Assessing the risk of turnover intention among hospital workers

Tsai, Yafang; Wu, Shih-Wang; Chen, Szu-Chieh Article Assessing the risk of turnover intention among hospital workers International Journal of Management, Economics and Social Sciences (IJMESS) Provided in Cooperation with: International Journal of Management, Economics and Social Sciences (IJMESS) Suggested Citation: Tsai, Yafang; Wu, Shih-Wang; Chen, Szu-Chieh (2017) : Assessing the risk of turnover intention among hospital workers, International Journal of Management, Economics and Social Sciences (IJMESS), ISSN 2304-1366, IJMESS International Publishers, Jersey City, NJ, Vol. 6, Iss. Special Issue, pp. 244-258 This Version is available at: https://hdl.handle.net/10419/173240 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by-nc/3.0/ 244 International Journal of Management, Economics and Social Sciences Special Issue-International Conference on Medical and Health Informatics (ICMHI 2017) 2017, Vol. 6(S1), pp.244 – 258. ISSN 2304 – 1366 http://www.ijmess.com Assessing the Risk of Turnover Intention among Hospital Workers Yafang Tsai1 *Shih-Wang Wu2 Szu-Chieh Chen3 1 Dept. of Health Policy and Management, Chung Shan Medical University, Taiwan 2 Dept. of Hospital and Health Care Administration, Chia Nan University of Pharmacy, Taiwan 3 Dept. of Public Health, Chung Shan Medical University, Taiwan Solving the shortage of hospital workers has become an increasingly urgent priority in recent decades. Understanding the turnover problem remains a major scientific challenge. The objective of this study was to assess the risk intention for hospital workers based on the probabilistic risk assessment concept. A cross-sectional study was conducted. The survey samples included nursing staff and hospital workers from one regional teaching hospital in Taiwan. Participants completed a questionnaire with measures of emotional labor, job stress (JS), internal marketing (IM), organizational citizenship behavior, and the perception of turnover intention (TI) in order to assess a risk-based model of the perception of TI based on a doseresponse relationship. The results showed that employees’ perceptions of JS influenced their perception of TI, and organizational commitment was a mediator between IM and the perception of TI. To represent the current knowledge of the predictive model, the present study was the first to incorporate the probabilistic and risk assessment concepts to assess the perception of TI. The proposed dose-response scheme may enable the early identification of the perception of TI among individuals, and help to maintain workflow stability in hospital environments. Keywords: Turnover intention, hospital workers, risk, job stress, healthcare management A shortage of hospital workers has become an increasingly urgent problem in recent decades and it has brought a series of challenges to hospital administration (Manzano-García and Ayala-Calvo, 2014). Hospitals need to invest additional time and money to fill the vacancies and to train newly hired nurses; as a result, the quality of the patient care they deliver is often reduced (Gieter et al ., 2011). Therefore, explaining the turnover problem remains a major scientific challenge. The purpose of this research was to assess the risk of turnover intention (TI) among hospital workers. LITERATURE REVIEW Manuscript received May 20, 2017; revised September 15, 2017; accepted October 28, 2017. © The Author(s); CC-BY-NC; Licensee IJMESS *Corresponding author: [email protected] Tsai et al. 245 Most previous research into the perception of TI has focused on the influence of personality or attitude among hospital workers, such as the predictors of organizational commitment or job satisfaction (Gieter et al ., 2011; Lu et al ., 2002; Tsai and Wu, 2010). However, to prevent turnover, hospital administrations tend to avert the leaving behavior before the perception of turnover and turnover behavior develops. Therefore, to explain the risk of TI among hospital workers, the present study instead incorporates more key predictors in a survey questionnaire, such as emotional labor (EL), job stress (JS), internal marketing (IM), and organizational citizenship behavior (OCB). EL, which is “the management of emotions as part of the work role” (Diefendorff and Richard, 2003), is believed to influence a company’ s well-being through customer satisfaction (Kim, 2008). A healthcare worker’ s emotions and expressiveness will greatly influence a patient’s feelings and experiences. Although the employees become emotional, they must hide their true emotions during healthcare delivery. Additionally, many healthcare workers have died from overwork and