This chapter outlines the research framework, research design, sampling, data collection, measurement, validity and reliability. Furthermore, readjustment of hypotheses was proposed and discussed.
Research Framework
The researcher investigated to what extent the power of personal social conformity influences the effect of perceived leaders’ intended humor toward an individual’s perception of pleasant climate. Figure 3.1 illustrates four leaders’ humor styles which are the independent variables; pleasant climate which is formed from employee’s perception of workplace humor, is the dependent variable; social conformity is the moderator, which is proposed to have a positive moderating effect on the relationship between perceived leaders’ humor behavior and pleasant climate.
Figure 3.1. Research framework
The hypotheses of the study are as follows:
Hypothesis 1. Employees’ perception of leaders’ humor behavior is related to the pleasant climate.
Hypothesis 1a. Employees’ perception of leaders’ affiliative humor is positively related to the pleasant climate.
Hypothesis 1b. Employees’ perception of leaders’ aggressive humor is negatively related to the pleasant climate.
Hypothesis 1c. Employees’ perception of leaders’ self-enhancing humor is positively related to the pleasant climate.
Hypothesis 1d. Employees’ perception of leaders’ self-defeating humor is positively related to the pleasant climate.
Hypothesis 2. Employee’s social conformity has a positive moderating effect on the relationship between perceived leader’s humor and pleasant climate.
Research Design
A quantitative approach was adopted for this study to achieve its objective to analyze the relationship between perceived four styles of leaders’ humor behaviors and the pleasant climate, as moderated by the employees’ personal social conformity. In the data collection, the leader’s leading humor style was obtained and tested from employee’s evaluation, and an employee self-report measurement of a pleasant climate at work was used to reveal the atmosphere at work in employees’ mind. The moderating effect of personal social conformity was investigated on the relationship of leaders’ leading humor behavior and pleasant climate. Questionnaire items were translated from English to Chinese. Back-translation and accuracy check of Back-translations were done by two Chinese native speakers who are fluent in English. Considering the convenience for the respondents, both online and paper surveys were adopted simultaneously. Statistical software SPSS v.22 and SPSS
In order to complete the study, eight steps of research procedure were conducted following the Figure 3.2. The first step was to narrow down and identify the research topic based on literature review of interested directions. The second step was to build background knowledge and review the literature to identify variables of interest related to the topic.
Based on the literature review the research purpose and questions were proposed. Then the research framework and hypothesis were developed to show the structure of the relationship between the variables. A questionnaire was designed with proper measurement to collect data for the study. A pilot test was conducted to ensure the validity and reliability of the measurement. Then data was collected and analyzed to test the hypotheses and interpreted to obtain the findings. Finally, the results, conclusions, and future suggestions were summarized and proposed.
Figure 3.2. Research procedure
Sampling
For the study purpose to explore the employees’ subjective interpersonal outcomes, the sample of this study focused on the employees working in various Taiwanese organizations and having more than six weeks collaborating with his/her manager, based on scholars’ suggestion that the minimum time to cultivate LMX can be expected in six weeks (Liden, Wayne, & Stilwell, 1993; Pundt & Herrmann, 2015). Therefore, the prerequisite for ensuring the inclusion is a tenure with the leader of longer than six weeks.
Besides, the cover letter had explained that to qualify as the sample the participant would need to have two or more colleagues to interact at work and one picked leader whom most familiar with and have most frequent contact at the workplace.
Sample Profile
The researcher collected a total of 596 completed questionnaires. To ensure the respondents were qualified to represent the phenomenon in their workplace, those answering to no facing Colleagues or Leader in Contact Type were excluded in the sample.
Therefore, a total of 521 responses remained: 73 (65 online, 8 hardcopy) questionnaires for the pilot test; 448 (407 online, 41 hardcopy) for the main study. Descriptive statistics was applied to present the key characteristics.
Pilot test.
A total of 73 responses were collected by the snowball contact with personal network.
