An empirical study on the job satisfaction and turnover intention of health-care professionals: Moderated moderation approach

Introduction

Turnover intention is a mental process and it needs a lot of thinking for decision-making. The decision of whether to leave the job or to stay at a job is very difficult.[1,2] There are direct and indirect costs associated with turnover intention. For example, the recruitment and selection process, training, and development process need time and a lot of cost paid by organizations to train new employees.[3] Meanwhile, it brings a lot of burden to existing employees for replacing the previous employee.[4] Organizations, especially in health care, are very difficult to retain employee[5] as it embroils a lot of thinking and decision-making how to reduce turnover intention. Allinson and Hayes[4] have developed cognitive style index (CSI). They have divided human mind into two parts; one is called left brain it is also called analytical and second part is called intuition or right brain. People with analytical style tend to collect information first and then they make decisions while intuitive people make decision on basis of their experience, judgment, and feelings.[6] Later on, Cools and Broeck[7] developed cognitive style indicator (CoSI). She has divided analytical style into three more parts and named those constructs knowing, planning, and creating for making decisions. It is believed that decision-making plays very important role in reducing turnover intention and increasing job satisfaction. Hence, it was very crucial to use a well-reported and validated instrument for decision-making in this study. Job satisfaction is an international phenomenon,[8] in the 1960s, more than 4000 articles were published on job satisfaction. However, it gains popularity after the theories given by Herzberg two factor theory, Vroom’s[9] theory, McGregor Theory, Theory X and Theory Y, and job characteristics model by Hackman and Oldham[10] Spector theory of job satisfaction.[11] All these researchers confirmed that there are few attributes of job satisfaction that is held constant, for example, salary, supervisor, promotion, colleagues, work, work environment, and job security.[12,13] Later on, researchers added more variables in theories of job satisfaction but for this study, researchers have chosen the above-mentioned variables.[7,13] There is an American researcher named Judge, Klinger, and Simon[13] conducted and reported many articles on job satisfaction. From the findings of this study, it was indicated from the results that instrument was found reliable and valid.[14,15] CoSI was only related with job and supervisor while CSI was not significantly related with any factor of job satisfaction. In addition, there was moderated moderating effect on satisfaction and salary only but no moderated moderation effect on other variables.[16,17] Themes emerged from qualitative interviews and findings also supported the quantitative findings. Based on the above literature, we hypothesized proposed the following: H₁: There is a negative relationship between predictors and criterion variables, that is, satisfaction, salary, supervision, promotion, coworkers, work environment, work itself, CSI and CoSI, and turnover intention. H₂: There is a moderated moderating effect of CSI, CoSI on factors of job satisfaction, and turnover intention.

Methods

Population and sampling

The population of the study consisted of health-care professionals in the Riyadh region of Saudi Arabia that included 124 administrative and 23,246 staff members, thus, N of the study was 23,370. Since the population was big so it was impossible to contact every member to obtain their responses, therefore, sampling was essential. Probability and non-probability sampling techniques are given by researchers in health care. Both have their strengths and weaknesses.

In this study, researchers have used simple random stratified sampling, whereas, convenience sampling technique was applied to collect responses from the respondents. Weiers[22] has developed a statistical formula [SD2/((E2/Z2) + (SD2/N))] for the finite population to determine sample size. Therefore, researchers have employed the Weiers formula. 95% confidence level is suggested by social scientists to determine the sample size for the finite population, which is equal to 1.96 z-value. Table 1 illuminates the procedure to determine the sample size of the study in hand.

Table 1 Sample size

Data analysis

SPSS was used for quantitative data analysis. Exploratory factor analysis, bivariate correlation, and hierarchical multiple regression (HMR) were run to investigate the best fit model. Further for moderation, a procedure given by Aiken and West[23] is used. First, the variables were mean centered to reduce the multicollinearity problem. Then, interaction term was created to test moderation results. Thereafter, all variables were entered into HMR according to theoretical significance. Then, interaction plots were plotted one standard above the mean and one standard deviation below the mean. Nvivo 8 was used for qualitative data analysis; nodes were developed, where from themes emerged [Figures 11 and 12]. Template analysis was used to analyze themes. Since, ideas and themes are parcel part of the respondents’ lives. Researchers also focus on the phenomenological flavor of themes. The current study was conducted in 2020.

