Some Factors Affect Turnover Intention of Information Technology Employees in Vietnam

This study was undertaken to explore the factors that could affect to the emotional exhaustion and turnover intention of Information Technology (IT) employees in Vietnam. This study was done through two steps including qualitative and quantitative research methods. In qualitative research method, nine information technology employees were interviewed to assess the clarity of words, the content of the questions and the ability to answer the questions of the readers. Formal quantitative research was executed by interviewing a sample collected by convenient method, using a questionaire, including 194 information technology employees in the areas of HoChiMinh city. The data was used to assess the scales and test the hypotheses. Cronbach Alpha coefficient analysis, exploratory factor analysis (EFA), regression analysis were used in this section. The results showed some differences compared to the initial theoretical model. Only some elements in the theoretical model affect the emotional exhaustion, in which IT job demands is the factor that have the most powerful affect. Study results contribute additions to the theoretical basis of human resource management, by understanding more about the factors that have impact on the emotional exhaustion and turnover intention of information technology employees. Thereby, the results also provide additional facilities for the management of information technology enterprises so that the managers can make appropriate decisions in order to preserve the IT human resources.


Turnover Intention
"Turnover Intention reflects the (subjective) probability that an individual will change his or her job within a certain time period" (Sousa-Poza & Hennehenger, 2004, p. 113).Acording to Ajzen (1985), the usage of Turnover Intention began from Theory of Planned Behavior (TPB) (Hein, 2006): an intention is the readiness of a person to execute a planned behavior, and considered as the factor that affect directly to behavior.
Turnover Intention was mentioned in many previous studies.Many studies showed that job satisfaction, organization commitment and job involvement affect directly to turnover intention (Elder, 2004;Cropanzano, Rupp, & Byrne, 2003).In addition, emotional exhaustion can affect to turnover intention (Geurts, Schaufeli, & De Jonge, 1998;Cropanzano et al., 2003).Some other factors, for example, ethical environment, quality of work and flexible working time can affect to turnover intention in other studies (Allen, Armstrong, Riemenschneider, & Reid, 2006).Recently, "Quality of working life", which includes job satisfaction and emotional exhaustion, was used in a variety of researches to described the influences of the factors to turnover intention (Korunka, Hoonakker, & Carayon, 2008), (Hoonakker, Carayon, Schoepke, & Marian, 2004).Many researches showed that there is a strong correlation between turnover intention and turnover behavior (Korunka et al., 2008).Therefore, in many researches, turnover intention was used instead of turnover behavior.As a result, this study will use turnover intention instead of actual turnover behavior.
To Maslach & Jackson, 1986, when burnout happens, individuals can not meet the jobs' needs, and ready to quit (Janssen et al., 1999).Hence, burnout will be considered as a factor that could affect strongly to turnover intention and turnover behavior.In addition, Maslach (1993) supports that emotional exhaustion is the major factor of burnout (Janssen et al., 1999).Therefore, emotional exhaustion represents for burnout in many researches (Janssen et al., 1999;Hoonakker et al., 2004;Korunka et al., 2008).

Emotional Exhaustion
Demerouti, Bakker, Nachreiner and Schaufeli (2001, p. 499) suggested: "Emotional exhaustion closely resembles traditional stress reactions that are studied in occupational stress research, such as fatigue, job-related depression, psychosomatic complaints, and anxiety" (Cropanzano et al., 2003).Emotional Exhaustion also described as the situation in which people are lack of energy severely and have the feeling that their resources were lost (Wittmer & Martin, 2010).In addition, according to Maslach (1982), emotional exhaustion appropriate for health care service (Wittmer & Martin, 2010).However, other researches used emotional exhaustion in many services which relates to human such as hotel service, banking and sales (Maslach & Jackson, 1981).In addition, Kalimo & Toppinen (1995) supports those technical jobs can cause emotional exhaustion severely because the employees have to adopt with the rapid change of the technology and they have to learn continuously to handle their job well (Korunka et al., 2008).Hence, emotional exhaustion was used in Information Technology (Hoonakker et al., 2004;Korunka et al., 2008).And then emotional exhaustion was used in post office (Wittmer & Martin, 2010).Researches of Hoonakker et al. (2004), Korunka et al. (2008) showed that emotional exhaustion is negatively associate with job satisfaction and positively associate with turnover intention.

