The AMBS was devised by (1) defining concept using hybrid model as concept analysis approach (2) item generation based on findings of concept analysis study (3) testing for reliability and validity of AMBS.
Concept Refinement By Hybrid Model
The concept of mutual benefits in nursing academic-service partnership was defined by using the three-phase hybrid model approach. In the first theoretical phase, a comprehensive literature review was performed on documents published from January 1st 2000 through September 25th 2015. This phase yielded to a clear and comprehensive definition of the concept based on the existing literature. The field work phase, data were collected through 18 semi-structured in-depth interviews. Inclusion criteria were having experience in clinical or academic management or academic-service partnership programs for at least two years, and willingness to participate in the research. For the analysis, each interview was recorded. Data collection was continued until data saturation and emergence of categories. The mean duration of the interview was almost 45 minutes. MAXQDA 10 was used for data organization. In final phase, the findings of two previous phases were combined to redefine and provide a final definition of the concept.
Item Generation
The findings of the concept analysis study were used to generate an item pool for AMBS. The item pool was assessed and negotiated by the research team in three sessions. Overlapping or repetitive items were either deleted or combined. The research team strived to choose the clearest and most relevant items.
AMBS Reliability And Validity
The face validity of the AMBS was evaluated both qualitatively and quantitatively. Qualitative face validity evaluation was performed by 10 participants. They were asked to read each item loudly and explain their understanding of it. Moreover, they were asked to comment on the difficulty, relevancy, and ambiguity of the items. Items were edited and reworded based on their comments. Then, the quantitative item impact method was used to identify the importance of each item. Impact scores of 1.5 or higher showed that the intended item was appropriate [16, 17]. Moreover, item clarity and comprehensibility were improved by striving to editing and rewording.
Fifteen experts in the areas of instrument development and partnership in nursing were invited to qualitatively and quantitatively assess the content validity of the AMBS. For qualitative content validity assessment, the experts were asked to assess the grammar, wording, item allocation, and scaling of the scale. On the other hand, quantitative content validity assessment was done by calculating the Content Validity Ration (CVR) and the Content Validity Index (CVI) for each item. CVR of each item was calculated by asking the fifteen experts to score the items by using a three-point scale: “essential”, “useful but not essential”, and “not essential”. According to Lawshe, when the number of experts is fourteen, items with a CVR value of 0.49 or higher are considered appropriate [18]. Afterwards, the CVI of each item was determined by using the Waltz and Bussel’s criteria [19, 20]. Accordingly, the experts were invited to determine the relevancy of the items on a four-point Likert-type scale (not relevant: 1; quite relevant: 2; relevant: 3; and completely relevant: 4). The CVI of each item was then calculated through dividing the number of experts who had considered the item as either relevant or completely relevant by their total number. Items with a CVI of less than 0.78 were removed from the scale. Finally, the mean CVI of all the items were used to calculate the scale-level CVI/Averaging Calculation Method (S-CVI/ Ave). Polit and Beck recommended that S-CVI/Ave of 0.9 or greater reflect excellent content validity [20].
Exploratory factor analysis was conducted to assess construct validity of 68-item AMBS. We used the Principal Axis Factoring with a Varimax rotation. The number of factors was established based on eigenvalues over 1.0. The internal consistency of the whole scale and subscales was evaluated using Cronbach’s Alpha and Omega coefficient. Subscales were considered acceptable as a consistent measure if Cronbach’s Alpha and Omega coefficient ≧ 0.70. Statistical analyses were conducted with IBM SPSS Statistics 25.0 software. The minimum sample size for factor analysis is equal to the number of items multiplied by [21]. Study subjects were recruited by the simple random sampling method. Primarily, a comprehensive list of all clinical directors, academic managers and some nurses in school of nursing and midwifery and educational hospitals affiliated to Mashhad University of Medical Sciences, Iran, was created and then a random sample of 400 subjects were asked to fill out the AMBS. Inclusion criteria were having a clinical or academic management experience or having experience in collaborative programs between clinical and academic nursing institutions for at least two years and willingness to participate in the research. Finally, 345 completely-filled scales were included in the final analysis. Researcher transferred the information into SPSS file. Missing values were substituted by the digit three, which was the median of the Likert scale. The exploratory factor analysis with Varimax rotation was performed. The Bartlett’s test of Sphericity, the Kaiser Meyer-Olkin (KMO) test, as well as the scree plot and eigenvalues were used to respectively determine the appropriateness of the factor analysis model, the sampling adequacy, and the number of factors. The minimum factor load of 0.3 was employed to maintain the items in the extracted factors.
The reliability of the AMBS was evaluated by the internal consistency and the stability assessment techniques. The result of internal consistency assessment is reported as Cronbach’s alpha and Omega coefficient [22, 23]. Stability assessment was performed by test-retest technique. Burns and Grove recommend a two-week interval for test-retest stability assessment [24]. The current study participants completed the AMBS twice with a two-week interval in between. The correlation between the test and the retest scores was evaluated by the Interclass Correlation Coefficient (ICC). ICCs of 0.8 or higher denote satisfactory stability [25].