Close

Figure 1

Measurement model. RMSEA: Root mean square error of approximation, GFI: Goodness of fit index, IFI: Incremental fit index, CFI: Comparative fit index, TLI: Tucker Lewis index, NFI: Normed fit index, RFI: Relative fit index. Chi-sq/df: Chi-square/Degree of freedom

Export to PPT

Close

Figure 2

Structural model. RMSEA: Root mean square error of approximation, GFI: Goodness of fit index, IFI: Incremental fit index, CFI: Comparative fit index, TLI: Tucker Lewis index, NFI: Normed fit index, RFI; Relative fit index. Chi-sq/df: Chi-square/degree of freedom

Export to PPT

Close

Figure 3

Moderation model

Export to PPT

Close

Figure 4

Satisfaction CoSI, CSI. CSI: Cognitive style index, CoSI: Cognitive style indicator

Export to PPT

Close

Figure 5

Salary CoSI, CSI. CSI: Cognitive style index, CoSI: Cognitive style indicator

Export to PPT

Close

Figure 6

Coworkers CoSI, CSI. CSI: Cognitive style index, CoSI: Cognitive style indicator

Export to PPT

Close

Figure 7

Promotion CoSI, CSI. CSI: Cognitive style index, CoSI: Cognitive style indicator

Export to PPT

Close

Figure 8

Supervisor CoSI, CSI. CSI: Cognitive style index, CoSI: Cognitive style indicator

Export to PPT

Close

Figure 9

Work environment CoSI, CSI. CSI: Cognitive style index, CoSI: Cognitive style indicator

Export to PPT

Close

Figure 10

Work CoSI, CSI. CSI: Cognitive style index, CoSI: Cognitive style indicator

Export to PPT

Close

Figure 11

Word cloud taken from NVIVO

Export to PPT

Close

Figure 12

Qualitative model taken from NVIVO

Export to PPT

Results

A total of 301 employees have participated in the current study. There were total 225 male participants, that is, 74.75%, 76 females have participated in the current study, that is, 25.2%. Majority of the participants belong to public sector health-care organization such as 198 (65.7%) while 34.2% belong to private sector organizations. Further analyses of results revealed that most of the professionals were foreigners and belong to India, Pakistan, Egypt, and Sudan, that is, 49% Indian, 33% Pakistani, 9.9% Egyptian, and remaining Sudan. In the first run of exploratory factor analysis, it is necessary to check the sample is adequate for EFA or not? This objective was accomplished by Kaiser-Meyer-Olkin (KMO) of all variables KMO for cognitive style was recorded KMO = 0.7 which is above 0.5 Hair et al.[24] for job satisfaction KMO = 0.775, for turnover intention KMO = 0.703. Bartlett’s’ test of sphericity for all variables was significant. Promax was used for rotation, as shown in Table 2.

Table 2 Factor loading pattern matrix

Table 3 presents the results of the correlation between predicting variables and the criterion variable. As expected, satisfaction is negatively related with turnover intention while positively but weakly related with CoSI and CSI, that is, r = 0.463, P < 0.05 and r = 0.91, r = 0.91, P > 0.05. Further examination of the relationship between salary and turnover intention is found moderate and negative but significant, that is r = −0.502, P < 0.05, but alary is not related to CoSI and CSI.

Table 3 Correlation matrix

Further examination of results shows that promotion is negatively but significantly related with turnover intention, that is r = −0.332, P < 0.05, but promotion is negatively related with CoSI, that is r = −0.063, P > 0.05 while promotion is positively but insignificantly related with CSI, that is r = 0.028, P > 0.05. In the same way, further examination of correlation results shows that job/work is negative but significant related with turnover intention but also significant related with CoSI but weakly related with CSI, that is r = −0.412, P < 0.05; r = 0.134, P < 0.05; and r = 0.024, P > 0.05. Furthermore, examination revealed that coworkers/colleagues are also negative but significant related with turnover intention but weakly and positively insignificant with CoSI and CSI, that is r = −0.250, P < 0.05; r = 0.070, P > 0.05; and r = 0.050, P > 0.05.

Likewise, results revealed that supervisor is negative and significant but weak related with turnover intention also significant related with CoSI but insignificant related with CSI, that is r = −0.373, P < 0.05, r = 0.112, P = 0.05, and r = 0.012, P > 0.05. Furthermore, work environment is negative and significant with turnover intention but insignificant with CoSI and CSI, that is r = −0.304, P < 0.05; r = 0.105, P > 0.05; and r = 0.035, P > 0.05, so on the basis of discussion from correlation results, it is clear that our hypothesis 1 is substantiated and accepted. This means that there is a significant positive relationship between the predictors and criterion variables of the study, Table 3.