the numbers are still increasing (Allegra et al ., 2005). Under high JS, many employees want to leave, which increases the turnover rate and damages the healthcare service quality. These reasons justify the inclusion of an employee’s EL and JS as predictors of the perception of TI. IM is one tool that can motivate employees in the service industry. Almost all hospitals in Taiwan try to enhance their efficiency, and cost control is one of the targets when hospital administrators want to sustain operation during changes in the National Health Insurance Policy. When hospitals feel intense pressure to control their operating costs, many administrators resort to staff reduction and redesigning other organizational structures. However, both solutions decrease employee morale and indirectly affect operating efficiency in the short term. Administrators should look for a more suitable approach that will not only maintain employee morale but also improve efficiency despite limited human resources (Tsai and Wu, 2010). OCB, which characterized by individuals voluntarily extending contributions that surpass their respective job duties, is regarded as a factor that influences an organization’s effectiveness (Organ, 1990). OCB is generally considered a positive behavior in organizations. OCB and the perception of TI influence employees’ service attitudes (Chan et al ., 2009), and the employees play a crucial role (González and Garazo, 2006) in the process of creating customer service value. Almost all of the past empirical studies International Journal of Management, Economics and Social Sciences 246 about the perception of TI have explored the influence of negative employee behavior. Tsai and Wu (2010) found that positive employee behavior affects the perception of TI in the medical industry. The results also demonstrated a significant negative correlation between OCB and the perception of TI. Many published papers use theoretical models (Gao et al ., 2014; Tourangeau et al ., 2014) or structural equation models (Gursoy et al ., 2011) to measure the complex relationships between different variables. However, a new concept was developed to understand the perception of TI. The United States Environmental Protection Agency uses risk assessments to characterize the nature and magnitude of health risks to humans and ecological receptors from chemical contaminants and other possible environmental stressors [14]. In general, the amount of risk depends on the following three factors: (i) how much of a chemical is present in an environmental medium, (ii) how much contact a person or ecological receptor has with the contaminated environmental medium, and (iii) the inherent toxicity of the chemical (The United States Environmental Protection Agency, 2016). Based on these concepts, we reconsider our approach and construct a dose-response curve for understanding social issues, such as the perception of TI among hospital workers. In this study, “dose” represents the external characteristics, job content, work environment, organizational support, and personal characteristics. “Response” represents the risk of TI. In other words, the most significant factor for assessing the risk model of the perception of TI is investigated. Hence, the two objectives of this study are (i) to conduct a cross-sectional study via a survey questionnaire and consider the key predictors of EL, JS, IM, and OCB to explain the risk of TI, and (ii) to assess the risk model of the perception of TI based on the doseresponse concept. METHODOLOGY -Study Design and Data Collection A cross-sectional study was conducted between September 1 and September 30, 2015. The survey samples included nursing staff from one regional teaching hospital in Taiwan. At the beginning of the questionnaire, participants were asked to provide information regarding their sex, marital status, age, education, job position, and seniority. Fifty questionnaires were distributed, and 44 valid questionnaires were received, yielding a response rate of 88.0%. The questionnaire contained a range of closed statements. Respondents Tsai et al. 247 were asked to rate their level of agreement on a five-point Likert scale. The response options from 1-5 represented “strongly disagree”, “disagree”, “neutral”, “agree” and “strongly agree”, respectively. For JS, a modified 35-question version of England’s Health and Safety Executive’s Management Standards Indicator Tool (Health and Safety Executive, 2001) was used (Edwards et al ., 2008). For EL, Wu’s scale that was modified to include 13 items (Wu and Cheng, 2006) was used. For IM and OCB, Tsai and Wu’s (2011) 14- item and 8-item scales were used, respectively. Tsai and Wu’s (2010) 