In Chinese society, individual via family socialization process, and learned about psychology or behavior process in family life and spread in group life, and show a strong family ideology in common business or government organization (Fan, 2004).
Based on above, our investigation of public sector for government; administrative organizations is a bureaucracy, manager has to under pressure from superior manager in organization, hence, an authoritative leader has to focus on task (Authoritative);
manager has to concern about subordinates appropriately, find out the need of subordinates initiatively, and solve the problem for subordinate then cohere centripetal force of subordinates (Benevolent); besides, manager should strictly observe discipline of job, administration by law, impartial, established trust with subordinates in public organization, and show an aspect of active of job, and win the admire and emulate;
manager has the responsibility for the illegal behavior of subordinate (Moral) by ability of professional, with an open mind attitude and integrity position of job.
As noted earlier, we can infer to there are some component of PL (paternalistic
leadership) in public sector. Therefore, the subjects in this study were sampled from qualified civil servants in Taiwanese public sector, for example, police agency of ministry of the interior, civil servants in township office, city government, forest bureau, coast guard administration and the unit of judiciary in Hualien, Taipei, Kaohsiung, and Taoyuan etc. The population for this investigation included all qualified civil servants and. This study expects have over 300 respondents.
3.5 Questionnaires Design
As mentioned early, a 48-item survey questionnaire was developed to obtain the responses from the Taiwan Public Sector about their opinions on various research variables. The questionnaire items were also translated into “Chinese”. The questionnaire of this study is consisted of two constructs: “Paternalistic Leadership (33 items),” “Organizational Ethical Climate (15 items).”
This research question and research method are sensitive, so it should not easily commit correct responses. However, for the safety, this research will take the following action to ensure its valid.
First, voluntary participation is important (Babbie, 1998). It means every participant has the right to join and quit a research. As a result, participants will be received an informed consent before they take part in the research (Wysocki, 2004). If they are not interested in it, they can reject.
Second, anonymity and confidentiality are also important issues (Babbie, 1998).
Therefore, the questionnaire does not request the participants to give their name. In addition, the researcher and the participants should not know one another, because the questionnaire distribution and submission are via third parties.
Third, at the beginning of the questionnaire, it is an introduction to explain the objectives of this research to participants. Although it may influence the response of the participants, it is necessary. It is because researchers have to be honest (Babbie, 1998). For the effectively, so we mixed the all items.
Finally, the negative wordings will be avoided in the questionnaire, because negative wordings may give psychological harm to participants. An ethical research should not harm the participant physically and psychologically (Bryman, 2004).
3.6 Measures
3.6.1 Preamble
For measured the leaders behavior, PL (paternalistic leadership) scale is the main index, in business sample (Erben & Güneser, 2008) (the rate of return was 84%), authoritarianism, benevolence, and moral Cronbach’s α was 0.84, 0.83, 0.91, respectively; according to Victor and Cullen’s (1988) taxonomy, ethical climates involve three classes of ethical criteria: egoism, benevolence, and principled (Koh &
Boo, 2001; Schminke et al., 2005). Egoistic climate, benevolent climate, and principled climate were assessed with Koh and Boo’s (2001) 12-item scale, four items for each climate type. Coefficient α of the scales were 0.67, 0.75, and 0.82, respectively.
Alpha coefficients (Cronbach, 1951) are computed to assess the reliability of PL and organizational ethical climate. According to Robinson and Shaver (1973), ifαis greater than 0.7, it means that it has high reliability and ifαis smaller than 0.3, then it implies that there is low reliability. Hence, these two scales (please refer to Appendix I and II for details) are suitable for this study.
Paternalistic Leadership. Paternalistic leadership has three distinct elements:
authoritative leadership, benevolent leadership, and moral leadership. The paternalistic leadership scale that was developed by Cheng, Chou and Farh (2000) was employed for measurement of this factor. Each of the three subtypes of leadership was measured with the 5 or 6 highest factor-loading items of the factor analysis result of Cheng et al.