There is different goodness of fit indices given by Hair et al. [24] for model fitness. Fit indices and their fitness range values are given in Table 5.

Table 4 Evaluating the model fit to fit before confirmatory factor analysis CFA

Table 5 Model fitness measurement model results

Table 4 and Figure 1 illustrate results for model 1 of the study. AMOS-SEM was used to develop a measurement model. In Figure 1, measurement model is given. All the factor loadings are higher than 0.5 except one construct, that is, job it is 0.49, two constructs are deleted from analysis because of low factor loadings. All the goodness of fit index values met the required level, that is, Chi-square =2.847, RFI = 0.889, NFI=0.921, TLI = 0.925, CFI = 0.947, and IFI = 0.947, P < 0.01, all the values are in the specified range so the model if fit. Further analysis of results revealed construct reliabilities and average variance extracted values. Construct reliability for job satisfaction and CoSI is 0.786 and 0.706 while AVE value is 0.433 and 0.461 below 0.5 but construct reliability is met criteria, so reliability and validity are established.[25]

Figure 2 was developed for the structural model. In this model, job satisfaction is independent; cognitive style is also independent (moderator is considered independent) and turnover intention is criterion variable. Table 6 shows the goodness of fit indices. It is found that model is fit, and all the goodness of fit indices are in the specified range, that is, χ² = 2.847, RFI=0.889, NFI=0.921, TLI=0.925, CFI=0.947, IFI =0.947, and RMSEA=0.078. Further analysis of results revealed that correlation between job satisfaction and cognitive style is r = 0.17, P < 0.01, and impact of job satisfaction on turnover intention is beta β = −0.64, P < 0.05 and β = 0.20, P < 0.05 is also significant and R²= 0.41 it means that job satisfaction and cognitive style show variance on turnover intention 41%. It means that there is a negative impact of job satisfaction on turnover intention and the positive impact of cognitive style on turnover intention.

Table 6 Model fit

Moderating effects of cognitive style on job satisfaction and turnover intention were also checked in AMOS-SEM. Results are highlighted in Figure 3, it is found that job satisfaction has a negative impact on turnover intention, that is, β = −0.81, P < 0.05, while cognitive style has positive effect β = 0.10, P < 0.05, an interaction term was created by multiplying job satisfaction and cognitive style β = 0.24, P < 0.05, R² = 0.37 it means job satisfaction, cognitive style, and interaction term shows variance on turnover intention 37%. Next HMR is used for testing hypothesis 8, that is, there is moderated moderating effect of CoSI and CSI on job satisfaction factors and turnover intention. This moderated moderation results are also known as three-way interaction. All seven independent variables and their three-way interaction results are presented in Table 7.

Table 7 Moderated moderation results

In this study, CoSI is used as a moderator and the CSI is used as a moderated moderator. First examination of three-way interaction done with satisfaction CoSI and CSI and turnover intention. The results revealed that the interaction term is significant and ∆R² is also significant it means that there is moderated moderation and CSI does act as a moderated moderator on the CoSI, satisfaction, and turnover intention. Three-way interaction was conducted using the same procedure and found that CSI does not act as a moderated moderator on CoSI, coworkers, and turnover intention, there is no significant interaction term and ∆R² is also not significant although there is changed in ∆R² is recorded.

Further examination of the results revealed that there is moderated moderation occurs on the relationship between CoSI, promotion, work, work environment, and supervisor because all interaction terms were insignificant and there is no significant change in recorded in ∆R² but when CSI is used with salary and turnover and CoSI it does acts as moderator because interaction term is significant and ∆R² is also significant so CSI acts as a moderated moderator on CoSI salary and turnover intention so our hypothesis 2 is partially accepted for satisfaction and salary. Table 7 highlights the moderated moderation result of this study.

To interpret the interactions, all three-way interactions were plotted, with cut values of one standard deviation below the mean and one standard deviation above the mean on each moderator variable. Examination of the first plot revealed that the CSI, that is, CSI is acting as a moderated moderator on CoSI, that is, COSI.

From the plot, it is also revealed that low CoSI and high CSI interact with high CoSI and high CSI. Hence, this plot [Figure 4] is consistent with the results shown in Table 3 for satisfaction. Further analysis of results shows that low COSI and low CSI fully moderate the relationship between salary, turnover intention, and CoSI. Hence, this plot [Figure 5] is consistent with the results shown in Table 7 for salary.

