Construct validity and internal consistency of Hall’s Professionalism Scale: tested on South African nurses

Objective In South Africa, appropriate criteria to measure the professional standing of professional nurses are essential. Internationally, there are professionalism scales by which to measure professionalism, but none could be identified that were particular to the South African context. Hall’s Professionalism Scale consists of 50 items and was specifically developed to measure the attitudes and ideologies held by professionals in various professional occupations by measuring five attitudinal components of professionalism, namely: Sense of calling to the field; Autonomy; Using professional organisation as major referent; Belief in self-regulation; and Belief in public service. In this study, the construct validity and internal consistency of the constructs of Hall’s Professionalism Scale were assessed among professional nurses in the South African context. Results Originally Hall’s Professionalism Scale comprises 50 items. This scale was reassessed by Snizek, who retained only 25 items of the original scale to measure professional standing. During preliminary analysis of the South African data, 23 items were included. Electronic supplementary material The online version of this article (10.1186/s13104-019-4515-6) contains supplementary material, which is available to authorized users.


Introduction
In South Africa, there are few studies that focus on professionalism amongst nurses, and there is no well-known instrument by which to gauge professionalism. However, many international studies have used Hall's Professionalism Scale (HPS) to measure professionalism [1]. This scale was developed by Professor Richard Hall, who aimed to determine suitable criteria and measure a person's professional standing [2]. Professional standing refers to a person belonging to a regulated professionsuch as South African professional nurses who are regulated by the South African Nursing Council-to validate the veracity and legitimacy of a person in a particular profession [3]. The 50-point Likert scale measures professionalism in 5 domains [2], namely: the use of a professional organisation as major referent, belief in public service, belief in self-regulation, sense of calling to the field, and autonomy. The Likert scale options are: very well (VW), well (W), neutral opinion (?), poorly (P) and very poorly (VP) [4]. The first domain 'Using a professional organisation as major referent' , refers to the importance of being affiliated with the professional community to ensure good standards, values and principles in the profession [1,4]. This includes having a sense of professional commitment to attend meetings, keeping abreast of developments in the field, and being willing to support and participate in committees [5]. Secondly, the 'Belief in public service' focuses on whether professionals believe that their profession is beneficial and indispensable to the communities they serve [4]. In some professions such as nursing, the public sometimes does believe in the indispensability of certain services delivered, resulting in professionals in a particular profession to be slow to develop this belief themselves [5]. Thirdly, 'Belief in self-regulation' refers to the "professional endorsement of the notion of colleagues' control", meaning that a person only feels comfortable if their work is judged by other professionals in the same field, and not by outsiders [4,5]. The fourth domain, 'Sense of calling to the field' , focuses on the dedication and commitment of a professional to his/her work. The professional person is motivated by a higher purpose than mere financial gain; there is an 'inward calling' [4,5]. The last item is ' Autonomy' , refers to the practitioner's ability to make their own decisions and render judgements about services independent of outside pressures [4,5]. Hall originally tested 50-item scale on 328 participants, including nurses Cronbach's alpha was 0.86 and Snizek's shorter version which included 25 of the original 50 items yielded a Cronbach's alpha of 0.79 [4].

Methods
A preliminary study with a quantitative, cross-sectional survey design was conducted. All-inclusive sampling (N = 2 158; n = 166) of professional nurses who were enrolled or completed their under and/or post graduate studies (2010-2017) at a particular university were included. Data collection took place via Survey Monkey, an online survey platform. Participants were invited to participate in the study via a computerized short message system (SMS) that communicated the following: "Dear Mr./Mrs./Miss, you are invited to participate in a web-based 50 item survey based on the professionalism of nurses in nursing practice environments in South Africa. Should you decide to participate, kindly follow the http://-link". This message was sent out on three different occasions (days 1, 2 and 6) in 2017. If respondents opened the http://-link, they were directed to an introductory page and informed consent form, if they ticked the agreement box on the screen, they continued to the questionnaires consisting of two sections (Section Ademographic profile and Section B-the 50 items of HPS). Data was analysed by means of the SPSS Version 25 computer program and included descriptive statistics for the demographic profile and exploratory and confirmatory factor analyses to determine construct validity of HPS.

Construct validity and reliability
Exploratory factor analysis The results of the Varimax and Oblimin rotations were very similar, confirming literature reports that the type of rotation used should not have a significant influence. Finally, Oblimin rotations were accepted because there were meaningful correlations between some factors. Loadings of below 0.30 were suppressed; however, smaller loadings of items of the original HPS were included in order to include as many items as possible ( Table 1).
Six of the ten items of the factor, named 'Sense of calling to the field' loaded correctly, whereas 2 items loaded under 'Belief in public service' . One item double-loaded: higher under 'Using professional organisation as major referent' and lower under 'Belief in self-regulation' . Lastly, one item loaded under ' Autonomy' .
Six of the ten items loaded correctly on factor 2, ' Autonomy'; two items loaded higher under 'Sense of calling to the field'; one item double-loaded higher under 'Sense of calling to the field' than 'Belief in self-regulation' and another under 'Belief in self-regulation' .
Factor 3, 'Using professional organisation as a major referent' , proved to be problematic because only 3 of the ten items loaded under this factor. One item loaded under 'Sense of calling to the field' , one item doubleloaded higher under 'Sense of calling to the field' than 'Belief in public service' . Two items did not load above 0.30 on any factor. Another item double-loaded, higher under 'Sense of calling to the field' than ' Autonomy' . There were also double-loadings under 'Sense of calling to the field' with the highest loading under 'Belief in public service' . Another item loaded under 'Belief in self-regulation' .
Seven of the ten items loaded correctly under, 'Belief in self-regulation' , with one item loading under 'Sense of calling to the field' and two items under ' Autonomy' .
Six of the ten items loaded correctly under, 'Belief in public service' . One item loaded under ' Autonomy' , one under 'Using the professional organisation as major referent' , and one under 'Sense of calling to the field' . One item had no loading above 0.30 on any factor. It was decided to perform a confirmatory factor analysis on these factors to determine the fit in a South African context. Figure 1 indicates items included in the South African model for confirmatory factor analysis.

