MSc Strategic Innovation Management
Master thesis March 2015
Work hard, laugh harder:
Does being in a humorous mood affect the individual’s
innovative and creative performance?
An experimental research study
By
Merel L. Koppen
Supervisor: Dr. W.G. Biemans Co-assessor: Dr. K.R.E. Huizingh
Faculty of Economics and Business Student ID: S1866494
WORK HARD, LAUGH HARDER
i
ABSTRACT
The effects of humor exposure on creative performance quantity and quality were assessed in
a scientific experiment. 151 participants generated ideas after being exposed to a neutral or
humorous film stimulus. As compared to the control condition, participants exposed to the
humor stimulus overall generated more ideas and the quality of their best idea was higher.
The humorous film stimulus brought participants in a significantly higher humor state than
the neutral film stimulus. The humor manipulation was found to have a significant positive
effect on the relationship between components of humor state, humor trait and creativity state.
ii
TABLE OF CONTENTS
1. Introduction………...……1
2.Theoretical Section………...…...2
2.1 What is Creativity and what is Creative Performance?...2
2.2 What is Humor?... ...4
2.3 What are the effects of humor?...7
2.4 Hypotheses and conceptual model………...…………...……..9
3. Methodology and Analyses………...10
3.1 Research Design………...10
3.2 Operationalization………....…11
3.3 Data collection………....….12
3.4 Procedure and Measurements………...…...13
3.5 Quality of research……….………...16
3.6 Data analysis overview……….………...…19
4. Results………..…...….20
4.1 Descriptive statistics………..……..20
4.2 Sample characteristics………..……21
4.3 Bivariate correlations………..…….21
4.4 ANCOVA analyses, regressions and hypotheses………..……..24
5. Conclusions and implications………..…29
5.1 Discussion………..……..29
5.2 Managerial Implications………..…...….30
5.3 Limitations………...…..31
5.4 Suggestions for future research………...…..32 References
Appendices
Appendix A: Instructions experiments
Appendix B: Experiment Questionnaire overview Appendix C: Recruitment text participants experiment Appendix D: Results Humor Pilot
Appendix E: Results Neutral Pilot
Appendix F: STCI Trait Questionnaire and scoring key Appendix G: STCI State Questionnaire and scoring key Appendix H: Humor State Component
Appendix I: Gough Creative Personality Scale (CPS) Appendix J: Individual creative performance Appendix K: Intraclass correlation
Appendix L: Factor-analyses State and Trait Scales Appendix M: Manipulation check
Appendix N: Demographics experiment participants Appendix O: Normality testing
Appendix P: One-way ANOVA mean comparison groups Appendix Q: ANCOVA analyses
Appendix R:OLS Regression Humor State
- 1 - 1. INTRODUCTION
The quote from Robert Iger states what every company knows: it is impossible to get away from innovation and creativity. Every business journal, news site or paper emphasizes the importance of innovation. Every corporation feels the need to integrate innovation in their corporate strategy plans. In this era of internationalization, e-commerce and the fact that many companies struggle to survive the pressure of growing competition, innovation seems to be more crucial than ever (Alhstrom, 2010). Since the innovation topic has caused major attention over the years, we find an extensive literature field on innovation in organizations.
In line with the quote of Iger, creativity is often designated in these studies as an important determiner and facilitator for innovation (Baron & Tang, 2011). Despite everything that has been written on this relationship, we still lack a clear picture of what factors influence the creativity in organizations. Is it enough to hire people that score high on creativity traits, or can the level of creativity and innovation of organizations be increased in different ways? That is to say: does a manager know how to influence the innovative and creative behavior of his employees?
From a psychological perspective, a large research field exists on the relationship between creativity and humor (Thorson & Powell, 1993). Linking this to the field of management and strategy, we see that
successful multinational companies as Ben & Jerry’s (Lager, 2011) and Kulula Airlines (Morreall, 2014) have made humor a core part of their organizational culture to create an innovative, competitive advantage. Still, in these cases, humor is often used as a customer approach or strategic asset instead of a catalyst for internal creative processes.
The humor concept demonstrates similarities with the innovation concept; both processes are characterized by a producer performing an action that creates an outcome that is new or unexpected (Romero and Cruthirds, 2006; Woodman, Sawyer, & Griffin, 1993). Though, when sifting the existing literature on the relationship between humor and innovation, it is rather remarkable that this research field appears to be largely understudied. We believe that both researchers and practitioners will benefit from as well enlarging this research field as having concrete tools to affect creative behavior.