IT Job Demands
"Job demands refer to those physical, social, or organizational aspects of the job that require sustained physical or mental effort and are therefore associated with certain physiological and psychological costs" (Bakker, Demerouti, & Euwema, 2005, 170).Job demands can cause stress extremely (Leung, Sham, & Chan, 2007).Especially, IT job demands results in emotional exhaustion (Hoonakker et al., 2004, Korunka et al., 2008). Fernet, Guay, & Senecal (2004) suggested that stress caused of job demands is the stress of the number of tasks need to be done and the difficulty of the tasks (Leung et al., 2007). Cox, (1978, p. 25) noted that stress caused of job demands was defined as "the perception happens when people compare the job demands and their ability to cope" (Leung et al., 2007).Job demands (For example, time pressure, hard working, the complex of the tasks) are the factors that cause stress severely (Leung et al., 2007).In addition, researches of Hoonakker et al. (2004), Korunka et al. (2008) showed the positive relationship between IT job demands and emotional exhaustion.

Supervisory Support
According to Kaplan et al. (1977, p. 50), Social support has been defined as a "relative presence or absence of psychosocial support resources from significant others" (Haddad, 1998).In addition, Kaplan et al. (1977) showed that Supervisory Support is a part of social support (Haddad, 1998).Supervisory support is the support of supervisor to employee with understanding and honest (Slocum & Hellriegel, 2009).Supervisory support includes sharing the problems with employees and helping them (Bultular & Oz, 2009).In addition, high degree of supervisory support can help to reduce emotional exhaustion (De Jonge, 1995;Korunka et al., 2008;Hoonakker et al., 2004).Korunka et al. (2008) found that supervisory support is negatively associated with emotional exhaustion.In addition, Hoonakker et al. (2004) found that supervisory support is negatively associated with emotional exhaustion for women.

Colleagues Support
Colleagues Support is also a part of social support.Schwab, Jackson, & Schuler (1986) supports that Support from Colleagues can help to reduce burnout.In addition, the high level of Support from Colleagues can lead to low level of emotional exhaustion in IT employees (Hoonakker et al., 2004;Korunka et al., 2008).

Role Ambiguity & Role Conflict
According to Spector (1997), roles in jobs are the models of behaviors that organizations expect from employees (Mulki, Jaramillo, & Locander, 2008).Moreover, to Quick & Quick (1984), while executing their jobs, employees usually have to meet the expectations of other employees in the organization (Mulki et al., 2008).In addition, Rizzo, House, & Lirtzman (1970) noted that the roles which cause stress in job include role ambiguity and role conflict (Mulki et al., 2008).Rizzo et al. (1970) defined that role ambiguity is the level of certainty about responsibility, the right, time allocation, and the relationships, the clarity of the instructions, policies (Tang & Chang, 2010).In addition, Rizzo et al. (1970) showed that role ambiguity is the result of the ambiguity in working responsibility, therefore the employees have the feeling that they can not control the tasks -for example, service provider employees have the responsibility for attracting new customers or not (Mulki et al., 2008) ? Rizzo et al. (1970) also illustrated that role conflict is the conflict between people' expectations.Role conflict occurs when the employees believe that their jobs have to meet the conflict requirements and expectations, result in the employees have trouble in selecting the expectations to handle and they could have stress -for example, satisfy customers while cutting the cost in the same time (Mulki et al., 2008).

Career Opportunities
There are many studies which showed the negative relationship between career opportunities and stress, burnout and emotional exhaustion in IT field.According to Korunka et al. (2008), Hoonakker et al. (2004), there is a correlation between career opportunities and emotional exhaustion.If an employee can not have their career opportunities need satisfied for a long time, he/she could have negative emotions which lead to stress.Quantitative research: the sample was collected through direct sampling by using the questionnaire.The sample is used to assess the scales and test the hypotheses.This study was conducted during October and November, 2014.Multivariate regression was used to test the hypothesis with the help of SPSS 16 software.