Similarly, examination of the plot [Figure 6] shows that CoSI and CSI interact with each other, but these two variables do not interact with coworker and turnover intention so three-way interaction does not exist in this Figure 6. Further examination of plot [Figure 7] indicated that high CoSI and high CSI and high CoSI and low CSI interact with promotion and turnover intention but low COSI and low CSI do not interact so there is no three-way interaction exists. Further examination of plots revealed that Figures 8-10 also consistent with regression results.

Discussion

In this study, we have used 35 interviews, whereas Creswell[27] Smith et al.[29] suggested the use of at least 10 interviews for qualitative analysis. Furthermore, Silverman[25] suggested that in qualitative research, qualitative data must be natural. There should be no coding and the researcher is not allowed to minimize data. Semi-structured interviews are the best source for qualitative study as in this study researchers aimed to understand the phenomena in-depth, additionally; this method allows the interviewees to ponder freely on the issues. In the first instance, researchers got the consent of all the interviewees then an interview guide was provided to them as Rubin and Rubin[26] considered an interview guide as the scaffolding of the interview. The date and time for the interview were fixed mutually as practiced by Smith et al.,[20] and interviews were recorded suggested by IPA. All interviews were transcribed verbatim. As asserted by Remenyi et al.,[27] the researcher has read transcription thoroughly and made necessary corrections to make sense of it. Below is the narration and analyses of the interview data. Different questions were asked from interviewees about different attributes of their current job. Interviewees respond in the following manner: “Regarding salary my package is according to government rules and you know in public sector rules are already defined for each cadre and scale. So, salary is not an issue.” Another respondent explained “My package is attractive, and I am happy with this package, but I am not satisfied from the workload in this organization that is beyond the capacity, sometimes, I have to complete the work of my colleagues.” Other interviewees highlighted the issue of inter-relationships. Their concerns are as under “Relationships with colleagues are fine but my senior colleagues are not supportive. If I need equipment and consumable office supplies, I have to wait a few weeks or even a month. Although I know they have their limitations as they have to follow the rules, and they cannot buy from their own money since they have to ask the purchase department to make things available that is why it takes time. Efficiency of work is concerned, comparatively; the progress of work is slower than the private sector.” Other attributes of job satisfaction are highlighted in this way as under: “Work environment is good, yet promotion policies are inflexible, you know you have to fulfill the criteria for promotion, once you fulfill it, then you will get it” one of the respondents discussed about job security that “Job security in the public sector is very high since these are tenured jobs… you can easily work for whole life…. and nobody can disturb you if your conduct is aligning with the rules of business.” Responses of respondents regarding turnover are mentioned below. “I don’t have the intention to leave that job in future I am satisfied with my current job.” Other respondents highlighted their concerns in this way… “Since I am working in public sector organization so I have no intention to leave that job, however, if I get a better opportunity, then I could decide to move.” Cognitive style is used for making decisions. Different cognitive styles were used by researchers to know their role in decision-making, therefore, CSI and CoSI were used in this study. Respondents were of the view that: ”First I try to find the reason of problem… and it is better to know the history and background of the problem then you are better able to find the better solution, so I used to analyze the situation first as you know, we all are human beings and we have emotions and feelings so sometimes we cannot control our feelings and, in such situation, I use to make spontaneous decisions. However, I am intuitive by nature.” This study has some limitations which can affect some potential areas. First, the sample size of this study is small and collected from health-care organizations only and that to in Riyadh region, thus findings of this research could not be generalized to other sectors such as banks and universities, therefore, it will be better if future studies may use large sample size. Second, this study is quantitative; future studies can use mixed methods and longitudinal data for more thought provoking results. The third limitation is methodological limitations. This study has used two moderators, whereas, future studies can use multiple or series of mediators to come up with novel results and findings. Other dimensions of cognition such as feelings, emotions, and judgment can be used in future studies.[28,29]

Conclusions

It is imperative for the health-care organizations to enhance the satisfaction of their professionals by applying different strategies for involving them in decision-making policies, this not only increases their job satisfaction but also reduces turnover intention of workforce. Furthermore, management and leadership of the health-care professionals play very important role in increasing the job satisfaction of their professionals.

Authors’ Declaration Statements

Authors’ Contributions

G.M.K and Y.H.M participated in manuscript drafting, conceptualization, the idea of study design, and methodology. F.A and S.A.P.: Perform review, editing, formal data analysis, and validation. All authors have read and agreed to the published version of the manuscript.