In this article, the researchers’ main aim was to report on the psychometric test of clinical learning environment test inventory. The clinical learning environment inventory consists of 42 items which are classified into six scales. The scales are task orientation, personalization, innovation, satisfaction, and individualization and student involvement. These scales were set up to check the perceptions of nursing students of learning during on placement in various clinical settings (Newton et al. 2010).
Clinical placements in undergraduate healthcare programmes are crucial for improving and development of skills, professional socialization and integrating learning (Newton et al. 2010). Various studies have been carried out to determine what determines a good learning environment for graduates and new students. Most students prefer working in clinical environments that they have positive experiences with. Australia has adopted models such as the MASH model – the model was an agreement between the clinical agency where students undertake their clinical placements in second and third years of their three year study. Secondly, there is the preceptorship model – it increases collaboration between universities and the healthcare through assigning students to an experienced nurse. This model is clearly different from the traditional model which a clinical teacher coordinates with the university to support nursing students in different healthcare facilities. Outcomes of the MASH model were examined to determine the impact the clinical placement’s model had on student’s perception. The clinical learning environment inventory (CLEI) was chosen due to the item’s reflection on dimensions of clinical settings. The CLEI facilitates the assessment of nurses’ perceptions of psycho-social aspects of the clinical learning environment (Chan 2003).
The researchers selected a sample of students in the year 2 or 3 who did not have any experience in the clinical environment. The students were Bachelor of Nursing degree students from two campuses that participated in the study. After the clinical placement, the students were given CLEI mid-year to complete. The clinical placements involved intensive care, accident and emergency and general and surgical wards. Both the traditional and the MASH models were involved in the study.
In the research, 767 undergraduate students completed the CLEI during 3 years collection period. Students who had completed a second survey in the preceding years were excluded to avoid accommodation. The final sample was therefore 659 first time responses. Five hundred and thirteen (513) students responded to all the questions, therefore they were eligible for factor analysis. “The CLEI is an instrument that has 42 positively and negatively worded items with seven items assessing each scale: personalization, student involvement, task orientation, innovation, satisfaction and individualization” (Chan 2002, p. 72). The student response was gauged using 4-point Linkert scale: “strongly agree, agree, disagree and strongly disagree.” In the original scale, the authors of this article scored items 5, 4, 2 and 1 according the order above. Omitted or invalid responses were set to be 3. All missing data was also excluded from the researchers’ analyses; the authors assumed that a missing response is not put at the midpoint of the scale, and negative items were reverse-scored to reflect the same direction (Newton et al. 2010, p. 632). The researchers’ CLEI were modified to include an extra ten parallel items called preceptor. The assumptions made by the researchers for exploratory factor analysis were that the variables demonstrated between 0.3 and 0.7, that all variables were reliably measured (the Cronbach’s Alpha should be >0.7), and that the data in normally distributed-factor loading were free to vary by using maximum likelihood extraction.
Approval of the research was authenticated by the university and participating health care organizations. The participation was voluntary and consent was implied by the completion of the questionnaire. The researchers then used SPSS version 16 to analyze their collected data. The Kaiser-Meyer-Olkin (KMO) index was also used by the researchers to detect whether there was sufficient covariance in the scale items to warrant factor analysis (Newton et al. 2009). They also used the Barlet test of Sphericity to determine whether the correlation matrix was an identity matrix and therefore unsuitable for factor analysis. To determine the factor analysis matrix, the researchers found that the KMO was 0.929 and compared to the required minimum of 0.5 and the Barlett’s test was 8390.0, p< 0.0001 indicated that there was sufficient variance for analysis (Newton et al. 2010). Through developing the correlation matrix, they extracted the variables with high inter-correlations.