The Study Hypotheses
There were 238 respondents to the questionnaire.After data screening, only data from 194 participants was analysed.Regarding gender, there were 66 subjects (34%) are females and 128 subjects (66%) are males.Regarding age, there are 2 subjects (1%) are under 24 years of age, 27 subjects (13.9%) are from 24 to under 26 years old, 40 subjects (20.6%) are from 26 to under 28 years old, 60 subjects (30.9%) are from 28 to under 30 years old and 65 subjects (33.5%) are 30 years old or older.Regarding occupation, 30 subjects (15.5%) are in hardware, 82 subjects (42.3%) are in software, 43 subjects (22.2%) are in the network, and 39 subjects (20.1 %) are in telecommunications.Regarding income, no subject (0%) had a monthly income of less than 2 millions VND, 14 subjects (7.2%) are from 2 to under 4 millions VND, 37 subjects (19.1%) are from 4 to under 6 millions VND, 42 subjects (21.6%) are from 6 to less than 8 millions VND, 101 subjects (52.1%) are from 8 millions VND or more.

Procedure
In formal quantitative interviews, 260 questionnaires were delivered, including 110 questionnaires were given directly and 150 questionnaires were sent via email through the support of google docs, result in 228 respondents.After the test, 34 were excluded due to many empty cells or most of the options have the same results (thus there is a suitable reason to believe that these answer sheets have not much value).

Measures
Scales in this study are based on theory and scales in the world.They were adjusted and supplemented in accordance with Vietnam's conditions based on the results of qualitative research techniques discussed bilaterally.On the basis of the theory, there are eight research concepts used in this study which are: (1) IT job demands (denoted JD), (2) Supervisory Support (SS), (3) Colleagues Support (CS), (4) Role Ambiguity (RA), (5) Role Conflict (RC), (6) Career Opportunities (CO), (7) Emotional Exhaustion (EE), (8) Turnover Intention (TI).

IT Job Demands
IT job demands: IT job demands, which was denoted JD, was measured by observed variables: JD1, JD2, JD3, JD4, JD5, JD6, JD7, JD8.The observed variable is based on the scale of De Jonge et al. (1995) used in the study of self-management in work, happiness and health: "Job autonomy, well-being, and health: A study among Dutch health care workers", including 8 observed variables, such as: You have to do the tasks under time pressure.These scales were translated into Vietnamese and tested through qualitative research with IT employees in HoChiMinh City.The observed variables are measured by 7-point Likert-type scale ranging from 1 (strongly disagree) to 7 (strongly agree).This scale showed a good reliability.Cronbach's alpha was 0.8749.

Supervisory Support
Supervisory Support, which was denoted SS, was measured by four observed variables SS1, SS2, SS3, SS4.The observed variables are constructed based on the scale of De Ruyter, De Jong, & Wetzels (2009) used in the study of the influence of environmental management on the expansion of the B2B group, such as: Your team can communicate with the manager if need.These scales was translated into Vietnamese and tested through qualitative research with IT staff in HoChiMinh City.
Through qualitative interviews, the concept of "Supervisory Support" was revised to "Manager's Support" to better fit the situation in Vietnam but remained significant.The scale was therefore also revised accordingly.The observed variables are measured by 7-point Likert-type scale ranging from 1 (strongly disagree) to 7 (strongly agree).This scale showed a good reliability.Cronbach's alpha was 0.8756.

Colleagues Support
Colleagues Support, which was denoted CS, was measured by four observed variables which are CS1, CS2, CS3, CS4.These observed variables are constructed based on the scales of De Jonge et al. (1995) which were used in the study of self-management in work, happiness, health: "Job autonomy, well-being, and health: A study among Dutch health care workers", such as: You have a good relationship with your colleagues.These scales were translated into Vietnamese and tested through qualitative interviews with IT staff in HoChiMinh City.The observed variables are measured by 7-point Likert-type scale ranging from 1 (strongly disagree) to 7 (strongly agree).This scale showed a good reliability.Cronbach's alpha was 0.8468.

Role Ambiguity
Role Ambiguity, which was denoted RA, was measured by 3 observed variables which are RA1, RA2, RA3.The observed variables are constructed based on the 3-item scale of Bettencourt & Brown (2003), such as: You know what are your responsibilites.This scale was translated into Vietnamese and tested through qualitative interviews with IT staff in HoChiMinh city.The observed variables are measured by 7-point Likert-type scale ranging from 1 (strongly disagree) to 7 (strongly agree).This scale showed a good reliability.Cronbach's alpha was 0.7642.