(2000). The internal reliability coefficient for the authoritarian leadership scale was 0.75, for the benevolent leadership scale was 0.87, and for the moral leadership scale was 0.86. Sample items include: “My supervisor is like a family member when he/she gets along with us,” “My supervisor does not take advantage of me for me for personal gain,” and “My supervisor asks me to obey his/her instructions completely”.
Each type was measured using a six-point rating scales, with 1 representing “strongly disagree” to 6 representing “strongly agree”.
Organizational Ethical Climate. Egoistic climate, benevolent climate, and principled climate were adopted from Victor and Cullen (1988) 9-item scale, three items for each climate type. Sample items of egoistic climate were: “My organization emphasizes the importance of furthering its interests’’ and “Employees in my organization are not expected to be concerned with the organization’s interests all the time.’’ Sample items of benevolent climate included: “Concern for employees is prevalent in my organization’’ and “My organization does not emphasize employee welfare.’’ Sample items of principled climate were: ‘‘Compliance with organization rules and procedures is very important in my organization’’ and ‘‘Employees in my organization are not expected to stick to organization policies strictly.’’ Each type was measured using a seven-point rating scales, with 1 representing “extremely disagree” to 7 representing
“extremely agree”. Therefore, a low score shows the absence of a climate and a high score shows the presence of a climate.
3.7 Data Analysis Procedures
In order to achieve the purposes of this research and test, SPSS and Partial Least Square (PLS) software were employed to help us analyze the collected data.
3.7.1 Descriptive Data Analysis
To better understand the characteristics of each variable, descriptive statistic analysis was used to illustrate the means, and standard deviation of each research variable.
3.7.2 Pearson Correlation Analysis and Partial Least Square Analysis
Pearson correlation coefficient and correlation coefficient were used to assess the relationship between all the dimensions. If the p-value is between 0.05 and 0.01, the hypothesis is strongly significant. And if the p-value is less than or equal to 0.01, the hypothesis is extremely significant. To analyze the relationship of the linear components between criterions and predictors, we used partial least square (hereinafter refers to as PLS). Because the PLS approach for its advantages over the covariance approach. The advantages of this soft-modeling approach include theoretical conditions, measurement conditions, distributional considerations, and practical considerations (Falk & Miller, 1992). PLS is an exploratory methodology that relies on the data. The PLS approach matches the researcher’s prediction-oriented objective, does not require normal data distribution, and accommodates small sample sizes (Chin & Newsted, 1999). However, PLS approach to Structural Equation Modeling (SEM) is a useful and flexible tool for statistical model building. The flexibility and scope of PLS facilitates the analysis and investigation of large and complex path models, particularly in the more exploratory fashion, as in this research.
CHAPTER FOUR
RESULTS OF THE RESEARCH
This chapter presents the empirical results of this study. The first section is the descriptive analysis of the respondents including the data collection, the attributes of the respondents, the results of the measurement variables and the reliability tests. The second section presents pearson’s correlation table is prepared to identify the interrelationships between independent variables and dependent variables. The third section presents a partial least square analysis is conducted to examine the significance and the relationships between the dimensions of PL (paternalistic leadership) and the dimension of organizational ethical climate.
4.1 Descriptive Analysis
Preliminary analyses were conducted in this section to provide information about the characteristics of respondents and the results of relevant research variables.
4.1.1
Data Collections
The data were gathered via questionnaire form over three month period. Due to limited time and resources we used convenient sampling and delivered the questionnaire through the relatives and friends. Totally 254 questionnaires were collected, and 254 of those were answered. The study sample consisted of 254 individuals working in Taiwan Public Sector. Respondents work in different sectors such as police agency of ministry of the interior, civil servants in township office, city government, forest bureau, coast guard administration and the unit of judiciary. Two hundred and seventy-nine questionnaires were distributed and 254 were returned for a response rate of 91.04%. After data cleaning, 4 incomplete questionnaires were abolished, due to incomplete answers or because it was clear that the respondents had
not treated the questionnaire seriously. For instance, some of them chose the same answer (score) for every question. There were no missing data among the 250 questionnaires, hence there are 250 effective samples participated in the study and valid retrieving rate is 89.61%. It is listed in Table 4-1
Table 4-1
Summary of Questionnaires and Response Rates
Questionnaire Distributed Collected Valid Response rate%
Total 279 254 250 91.04 %
4.1.2 Characteristics of the Respondents
Table 4-2 show the basic attributes of the respondents, including 3 major items in this study: (1) gender, (2) age, and (3) tenure. Data were collected by convenience sampling. From the graph below, we notice that 72.8% of respondents were male.
