CHAPTER III RESEARCH METHODOLOGY
Transcription
CHAPTER III RESEARCH METHODOLOGY
CHAPTER III RESEARCH METHODOLOGY A cross-sectional study design was used in order to determine the level of job stress, role conflict, role ambiguity, and job satisfaction; investigate the relationships between job stress, role conflict, role ambiguity and job satisfaction; and examine the predictive effects of job stress, role conflict and role ambiguity on job satisfaction. Study setting, population and sample, sampling method, instruments, data collection procedures, ethical consideration and data analysis are presented in this chapter. Research design A cross-sectional study design was used to accomplish the purposes of this study. Setting This study was conducted in the Thai Nguyen provincial general hospitals. These hospitals are public hospitals, including the A hospital, the C hospital and the Gang Thep hospital. They are at the secondary care level in health care system, and are quite similar about organizational structure and policies, but only differently about total number of staffs, and number of the beds. They were managed under the jurisdiction of the Provincial Department of Health in Thai Nguyen Province. There are about 20-24 functional department in each hospital. Total staff nurses have about 150-200 employees. Total of the beds have about over 300-350 (Department of Planning and Investment, 2010). They provide to treat for outpatients and inpatients with all kinds of disease, illness and injury (Department of Health, 2009). Population and sample Population: the target population for this study consists of 239 staff nurses working in three clinical units (medical unit, surgical unit, and intensive care unit 45 [ICU]) of three provincial general hospitals (the A hospital, the C hospital, the Gang Thep hospital) in Thai Nguyen province, Vietnam. The population is shown in table 1. Table 1 The detail of population in Thai Nguyen provincial hospitals Staff nurses of three units Hospital Population Medical Surgical ICU A hospital 82 40 30 12 C hospital 83 33 35 15 Gang Thep hospital 74 35 25 14 239 108 90 41 Total Sample was selected from three clinical units (medical unit, surgical unit, and ICU) in three Thai Nguyen Provincial General Hospitals (A hospital, C hospital, Gang Thep hospital) from September through October, 2010. The participants are officials, they directly provide patient care. Those nurses have met the eligibility criteria are as follows: Inclusion criteria 1. Currently working full time in the three clinical units (medical unit, surgical unit, and intensive care unit). 2. Have been working as a nurse for at least 09 months after recruited. 3. Providing direct patient care. 4. Willing to participate in the study Exclusion criteria 1. Head nurses. 2. The director nurse. Full time employees are used due to potentially different stress experienced by both novice time and main working time. The 9 months are chosen because the respondents were well-oriented to the organization, and passed the initial stress of working in a new environment. Nurses in management positions are excluded because there are differences in stressors related to job responsibilities. 46 Sample size: Yamane (1973) provides a formula to calculate sample sizes with a 95 % confidence level and e = .05. When know size of population the sample size was determined based on the formula as follows: N n= 1 + N(e)2 n = sample size N = population size e = level of precision The values were set for formula: N = 239 e = 0.05 239 n= 1 + 239(0.05) 2 = 150 Therefore, the sample size was 150 nursing staffs. Sampling technique This study was applied the multistage sampling technique to determine the number of participants following step. 1. The first step was applied the purposive sampling technique to determine the hospitals in this study. The secondary care level of hospitals in Thai Nguyen province is classified into 3 public health facilities, and 1 military health facility. There is a difference between public health facilities and military health facility. The public health facilities provide medical services for large population and all social strata. In contrast, the military health facility serves only to specific groups. The researcher selected 3 provincial hospitals from the public health facilities to obtain representative samples. 2. The second step was applied the stratified random sampling technique to determine 150 nursing staffs from the 3 provincial hospitals at different hospitals. The researcher calculated sample size by applying the formula of Cochran (1977). The numbers of participant in each hospital are detailed in Table 2. 