Role Conflict
Role conflict, which was denoted RC, was measured by three observed variables RC1, RC2, RC2.The observed variables are constructed based on the scale of Bettencourt & Brown (2003), such as: You have to violate some rules to execute some tasks.This scale was translated into Vietnamese and tested through qualitative interviews with IT staff in HoChiMinh city.The observed variables are measured by 7-point Likert-type scale ranging from 1 (strongly disagree) to 7 (strongly agree).This scale showed a good reliability.Cronbach's alpha was 0.6559.

Career Opportunities
Career Opportunities, which was denoted CO, was measured by three observed variables CO1, CO2, CO3.The observed variables are constructed based on the 3-item scale of Hill, Ferris, & Baker (2004), such as: You feel optimistic about your future at your company.This scale was translated into Vietnamese and tested through qualitative interviews with IT staff in HoChiMinh city.
Through qualitative interviews, the observed variable CO3 was adjusted from "Director" to "Manager" in order to fit more with Vietnamese conditions.The observed variables are measured by 7-point Likert-type scale ranging from 1 (strongly disagree) to 7 (strongly agree).This scale showed a good reliability.Cronbach's alpha was 0.8909.

Emotional Exhaustion
Emotional exhaustion, which was denoted EE, was measured by 8 observed variables E1, EE2, EE3, EE4, EE5, EE6, EE7, EE8.The observed variables are constructed based on the scale of the Maslach & Jackson (1981) with 9 observed variables, however the variable "Working directly with people gives me a lot of stress" which was ruled out in many previous studies sufficiently value (Geurts et al., 1998), thus remaining 8 observed variables.The scale was translated into Vietnamese and tested through qualitative interviews with IT staff in HoChiMinh city.
The observed variables are measured by 7-point Likert-type scale ranging from 1 (strongly disagree) to 7 (strongly agree).This scale showed a good reliability.Cronbach's alpha was 0.9220.Cronbach alpha test results showed that the scales (except the scale for dependent variables TI which has only one observed variable) achieved the reliability.The corrected item-total correlations are high (The lowest is of the variables RC = 0.4007).Cronbach alpha of the scales are high, the lowest is of the scale of role conflict (0.6559).

Descriptive Statistics
Specifically, Cronbach alpha of the scale for IT job demands is 0.8749; the scale of supervisory support is 0.8762; the scale of colleagues support is 0.8677; the scale of role ambiguity is 0.7642; role conflict is 0.6559; career opportunities is 0.8909; and the emotional exhaustion is 0.9220.The observed variables have corrected item-total correlation > 0.3.Cronbach Alpha of the scales are greater than 0.6.Therefore, all the observed variables will be used in subsequent analysis which is EFA.

Exploratory Factor Analysis (EFA)
KMO was checked and met the requirements.There were 6 factors which were extracted at eigenvalue 1.082 and total variance extracted was at 68.974%.The weights of the scales met the requirement (>0.50).The lowest weights are of the variables which are JD2 and JD8 (0.547) of JD scale.Therefore, these scales met the requirements and will be used in next quantitative step.
EFA result of dependent variable is as below: As shown in Table 2, Total variance extracted of the dependent variable was at 64.764% and the Cronbach Alpha was at 0.922.
After EFA step, the hypotheses are not changed.Therefore, there are 7 hypotheses were recommended.
Before all proposed relationships were tested, a correlational analysis was conducted (Pearson correlations) among the variables mentioned in this study.As shown in Table 3, the hypothesized pattern of relationships holds true and the relationships point in the expected direction.

Model 1 EE = f (JD, SS, CS, RA, RC, CO)
In model 1, adjusted R 2 is 0.476, therefore model 1 is fit.F value is 30.278 with a tiny Sig.(< 0,000) shows that it is safe to eliminate the H 0 hypothesis in which all the coefficients of the independent variables are zero.Therefore the multiple linear regression is fit to the data and can be used.
There are 6 independent variables in which 3 variables which are SS, CS, CO are negatively associated with EE, 3 remain variables which are JD, RA and RC are positively associated with EE.Sig. of 2 variables SS and CS >0.05, therefore these variables do not affect in meaningful statistics to EE.All 6 independent variables have VIF<2, therefore there is no multicolinearity.In model 2, adjusted R 2 is 0.227.F value is 57.696 with a tiny Sig. which is <0.000 shows that it is safe to reject the H 0 hypothesis in which the coefficient of EE variable is zero.Therefore the multiple linear regression is fit to the data and can be used.