In this research, we strive to fulfill this need by contributing to the existing literature on three major points. First, we will concretize the extensive expertise and literature from the psychological research field by identifying humor and its effects on creativity in a business perspective. This way, we will not only connect psychological research to business research, but we will also connect the academic field with the more applied business level knowledge. Second, since we strive to move away from the pure theoretical definitions to a more pragmatic business perspective, we will set up a scientific research experiment to test and manipulate humor and
- 2 - measure its effect on creative and innovative
performance. As far as we know, an experiment as such was never conducted before. Third, we will explore the extent to which humor , creativity traits and states are a reliable predictor for creative performance. We will make these contributions by answering the main research question:
Does being in a humorous mood affect the individual’s innovative and creative
performance?
This paper is structured as follows. In the theoretical section, we will elaborate on the existing research on humor, creativity and its effects and potential relationships. Subsequently, the hypotheses and conceptual model to test our research question are drawn. This theoretical section is followed by a operationalization of these concepts in the methodology section. In the results section, the hypotheses are tested. The conclusion will summarize our findings and provides the reader with suggestions for future research.
2.THEORETICAL SECTION
In the subsequent section, we will elaborate on the different forms of humor and creativity and their potential reciprocal relation. We use the identified research gap as the starting point for our research. In section 2.1, we will first put emphasis on creativity, creative performance and innovation. After that, it will present the hypotheses derived from theoretical literature. In section 2.2, we elaborate on the concept of humor, its definitions and the two types of
humor we will apply in this research; humor trait and humor state. In section 2.3, we zoom in on the effects of humor, both in general and related to creative performance. This leads to the relevant theoretical hypotheses. Lastly, in section 2.4, the hypotheses will be schematized by means of table 1 and visualized in the conceptual model (Figure 2).
2.1 What is Creativity and what is Creative Performance?
Plucker et al. (2004) argue that creativity is the interaction among aptitude, process and environment by which an individual or group produces a perceptible product that is both novel and useful as defined within a social context. Mumford (2013) evaluated existing research more concisely by concluding that over the course of the last decades, a general agreement has been reached on the fact that creativity involves the production of novel, useful products and ideas. In our research, creativity is divided into creativity state and creativity trait.
WORK HARD, LAUGH HARDER
- 3 - creative to the extent that appropriate observers
independently agree it is creative (Amabile,1996). Creative
performance positions creativity as the “generation of domain-specific, novel, and useful outcomes” (Amabile, 1988; Oldham & Cummings, 1996; Tierney et al, 2002). Subsequently, we need to concretize this creative performance to an applied level. Poetz and Schreier (2012) contend that this concretization can be executed by means of the identification of four creative performance levels; novelty (to what extent a certain product or process is new compared to an already existing one), customer benefit (to what extent the ability of the idea solves an existing problem), feasibility (to what extent the idea is commercializable), and number of ideas. In this research, these four concepts will be applied to define creative performance.
2.1.1 Creativity and Innovation
When we apply this creative performance definition to a business environment, we learn that in lieu of creativity or creative performance, the concept of innovation is often applied. The terms are repeatedly used interchangeably in research studies (West & Farr, 1990). Creativity is characterized by the production of useful and novel ideas (Mumford & Gustafson, 1988), and innovation is about the production or adoption of useful ideas and idea implementation (Kanter, 1983; Van de Ven, 1986). Creativity might be framed as “doing something for the first time anywhere or creating new knowledge” (Woodman, Sawyer, & Griffin, 1993) and therefore might
sound similar to definitions of innovation, but does not take into account the adaption and implementation of products or processes into account. Therefore, we argue that creative performance is a very crucial, but partial component of innovation (Kanter, 1983).
2.1.2 How is Creativity Trait related to Creative Performance?
- 4 -
H1: The positive relationship between a
participant’s Creativity Trait and participant’s Creative Performance will be stronger when exposed to a humorous stimulus than when exposed to a neutral stimulus.
2.1.3 How is Creativity State related to Creative Performance?
Following the literature review on creative trait above, we would like to extend this view by arguing that creative performance is not only being predicted by an individual’s creative trait, but as well by the more temporary individual creativity state (Shibata & Suzuki, 2004). This momentary “out of the box thinking”, beyond the conditioned boundaries of our mental assumptions and preconceptions, leads to a higher creative performance (Lizotte, 1998). To seek for the effect of humor on this relationship, the following hypothesis will be tested:
H2: The positive relationship between a
participant’s Creativity State and participant's Creative Performance will be stronger when exposed to a humorous stimulus than when exposed to a neutral stimulus.