47 Nh*n nh = N nh = Sample in each hospital Nh = Population of staff nurses in each hospital N Population (N = 239) = n = Sample size (n = 150) Table 2 Population and sample size of subjects from different hospitals in the three Thai Nguyen provincial hospitals Hospital Population Sample of hospital A hospital 82 52 C hospital 83 52 Gang Thep hospital 74 46 Total 239 150 3. The third step was applied the quota sampling technique to achieve appropriate sample representation at different units. The researcher calculated sample size for the staff nurses by using the formula of Cochran (1977). The numbers of participant in each unit are detailed in Table 3. Table 3 Number of subjects from each of clinical units Hospital Popu. Samp. Popu. Samp. Popu. Samp. Med Med Sur Sur ICU ICU A hospital 40 25 30 19 12 8 C hospital 33 21 35 22 15 9 Gang Thep hospital 35 22 25 15 14 9 Total 68 56 26 48 4. The fourth step, the participants were selected by using a simple random sampling technique. The researcher met the director nurse of each hospital to get the list name of nursing staff, and then selected participant by using simple random sampling technique. The name of staff nurses were established the lists on papers with each unit of each hospital. Each name was ordered a number. Number order in each of the lists was sorted in ascending numerical order starting with 01. And then, the researcher wrote one numerical label on a small piece of paper, number of pieces of paper are consistent with the number of staff nurses of each unit, after that, put all of them in one box and mix well. At the final, the researcher picked up the random one piece at the time, subject who assigned number was matched with the random number, this process continues until the desired sample size has been reached. Instrumentations Questionnaires This study used the self - administrated questionnaires to collect nurse opinions comprising of 4 sections including the Demographic Questionnaire, the Expanded Nursing Stress Scale, the Role Conflict and Ambiguity Scale and the Job Satisfaction Survey. Section 1: Demographic questionnaire The demographic data questionnaire was constructed by the researcher. It was designed to obtain information regarding: age; gender; marital status; number of children, educational level; name of hospital; working units; years of working in nursing profession; and years of working in the current unit; income; number of patient per shift. It includes 12 items (Appendix 6). Section 2: Expanded nursing stress scale Expanded Nursing Stress Scale (ENSS) was used to measure job stress. A final version ENSS was developed by French and her colleagues in 1995 (French et al., 2000). ENSS using in this study contains 54 items in eight subscales: 1) Death and Dying with seven items, 2) Conflict with Physicians with five items, 3) Inadequate Emotional Preparation with three items, 4) Problems Relating to Peers with six items, 5) Problems Relating to Supervisors with seven items, 6) Work Load with nine items, 7) Uncertainty Concerning Treatment with nine items, and 8) Patients 49 and their Families with six items (French et al., 2000). Discrimination subscale with three items was not use in this study, because the difference about the cross - culture, and it is not appropriate for measuring. Internal consistency reliability was assessed by using Cronbach’s coefficient alpha. The whole ENSS demonstrated the reliability (α = .96) (French et al., 2000). The current study obtained reliability to be 0.94. The 54 items were ranked on a 4-point Likert response scale from 1 - never stressful to 4 - extremely stressful. The total score ranged from 54-216. Responses are summed and divided by 4 to provide a mean score. The high score means higher level of job stress. Responses were some and divide by 3 to provide a mean score. Higher scores indicate the high level of job stress. Score transformation: the highest mean score was 4 and lowest mean score was 1. Dividing the result of highest mean score minus the lowest mean score by 3 gives an interval of 1 (Polit & Hungler, 1999). The written permission was obtained from the author to use this instrument. Table 4 Scales of ENSS and item number (French et al., 2000) Subscales Item numbers Death and Dying 1,8,15,24,34,44, and 50 Conflict with physicians 2, 9, 25, 35, and 45 Inadequate emotional preparation 3, 10, and 17 Problems relating to peers 4, 11, 18, 19, 20, and 47 Problems relating to supervisors 5, 27, 28, 37, 43, 46, and 51 Workload 12, 21, 29, 38, 39, 42, 48, 52, and 54 Uncertainty concerning treatment 6, 13, 16, 22, 26, 30, 33, 36, and 40 Patients and their families 7, 14, 23, 31, 32, 41, 49, and 53 50 Table 5 Interpretation of job stress (Polit & Hungler, 1999) Mean Score Level of job stress 1.00 – <2.00 Low job stress 2.00 – 3.00 Moderate job stress >3.00 – 4.00 High job stress Section 3: Role ambiguity and role conflict scale Role stress was measured by translated Role Conflict and Ambiguity Scale (RCAS) (Appendix. 