These sample data helped to provide a primary understanding of the relationships among study variables and ensured the face validity and reliability of the questionnaire.
Those respondents had worked in service (61.6%) and non-service (38.4%) jobs with a tenure with a current leader from 2.5 months up to more than 35 years. Most of them aged 21~25 years old (35.6%) and had an age disparity ranged in 6~10 years (30.1%). Weekly immediate work contact of 1~5 hours with the leader (39.7%) was the majority.
Main study.
From the 448 respondents, there were 157 (35%) males and 291 (65%) females; their leaders included 284 males (63.4%) and 164 females (36.6%). The age range of 21~26 was the major group (34.4%); age disparity with leader was in majority of 1~5 years (23.9%).
Above half of the participants hold a bachelor degree (62.5%).
Depicting the context of the workplace, 291 (65%) respondents were in service work and 157 (35%) were in non-service work. The order of frequency contact with client and suppliers ranged from no contact (0) to last contact (4); with colleagues and supervisor from first frequency contact (1) to last contact (4).
Capturing the familiarity with leader, tenure with leader had lasted from 1.5 months up to 33 years long. The majority of the respondents had immediate 1~5 hours a week (33%) of work contact with the leader. Below shown the summary of the sample features.
Table 3.1.
Descriptive Statistic on Sample Characteristics of Main Study (N=448)
Variable Item Frequency Percentage (%)
Gender
Table 3.1. (continued)
Variable Item Frequency Percentage (%)
Age Disparity
Table 3.1. (continued)
Variable Item Frequency Percentage (%)
Workplace
The data was collected in June and July of 2017, three weeks after the pilot test. Initial contact with the potential candidates was made via the researcher’s personal contacts to working people. Snowball sampling method was applied as a tool for acquiring more data by including in the appreciation letter an invitation to send the survey link to other employees that may be interested in participating.
In the invitation, the study purposes were briefly explained. Participant were told that the survey is related to employee perception of interpersonal interaction at work, the importance of the participant’s contribution, the value that the results are going to provide,
and around 20 minutes was needed to complete the questionnaire. The prerequisite question was designed on the cover page of questionnaire. It helped the research filter the qualified candidates. Moreover, “focusing on one picked leader” was stressed throughout all the questionnaire texts. While employee submitted the questionnaire, a small gift was provided to hard-copy respondents; a drawing of the gift was conducted for on-line respondents who provide the personal contact to participate.
Measurement
The complete questionnaire can be found in Appendix A. Totally, it contains ten control variables and three main variables with 48 items: Perceived leader’s humor style with 29 items; pleasant climate with 8 items; social conformity with 11 items.
Leaders’ Humor Styles
Humor Style Questionnaire (HSQ; Martin et al., 2003), a self-report instrument with four dimensions was used to measure leaders’ humor styles from the subordinates’ point of view. It is designed to assess an individual’s particular use of humor style daily and naturally. In the present study, the measurement adopted the receiver version of HSQ (Scott, 2015). As this research targets the employees’ perspectives, it is rigorous to apply the scale on receivers, not on initiators (Sosik, 2012). The research states that the leader inclines to exhibit one of the four humor behavior styles mostly at the workplace and the receiver HSQ is a suitable measurement.
The receiver HSQ contains 29 items. There are 16 items in interpersonal humor (8 in Affiliative and Aggressive styles respectively), and 13 items in intrapersonal humor (6 in Self-enhancing and 7 in Self-defeating style). A 7 point Likert scale will be used from totally disagree (1) to totally agree (7). Reliability of the four scales was .731, .762, .693, and .777. The same factor structure of Traditional Chinese version of the Humor Styles Questionnaire (HSQ-TC; Chan et al., 2011), which was revised and served as a means to
investigate humor styles in Taiwan’s context, has an acceptable internal consistency (α= .73 - .88) and a good test-retest reliability.
