This study aimed to further validate the reliability and validity of the core competency evaluation instrument for gerontological nurse specialists to improve its promotion and application nationwide. At first, the original scale was poorly fitted to the national sample data, which reminded us that we should be careful when applying a new evaluation instrument in different scenarios and samples. After revision, the scale can be applied nationwide because the revised scale has excellent reliability and validity and a clear, stable structure. In addition, the second-order confirmatory factor analysis showed a good model fit to sample data, which validated the competency model—Attitude-Skill-Knowledge model proposed by Baartman32. This model decomposed the competency into three components (i.e., attitude, skills, and knowledge), which made the competency manageable and assessable. It can also be used to guide the setting of training course systems for gerontological nurse specialists.
There are several possible reasons for the poor model fitness of the original version of the scale in the initial confirmatory factor analysis. Firstly, although we adopted multiple methods to develop the scale to ensure it can be applied nationwide, we only tested its reliability and validity in one province of China. This might lead to the factor structure obtained through exploratory factor analysis only applicable to the samples in Sichuan Province. Secondly, there might be some problems with the content or structure of the original scale itself. When firstly performing the confirmatory factor analysis, the factor loading of most items was excessively high (> 0.95), and high modification indices (> 100) exist among some items and the residuals of some items. This indicated that these items were not independent and with a high degree of collinearity. In addition, the correlation coefficient value between the secondary dimensions was also too large (> 0.95), which indicated that the collinearity between these dimensions was also significant. Considering that even if the sample was divided into two subsamples evenly, the sample size was still sufficient to perform confirmatory factor analysis again. Therefore, we re-explored the structure of this scale and then tested the revised scale’s reliability and validity.
The proposed factor structure still held for the majority of the items in all three subscales when performing exploratory factor analysis in subsample 1. In the Attitude subscale, the factor structure was similar to the original ones, except those items A-5 to A-8 were cross-loaded on factors of professional self-identity and learning enthusiasm. Thus, items from A-5 to A-8 were deleted because the remaining entries were sufficient to reflect these two dimensions’ connotations. Two factors extracted from the revised Attitude subscale explained 87.02% of the total variance. This was much higher than the variance explained by the original factors (68.58%), indicating that the structure of the revised Attitude subscale was much clearer and can better reflect and capture gerontological nurse specialists’ attitudes towards geriatric nursing.
In the Skill subscale, several differences in the factors structure existed between the original subscale and the current subscale when performing exploratory factor analysis. Firstly, the analysis/judgment decision-making skill’s eigenvalue was lower than one, and items originally belonged to this factor were miss-loaded on other factors. The reason for this result might be that the majority of the current gerontological nurse specialists were bachelor or college degrees. Nevertheless, this skill was often trained in master's degree and above education. Furthermore, doctors still dominate the right to decide on clinical practice, which might limit the cultivation and development of this skill for gerontological nurse specialists. Therefore, the gerontological nurse specialists might have a certain deviation understanding of the content of the items in this dimension. What had been mentioned above probably caused items of this dimension to be loaded to other factors. Finally, we decided to retain this factor by considering the following reasons. First of all, the weight coefficient of this dimension ranked fourth in Skill subscale12, which showed that it was one of the essential core competencies for gerontological nurse specialists. Secondly, this skill was also one of the core competencies of other clinical nurse specialists in China, which reflected the importance of this skill to clinical nurse specialists. In addition, doctors and nurses always need to weigh the pros and cons of various therapeutic and nursing measures due to the health problems of elderly patients being relatively complex and unique, especially for those elderly patients who do not have abilities to make decisions. Therefore, gerontological nurse specialists should master this kind of skill to make accurate clinical judgments and decisions and then safeguard the rights of elderly patients. Finally, with the development of the medical-nursing collaboration model, especially for elderly patients who need more nursing than treatment, the conditions of nurses in making decisions will become more and more. Therefore, it was necessary to retain this dimension and train gerontological nurse specialists to be with this kind of skill for advancement.
In addition to this difference, items S-26 to S-28, originally loaded on communication management and research skill, were loaded on the dimension of analysis/judgment decision-making skill in the current exploratory factor analysis. It was considered that the content of these items reflected the connotation of research skill which was considered to share common intrinsic properties with the analysis/judgment decision-making skill. Furthermore, both skills were usually cultivated in postgraduate nursing education more comprehensively and systematically. Therefore, we adjusted these items to the dimension of analysis/judgment decision-making skill and renamed it research/analysis decision-making skill. After modifying the Skill subscale, five factors were extracted, explaining 81.70% of the total variance, which was much higher than the result of the original subscale (69.90%). This indicated that the revised Skill subscale had a more transparent structure, and it can reflect and the skills of the gerontological nurse specialists better.
In the Knowledge subscale, if the authors only extracted factors whose eigenvalues > 1, only one factor could be extracted, which was inconsistent with the experts’ opinion at the phrase of developing this scale. Scholar Polany (1964) divided knowledge into tacit knowledge and explicit knowledge. Tacit knowledge is an important raw material for any professional knowledge, referring to internal and automatic declarative knowledge that people are no longer aware of. The tacit knowledge corresponds to the dimension of basic knowledge in our original scale. Explicit knowledge refers to a kind of academic and professional knowledge that can easily be transferred from one owner to another owner33. It corresponds to the dimension of professional knowledge in our original scale. Therefore, the authors also preset to extract two factors in exploratory factor analysis. The extracted two factors explained 81.70% of the total variance in the revised Knowledge subscale, higher than that of the original subscale (75.87%). This indicated that the structure of the revised knowledge subscale was much clearer, and it could capture the characteristics of the knowledge of gerontological nurse specialists.
The revised scale was cross-validated by confirmatory factor analysis. The first-order and second-order confirmatory factor analysis provided reasonable and sufficient evidence to support the construct validity of the revised three subscales (Attitude, Skill, and Knowledge). The factor loadings of all items on their corresponding factors were great than 0.7. Among them, only two items' factor loading (A-2 and S-34) were too large (> 0.95). However, they did not affect the model fitness. The CR values of nine dimensions were greater than 0.7, ranging from 0.86 to 0.98. And their AVE values were greater than 0.5, ranging from 0.68 to 0.87, indicating that items in each dimension were highly homogeneous and can reflect the characteristics of their corresponding dimensions. In addition, various indices were utilized to evaluate the overall model fit, of which χ2/df, RMSEA, SRMR, GFI, AGFI, NFI, TLI, and CFI were the most frequently reported indices34. Thus, we used these eight indices to evaluate the model fit of our study. Fortunately, the revised scale showed an excellent model fit as the values of χ2/df < 5, RMSEA and SRMR < 0.08, and GFI, AGFI, NFI, TLI, and CFI > 0.9, indicating that the revised scale fit the sample from all over the country and had a good external validity. The results of the multi-sample analysis showed that subsample 1 and subsample 2 were congruent in the four aspects of the factor loading, intercept, covariance and variance when testing the cross validity of the revised scale. This indicated that the revised scale could be applied to different samples from the same population, and it also indicated that the factor structure of the revised scale was steady.
The discriminant validity among the three dimensions of "clinical nursing skill", "communication and management skill", and "research/analysis decision-making skill" was relatively poor, suggesting that these three dimensions needed further revision in the following research. The internal consistency reliability of the revised scale was 0.98, and the test-retest reliability was 0.87. Both were higher than 0.7, demonstrating that the revised scale can evaluate the core competency of gerontological nurse specialists stably.