Conclussion
This study has recognized some factors that affect to emotional exhaustion and turnover intention of IT staff and checked the influences of every factor.At first, a research model was built based on theory base.Qualitative research was executed in order to check the scales which were translated into Vietnamese about the clarity and whether participants could understand exactly or not.Then CFA and EFA showed that the scales met the requirements.After these above steps, research model and hypotheses keep unchanged.Test results accepted five hypotheses: H1, H4, H5, H6 and H7, in which the factor affects most powerful to emotional exhaustion is IT Job Demands (beta = 0.401; p = 0.000).The next factors are Role conflict (beta = 0.322; p = 0.000); Career opportunities (beta = -0.257;p = 0.000) and Role ambiguity (beta = 0.195; p = 0003).In addition, emotional exhaustion is positively associate with turnover intention (beta = 0.481; p = 0.000).
The study also showed that two factors which are Supervisory support and Colleagues support did not affect to Emotional exhaustion (SS: beta = -0.091,p = 0.192; CS: beta = -0.035,p = 0.601).This result can be explained as two mentioned factors both have high mean (SS: 4.84; CS: 5.41).Therefore, there were variances because of people' perception and the own situation of each company, but they do not affect to Emotional exhaustion.

Managerial Implication
This study has the positive contribution to management in practical.Research model was tested and fit to the population.Five hypotheses accepted bring to managers some useful information in management.Accordingly, managers can care more about these factors to reduce emotional exhaustion and turnover intention in their company.
The factors that affect to emotional exhaustion in sequence are IT job demands, role conflict, career opportunities and role ambiguity.However, two factors which are supervisory support and colleagues support are still important and the managers need to care about them.In addition, emotional exhaustion affects strongly to turnover intention, therefore it needs to be tracked.Accordingly, the managers should have the appropriate policies that fit to their company characteristics, their departments, time and strategies in order to improve weak factors.

Limitations
The study was only conducted with the object of study in the area of HoChiMinh City with convenient sampling methods, despite wide selection of special groups of people with different demographic, but not yet collected all the comments the IT staff groups.The ability to generalize the results will be higher if the data collection is done on a wider geographical areas, with sampling more representative, such as probability sampling, with the using all possible groups represent the entire IT staff in the territory of Vietnam .
Second drawback of the study is only considering the general impact of the entire sample collection and analysis unrealized differences between various objects on the demographic characteristics.In addition, the study considers only factors affecting the turnover intention rather than the turnover behavior, although turnover intention is the best predictor for the turnover behavior.Further research should de directed at determining the differences between IT staff and considering turnover behavior of the IT staff.
Study results will define clearly what parts of job characteristics, organization characteristics and human resources policies can affect to emotional exhaustion and turnover intention.Based on past literature, these hypotheses were formed: H1: IT job demands is positively associated with emotional exhaustion.H2: Supervisory support is negatively associated with emotional exhaustion.H3: Colleagues is negatively associated with emotional exhaustion.H4: Role ambiguity is positively associated with emotional exhaustion.H5: Role conflict is positively associated with emotional exhaustion.H6: Career opportunities is negatively associated with emotional exhaustion.H7: Emotional exhaustion is positively associated with turnover intention.

Figure
Figure 1.Study hypotheses Final sample size was N = 194.The sample in formal quantitative was collected from: (1) two Master of Business Administration classes of Polytechnic University of HoChiMinh City; (2) two System Management classes of Polytechnic University of HoChiMinh City; (3) two Master of Business Administration classes of University of Economics HoChiMinh City; (4) officers who are working at the IT companies in HoChiMinh City.

Table 1 .
Descriptive statistics Table 1 showed the Min, Max, Mean and Standard Deviation of the variables with 194 samples.Highest means are of Colleagues Support and Supervisory Support (5.4111 and 4.8432).

Table 2 .
EFA result of the dependent variable