Four subscales and revised example items are as below:
1. Affiliative humor
“My leader appears to enjoy making people laugh”, and “My leader rarely makes other people laugh by telling funny stories about himself/herself (reverse-scored)”.
2. Aggressive humor
“If someone makes a mistake, my leader will often tease them about it”, and “My leader never participates in laughing at others even if others are doing it (reverse-scored)”.
3. Self-enhancing humor
“My leader is often amused by the absurdities of life”, and “When my leader is feeling sad or upset, he/she usually lose his/her sense of humor (reverse-scored)”.
4. Self-defeating humor
“My leader will often get carried away in putting he/her down if it makes everyone laugh”, and “My leader lets everyone laugh at him/her as a way of keeping us in good spirits (reverse-scored)”.
Pleasant Climate
The measurement of Pleasant Climate Scale was initially selected from Humor at Work (HAW; Rawlings & Findlay, 2016) scale, a dynamic indicator of humor relationships among workers. Because of the objectives of the present study, the Unpleasant Climate Scale in HAW which investigates personal job satisfaction when exposed to the others’
behavior is not a suitable measurement tool.
Pleasant Climate Scale assesses the humorous behavior in the natural workplace environment or social groupings, disclosing humor in the context of interpersonal interaction such as the frequency of humor used by an individual preference and the perception of the use of others’ humor in the workplace. In order to fully depicting the
humor climate, the present research revised the scale into capturing the employee’s perception of the humorous atmosphere at the workplace.
8-items with two reverse scoring items are included with a 5 point Likert scale from strongly disagree (1) to strongly agree (5). It has a moderate reliability of .78. Revised sample items are “People in my unit like to share funny things that happen to them with the colleagues they work with”, and “People in my unit just like to do their jobs without humorous distraction (reverse-scored)”.
Social Conformity
The measurement adapts the Conformity Scale (CS; Mehrabian, 2005), an assessment that measures the tendency to be the follower. It exposes the degree to which a person has a characteristic or willingness to identify, comply with, and emulate others such as to agree with other’s ideas, values, and behaviors in the purpose to avoid negative interaction (Mehrabian, 2005).
The CS scale consists 11 items with four reverse items. And used a 9 point Likert scale from “very strong disagreement (1) to very strong agreement (9). It has a satisfactory reliability coefficient of .77 and a test-retest intercorrelation of .73 (Mehrabian & Stefl, 1995). Samples items are “I tend to rely on others when I have to make an important decision quickly’’, and “I don’t give in to others easily (reverse-scored)”.
Control Variables
A total of ten specific variables were included as controls in the analyses. Two major reasons underlie the inclusion of demographic controls. Firstly, because humor is a social interpersonal interactive group product (Owen, 2009), it takes time to cultivate especially in the workplace which comprises managers level and employees level (Robert et al., 2016).
Secondly, social conformity exists when a person follows the seemingly right people or only being attracted to follow by nature (Asch, 1951). All these variables were measured
Personal background. Five variables are related to personal information of the respondents and the leader themselves.
Gender (self) and Gender (leader). Gender may have an effect on the preference of different usages of humor behavior. For example, male may be more likely to follow the informational source, while females follow the group pressure (Eagly & Chrvala, 1986). A nominal measure with two options (male/female) was applied to both leaders and respondents.
Age (self) and Age discrepancy. Age indicates the personal tendency to feel closer to the people with narrower age gap (Scott, 2015). The ordinal questions were adopted. Ten options were provided to ask the respondent’s age from below 20 years old, 21-25 years old, 26-30 years old, 31-35 years old, 36-40 years old, 41-45 years old, 46-50 years old, 51-55 years old, 56-60 years old, to above 61 years old. Nine options were provided for
“age disparity between leader and member” from same year, 1-5 years, 6-10 years, 11-15 years, 16-20 years, 21-25 years, 26-30 years, 31-35 years, to above 36 years.
