Design
This cross-sectional study was performed to evaluate the effect of SDL on CC with the mediating role of CDL by using structural equation modeling (SEM) at Shahid Beheshti University of Medical Sciences (SBMU) in Tehran, Iran during 2020-2021.
Participants and setting
The participants included nursing students undergoing the internship program in the 7th and 8th semesters at SBMU. Titles of clinical courses in the seventh semester of nursing education were medical-surgical nursing (eight credits) and critical care nursing (three credits). In addition, the titles of courses in the eighth semester were emergency nursing care (two credits), maternal health care (two credits), pediatric nursing (two credits), community nursing care (two credits), and nursing management (two credits). With the exception of community nursing care, training for the rest of the clinical courses was provided in various departments of hospitals affiliated with SBMU.
SBMU annually accepts 220 applicants to study nursing in the national centralized annual exam for the fall and spring semesters (100-120 per semester). Therefore, every year, between 200 and 240 nursing students are studying in the last year of nursing education (i.e., the 7th and 8th semesters). In this study, for two consecutive years, 344 internship nursing students who were in their final year of study and were undergoing the internship course, were included in the study. Following eliminating incomplete tools, a total of 300 internship nursing students were enrolled in the study. In fact, the tool completion rate was reported at 87%.
Instruments
Demographic information questionnaire
In this study, a demographic characteristics questionnaire was used to assess five criteria of age, gender, mean grade point average, nursing work experience as a student job, and duration of employment.
Self-Directed Learning Readiness Scale for Nursing Education
At this stage, the Self-Directed Learning Readiness Scale for Nursing Education was used to measure SDL. The tool was introduced by Fisher et al. in 2001 for the first time with 40 phrases. In 2010, Fisher and King conducted a psychometric assessment of the tool, which led to the extraction of a 29-item tool with three subscales of self-management (10 items), desire to learning (9 items), and self-control (10 items). The items are scored based on a five-point Likert scale from completely agree (score 5) to completely disagree (score 1). Notably, items 2, 15, and 21 are scored reversely [26]. In addition, the lowest and highest mean scores of the tool are one and five, respectively. It is also worth noting that the reliability and validity of the Farsi version of the tool were assessed by Nadi and Sadjadian for medical and dental students, and all three subscales had high internal consistency coefficients. The Cronbach’s alpha, the Spearman-Brown coefficient, Guttman scale, and retest coefficient were reported at 0.913, 0.899, 0.898, and 0.861, respectively, all of which confirmed the reliability of the tool [27].
Undergraduate Clinical Education Environment Measure
At this stage, Undergraduate Clinical Education Environment Measure was applied to measure CLE. The measure was first designed at Lund University in 2012 based on theories of experiential learning and social participation. The tool encompasses 25 items and four subscales of preparedness for student entry (6 items), opportunities to learning in and through work and quality supervision (11 items), workplace interaction patterns and student inclusion (6 items), and equal treatment (2 items). The items are scored based on a five-point Likert scale from completely disagree (one score) to completely disagree (five scores). A higher score is indicative of a higher quality of educational environment [28]. In addition, the lowest and highest mean scores of the tool are one and five, respectively.
In 2015, Abbasi et al. carried out the psychometric assessment of the Farsi version of the instrument, and the results confirmed the tool’s reliability at a Cronbach’s alpha of 0.93. The construct validity of the tool was evaluated by exploratory factor analysis and Pearson’s correlations, and the four factors above were extracted based on the original version. Therefore, the reliability and validity of the tool were approved [29].
Clinical Competence Questionnaire
The Clinical Competence Questionnaire, which is used to measure CC in internship nursing students, was first developed by Cheng and Liou. The instrument includes 46 items and four subscales of nursing professional behaviors (16 phrases), skill competence: general performance (12 phrases), skill competence: core nursing skills (12 items), and skill competence: advanced nursing skills (6 items). The items are scored based on a five-point Likert scale, as shown follows: “Do not have a clue” (score 1), “Know in theory, but confident at all in practice” (score 2), “Know in theory, can perform some parts in practice independently, and needs supervision to be readily available” (score 3), “Know in theory, competent in practice need contactable sources of supervision” (score 4), “Know in theory, competent in practice without any supervision” (score 5). In this tool, a higher score is indicative of greater CC [30]. Moreover, the lowest and highest mean scores of the tool are one and five, respectively. This Clinical Competence Questionnaire was translated and psychometrically assessed in Iran in this research for the first time. After translation and ensuring consistency between two translations, the tool was provided to 10 faculty members of the nursing and midwifery school to assess qualitative validity and the necessity of questions by using the content validity ratio formula. According to the suggested values of Lawshe’s table and scoring more than 0.64 for each of the questions, all of the items were kept in the tool. Moreover, the relevance of the questions to the purpose of the questionnaire was also checked using the content validity index [31] and its total mean was 0.99. Therefore, all the questions were confirmed in terms of relevance.
