Nurse Professionalism Scale: Development and Psychometric Evaluation

Professionalism is a key trait connecting the nurse and patient, and code of professional conduct, a professional legitimacy in considering nursing as a profession and an essential tool that facilitates nurse practice. This study sought to develop Nurse Professionalism Scale and test the psychometric properties. Data were collected through self-report from registered nurses working in various health care settings selected using strati (cid:192) i ed random sampling and as (cid:192) multi-source feedback from supervisors and colleagues using Nurse Professionalism Scale. Reliability estimate of the 38 item scale was 0.910 (self-report), 0.951 (supervisor-feedback) and 0.952 (colleague-feedback). Exploratory factor analysis using self-reports extracted (cid:192) ve (cid:192) factors with 22 items. Con (cid:192) n rmatory Factor Analysis using supervisors and colleague feed- (cid:192) backs yielded acceptable model (cid:192) t indices con (cid:192) (cid:192) n rming psychometric properties. The scale can (cid:192) be used to evaluate professionalism among nurses across settings. Multisource feedback from stakeholders can be considered as an effective method of gathering data on this construct.

used the "Code of Ethics for Pharmacists and the American Association of Colleges of Pharmacy" and "The American Council on Pharmaceutical Education Accreditation Standards" in describing behavioural components in pharmaceutical care. Miniggio (2015) considered "The College of Medical Laboratory Technologists of Ontario's Code of Ethics and Standards of practice" to measure MLT's professionalism, whereas (Irvin, 2012) evaluated professionalism among lawyers using American Bar Association and the judicial statements.
Professional code of conduct is viewed as a professional legitimacy for considering nursing as a profession and an essential tool that facilitates nurse practice while handling ethical challenges (Balang & Burton, 2014). The code has been considered across nurses and as multi-source feedback from their supervisors and colleagues. Following entry of 830 self reports, 687 supervisor and 747 colleague responses, data were checked for missing and incomplete responses and outliers. Complete set of self-supervisor-colleague responses contained in 644 data sheets from respondents working across work areas (medicine=130, Surgery=136, Obstetrics & Gynae-cology=42, Paediatrics=88, Emergency and Intensive Care Unit=144, Psychiatry=28 and Community=76), across levels (Tertiary=400, Secondary=195 and Primary=49) and across sectors (Private=151, Gov-ernment=479 and Autonomous=14) was used for analysis. Reliability estimate for internal consistency of 38-item NPS using Cronbach was 0.910 (self-report),0 .951 (supervisor-feedback) and 0.952 (colleague-feedback).
Extraction of Factors: Exploratory Factor Analysis (EFA) is used to inspect the item set underlying dimensionality and extracted factors explaining maximum variance in the scale. Thus, a large set of items can be grouped into meaningful subsets gauging different factors (Worthington & Whittaker, 2006). EFA was performed to reduce and group items together so that each factor would represent consistent content area. Factors extracted with Eigen values greater than 1 and items with communalities above 0.4 which con rms common variance shared by each measured item with other items of the construct on which it loads were retained. The Scree test which identi es optimal number of factors can be extracted as per graphical presentation indicated ve factors above one (Fig. 1).
Factor loadings of ± 0.5 and greater are measured as practically signi cant (Hair et al, 2010). Factor loadings obtained are between 0.84 and 0.50 (Table  1). Four factors are explained by 4-6 items. Two item factor can also be retained and considered acceptable if the items are strongly correlated (r > 0.70; or >0.60) and reasonably uncorrelated with other variable (Worthington & Whittaker, 2006). Factor with two items (r=0.62) was retained in this study. In social sciences a factor solution accounting for total variance extracted up to 60 percent (Hair et al, 2010) or at least 50 percent is acceptable (Streiner, 1994). Five factors measured by 22 items explained the total cumulative variance extracted at 51 percent. Kaiser-Meyer-Oklin measure of sample adequacy (MSA) value of 0.893 indicated sample adequacy. Bartlett Test of Sphericity (BTS) was signi cant (x 2 = 3318, df= 231, p<0.000) and indicated suf f f ciently large correlations among items.
Labelling factors or dimensions: Variables with higher loadings on a particular factor are considered as more signi cant and representative of the factor. professions and across countries to evaluate professionalism. Hence the researcher felt the need to develop the nurse professionalism scale on the basis of the national code of professional conduct which can be used in a developing country.

Development of the Nurse Professionalism Scale (NPS):
The process followed in the development of the scale is based on steps enlisted by Boateng et al (2018).

