This study, as a cross-sectional survey structural model study, examines factors that affect the relationship between nurses'' turnover intention and self-efficacy, career plateau, job embeddedness and organizational commitment based on literature review. Also, it attempts to derive a hypothetical model from these factors and to verify whether the model has validity as a means of explaining and anticipating turnover intention of nurses. Finally, this study tries to provide the basic data required to establish methods for reducing nurses'' turnover intention in terms of managing human resources and hospital organization.

Data was collected from six randomly selected hospitals: a university hospital located in city J in Kyungnam and five general hospitals located in two cities (cities B and C), each having 400 to 720 beds. Also, 318 nurses were surveyed (incoming nurses with clinical experience of less than one year and senior nursing officers were excluded), and the data was collected between February 20, 2014 and March 10, 2014.

The data collected was analyzed with each goal of the study in mind. First, general characteristics and characteristics related to turnover intention of the subjects were analyzed by using SPSS 21.0 (IBM Corporation, Armonk, NY). Second, hypothesis testing and verification of suitability of the hypothetical model was conducted by using SEM (Structural Equation Modeling) generated by AMOS 18.0 (IBM Corporation, Armonk, NY).

The hypothetical model of the study included five different variables. The exogeneous latent variables are self-efficacy, career plateau, and job embeddedness; The endogenous latent variables are organizational commitment and turnover intention. Fitness of the model and significance of direct/indirect paths were verified.

The results of this test can be summarized as follows:
As a result of goodness-of-fit verification of the hypothetical model, GFI(.977), AGIF(.921), RMR(.008), CFI(.984), NFI(.977), and TLI(.957) were shown as suitable, but Q-value(X2/df=3.393) and Chi-square value(X2=27.141, p<.001) did not reach the recommended level. In addition, fitness of RMSEA(.087) was slightly insufficient.

To improve the fitness of the hypothetical model in the modified model, the existing latent variables were maintained and free parameters were added. In order to add the free parameters, modification indices were used. The modification index(5.557) between the error clauses of the organization link and organization sacrifice, which is a measurement variable of job embededness is connected. As a valuation result of fitness of the modified model, Q-value(2.941), GFI(.982), AGFI(.928), CFI(.988), NFI(.983), TLI(.965), RMR(.006), and RMSEA(.079) appeared to be suitable, but Chi-square appeared to be unsuitable, X2=20.587(p=.004). In the verified model goodness-of-fit is compliant with the actual data. Among the seven paths six hypotheses were supported and one hypothesis was rejected.

Job embeddedness and Career plateau appeared to significantly affect organizational commitment, but self-efficacy did not appear to significantly affect organizational commitment. Organizational commitment was explained 66.3% by job embeddedness and career plateau support.

Organizational commitment, job embeddedness and self-efficacy statistically appeared to significantly affect turnover intention, but career plateau did not appear to significantly affect turnover intention. Turnover intention was explained 59.0% by organizational commitment, job embeddedness and self-efficacy support.

As a result, job embeddedness and organizational commitment were the most likely path to reach turnover intention. Also, organizational commitment directly affects turnover intention the most, then job embeddedness and then self-efficacy, but career plateau indirectly affects turnover intention. Since turnover intention is greatly affected by organizational commitment and job embeddedness, methods to improve organizational commitment and job embeddedness should be actively developed to reduce turnover intention among nurses.