Education. It has a potential influence on social conformity, which is a tendency to follow a group of more highly educated whom may create a demand for others to conform to their expertise (Ganser & Zwiefelhofer, 2006), or of a higher level of cognition to interpret the humor when interacting with other in the humorous event (Scott, 2015). An ordinal question with four options from high school and below, bachelor degree, master degree, to PHD degree.
Workplace context. Two variables were designed to understand the context of the workplace. Due to they may potentially guide the humor usages in interpersonal interaction.
Work nature. It attempted to capture the nature of the respondent’s work, and was measured using two options: service or non-service.
Leader most and Colleague most. Contact type attempted to capture the type of people in the current work the respondent mostly faces, and was measured with a ranking
order of four groups, namely customers, suppliers, peers, and supervisor. Five Arabic number were provided for the ranking: 1 (most frequently facing group), 2 (second most frequently facing group), 3 (third most frequently facing group), 4 (least frequently facing group), and 0 (no facing that certain group).
Work relationship with the manager/leader. Two variables are related to the intensity of relationship between the respondent and the leader.
Tenure with leader. It shows how familiar the respondent is with the manager and potentially the leader’s humor styles (Cooper, 2002; Pundt & Herrmann, 2015; Robert et al., 2016). This was measured using an open-ended question which asked the time (in years and months) the respondent has worked for the current manager.
Weekly contact. Frequency of immediate contact with the manager at work shows the potential of leader’s humor affecting the employee’s humor style (Pundt & Herrmann, 2015;
Scott, 2015), and was measured using an ordinal question with eight choices from 1-5 hours/week, 6-10 hours/week, 11-15 hours/week, 16-20 hours/week, 21-25 hours/week, 26-30 hours/week, 31-35 hours/week, to more than 36 hours/week.
Validity and Reliability
This section contains three parts: The pilot test section explored the reliability of instrument design. The main study section presented the result of construct validity, reliability, and common method variance. The summary of validity and reliability section proposed the changes in research framework and hypothesis. Besides, content validity was ensured by using mature measurement scales, the accuracy in translation, and expert review.
Pilot Test
The variables’ Cronbach’s Alpha was observed to assess the reliability of the measurement usage. Table 3.2 shows a lower Cronbach’s alpha of Social Conformity (.64) below the .7 acceptable value (Nunnally & Bernstein, 1994). It is considered reasonable
when analyzing a small sample. Additionally, Cronbach’s alpha coefficient of .6 - .7 is still considered acceptable (George & Mallery, 2003).
Table 3.2.
Reliability Analysis of Pilot Test (N=73)
Scales Number of Items Cronbach’s Alpha
Leader Humor Style 29 .83
Affiliative 8 .89
Aggressive 8 .72
Self-enhancing 6 .71
Self-defeating 7 .77
Pleasant Climate 8 .88
Social Conformity 11 .64
Main Study
Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA) were applied collectively to ensure the construct validity. First, items of each variable were run through CFA using AMOS to see whether the data fit the theoretical measurement model.
When the CFA resulted in less-than-satisfactory fit, the measurement model was modified based on a procedure described below. The modified measurement model was subsequently cross validated to ensure the construct validity. To process this procedure, the sample was split in half randomly. Data from the first half of the sample was used to conduct an EFA.
Measurement of sampling adequacy was checked by Kaiser-Meyer-Olkin (KMO) above .5 and Bartlett’s test of sphericity significant (Kaiser, 1974). Any item with a factor loading less than .4 (Costello and Osborne, 2005) and cross loaded items were removed to purify of the scale. Then, data from the second half of the sample was used in a CFA to test whether the modified measurement model achieves an acceptable goodness of fit. The criteria of fit indices are listed in Table 3.3. In short, except Pleasant Climate which has a satisfactory
model fit in CFA, measurement of Leader Humor Style and Social Conformity were modified and went through abovementioned process for cross validation.
Finally, the variables with their revised items were tested on internal consistency of reliability via Cronbach’s alpha coefficient. Common method variance (CMV) was tested to understand the threat of type I or type II errors in the variances.