Phase 1
Item Generation: The national "Code of Professional : Conduct for Nurses in India" consisting of 38 items and six dimensions, was identi ed as a comprehensive measure to identify professionalism among nurses. The code was reviewed by four nurse educators, two clinical nurses and three management faculty for readability, comprehensiveness and appropriateness of items. It was decided to use the code as a Nurse Professionalism Scale (NPS) on six-point Likert scale with 0=Not Applicable, 1=never, 2=rarely, 3=sometimes, 4=mostly, and 5=always. Two versions of the instrument were devised; self-assessment for clinical nurses and multi-source feedback for supervisors and colleagues to evaluate the nurses' behaviour.

Content validation: 'A measure has content validity
: when its items accurately represent the thing being measured' (Baumann & Kolotylo, 2009). The tool was validated by 16 nursing and 7 health care management experts for relevance and clarity on a 4-point rating scale. The calculated I-CVI and the S-CVI were above 0.9.
Ethical considerations: Approval was obtained from : concerned ethical review committees. Informed consent was obtained from respondents after explaining the purpose, bene ts, and risks and con dentiality assurance.
Phase II: Scale Development: Pre-testing: "Pre-testing helps to ensure that the items are meaningful to the target population before the survey is actually administered" (Boateng et al. 2018). The tool was administered on conveniently selected 55 clinical nurses. The participants indicated no dif f f culty in providing responses.
Survey administration and sample size: Various clinical settings which permitted and had more than ve registered nurses were included in the study. Nurses were selected using strati ed random sampling. The tool was administered to 1054 registered nurses and their supervisors and colleagues personally. To avoid researcher presence bias and considering their demanding work schedule, participants were given one week period to complete their responses. Data collection period was from April to October 2018.
Data were collected through self report from error of approximation (RMSEA) at or less than 0.05" indicates secure model t (Worthington & Whittaker, 2006 CFI=0.923 and RMSEA=0.058) using colleague data con rmed the factors in NPS.
This study found that the Nurse Professionalism Scale is a valid and reliable tool for measuring professionalism among nurses. EFA using self reported data resulted in extraction of ve factors. Supervisor and colleague feedback regarding professional behaviours of registered nurses con rmed the items and factors through con rmatory factor analysis. Miller

Implications
The code guides assists nurses at every phase of practice from carrying out responsibilities of prevention of illness, promotion and restoration of health and alleviation of suffering among individuals, families and communities. It is a vehicle for self and peer-evaluation of the care quality delivered to consumers. It provides ethical framework and standards for practice. Nurses need to be aware of the important professionalism accents, attitude and behaviours that will aid in the formulation of their identity as indispensable health care providers. Limitations: Self reported data from respondents could involve social desirability bias although multisource feedback was obtained. Busy schedule Hence factor is labelled with reference to variable with higher factor loading (Hair et al, 2010). In this study, in Factor I, variable PA2 with highest loading and PA5 was originally from the dimension "Professional advancement". Items MAN9 and MAN8 re ect development of the profession through working with other stake holders and participating in policy decisions. Hence, the factor is labelled as "Professional Advancement /Development". In Factor II, two variables with higher factor loadings (MAN4 and MAN3) are originally from dimension "Management", variable VHB2 re ects decision making which can be considered as a management function. Hence the second factor is labelled as "Management" and includes items PA1, PA4 and PRA8. Factor III is basically a re ection of "Nursing Practice". Factor IV is explained by three variables; PRA5, PRA3 and PRA1, from the original dimension "Professional Responsibility and Accountability". Factor V is explained by two variables from the original dimension "Valuing Human Being".

Phase III: Scale Evaluation
Tests of Dimensionality through Con n rmatory Factor Analysis: "Tests of dimensionality determine whether the measurement of items, their factors, and functions are the same across two independent samples or within the same sample at different time points. Such tests can be conducted using independent con rmatory factor analysis" (Boateng et al 2018). Obtaining a good model t to the data in a different sample supports the factor structure reliability and validity of scale (Worthington & Whittaker, 2006). Con rmatory factor analysis using AMOS version 22 was conducted using supervisor and colleague feedback data separately. Con rmation of factors is based on t indices which range from 0 to 1. Values closer to 1 suggest good model t (Hair et al, 2010). Structural equation modeling researchers advocate 0.95 as a more desirable level. "Root mean square or as multi-source feedback of professional behaviour within varied practice settings.
amidst the shortage of nurses and the complexity in wording of items could have led to some amount of response error.
Recommendations: Similar study can be conducted in settings outside the state. Comparative study on nurses' professionalism in private and public settings, or different areas can be conducted using the scale.
Every practicing nurse is expected to share the responsibility of self-regulation and practice in accordance with the professional standards and code of ethics as these de ne values and beliefs in nursing profession. The scale can be used to explore professionalism through individual nurses' self-re ection