Development of the Assessment of Belief Conflict in Relationship-14 (ABCR-14)

Purpose Nurses and other healthcare workers frequently experience belief conflict, one of the most important, new stress-related problems in both academic and clinical fields. Methods In this study, using a sample of 1,683 nursing practitioners, we developed The Assessment of Belief Conflict in Relationship-14 (ABCR-14), a new scale that assesses belief conflict in the healthcare field. Standard psychometric procedures were used to develop and test the scale, including a qualitative framework concept and item-pool development, item reduction, and scale development. We analyzed the psychometric properties of ABCR-14 according to entropy, polyserial correlation coefficient, exploratory factor analysis, confirmatory factor analysis, average variance extracted, Cronbach’s alpha, Pearson product-moment correlation coefficient, and multidimensional item response theory (MIRT). Results The results of the analysis supported a three-factor model consisting of 14 items. The validity and reliability of ABCR-14 was suggested by evidence from high construct validity, structural validity, hypothesis testing, internal consistency reliability, and concurrent validity. The result of the MIRT offered strong support for good item response of item slope parameters and difficulty parameters. However, the ABCR-14 Likert scale might need to be explored from the MIRT point of view. Yet, as mentioned above, there is sufficient evidence to support that ABCR-14 has high validity and reliability. Conclusion The ABCR-14 demonstrates good psychometric properties for nursing belief conflict. Further studies are recommended to confirm its application in clinical practice.

Introduction present, however, there are no comprehensive, validated assessments for measuring nurses' ideas regarding belief conflict.
Therefore, we organized a research group to develop a new, belief-conflict-specific, psychometrically valid, self-reported measure, named the Assessment of Belief Conflict in Relationship (ABCR), on the basis of DAB. This model's perspective can be identified with a comprehensive concept framework of belief conflict in human relationships at work, and from this, it can create problem-related item pools. This development is specifically significant for healthcare professionals, because it enables quantitative evaluation of belief conflict.

Methods
The ABCR development methodology included informed consent, development of a construct framework and an item pool, item reduction, and development of assessment.

Ethics statement
The Ethics Committee of Kibi International University approved this study . In this study, along with a survey form, we enclosed a letter explaining the purpose and method, and informed consent. All participation was voluntary, and participants had the right to leave the study at any time without providing a reason. We regarded the return of the survey form as consent for participation. The survey form was returned anonymously in sealed submission envelopes.

Development of framework concept and item pools
The ABCR item pool was generated through a multi-step process: (1) constructing a framework concept, (2) reviewing relevant assessments, and (3) convening an expert panel.
DAB was used as a construct framework concept in ABCR development, because it has a theoretical and practical foundation for resolving belief conflict. The framework concept for ABCR was defined on the basis of a DAB literature review by literature researcher [13][14][15]. For over a decade in total, this literature researcher had a background as an educator and researcher of DAB in Graduate School of Nursing Master and Doctor Courses in two universities, and had a clinical experience in hospital and facility. That is, the reviewer of literature was expert in nursing related belief conflict and DAB. As a result, three belief conflicts that arise in human relationships were considered: (1) among the same healthcare professionals, defined as conflict arising from a perception gap among peers; (2) between healthcare workers and other staff, defined as conflict arising from a perception gap among different healthcare practitioners; (3) in therapeutic relationships, defined as conflict arising from a perception gap between the patient/family and the healthcare professional.
Moreover, the above literature researcher and three experts gathered an initial pool of 45 items by using this framework: (1) belief conflict among the same healthcare workers (15 items); (2) belief conflict between healthcare workers and other staff (15 items); and (3) belief conflict in therapeutic relationships (15 items). The newly involved three experts had clinical experience of nurse home, rehabilitation, and DAB in hospital (average years of clinical experience (standard deviation, SD) = 10.667 (5.859)). The desired number of items for the final ABCR scale was 10 to 15. Relevant assessment included the Emotion Labor Inventory [29], Stressor/Stress Reaction Scales [30], Scale of Nurses' Job Satisfaction [31], Nursing Stressor Inventory [32], and many other assessments. The ABCR item design was based on a 7-point Likert scale in which 1 corresponded to strongly disagree; 2 to disagree; 3 to slightly disagree; 4 to neither agree nor disagree; 5 to slightly agree; 6 to agree; and 7 to strongly agree.
To assess the instrument's face and content validity, a panel of seven DAB experts reviewed the item pool. The expert panel had the licentiate of five occupational therapists, and two nurses. All paneler had clinical experience in hospital such as emergency medicine, orthopedic, psychiatry, palliative care (average years of clinical experience (SD) = 12.857 (7.559)). The ABCR was developed using the expert consensus method in three steps: Step 1: item pool screening; Step 2: item adjustment; Step 3: final review. In Step 1, the item pool was screened to determine the question type and the content validity of the framework concepts. Using a Likert scale of 1-2 (1 = least plausible; 2 = most plausible), 45 items were rated for importance. For inclusion in the ABCR item pool, 85% consensus of the expert panel was needed. If the Likert rating was 1, the panelist was asked to suggest additional exemplars, modifications of wording, or deletions, as appropriate.
Step 2 included language modifications of the collected questions to reflect the results of the previous step. In Step 3, the expert panel again reviewed the 45 items to ensure that the belief-conflict-related framework concept was adequately represented in the draft. After Step 3, nurses and literature researcher checked final details on framework concept and item pools. The 45 items successfully screened through this process were sent for field-testing. The final ABCR pool contained 45 belief-conflict items.