Table 3.3.
Summary of Model Fits Indices
Index Good fit Acceptable fit
χ2/DF 2-5 <5
GFI >.95 >.90
AGFI >.90 >.85
RMSEA <.08 .08-.1
SRMR <.05 ≦.08
AVE >.7 >.5
CR - ≧.7
Note. Summary according to Hooper, Coughlan, and Mullen (2008) and Fornell and Larcker (1981) (Last two rows).
Leaders’ humor style. The 29-item leader humor style scale underwent initial CFA based on the theoretical factor structure. The result yielded a poor model fit. Therefore, the research proceeded to draw a modified model by splitting the sample into two, dataset A and B.
First, dataset A was put to run EFA. The EFA result showed that items of Affiliative, Self-enhancing, Self-defeating mainly loaded onto one component; items of Aggressive loaded onto another component. It indicated the respondents of the main study could only
distinguish two humor styles, the aggressive leader humor styles and others. Other three leader styles (i.e. Affiliative, Self-enhancing, and Self-defeating) were perceived to function similarly from the respondents’ point of view. Thus, the researcher combined Affiliative, Self-enhancing, and Self-defeating and renamed them Jovial leader humor style (JO). Moreover, fourteen items with factor loading less than .4 or items with cross-loading were deleted: AF1, AF3, AF5, AF7, AF8, SE5, SD2, SD4, SD5, SD6, SD7, AG3, AG4, AG5.
Table 3.4.
Exploratory Factor Analysis of Leader Humor Style (Dataset A) Factor
Table 3.4. (continued)
Note. Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. KMO measure of sampling adequacy: .928. Bartlett’s test of sphericity: .000. JO= Jovial Leader Humor Style; AG= Aggressive Leader Humor Style.
Then, dataset B was used in CFA to test the fit of the modified model. The revised model reports better fit. So, the LHS modified model replaced the original model and was utilized. Table 3.5 shows the model fit summary. Figure 3.3 displays the CFA result of the modified measurement model.
Table 3.5.
Leader Humor Style Model Fit Summary
χ2 df P χ2/df GFI AGFI RMSEA SRMR AVE CR Original
29 items 1536.08 371 .000 4.14 .78 .74 .08 .91 .23 .72 Modified
15 items 185.71 89 .000 2.09 .90 .87 .07 .07 .51 .94
Figure 3.3. Leader humor style modified CFA measurement model (Dataset B).
Pleasant climate. The 8-item PC scale was tested in the initial CFA based on the theoretical factor structure. The result yielded a satisfactory model fit: χ2/df (4.49) and RMSEA (.09) show acceptable fit; GFI (.95), AGFI (.91), and SRMR (.05) are a good fit;
CR (.84) meets the acceptable criteria. However, AVE (.40) is lower than the standard .5. It is one of the limitations of the current study. Figure 3.4 and Table 3.6 presents the measurement model and model fit summary.
Table 3.6.
Pleasant Climate Model Fit Summary (N=448)
χ2 df P χ2/df GFI AGFI RMSEA SRMR AVE CR 8 items 89.836 20 .000 4.49 .95 .90 .09 .05 .40 .84
Figure 3.4. Pleasant climate CFA measurement model.
Social conformity. The 11-item SC was tested in the initial CFA based on the theoretical factor structure. The result showed a poor model fit. Therefore, modification and cross validation were performed through splitting sample in two randomly, and using EFA result from dataset A to modify measurement and CFA with dataset B to cross validate the modified model.
First, three components emerged in the EFA result of dataset A. Item SC2 (-.78) was dropped because of significant negative factor loading. Then the EFA test was processed again. Two components emerged in the EFA result: Component 1 contains items expressing
First, three components emerged in the EFA result of dataset A. Item SC2 (-.78) was dropped because of significant negative factor loading. Then the EFA test was processed again. Two components emerged in the EFA result: Component 1 contains items expressing