7-item self-reporting instrument was used to measure the perception of TI. This survey questionnaire was approved by the institutional review board of the Committee of Kaohsiung Armed Forces General Hospital (KAFGHIRB 105-031). -Data Analysis The data were analyzed using SPSS 19.0, with descriptive statistics indicating the demographics of the sample. To understand the relationships between the demographic characteristics of the hospital workers and their perceptions of EL, JS, IM, OCB, and TI, a one-way analysis of variance was conducted with equal variance assumed (Macnee and McCabe, 2007). Furthermore, Scheffe’s post-hoc comparison was conducted, focusing on results with statistically significant differences. Linear regression was used to model the relationship between the perception of TI and predictors of JS, IM, OCB, and EL. Cronbach’s α was also used to measure the internal consistency (reliability) and it was most commonly used with multiple Likert questions in the questionnaire. All Cronbach’s coefficients exceeded 0.70, which is regarded as acceptable (DeVellis, 2011). -Probabilistic Density Functions for the Variables Figure 1 illustrates the model framework for constructing the predictive model in this study. Briefly, based on the survey questionnaire, the mean and standard deviation for each item among the five variables (IM, OCB, JS, EL, and TI) were analyzed (Figure 1A). Probabilistic density functions (PDFs) for each variable were constructed using the following equation:      7 1)( i ii TIPTIP (Eq. 1) Where )( 1 TIP to   7 TIP represent the PDFs for item 1 and item 7 in the perception of TI section, respectively. After calculating each score, the total score, in conjunction with a Monte Carlo (MC) analysis, was used to International Journal of Management, Economics and Social Sciences 248 incorporate variability among the different participants. To explicitly quantify the uncertainty/variability of the data, a MC simulation was performed and repeated 10,000 times via the random sampling method (stability condition) to obtain a 95% confidence interval (CI). The process of repeatedly sampling from probability distributions was used to derive a distribution of outcomes. The MC simulation was implemented using the Crystal Ball software (version 2000.2, Decisioneering Inc., Denver, CO, USA). Log-normal (LN) distributions were also assigned for total items, since the estimation must be positive. The PDFs for IM, OCB, JS, and EL were constructed using the same method as for the PDF for each variable (Figure 1B). . Figure. 1. Study Framework and Illustration Explanation used in this Study -Risk Model of the Perception of Turnover Intention C. Individual-based dose-response curve B. Probabilistic density function (pdf) A. Questionnaire survey  Internal Marking (IM)  Organization Citizenship Behavior (OCB)  Job Stress (JS)  Emotional Labor (EL)  Turnover Intention (TI) Score P(IM) Probability P(JS) P(OCB) P(EL) Score P(TI) Score TI Score Significant variable (Sv) P(TI｜Sv) D. Predicted risk of TI Threshold Cumulated Probability TI Score Tsai et al. 249 Based on the cross-sectional study and statistical analysis, the best significant variable (Sv) of TI was chosen to construct the best-fitting dose-response curve. Dose and response were used to illustrate the Sv and TI, respectively. The TableCurve 2D software (version 5.01, SYSTAT Software Inc.) was used to perform the curve fitting techniques. TableCurve 2D provided the best-fitting model with coefficients of determination and 95% CIs (Figure 1C). Therefore, the joint probability technique was used to connect the PDF of the Sv and the dose-response curve, employing the following equation: R (TI) = P (Sv)* P (TI｜Sv) (Eq. 2) where R (TI) represented the predicted risk of TI at a specific score and P (Sv) represented the PDF of the Sv. The relationship between a one-unit estimation of the Sv and the estimation of TI could be expressed as P (TI ｜Sv). Finally, the cumulative probability function was used to express the predicted risk of TI (Figure 1D). RESULTS -Descriptive Statistics The majority of the participants were female (88.6%). Of the participants, 59.1% were married, 31.8% were between the ages of 36 and 40 years, 79.5% had a college or university education, 43.2% held positions as general employees, and 56.8% had seniority over 10 years (Table 1, Appendix-II). The mean value of the employees’ EL responses ranged from 3.05 to 3.98. The mean value of employees’ JS, IM, OCB, and TI ranged from 2.75– 4.02, 3.41– 4.00, 3.77– 4.30, and 2.39– 3.00, respectively. The Cronbach’s α for JS, EL, IM, OCB, and TI was 0.874, 0.830, 0.913, 0.877, and 0.850, respectively. All Cronbach’s