4). Role conflict and Ambiguity scale was developed by Rizzo and his colleagues in 1970 (Rizzo et al., 1970). It consists of 14 items in 2 subscales: role conflict with 8 items, and role ambiguity with 6 items. Lu et al. (2007) reported Cronbach’s alphas of 0.82 for role conflict, and 0.80 for role ambiguity. In the current study, internal consistency coefficient alpha was 0.80 for total scale. Each item was rated on five-point Likert scale ranging from 1 - never to 5- very often and the total scale ranged from 14-70. Tunc and Kutanis (2009) have noticed six items for role ambiguity should be reverse scored before summing the score. Responses are summed and divided by 5 to provide a mean score. Higher scores represent higher level of role stress. Score transformation: the highest mean score was 5 and lowest mean score was 1. Dividing the result of the mean score minus the lowest mean score by 3 gives an interval of 1.33 (Polit & Hungler, 1999). Table 6 Interpretation of role stress (Polit & Hungler, 1999) Mean Score Level of role stress 1.00 – < 2.33 Low role stress 2.33 – 3.67 Moderate role stress > 3.67 – 5.00 High role stress 51 Section 4: Job satisfaction survey (JSS) The translated of Job Satisfaction Survey (JSS) was used to measure job satisfaction (Appendix. 6). The version of Job Satisfaction Survey was developed by Spector in 1994 (Spector, 1997). The questionnaire consists of 36 items, it is separated nine facets of job satisfaction. Those facets are 1) pay and pay rises, 2) promotional opportunities, 3) fringe benefits, 4) contingent reward (appreciation and recognition), 5) supervision (the person’s immediate boss), 6) coworker, 7) nature of work (type of work done), 8) communication within the organization and 9) operating procedures (rules and procedures) to assess employee attitudes about the job and aspects of the job. Each aspect was assessed with 4 items, and the total score is computed from all items. A summated rating scale was used, each items is rated on six-point Likert scale ranging from 1- Disagree very much to 6-Agree very much and the total score ranged from 36-216. The JSS has some of its items written in positive and negative directions. The negative worded statements have to be reverse scored before summing the score. The level of job satisfaction as measured by JSS, the higher the mean score, the higher the level of job satisfaction. Spector (1997) reported reliability (coefficient alpha) of the JSS ranging from 0.66 to 0.91 (Spector, 1997). Norbu (2010) reported reliability to be 0.90. In this study, Cronbach’s alpha for the total score was 0.80. The written permission was requested from the author to use his instrument. Below are reliability results on 9 facets of job satisfaction (Spector, 1997). Table 7 Scale of Job Satisfaction Survey, descriptions and item number (Spector, 1997). Facets Alpha Description Item number Pay .75 Pay and pay raises 1, 10r, 19r, 28 Promotion .73 Promotion opportunities 2r, 11, 20, 33 Supervision .82 The person’s immediate supervisor 3,12r, 21r, 30 Fringe benefits .73 Fringe benefits 4r,13, 22, 29r Contingent .76 (not necessarily monetary) given 5,14r, 23r, 32r rewards for good performance 52 Table 7 (Continue) Facets Alpha Operating .62 conditions Description Item number Satisfaction with rules and 6r, 15, 24r, 31r procedures Coworkers .60 Satisfaction with coworkers 7, 16r, 25, 34r Nature of Work .78 Satisfaction with the type of work 8r,17, 27, 35 done Communication .71 Satisfaction with communication 9,18r, 26r, 36r within the organization Total .91 Total of all facets NOTE: followed by “r” should be reverse scored. Table 8 The JSS scoring of responses to the items of the scale Responses Positive items Negative items Disagree strongly 1 6 Disagree moderately 2 5 Disagree slightly 3 4 Agree slightly 4 3 Agree moderately 5 2 Agree strongly 6 1 Table 9 Interpretation of job satisfaction (Spector, 2007) Mean score Level of job satisfaction 1.00-2.99 Low 3.00-4.00 Moderate 4.01-6.00 High 53 Translations, validity and reliability of instrument Instrument translation The original instruments using for this study are English version, hence, these instruments were translated into Vietnamese version. The translation and backtranslation method used a combination and modified of Brislin’s back-translation model and a committee approach (Brislin, 1970; Lu et al., 2006). The specific translation procedures were as follows: Step 1. Two independent bilingual experts, including 1 bilingual researcher aware of the objective of the questionnaire and 1 other bilingual person without any knowledge of the instrument (professional translator) translated the instruments from English into Vietnamese. Step 2. The 2 translators synthesized the results of the translation by consensus. The researcher was observed and recorded a Vietnamese version. Step 3. The synthesis of the two translations was independently translated back into English again by 2 different bilingual persons including 1 English teacher and 1 health specialized translator. And then, they synthesized and recorded the result of the back-translation by consensus. Step 4. The panel of three experts including 1 psychologist, 2 nursing administrator were the bilingual experts, as well as the experts in healthcare management compared the original English version and the back-translation version with the various translations and developed the prefinal version by consensus. The role of expert committee was to assured the face validity of translated questionnaire. The validity testing According to Burns and Grover (2005), the valid of instrument may be valid in situation, but not valid in another. Thus, the validity of instrument should be reexamined. The three experts, including 1 nursing management educators, 1 nursing educator and 1 medical psychologist were selected for checked validity. They are experts in research and English. A content validity index (CVI) was calculated by S-CVI/UA (Universal agreement) approach method (Polit & Beck, 2008). The result of CVI reported to be 1.00 in this study. 