Confirmatory factor analysis
All items loaded statistically significantly on a 5% level on the latent variables, and item 39 on a 10% level; while items 7 and 37 did not load statistically significantly. Factor 1, 'Sense of calling to the field' , negatively correlated    .48** Item 24 Fig. 2 Modified factor structure according to South African data. **Indicates statistical significance on a 5% level with both factor 4, 'Belief in self-regulation' , and factor 5, 'Belief in public service' , implying that the higher a person's 'Sense of calling to the field' , the lower his/her 'Belief in self-regulation' and 'Belief in public service' . Factor 3, 'Using the professional organisation as major referent' correlated negatively with both factor 4, 'Belief in self-regulation' , and factor 5, 'Belief in public service' , indicating that the higher a person's belief in 'Using the professional organisation as major referent' , the lower his/ her 'Belief in self-regulation' and 'Belief in public service' .
The Chi square test is viewed as an overly strict indicator of model fit, given its power to detect trivial deviations from the proposed model [9]. The Chi square test statistic could be divided by degrees of freedom [10]. The factor model yielded a Minimum Sample Discrepancy divided by Degrees of Freedom (CMIN/DF) value of 1.80. Interpretation of the size of this value depends to some extent on the viewpoint of the investigator, but in practice a value of 2 indicates a good model fit [10]. It is, however, considered good practice to report multiple fit indices, typically from three broad classes [9]. The values above 0.9 indicate a good overall fit for a Comparative Fit Index (CFI) [10]. An unacceptable CFI of 0.59 was found for the model, while a Root Mean Square Error of Approximation (RMSEA) value of 0.07 with a 90% confidence interval of [0.062; 0.077] was obtained, which is considered acceptable. The models with RMSEA values of 0.10 and greater are unacceptable [11]. Although, two of the three fit indices indicate acceptable fit, it was decided to conduct a modified confirmatory factor analysis where items loading below 0.3 on their respective factors (items 3, 7, 27, 37 and 39) were removed.
After exclusion of these items, and the fit indices changed to a CMIN/DF value of 1.90, a CFI of 0.65, and the RMSEA value of 0.074 with a 90% confidence interval of [0.063; 0.084]. This also indicated that two of the three fit indices indicate acceptable fit, indicating construct validity in a South African context. However, rephrasing of currently excluded items and/or development of additional items for two factors namely factor 3, 'Using the professional organisation as major referent' and factor 5, 'Belief in public service' , which both only included 3 items should be explored in future studies.

Reliability
The overall Cronbach's alpha coefficient for the 23-item scale with South African data was 0.53. The acceptable Cronbach's alpha for cognitive tests is 0.70, but when dealing with psychological constructs such as professionalism, values below 0.70 could realistically be predictable due to the diversity of the constructs being measured [6]. The Cronbach's alpha for the sub-factors were acceptable, ranging between 0.52 and 0.67. The mean inter-item correlation was also determined, as this is advantageous when there are less than 10 items per factor [7]. The mean inter-item correlation should range between 0.15 and 0.55 [8]. The inter-item correlation ranged between 0.23 and 0.30, confirming reliability in the South African context. The means of 'Belief in public service' scored the lowest (2.16) while 'Using the professional organisation as major referent' and 'Belief in self-regulation' scored the highest (2.86) in the South African context (see Additional file 1: Table S1).

Discussion
From the first preliminary South African study, it can be concluded that 23 items of HPS were sufficient for measuring the five attitudinal components of nurses' professionalism. The confirmatory analysis (Fig. 1) indicated that 5 items (numbers 3, 7, 27, 37, 39) had loadings lower than 0.3, therefore a modified confirmatory factor analysis (Fig. 2) were done excluding these items. Satisfactory reliability of the HPS sub-scales were obtained, which included Cronbach's alpha and mean inter-item correlations.
Five items (numbers 4, 9, 14, 19, 24) loaded under 'Sense of calling to the field'; six (numbers 10, 15, 25, 30, 40, 50) under ' Autonomy'; three (numbers 1, 36, 46) under 'Using professional organisation as major referent'; six (numbers 13, 18, 23, 33, 38, 43) under 'Belief in self-regulation'; and three (numbers 2, 17, 32) under 'Belief in public service' . Two factors, namely factor 3, 'Using professional organisation as major referent' (numbers 1, 36, 46) and factor 5, 'Belief in self-regulation' (numbers 2, 17, 32), seemed to be the most problematic, as it only consisted of 3 items each with relatively lower internal consistency. South Africa is culturally diverse with 11 official languages. This led to the assumption that a number of the items in HPS could have been interpreted differently due to cultural misunderstandings. As a result, special attention should be given to developing items for these two factors in future studies with more participants.

Limitations
• Data was collected from a specific group; therefore results could only be used as guide for the greater South African professional nurse population. • A larger study should be conducted with more participants to test the construct validity and internal consistency of HPS. • Usage of SMSs as data collection method is not conducive due to a low response rates.