There is an imbalance of male gender representation in our sample. However, due to the conservative cultural conception of man are breadwinner and women are homemakers in Chinese society; there was nothing that could be done to correct this imbalance. According to Ken (2007) assumed that there may be lingering societal prejudice to assume that policing is “man’s work”, and that when help is needed “we want a policeman”. The age of collected sample is also quite young (37.6%) ranging from 27-36 and 34% ranging from 37-46 years old and the less of older in age (10.4%). The majority (20%) of the sample has a tenure that falls under the category of 10-15 years. Most of the tenure of respondents has a long working history and the less of the respondents have the short working history (7.6%).
Table 4-2
Characteristics of the Respondents for Taiwan Public Sector
4.1.3 Descriptive Statistics and Cronbach’s Alpha
Table 4-3 provides descriptive statistics by questionnaire items for the respondents of Taiwan public sector. These include 33 items of paternalistic leadership, 15 items of organizational ethical climate.
The results of means and standard deviations as shown in Table 4-3 indicated that, for the construct of paternalistic leadership, respondents identified the presence of authoritative leadership (mean= 3.84). This was followed by the moral leadership (mean= 3.73) and benevolent leadership (mean= 3.44). According to the high means of authoritative leadership (3.84), this is tally with the research of Chen & Farh (1996):
Authoritative leadership is one of the three main components of PL. While benevolence and moral leadership are integral, authoritarianism stands out, at least in
Question Frequency Percentage (%)
Gender
Male 182 72.8%
Female 68 27.2%
Age
18-26 26 10.4%
27-36 94 37.6%
37-46 85 34 %
47-56 38 15.2%
More than56 7 2.8%
Tenure
Less than 1 19 ..7.6%
1-5 (below) 43 17.2%
5-10 (below) 34 13.6%
10-15 (below) 50 20%
15-20 (below) 40 16%
20~25 (below) 45 18%
More than25 19 7.6%
the assessment by Western researchers, as the most prominent and most representative of the Chinese leadership tradition.
Table 4-3
Descriptive Analysis and Reliability for Paternalistic Leadership Types
Finally, for the construct of organizational ethical climate, Table 4-4 shows that respondents were identified the presence of organizational ethical climate, egoistic climate (mean=4.95). This was followed by the principle climate (mean=4.89) and benevolent climate (mean=4.11). If the Cronbach’s alpha is 0.7, the factor is reliable (Nunnally, 1978). These findings suggest that the majority of the respondents believed that their organizations major concern is about what is best for itself and subordinates are expected to comply with the law and professional standards, and to stick by organization rules and procedures.
Table 4-4
Descriptive Analysis and Reliability for Organizational Ethical Climate Types Taiwan Public Sector
(0.000)
Table 4-5 displays the pearson correlation coefficients for the model constructs of Taiwan public sector. According to Sharma and Patterson (1999), comparing the alpha coefficients of each construct and its correlation coefficients with other constructs aids in assessing discriminant validity. If the alpha coefficient is greater than the across correlations coefficient, evidence of discriminant validity has been demonstrated.
Table 4-5
Pearson Correlation for All Constructs
This study finds that dimensions of PL (paternalistic leadership) and dimensions of organizational ethical climate are significantly correlated. From the Table 4-5, except for H5, all of the Pearson coefficients are significant so hypotheses are supported, and the coefficient of authoritative leadership and egoistic climate, benevolent leadership and benevolent climate, moral leadership, principled climate
Authoritative
* Correlation is significant at 0.05 level (2-tailed).