2.2 What is humor?
Throughout history, different philosophers, researchers and psychologists, ranging from Aristotle and Freud to Bergson and Leacock (Walker, 1937), have tried to answer the questions ‘what is humor’ or ‘why do people laugh’. Later, researchers have repeatedly attempted to develop brief taxonomies of humor. We find that many of these definitions
of humor are tautological in that they use humor outcomes (e.g. laughter) as an explanatory mechanism of the humor construct (Roeckelein, 2002). Through the twentieth century, Eysenck’s (1942) typology of affective cognitive and conative theories is seen as one of the most succinct and most useful psychological models and served as basis for the research of other scholars. According to Martineau (1972), any communicative instance which is perceived as humorous is humor, while Crawford (1994) states that it should also consist of nonverbal and verbal communications which produce a positive cognitive or affective response from listeners. Robert and Yan (2007) define humor as “an intentional form of social communication delivered by a ‘producer’ toward an ‘audience”, while Romero and Cruthirds (2006) explain that it consists of amusing communications that produce positive emotions and cognitions in the individual, group, or organization.
WORK HARD, LAUGH HARDER
- 5 - exhilaration).
The reason for this struggle in isolating a definition for humor and to describe its meaning, is mainly the complexity of the humor construct, which causes many complications in construing a coherent understanding (Allport, 1961; Foot, 1991; Freud, 1928; Kuiper and Martin, 1998; Martin, 2001; Martin et al., 2003; Maslow, 1954; Robert and Yan, 2007; Vaillant, 1977; Warnars-Kleverlaan et al., 1996). In this research, we will clarify this broad understanding by dividing it into two more understandable, demarcated definitions; humor trait and humor state.
2.2.1 Humor Trait
When we review this broad research field on the humor concept, we see that the vast majority of those studies clearly focuses on ‘sense of humor’; a personality trait that enables a person to recognize and use successful humor as a coping mechanism and/or for affiliate/social communications /interactions (Lynch, 2002; Martin, 1996; Martin et al., 2003; Thorson and Powell, 1993). Literature tells us that, some people tend habitually to appreciate, initiate or laugh at humor more often or more intensely than others do. These enduring dispositions
(relatively stable over time) are considered to be the individual’s humor trait (Martin et al., 2003; Ruch, 1993). Ruch and Köhler (1998) concretize this typology by elaborating on the fact that humor differs from person to person; individuals differ in human behavior; thoughts, feelings and actions; not only in one situation, but habitually. In case these aspects are correlated, this might lead to the existence of a personality concept; sense of humor or humor trait, of which we will apply the latter term in this research.
- 6 - 2.2.2 Humor State
Opposed to the considerable mature literature field on humor trait, the identification of the temporary humor state is characterized by a largely understudied field with incoherent findings. Vance (1987) defines humor state as an emotional experience resulting from the perception of an incongruity or discrepancy based upon present expectations, and characterized by pleasure and increased cognitive arousal. Humor state may also include an emotional experience resulting from the resolution of the same discrepancy, and be characterized by pleasure and a decrease in arousal. Bennett and Lengacher (2008) attest that humor involves cognitive, emotional behavioral, physiological and social aspects and mention that humor can refer to a stimulus, which is intended to produce a humorous response (such as a humorous video), a mental
process (perception of amusing incongruities) or a response (laughter, exhilaration). Davis (1995) believes that those in the humorous mood find ambiguity, contradiction, and paradox a given rather than a problem, which relates to Mulkay’s (1988) point of view that the humorous mode is “consistently inconsistent or inconsistently consistent”.
Whether we use the appellation humor appreciation (Weisfeld, 1993), laughter, humor enjoyment (Ziv, 1976), incongruity humor (Vance, 1987), mirth (Gavanski, 1986), amusement (Martin, 2000) or humorous mood (Weisenberg et al., 1998), the terms all have the same common denominator; they all refer to a temporary, perishable state or emotion, instead of an attribute or anchored nature. One of the first researchers to address this specific characteristic in detail, is Willibald
WORK HARD, LAUGH HARDER
- 7 - Ruch. He founded the term ‘exhilaration’ to
describe the emotion of behavioral, physiological and experiential responses to stimuli such as humor, tickling and laughing gas (Ruch 1993). The term is derived from the Latin word ‘hilaris’ (=cheerful) to connote either the temporary rise and fall of a cheerful state or the process of making cheerful (Ruch, 1990, 1993). Exhilaration might be interpret as a facet of the positive emotion of “joy” or “happiness, but is also very strongly aligned with “laughter”. While striving for an overarching definition of this exhilaration or humor state, antagonistic factors should be contemplated as well. In other words; states that debilitate the induction of laughter and smiling. A prevalent bad mood and a serious frame of mind were hypothesized to be these right factors (Figure 1, Ruch, 1996). We will use these three terms by Ruch to define humor state and, interchangeably, humorous mood.