The authors of the article had initially explored the data and they realized they had no constraints using principle components analysis. They found the Eigen values which explained the total amount of variance explained by the factor and weight/loadings of each variable on each factor. The matrix was factor analyzed and resulted into a factor matrix. They further applied a Varimax rotation with Kaiser Normalization since the original dimensions implied orthogonal factors. A second analysis was done in which the PCA was constrained to six factors. Loadings <0.3 were suppressed. They also conducted Cronbach’s Alpha analysis to determine the consistencies in each scale.
The authors of this article present a table showing the results that were accrued from the analysis. In their results, 76% were participating in non-MASH models of clinical education and 53% had one or more preceptors. Non-constrained possible factors in the first PCA yielded nine factors explaining 56% of variance. However factor 1 suggested no coherent theme. Two factors were also expressed by three items. The second analysis carried out by the author also showed that data constrained with a 6-factor solution initially explained 49% of the variance.
Factor 1 was used by the authors to represent self centeredness; it contains all the seven Chan’s personalization items along six other items from various scales. Factor 2 is the affordance and engagement containing all seven items of Chan’s satisfaction. Factor 3 is individualization, students having some control on clinical experience and facilitating individual learning needs. Other factors include factor 4 – “fostering workplace learning”, factor 5 – valuing the work of nurses and factor 6 –adaptive and innovative culture.
The researchers’ limitation was due to modification of the CLEI which might have affected the student’s response. To determine the factor analyses, the factors were grouped into six by the researchers: Factor 1 and 2 accounted for more than 70% indicating that they were very reliable while factors 3 and 4 had more than 60% indicating that they were reliable. Factors 5 and 6 were less than 60% indicating slightly less reliability; these are the common factors influencing the responses. Factor analysis requires that the estimated factors extracted should account for at least 60% of the total variance and any factor to account for at least 5% of the variance. In the analysis of this research, factors only explain 51% of variance in the data and the maximum variance explained by factor analysis of other scales due to the perception of learning environment was 64%.
The correlation matrix determines whether the data is suitable for factor analysis or not; if the correlation matrix is an identity matrix through applying the Bartlett’s test of Sphericity, then factor analysis would be unsuitable for that case. The number of the sample used was also enough to use factor analysis effectively- the ratio of sample to item was 12.2:1. From the article, the data is evidently normally distributed because of the large number of random variables. Variables were reliably measured appropriately from the sample size collected by the researchers. However there were a few factors which were less reliable such as factors 5 and 6. Reliability is measured through the Cronbach Alpha which measures consistency of the collected data, the Cronbach should be normally > 0.7. Their conclusion is correct and based on their collected data. They concluded that the factor analysis offered alternative scales to the original CLEI; their scales took account of nuance in the workplace learning (Billet 2004).
Confirmatory factor analysis is used to test a specific hypothesis and significance to the observed variables. Exploratory factor analysis is used to test the interdependence among the observed variables in some set of data. Exploratory is mostly used if there is no underlying hypothesis. The advantage of confirmatory factor analysis over exploratory factor analysis is that a specific factor structure must be specified. Where item loads are indicated, CFA allows researchers to perform statistical comparisons of factors models and CFA provides a model fit to the hypothesized factor to the observed data.
References
Billet, S. (2004). Workplace participatory practices conceptualizing workplaces as learning environments. The Journal of Workplace Learning, 16(6), 312–324.
Chan, D. (2002). Development of the clinical learning environment inventory: Using the theoretical framework of learning environment studies to assess nursing students’ perceptions of the hospital as a learning environment. Journal of Nursing Education, 41(2), 69–75.
Chan, D. S. K. (2003). Validation of the clinical learning environment inventory. Western Journal of Nursing Research, 25(5), 519–532.
Newton, J. M., Jolly, B. C., Ockerby, C. M. & Cross, W. M. (2010). Clinical Learning Environment Inventory: factor analysis. Journal of Advanced Nursing, 66(6), 1371–1381.
Newton, J. M., Billet, S. & Ockerby, C. (2009). Journeying through clinical placements: an examination of six student cases. Nurse Education Today 29, 630–634.