**Correlation is significant at 0.01 level (2-tailed).
─
and authoritative leadership and benevolent climate are 0.602 (p<0.05), 0.694(p<0.05), 0.267(p<0.05), and -0.445(p<0.05) respectively. Therefore, H1, H2, H3, and H4 are established.
Validity is the degree to which a measure accurately represents what it is supposed to. Different validity terms are used to illustrate various aspects of construct validity (Sureshchandar et al., 2002). Ensuring validity starts with a thorough understanding of what is to be measured and then making the measurement as
“correct” and accurate as possible. There are two most widely accepted forms of validity and these two validity analyses would be used in this study.
1. Convergent validity. It is the degree to which two measures of the same concept are correlated. Each item in a scale can also be treated as a different approach to measure the construct (Ahire et al., 1996)
2. Discriminant validity. It is the assessment of the correspondence of the variables to be included in a summated scale and its conceptual definition. It is ensured by showing that a measure does not correlated very highly with other measures from which it is supposed to differ.
4.3 Partial Least Squares (PLS) for the Interrelationships among Paternalistic Leadership and Organizational Ethical Climate
A partial least squares (PLS) analysis was conducted to examine the relationship between multidimensional constructs of paternalistic leadership and organizational ethical climate. PLS was selected to analyze the overall model because:
˙ it is able to model multiple dependent as well as multiple independence variables;
˙ it is appropriate when multicollinearity is present;
˙ it can robust despite data noise and missing data;
˙ Creates independent latents directly on the basis of cross products involving response variable(s) = stronger predictions
˙ Allows for reflective and formative latents;
˙ Applied to small sample;
˙ it can distributional free; and
˙ it can handle range of variables: nominal, ordinal, continuous (Pirouz, 2006).
Because this study have multiple dependent as well as multiple independent, hence, we used WarpPLS 1.0 (Kock, 2009) to conduct a PLS (partial least squares) regression for examining our research hypotheses. WarpPLS is a powerful new structural equation modeling (SEM) software. Since most relationships between numeric variables are nonlinear, one could argue that WarpPLS finds the “real”
relationships between latent variables in an SEM analysis. Typically path coefficients are increased, in some cases going from non-significant to significant at the p lower than 1 percent level. The underlying algorithm employed by WarpPLS is PLS regression, whose main characteristic is its ability to minimize multicollinearity among latent variables (even in the presence of overlapping manifest variables, or indicators) (Kock, 2009).
It cannot distributed by types of variable and limited by sample size and it has a better ability of predictor and explanatory. Specifically, WarpPLS simultaneously assesses the psychometric properties of the measurement model (i.e. the reliability and validity of the scales used to measure each variable), as well as estimates the parameters of the structural model (i.e. the strength of the path relationships among the model variables). The measurement model and the structural model results are discussed below.
Table 4-6
Note: AVA=authoritative leadership; AVB= benevolent leadership; AVM=moral leadership; AVEC=egoistic climate; AVBC=benevolent climate; AVPC=principled climate
Reliability results from testing the measurement model are reported in Table 4-6.
The data indicates that the measures are robust in terms of their internal consistency reliabilities as indexed by the composite reliability. The composite reliabilities of the different measures in the model range from 0.81 to 0.94, which exceed the recommended threshold value of 0.70 (Nunnally, 1978). In addition, consistent with guidelines promulgated by Fornell & Larcker (1981), the average variance extracted (AVE) for each measure well exceeds 0.50 (the minimum AVE is 0.52).
Table 4-7
Discriminant Validity (Intercorrelations) of Variable Constructs
Latent Variables 1 2 3 4 5 6
1. Authoritative Leadership 0.72
2. Benevolent Leadership -0.44 0.76
3. Moral Leadership -0.27 0.59 0.81
4. Egoistic climate 0.54 -0.25 0.01 0.77
5. Benevolent Climate -0.40 0.63 0.40 -0.14 0.79
6. Principled climate 0.17 0.10 0.29 0.33 0.11 0.85
Note: Diagonal elements (bold) are the square root of average variance extracted (AVE) between the constructs and their indicator(s). Off-diagonal elements are correlations between constructs. For convergent and discriminant validity, diagonal elements should be at least .707 (i.e., AVE>.50) and larger than off-diagonal elements in the same row and column.