Item reduction and development of scales
Participants. Participants were recruited through cooperation with research collaborators in Japan. A non-random sample of 2,951 nurses and allied health professionals were contacted in 11 hospitals.
Additional Measures. In addition to the 45-item ABCR version, participating nurses were also assessed using the following measures.
Participant Profile. Demographic data were obtained from participants. We assessed gender, age, years of clinical experience, licensing, position, working arrangements, vacation, leaves of absence, health condition, team approach to healthcare, marriage, smoking, and drinking alcohol.

Sample Characteristics
The participants' demographics and JCQ were summarized using descriptive analyses in SPSS Statistics 22.
Item reduction. Of the 45 items, some were deleted according to certain conditions of item analysis: (1) item floor effect or ceiling effect was confirmed in SPSS Statistics 22; (2) entropy of item was much smaller in Exametrika 5.3 (entropy < 1.5); (3) polyserial correlation coefficient between each item and the total score, excluding the item, was very low compared with that of other items in Exametrika 5.3 (polyserial correlation coefficient < 0.2).
Construct validity. The ABCR's item reduction and factor structure was determined by performing exploratory factor analysis (EFA), using a robust weighted least squares factoring method with missing data (WLSMV) in Mplus 7.2 [35]. Items were also considered for deletion on the basis of framework concepts: (1) belief conflict among the same healthcare workers; (2) belief conflict between healthcare workers and other staff; (3) belief conflict in therapeutic relationships. Items not loading on a factor (factor loading < 0.4) or loading on more than one factor were deleted from the scale; analysis was then conducted with the reduced item set. Percentage of variance in the items accounted for by a factor was estimated using eigenvalues. We used three indexes to assess model data fit [36,37]. The first index was the root mean square error of approximation (RMSEA). The critical values of RMSEA from 0.08 to 0.10 show a mediocre fit, and below 0.08 indicates a good fit [38]. The second and third indexes were the comparative fit index (CFI) and the Tucker-Lewis index (TLI), both with critical values above 0.95 [39].
Structural validity. We performed an EFA followed by a confirmatory factor analysis (CFA). The CFA was performed with WLSMV in Mplus 7.2. Model data fit assessment of the CFA was based on multiple indicators, including RMSEA, CFI, and TLI. Values smaller than 0.08 for RMSEA are considered an acceptable and ideal fit [38]. Values greater than 0.90 for CFI and TLI are considered adequate although values approaching 0.95 are preferable [39].
Hypothesis testing. We analyzed to address the discriminant and convergent validity, based on the factor structure supported by CFA of ABCR-14. To substantiate the evidence of convergent and discriminant validity, we calculated the square of correlation between factors, average variance extracted (AVE) for ABCR in SPSS Statistics 22 and Mplus 7.2. AVE measures the explained variance of the construct [40]. To check convergent validity, we tested whether the square root of every AVE value belonging to each latent construct was greater than 0.5. Discriminant validity was assessed by comparing the squared correlation between each pair of constructs against the average of the AVE for the ABCR factor structure.
Internal consistency reliability. Internal consistency reliability was assessed using Cronbach's alpha in SPSS Statistics 22.
Concurrent validity. We analyzed the Pearson product-moment correlation coefficient of the between factor score of both ABCR and JCQ in SPSS Statistics Version 22.
Item response. The statistical models used in our analyses are based on the multidimensional extension of a graded item response theory. We used multidimensional item response theory (MIRT) with maximum likelihood robust (MLR) in Mplus 7.2. Herein, we used the Monte Carlo integration algorithm. The MIRT estimated the item slope parameters and item difficulty parameters in the ABCR final item pool. In addition, the MIRT estimated the ABCR-14 total information curve (TIC), item information curve (IIC), and the item response category characteristic curve (IRCCC) that indicates the belief-conflict level at which a response in a given category or higher becomes probable. The MIRT was employed to estimate Akaike's information criterion (AIC) and Bayesian information criterion (BIC). Then we compared two different IRT models with the 14 items for further consideration of the IRT model by AIC and BIC: (1) uni-dimensional model and (2) between-item multidimensional model (see Fig 1). Uni-dimensional models require a single factor dimension that is the most restrictive [41]. When an item can discriminate more factors, the assessment is said to be a between-item multidimensional model [42], which is the most general MIRT [43]. We assumed that ABCR is a between-item multidimensional model.

