2016). The standardized beta, t-test, p-value, and F square were estimated using the bootstrapping technique, specifying the hypotheses’ results. The direct effects of all the independent variables (PER, INV, ORI, INO, and IND) were considered while determining the impact of various constructs on the dependent variable (student satisfaction). Task orientation showed the greatest direct influence on student satisfaction (β = 0.29, t = 4.94, f² = 0.10), followed by involvement with a weak negative significant effect (β = -0.14, t = -1.96, f² = 0.02). Thus, task orientation might be the most important element influencing the students’ satisfaction in CLE. The beta value of 0.29 suggests a 29% variation in nurse students’ satisfaction explained by the ORI subscale. Meanwhile, PER, INV, INO, and IND showed no significant relationship with students’ satisfaction (f² = 0.01). This gives support for the H01, H02, H04, and H05 hypotheses but fails to support the H03 hypothesis.

Table 5: Path coefficients and hypothesis testing

Hypotheses		β	2.5% - 97.5% CI	t-value	p-value	f2	Effect size	VIF
H01	PER→SAT	-0.09	-0.28, 0.21	-0.70	0.24	0.01	No effect	1.13
H02	INV→SAT	-0.14	-0.27, 0.00	-1.96	0.03*	0.02	No effect	1.10
H03	ORI→SAT	0.29	0.17, 0.40	4.94	0.001***	0.10	Small	1.07
H04	INO→SAT	-0.10	-0.27, 0.25	-0.65	0.26	0.01	No effect	1.14
H05	IND→SAT	-0.09	-0.25, 0.13	-0.88	0.19	0.01	No effect	1.03

Discussion

The CLEI reliability was examined using CA and CR, which were computed from EFA and PLS-SEM models. The convergent validity was examined using cross-loadings and AVE, while the discriminant validity was tested by the Fornell-Larcker and Heterotrait-Monotrait tests. All the subscales indicated acceptable CA values, ranging from 0.74 to 0.94, in agreement with previous studies (Chan, 2001, 2002a). The CR of the scale, estimated by computing rhoC, revealed that all the values lie above the standard value of 0.7, which reflects good reliability. Moreover, the levels of internal consistency are higher than those found in a previous study (Woo & Li, 2020). Of note, none of these previous studies, including those of Chan (2002b), has examined the factorial dimensions of the CLE scale using SEM techniques. In addition, very few studies reported factorial analysis of the CLEI scale, but surprisingly they computed CA from factors loadings as a test of reliability (Ergezen et al., 2022; Hudacek et al., 2019). However, contrary to CR, the CA coefficient does not take into consideration the varying factor loadings of the items. Thus, it is no longer warranted to rely solely on CA as a reliability indicator (McNeish, 2018).

The main purpose of this study was to validate the CLEI scale in the Moroccan context following Chan’s scale structure, which considered the satisfaction scale the dependent variable (Chan, 2002a). Then, the five-factor scale was extracted through EFA and validated using a PLS-SEM approach because of the non-normal distribution and the sample size (Hair Jr et al., 2021).

Interestingly, the results showed that a PLS-SEM using CCA generated the best model fit in terms of factor loadings for this version of the CLEI scale. The scale’s developing structural model was then verified using a Bootstrapping analysis approach. The current scale version showed good structural model fit characteristics and convergent and discriminant validity. However, Chan’s study (Chan, 2003) assesses the scale validity using only items and factor correlations. However, these first-generation methods are criticized due to the assumption that every variable is observable, error-free, and has a simple model structure (Hair et al., 2019).

Associations between the satisfaction subscale and the CLEI-five subscales were investigated using PLS-SEM path analysis. The associations were not significant between students’ perceptions of personalization, innovation, and individualization, and their level of satisfaction, while there was a slightly and strongly significant association, respectively, for involvement and task orientation. Nevertheless, students who placed high importance on task orientation in their clinical environment reported much higher student satisfaction (β = 0.29, p <0.001). These results are partly in agreement with those of Chan (2002b), reporting that students who gave task orientation, involvement, individualization, and innovation a high priority in their clinical placement experienced significantly higher satisfaction levels (β = 0.17-0.37). The latter demonstrated that student satisfaction is significantly greater in students who highly valued task orientation, involvement, individualization, and innovation in their clinical placement (β values ranged from 0.17 to 0.37). Whereas, and in congruence with our findings, they showed the highest standardized path coefficient of direct effect from Task Orientation to students’ satisfaction (β = 0.37, p <0.001). This ascertains that nurse students put more emphasis on the importance of task orientation, which reflects the extent to which guidelines for clinical activities are well planned. Several studies on this topic confirmed that many nursing students found clinical experiences to be anxiety-inducing and frequently felt exposed to these settings due, in part, to the lack of clear instructions given to students (Kushnir et al., 2014). Therefore, clinical educators must examine how vulnerable nursing students are on the clinical wards as a potential means of lowering their anxiety (Chan, 2002a).

On the other hand, and in agreement with a previous study (Henderson et al., 2012), no association had been demonstrated between innovation and student satisfaction. This pleads in favor of possible students’ disinclination towards innovation, which may indicate that educators adopt a teacher-centered approach rather than a student-centered approach (Abualhaija, 2019). In addition, these findings are in dissonance with those of Chan (2002b) concerning other construct associations. This could be attributed to the caveats that underlie the first-generation multivariate methods used by previous studies, including multiple regression analysis. The latter assumes a simple model structure involving one layer of dependent and independent variables that can only be estimated piecewise rather than simultaneously with SEM, which could, subsequently, distort the quality of the result (Sarstedt et al., 2020).

Additionally, since students’ perceptions may fluctuate depending on the unique traits of each country, various factorial designs might reveal cultural variations through different settings. For instance, Moroccan students might be less sensitive to authorities compared to European people (Ciavolino et al., 2022; Mohammadi et al., 2020). This might explain why Moroccan students did not place much emphasis on the personalization dimension when compared to previous studies (Chan, 2001). But this does not exclude the positive impact of interpersonal relationships between clinical ward participants on the students’ outcomes (Ergezen et al., 2022).

Conclusion

The results of this study reflect the prior work of Chan (2001) in terms of the psychometric properties of the CLEI scale. Task orientation was found to be strongly connected with students’ levels of satisfaction following their clinical placement. Consequently, the students showing high expectations for task orientation demonstrated a highly positive perception of their clinical placement outcome. Contrary to other cultural backgrounds, personalization, involvement, innovation, and individualization subscales showed insignificant direct effects on the students’ satisfaction in the Moroccan context. Therefore, it is of paramount importance that clinical educators tailor their instructions to the students’ needs to enhance learning and reduce anxiety in clinical settings. Overall, this study is the first to use SEM techniques to validate the CLEI instrument in Moroccan clinical wards. Furthermore, because this short CLEI version is a rapid, accurate, and reliable tool for evaluating CLE, nursing facilitators may use it to assess student outcomes and improve clinical supervision.