Table 4-7 provides evidence of the discriminant validity of the measures used in the study. As strong evidence of convergent and discriminant validity, the bolded elements in the matrix diagonals, representing the square roots of the AVEs, are greater in all cases than the off-diagonal elements in their corresponding row and column, supporting the discriminant validity of our scales. In short, the square root of the AVE for each construct was greater than 0.707 (i.e., AVE> .50) and greater than the correlation between that construct and other constructs, without exception (Chin 1998; Compeau et al., 1999; Fornell & Larcker, 1981).
Table 4-8:
Factoring Loadings (Bolded) and Cross Loadings
Note: AVA: authoritative leadership; AVB: benevolent leadership; AVM: moral leadership; AVEC: egoistic climate; AVBE: benevolent climate; AVPC: principled climate.
We tested convergent validity using WarpPLS by extracting the factor loadings (and cross loadings) of all indicator items to their respective latent constructs. These results, presented in Table 4-8, indicate that all items loaded: (1) on their respective construct (i.e. the bolded factor loadings) from a lower bound of 0.53 to an upper bound of 0.93, with the exception of AVB8 (the loading was below threshold of 0.5, so we deleted it), all of these reliabilities far exceed the recommended threshold of 0.5 suggested by Hair et al., (1987); and (2) more highly on their respective construct than on any other construct (i.e. the non-bolded factor loadings in any one row).
Furthermore, each item’s factor loading on its respective construct was highly significant (p < 0.001). The loadings presented in Table 4-8 confirm the convergent validity of the measures for these latent constructs. Overall, the measured scales show excellent psychometric properties with high reliability and appropriate convergent and discriminant validity.
The Structural Model
Note: AVA: authoritative leadership; AVB: benevolent leadership;
AVM: moral leadership; AVEC: egoistic climate; AVBE: benevolent climate;
AVPC: principled climate
* Correlation is significant at 0.05 level.
** Correlation is significant at 0.01 level
Figure 4-6 Result from the PLS Analysis
Figure 4-6 visually presents the results of the structural model. The beta values of the path coefficients, indicating the direct influences of the exogenous constructs on the endogenous constructs, are presented. Each path’s significance was estimated by a bootstrapping procedure (Ravichandran & Rai, 2000) using 500 resamples, which tends to provide reasonable standard error estimates (Chin, 2001).
Authoritative leadership (AVA) exhibited a strong positive direct influence (beta =
0.54, p < 0.01) on egoistic climate (AVEC). Benevolent leadership (AVB) exhibited a positive direct influence on benevolent climate (AVBC) (beta = 0.56, p < 0.01). Moral leadership exhibited a strong positive direct influence (beta= 0.35, p<0.01) on
principled climate. AVA exhibited a strong negative direct influence (beta=-0.15,
p<0.01) on AVBC. The effect of AVB on AVPC (beta=-0.11, p=0.43) was not significant and is therefore not represented in the structural model results.
The direct influences of AVA, AVB, and AVM, accounted for 29%, 41%, and 9%
of the variance in AVEC, AVBC, and AVPC, respectively. Although not all of the goodness of fit measures met the guidelines, overall fit for each measurement model was considered acceptable.
TABLE 4-9
Conclusions with Respect to the Hypotheses
HYPOTHESIS CONCLUSION
H1: There is a positive relation between authoritative leadership on organizational egoistic climate.
Supported (beta = 0.54, p < 0.01) AVA found to affect AVEC significantly H2: There is a positive relation between
benevolent leadership on organizational benevolent climate.
Supported (beta = 0.56, p < 0.01)
AVB found to affect AVBC significantly H3: There is a positive relation between
moral leadership on organizational principled climate.
Supported (beta= 0.35, p<0.01)
AVM found to affect AVPC significantly H4: There is a negative relation between
benevolent leadership on organizational principled climate.
Not Supported
AVB did not affect on AVPC
AVB did not affect on AVPC