Sample Characteristics
In this study total, 1683 nurses (57.1% response rate) participated, including 93.6% female, 6% male, and 0.4% other. Participants had licenses including 80.5% nursing, 3.3% midwife nurse, 9.2% public health nurse, 6.2% vocational nurse, and 0.9% others. Details on the study sample and JCQ score are reported in Table 1.

Item reduction
As a result of the 45-item analysis, floor and ceiling responses were not observed in the distribution of answers (Table 2). Entropy value ranged from 2.007 to 2.635. Polyserial correlation coefficient between each item's score and the total score ranged from 0.203 to 0.702. In these analyses, all items were not reduced.

Construct validity
EFA determined the underlying factor structure of the item set. The three factors and 14 items were retained after the EFA procedure, as displayed in Table 3. ABCR-14 provided an optimal fit to the data (RMSEA = 0.067 [0.061-0.073]; CFI = 0.986; TLI = 0.976).

Hypothesis testing
ABCR-14 had a high level of convergent and discriminant validity, as shown in Table 5. Excellent convergent and discriminant validity were demonstrated in relation to evaluations.

Internal consistency reliability
Internal consistency reliability coefficients of ABCR-14 (total score and all subscales) were all in the acceptable range, from 0.831 and 0.884 (Table 6).

Concurrent validity
ABCR-14 was confirmed by the correlation between ABCR-14 factor scores and the JCQ. The same professional belief conflict of ABCR-14 factor scores shows a strong negative correlation among JCQ supervisor support, coworker support, and workplace social support. The ABCR-14 factor scores also show a weak positive association between job strain and job demand on the JCQ. Details on concurrent validity are reported in Table 7.  Item response All ABCR-14 items' parameters were estimated using the MIRT. The information criteria clearly supported the between-item multidimensional model (AIC = 66379.782, BIC = 66857.475) over the uni-dimensional model (AIC = 70015.450, BIC = 70476.859), so that the between-item multidimensional model was more appropriate for describing the ABCR-14 than the uni-dimensional model. As Table 8 shows, items were indicated based on the magnitude of the slope parameter, item difficulty parameters, and information criteria. S1 Fig presents the TIC plots for all three factors. The results showed ABCR-14 to be broadly applicable from mild to severe belief conflict; factor 1 provided much more information than the other factors.