2.3 What are the effects of humor?
2.3.1 Existing research on the effects of humor From a psychological and medical field, humor is manifold being used as an -experimental- treatment to explore the effects on a mental or physical state or health level. (Chik et al., 2005; Dunbar et al., 2011; Konradt et al., 2013; Mahony et al., 2001; Weisenberg et al., 1995, 1998). The research field that focuses on the influence of humor (stimulus) on physical and mental attributes has caught large attention over the recent years; Mahony et al. (2001) found that humor raised the individual’s discomfort thresholds and lower humor traits caused more vulnerability and need for
assitance, while Weisenberg et al. (1995, 1998) state that humor not only causes the release of endorphins, the lowering of tension, and (positive) distraction, but also induced a significant increase in pain tolerance. Konradt et al. (2013) even showed that humor treatment can enable increased satisfaction with life for people with depression. The focal point of those experiments usually finds it footing in a medical or psychological quest or affair. By zooming in to the relationship between (the effects of) humor and creativity and innovation in an experimental design, we strive for making the transition to the business field.
2.3.2 The effects of humor linked to creativity Psychological literature has yielded a substantial extent of research on the effects of humor on creativity. A considerable amount of research has found support for a close relationship between humor and creative abilities (Babad, 1974; Brodzinsky & Rubien, 1976; Fabrizi & Pollio, 1987). The two concepts are closely related: sense of humor is partly made up of creativity and therefore an individual’s humor characteristics can predict for one’s creative performance (Thorson & Powell, 1993). In the hypotheses, we will elaborate on this link by testing the effect of humor exposure on the relationship between respectively humor trait, humor state and creative performance.
2.3.3 How is Humor Trait related to Creative Performance?
- 8 - openness to new ideas by making people less
likely to criticize mistakes or new ideas, which leads to risk taking, which is the basis of creative thinking and creative behavior (Morreall 1991). Furthermore, humor trait is positively related to favorable emotions and negatively related to neuroticism (Romero & Cruthirds, 2006). The individual’s humor trait enhances creativity by enabling openness an improve communication. Psychological studies show that a higher humor trait is related to higher creativity and creative outputs ( Ziv, 1976).
Since we are striving for exploring the relationship between humor trait and creative performance by means of humor manipulation, this leads to the following hypothesis:
H3: The positive relationship between a
participant’s Humor Trait and participant’s Creative Performance will be stronger when exposed to a humorous stimulus than when exposed to a neutral stimulus.
2.3.4 What are the effects of Humor State on Creative Performance?
Not only for the humor trait, but as well for humor state, we find evidence in literature for its relationship with and effect on creative performance. A “fun mood” leads to increased individual creativity in which pioneering ideas are likely to emerge (Ziv, 1983). As well, individuals in a humorous mode are more likely to engage in creative problem solving. Research reveals that being in a humorous state has a positive effect on creative problem solving (Isen et al. 1987).
In contrast to the humor trait, a humor state is characterized by single incidents of exhilaration, and fluctuates over time; we are all inclined to appreciate, initiate of laugh at humor more at given times and less at others. (Ruch 1993). Since it is very sensitive to change, the measurement of this state requires a scale that is appropriate for perceiving those potential differences. Ruch, Köhler and Van Triel (1996) were the first ones to develop a self-report humor concept that is able to measure this. The STCI theory for state argues that for individuals in a cheerful state, the elicitation of exhilaration/amusement will be facilitated, while individuals in a more serious frame of mind or in a bad mood will be less readily inclined to laugh or smile (Ruch, Köhler and Van Triel, 1996). In line with the existing theories that confirm that humor state will be a predictor of creative performance, we will test the following hypothesis:
H4: The positive relationship between a
WORK HARD, LAUGH HARDER
- 9 - 2.4 Hypotheses and conceptual model
Figure 2: Conceptual model
Table 1: Hypotheses derived from theoretical literature
- 10 - 3. METHODOLOGY
In the methodology section, we will operationalize our theoretical concepts and give a substantiated overview of the methodological choices. Section 3.1 gives insides in the choice for an experimental design for this study. Section 3.2 operationalizes the relevant variables. Section 3.3 elaborates on the data collection, while section 3.4 illuminates the conducted procedure and measurements. Section 3.5 provides an overview of the prerequisites for this experimental study. The chapter is concluded by the data analysis overview in 3.6.