At this stage, the internal consistency method (Cronbach’s alpha) was used to confirm the reliability of the tools. To this end, all three tools were completed by 20 eligible internship nursing students who were not included in the research. In the end, reliability of the Self-Directed Learning Readiness Scale for Nursing Education, Undergraduate Clinical Education Environment Measure, and Clinical Competence Questionnaire was confirmed at a Cronbach’s alpha of 0.92, 0.88, and 0.96, respectively.
Data Collection
The data was collected at one stage and at the place of access to internship nursing students, which were different departments of hospitals affiliated with SBMU. The corresponding author of the study, who was the planner and supervisor of internship nursing students and directly interacted with them, collected data with the help of the second researcher over four academic semesters and in the morning and night shifts. In this respect, the tools were distributed among the students, and they were asked to complete and return them as soon as possible.
Data Analysis
In this study, the partial least squares-SEM (PLS-SEM) was used to test the hypothetical research model. The basis of PLS-SEM is a regression technique that, in addition to exploring the linear relationship between several independent variables and one or more dependent variables, measures the relationship networks between structures as well as the relationship between structures and their measures. The basis of model testing in SEM-PLS is to determine the fit of the measurement model and the structural model, and the data analysis was performed after obtaining assurance of the suitability of the two mentioned fits. The measurement model examines the assumed relationships between indicators and latent structures, while the structural model evaluates the assumed paths between endogenous latent variables and exogenous latent variables [32].
Measurement Model
The following steps were taken in the measurement model: indicator reliability, internal consistency reliability, convergent validity, and discriminant validity [33].
The first step (indicator reliability) determines how much of the variance of each indicator is explained by its construct and is performed by indicator loading. Indicators with values less than 0.4 should be removed, and indicators from 0.4 to 0.708 should be considered for removal only when the removal of the index leads to an increase in internal consistency reliability or convergent validity with values higher than the threshold [33].
The two measures of composite reliability and Cronbach’s alpha were used to assess internal consistency reliability, which shows the relationship between variables. For both measures, values 0.6-0.7, 0.7-0.9, and >0.9 were considered “acceptable”, “satisfactory to good” and “problematic”, respectively [33]. Convergent validity as a third step is the extent to which the construct converges in order to explain the variance of its indicators. The metric used to evaluate convergent validity is the average variance extracted (AVE) for all indicators in each structure, and its minimum acceptable value is 0.5 [33].
Discriminant validity as the fourth stage measures the degree of empirical differentiation of a construct from other constructs in the structural model. The foregoing concept can be assessed by three methods of Fornell-Larcker, cross-loading, and heterotrait-monotrait ratio of correlations (HTMT). In Fornell-Larcker, the square root of AVE is compared with the correlation of hidden variables and its value should be higher than the correlation of construct with hidden variables [33]. In examining cross-loadings, discriminant validity is shown when each indicator has a weak correlation with all other constructs except the construct that is theoretically related to it. The HTMT is defined as the mean value of the indicator correlations across constructs relative to the mean of the average correlations for the indicators measuring the same construct, and its value must be below 0.85. A value above 0.9 is indicative of a lack of discriminant validity in the path model [34].
Structural Model
The structural model in PLS-SEM is evaluated by focusing on evaluating the significance and relevance of path coefficients, followed by the model’s explanatory and predictive power. Moreover, significance assessment is carried out by calculating the t-value for path coefficients. In terms of relationship, path coefficients are normally between -1 and +1, coefficients closer to -1 indicate strong negative relationships and coefficients closer to +1 indicate strong positive relationships. The next stage is the coefficient of determination (R2) related to endogenous construct(s). in general, R2 varies from zero to 1, and higher values are indicative of higher explanatory power [35]. Considering that the basic assumption for analyzing with the SEM method is the normality of the data, the normality of the data was first performed using Kolmogorov–Smirnov test. In addition, data analysis was performed in SPSS version 21 to describe demographic characteristics and assess the main variables in terms of mean and standard deviation, and Smart-PLS was applied for path analysis of research variables.
Ethical Approval
All the procedures performed in this study were approved by the Ethics Committee of XXX (the names of university, city, country, and ethical code), in accordance with the ethical standards of the 1964 Declaration of Helsinki and its later amendments or comparable standards. Informed consent was also obtained from all the study participants.