Discussion
Psychometric properties of the ABCR-14 We developed and examined the psychometric properties of ABCR-14 as a new, belief-conflict-specific, self-reported measure for evaluating nurses. Overall, the validity and reliability of ABCR-14 were satisfactory. To the best of our knowledge, this is the first study on the development and psychometric properties of an assessment scale for belief conflict. Several results suggested evidence for the validity and reliability of ABCR-14 (see Tables 3  to 7). The construct and structural validity assessed by EFA and CFA were acceptable for the 14 items consisting of a three-factor structure (Tables 3 and 4). That is, ABCR-14 has goodness of fit indices to the validity of the factor. Table 5 shows that hypothesis testing of this study's result was acceptable for ABCR-14's convergent and discriminant validity. These results are identical to that of the DAB-based framework concept of ABCR. This fact clearly proves that the model fits the data. Assessed by Cronbach's alpha, internal consistency reliability was acceptable for the 14 items consisting of the three-factor structure ( Table 6). ABCR-14 was found to show good reliability and therefore has high validity and reliability.
Additionally, concurrent validity was assessed by comparison of ABCR-14 and JCQ participant characteristics (including leave of absence and health condition) ( Table 7). A modest correlation (0.182 from 0.238) between ABCR-14 and JCQ job strain was indicated. Job strain refers to a combination of job demand and job control. Job demand refers to the amount, pace, and difficulty of the work, while job control refers to the ability to make work-related decisions or be creative at work. ABCR-14 has a modest connection with work-related stressor and a deep negative connection with JCQ support. Specifically, the same professional belief conflict of the factor in ABCR-14 has strongly negative connection with JCQ support, which refers to assistance received from co-workers and supervisors. This result shows that ABCR-14 is associated with interpersonal problems and poor teamwork in a workplace. Thus, we can say with fair certainty that this result shows an association between belief conflict and job stress. The MIRT was used to examine individual item characteristics. From the results of information criteria, we concluded that the hypothesis of a between-item multidimensional model was supported. The result of the MIRT resembled that of results of various validities. The ABCR-14 scores proved to have high item slope parameters, in the range of 1.236 and 1.722. Difficulty parameter scores were wide, ranging from -1.794 to 2.409. These data offered strong support for good item response of item slope parameters and difficulty parameters ( Table 8). The ABCR-14 amount of information was sufficiently identified (see S1 and S2 Figs). However, the evidence for supporting this model's information criterion and IRCCC might be exiguous (see S3 to S5 Figs). ABCR-14 item design was based on a 7-point Likert scale. This scale might need to be explored from the MIRT point of view. Yet, as mentioned above, there is enough evidence to show that ABCR-14 has high validity and reliability. From this viewpoint, one may state that the ABCR-14 Likert scale design is correct.

Practice implications
There are some practical implications for developing a scale to measure nurses' belief conflict to clarify the problems arising from belief conflict. Valid and reliable assessment is needed for designing and evaluating management programs based on DAB. Interventions to resolve belief conflicts can target the beliefs that guide conflict behavior among nurses and other healthcare practitioners. In other words, ABCR-14 could lead to development of programs effective in achieving collaboration among team members. Additionally, this assessment provides individual belief conflict assessment and clarification of problems and improvements in multidisciplinary care. Particularly, healthcare providers who work with job management can use ABCR-14 to identify problems that prevent positive collaboration and, in turn, intervene to delay or prevent belief conflict.
For example, if nurses have high risk of belief conflict with patients' families from the result of ABCR-14, we should perform an analysis on the belief conflict's structure. Analysis of belief conflict by DAB involves the following questions: why, who, when, how, and where? What are the causes of the belief conflict? We have used data from unstructured assessment to analysis of nurse with belief conflict: including narrative interviewing and participant observations with the nurse and patients' families. After analyzing belief conflict through ABCR-14, we should choose and practice the best dissolution of the problem by following these DAB rules: 1) Accomplishing a goal is the first priority, 2). The method is a means to an end. 3) The effectiveness of a method depends on the circumstance and goal, and the method's effectiveness is judged ex post facto. To practice these rules, we reflect deeply on a circumstance and purpose of nurse and patients' families, verify the team's condition in practice, and reconfigure the team's common goal. These can be held at a meeting to discuss common circumstance and purpose. Or, we can talk about common circumstance and purpose in a natural situation. In addition, we can optimize the method of achieving the goal and become able to work   constructively together. The method of achieving a purpose is able to give by chose the most practical option: including stepwise and realistic goal setting, purpose-designed planning, increases options, unbiased consideration, unconventional thinking, meditation, reflection on the action, and evidence-based practice. Thus, the result of ABCR-14 can shape the direction of DAB-based practice.

Limitations
This study had a few limitations. For example, to reduce participant burden, we did not perform test-retest reliability. Next the survey was conducted at only 11 hospitals, raising the issue of the findings' generalizability. Furthermore, the study did not investigate healthcare practitioners other than nurses. However, based on the findings of this study, ABCR-14 is a potentially useful tool for estimating belief-conflict-related problems and for monitoring stressrelated problems as part of team management.

Conclusions
Overall, the study findings suggested that ABCR-14 is a valid, reliable scale for assessing nurses' belief conflicts. Further studies are recommended to confirm its application in clinical practice.