54 The reliability testing The version of the Vietnamese instruments were tested for internal consistency reliability before conducting the actual study. Internal consistency used the Cronbach’s alpha coefficients to determine the reliability of each part of the questionnaire. The primary reliability testing was carried out thirty staff nurses working in Bac Kan Provincial General Hospital in September, 2010 where had similar characteristics to the Thai Nguyen Provincial General Hospital such as structure, organizational system but only different about total number of staff nurses, total number of beds (Vietnam Association Young Physician [VAYP], 2009). These thirty staff nurses were selected from the three clinical units including medical unit, surgical unit, and intensive care unit. These thirty staff nurses who had similar characteristic to the subject and not included in the main study used to test the reliability. In this study, the Cronbach’s alpha obtained for ENSS, RCAS, JSS were 0.94, 0.80, 0.80. Data collection procedures The data collection was carried out from September through October, 2010. The procedure was as follows: 1. The thesis proposal was granted ethical approval from Burapha Institutional Review Broad (Burapha-IRB) by submitting research approval document. After obtained the permission to collect the data. A separate letter from the Dean of Faculty of Nursing, Burapha University (Appendix 2) was sent to the Director of Health Department of Thai Nguyen Province. 2. The researcher was obtained the permission from Director of Provincial Department of Health in Thai Nguyen province for all the three hospital authorities to collect the data by submitting approval document and the introducing letter to conduct the research. 3. The researcher obtained the permissions from director of hospitals as well as nursing administrator to collect the data. 4. The researcher obtained the lists of the potential respondents from the Department of Personnel and Organization of the three hospitals. And then, the 55 researcher selected participants by using simple random sampling method from the lists. 5. The researcher met the head nurses in the three units in each of hospitals for an orientation to the study, informed the objectives of the study, and required them about the supports to establish the meeting with participants. 6. The researcher contacted with the participants in one hour at the conference room after the annual meeting. And then, explains the objectives of study, and invites them to participate in the study, ask for their voluntary. Then give consent form to each subject who was willing to participate. The cover letter explains the nature of the study, invitations for participation in the study, and provided assurances of the anonymous, voluntary, and confidential nature of the responses, method for insuring confidentiality and assurance that participation is voluntary. 8. After receiving the written consent forms, the questionnaires were handdelivered direct to the participants by the researcher. The participants were asked to mark the section immediately by using paper and pencil self-rating questionnaires, and return them direct to the researcher. The researcher waited for complete, and collected the responses. 9. The researcher reviewed all the data, and asked the subjects do completely their questionnaires, which give incomplete answers. Then the researcher prepares the data for analysis. 10. The researcher codes the questionnaires. 11. Entering data into the computer. Data analysis The data was coded and was entered into statistical program. Data analyzed using descriptive and inferential statistics as determined by the level of measurement. An alpha level of 0.5 was considered as statistical significance. Descriptive statistical tool including means ( X ), percentage (%), standard deviations (SD), and frequency distribution (n) were used to describe the demographic data of the sample; levels of the job stress score and eight subscales scores; levels of 56 the total job satisfaction score and nine subscales scores; levels of role stress and two subscales scores. The Pearson product-moment correlation coefficients (r) were calculated to determine the relationships between job stress, role conflict, role ambiguity and job satisfaction. The correlation coefficient ranges from +1.00 through 0.00 to -1.00. If the r-value is +1.00, there is a perfect positive linear relationship. If the r-value is 1.00, there is perfect negative linear relationship or a perfect reverse relationship. If the r-value is 0.00, there is no correlation. Table 10 The correlation values and magnitude of correlations (Cohen, 1988) Correlation