coefficients exceeded 0.70, which means that the questionnaire had acceptable reliability. -Inferential Statistical Analysis Male individuals displayed stronger perceptions regarding JS than the female individuals. Age was found to affect employees’ perceptions of IM and TI. Participants between the ages of 36 and 40 years scored significantly higher on IM than employees aged 31-35 years. This study also found that employees aged 31- 35 years have markedly higher perceptions of TI than those aged 36-40 years. International Journal of Management, Economics and Social Sciences 250 Seniority was found to influence employees’ perceptions of TI. Those who had worked in the profession for 5-10 years showed stronger perceptions of TI than those who had worked in the field for over 10 years. Seniority was also found to influence employees’ perceptions of IM. Those who had worked in the profession for over 10 years showed stronger perceptions of IM than those who had worked in the field for 5-10 years. Employee position influenced their perceptions of JS, TI, IM, and OCB. Those positioned as frontline employees showed stronger perceptions of JS than those who were general employees, followed by first line managers and middle managers. First line managers showed stronger perceptions of IM and OCB than frontline employees (Table 2, Appendix-III). According to the linear regression analysis, employees’ perceptions of JS influenced their perceptions of TI ( β = 0.823, p = 0.000) (Table 3). β (t value) R 2 Adjusted R 2 F value freedom EL→JS -0.053(-0.409) 0.004 -0.021 0.167 1,40 EL→TI 0.186(0.766) 0.014 -0.010 0.586 1,41 JS→TI 0.823(3.070***) 0.191 0.170 9.424 1,40 IM→TI -0.181(-1.165) 0.033 0.009 1.357 1,40 IM→OCB 0.648(4.924) 0.377 0.362 24.242 1,40 OCB→TI -0.032(-0.204) 0.001 -0.023 0.042 1,42 Note. †p<.10 *p<.05 **p<.01 ***p<.001 Table 3. Linear Regression Analysis -Probability Distribution Function of the Five Variables The PDFs for the five variables are presented in Appendix-I (A-E) using the MC simulation technology. Independent runs with 10,000 iterations for each parameter sampled were conducted independently from the LN distribution. Hence, the median scores for TI, IM, OCB, EL, and JS were estimated to be 18.42, 51.65, 32.34, 47.20, and 88.92, respectively, whereas the scores ranged from 7-35, 14-70, 8-40, 13-65, and 35- 175. A box-and-whisker plot was used to represent the uncertainty with 95% CIs for the five variables. The ranges were calculated by multiplying the number on the five-point Likert scale and the item numbers of each variable. -Perception of Turnover Intention Tsai et al. 251 Based on the statistical analysis, JS was the most Sv correlating to TI compared to the other variables. Hence, the dose-response profile was implemented as P (TI｜JS) in Figure 2 using TableCurve 2D. The adequate fit model for the data points of employees were shown as: 2 )(JSbaTI  ( r 2 = 0.133) (Eq. 3) where the two parameters of a and b were estimated to be 11.36 ± 2.78 (mean ± standard error [SE]) and 0.00086 ± 0.00033 (mean ± SE), respectively (Figure 2). Figure 2. The dose-response profile was illustrated by Table Curve 2D. Original data and fitting model with a 95% confidence interval are shown Figure 3A (Appendix-IV) shows the histograms for the predicted cumulative distribution function (CDF) of the TI scores. The exceedance risk curve is shown in Figure 3B, which was estimated as 1-CDF of the TI scores (Figure 3A). The results demonstrated that the probability that 50%, 25%, 5%, and 2.5% or less of the TI is approximately 18.18, 20.25, 23.32, and 24.22, respectively. DISCUSSION The purpose of this study is to assess the risk of TI among hospital workers. An interdisciplinary method for assessing the probability of the perception of TI is used, based on a cross-sectional study combined with a risk-based framework. The results indicate that an employees’ perception of JS influences their perception of TI. However, this pilot study only has 44 valid questionnaires. A larger-scale investigation of hospital workers 0 5 10 15 20 25 30 35 40 60 70 80 90 100 110 120 Job stress (JS) scores Turnover intention (TI) scores P(TI｜JS) Data Model 95% CIs International Journal of Management, Economics and Social Sciences 258 Appendix-IV Figure 3. (a) Cumulative Probability and (b) Exceedance Risk for Predicted Perception of Turnover Intention Scores 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 10.6 12.7 14.9 17.0 19.2 21.3 23.5 25.6 Cumulated probability Predicted Turnover intention (TI) scores 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 10.6 12.7 14.9 17.0 19.2 21.3 23.5 25.6 Exceedance Risk Predicted Turnover intention (TI) scores A B