Positive Negative Small 0.10 to .29 - 0.29 to -.10 Moderate 0.30 to .49 -0.49 to -.30 High 0.50 to 1.00 -1.00 to -.50 If the r-value equals 0.10 to 0.29 or -0.29 to -0.10, correlation was interpreted as small. If the r-values is 0.30 to 0.49 or -0.49 to -0.30, there was a moderate correlation and if the r-values equals 0.50 to 1.00 or -1.00 to -0.50, there was a high correlation between the two given variables. An alpha (α) level of ≤ .05 was deemed statistically significant for all inferential test utilized: Pearson’s productmoment correlations and Spearman rank-order correlations. Multiple Regression Analysis: In response to the objective, to examine which factors among job stress, role conflict and role ambiguity are able to predict job satisfaction. The Stepwise Multiple Regression Analysis is used to predict the relationships between independent variables (job stress, role conflict, role ambiguity) and one dependent variable (job satisfaction). This procedure can find which variable has most influence on the variance of dependent variable. The analysis result is presented with an equation which consists of independent variables that have the significant effect on with the dependent variable. This is the best equation for value estimation of dependent variable. The assumptions of regression analysis were tested including normality of dependent and independent variables, multicolinearity, 57 linearity, autocorrelation and homocedasticity. The basic assumptions for data analysis consisted of: 1. Normal distribution Using a histogram of standardized residuals, assessed relationships are linear and the dependent variable is normally distributed for each value of the independent variable. The distribution of the residuals should be approximately normal (Munro, 2001). Furthermore, the researcher had tested the normal distribution of standardized residuals by using Kolmogrove-Sminov test, the result of Kolmogrove-Sminov test has to be non-significance. These indicated the standardized residuals were normal distribution. In addition, the researcher also used Fisher skewness coefficient (Skew/S.E) to test the normal distribution of dependent and independent variables. When the result of Fisher skewness coefficient lies between -1.96 and 1.96 meaning that is significant and the normal distributions were found by statistic. 2. Multicollinearity The Pearson’s product-moment correlation coefficient was used to test the relationships between independent variables. This assumption obtains when the correlation among the independent variables should not multicollinearity meaning that the correlation coefficient among the independent variables should not higher than 0.80 (Munro, 2001). In addition, tolerance of a variable is used as a measure of collinearity, which variables with high tolerances have small variance inflation factors, and vice versa (Munro, 2001). 3. Homoscedasticity Scatter plot was selected to test the homoscedasticity. The residuals can be plotted against the predicted values and against the independent variables. When standardized predicted values are plotted against observed values, the data would form a straight line from the lower-left corner to upper right corner, if the model fitted the data exactly. In general, they cluster fairly close around the prediction line (Weisberg, 1980). 58 4. Linearity When the residuals are from a normal distribution, the plotted values fall close to the line in the normal probability plot. Thus, the probability was selected to check this assumption. 5. Autocorrelation The Durbin-Wastson was selected to examine the autocorrelation of residual. If Durbin-Wastson value lies between 1.5 and 2.5, it means that the residual was no autocorrelation. Ethical consideration Before collecting the data, approvals for conducting the study were obtained from Burapha University, Thailand and Provincial Department of Health of Thai Nguyen Province, Vietnam. Permissions conducting the study were obtained of the three hospitals. Following the completion of these formalities, letter of introduction to research and the requirements of participation as well as the questionnaires were handed out to nurses. To ensure the anonymity and confidentiality of responses, no names were attached to survey; instead their responses were coded so that they would not be identified. The participants were informed that participations in the study to be purely voluntary and that no remuneration would be given. The participants were told that information they provide would be kept confidential and would only used for the purposes intended, and was subject to all the legal requirements regarding data collection. The participants were also assured that only the overall results would be shared with ministry, health center and the nursing administrators for the purpose of designing the needed managerial interventions. During to data collection, the subjects want to refuse or withdraw from the study, the researcher will respect their decision, and to assure anonymity and confidentiality, and no penalty for withdrawal or termination from the study.