3.1 Research design
Experimental design in business research? The objective of this study was to investigate the effects of a changing humor mood on creative performance. To do so, an appropriate data set is required. Within the scientific method, we can distinguish four common ways of collecting data: observation, surveys, computer simulation, and experiment (Srinagesh, 2006; Zikmund et al., 2012). The experimental approach has been a cornerstone of the scientific method for centuries (List, 2011) and has therefore been a persistent measure in especially psychological, sociological and medical research on (the relationship of and manipulation of) humor and creativity (Abelson & Levine,1958; Cattell & Luborsky, 1947; Dunbar et al., 2011; Eysenck, 1942; Konradt et al., 2013; Mahony et al., 2001; Martin et al., 2008; Ruch, 1992; Weisenberg et al., 1995,1998; Zweyer et al.,
2004). However, many economics and business researchers have been rather pessimistic that an experimental approach could be of that much added value and offer certain vivid illustrations of cause and effect in their field as it does in the psychological, sociological and medical research fields (List, 2011). It is argued that the business world often is too complex to control, while this control is one of the first and most important prerequisites of experimental research (Kirk, 1982). Though, a form of research that has offered a variety of insights in business settings, are laboratory experiments; these have been used as valuable tools in empirical economic analyses (List, 2011; Nobelprize.org, 2011).
Experimental design
WORK HARD, LAUGH HARDER
- 11 - stimulus, after identification of the treatment
unit and type of variables and identification of the phases, we decided the experimental design should be based on the two-group posttest experimental design (Campbell et al.; 1963, Crosier; 2000, Fraenkel & Wallen: 1993; White, 2000). The experimental design is used to establish cause and effect by manipulating (influencing) an independent variable to see its effect on a dependent variable. A true experiment requires randomization to eliminate threats to the experiment from extraneous data. The two-group posttest experimental design has randomization as its main advantage. The post-test comparison with randomized subjects controls for the main effects of history, maturation, and pre-testing; because no pre-test is used to measure the
same variable, there can be no interaction effect of pre-test and X. Moreover, the Solomon four-group design was used as an example to structure the different components of the experiment. This form of experiment has the advantage to assess the presence of pre-test sensitization in a sense that exposure before manipulation increases the subjects sensitivity to the experimental treatment, thus preventing generalization of results from the pretested sample to an unpretested population (Huck & Sandier, 1973). By using this approach, we as well strived for reducing the impact of the Hawthorne effect (Wickström & Bendix, 2000). We altered the design by using six groups instead of the original four groups. (Helmstadter, 1970).
3.2 Operationalization
3.3
- 12 - Data Collection
Research setting and participants
3.3.1 Experimental setting
Data was collected during six experimental sessions in a time period of two weeks. During three of the sessions, the participants were exposed to a humorous stimulus (treatment group) and during the other three sessions, the participants were exposed to a neutral stimulus (control group). The sessions in the first week took place at 10 AM and the sessions in the second week at 1 PM in a secluded computer lab. All sessions had a duration of approximately 40 minutes. The experiment was carried out with a fixed script for every session. When participants arrived, they were registered by one of the researchers, directed to
a seat and to wait for instructions. Every session started with a similar, pre-written introduction text, read out by the same researcher during all sessions (Appendix A). This way, each participant received exactly the same instructions. Data was entered by making use of computers and was recorded with Qualtrics survey software. The actual experiment consisted of , in chronological order, the following elements: (a) general questions, (b) a creative task, (c) a humorous trait test, (d) a creative personality test, (e) a
video stimulus (neutral or humorous), (f) a humorous state test, (g) an idea generation exercise, (h) a second creative task, and (i) a comments section (Appendix B). The instructions after each section stated whether participants were allowed to proceed to the next section, or had to wait until all other participants finished that particular section. This way, it was possible to expose all participants simultaneously to the stimulus, which was broadcasted on a big screen. The experiment ended with several closing remarks and the handing out of the lunch voucher that was provided to all participants. A schematic overview of the experiment is displayed in Figure 3.
Figure 3: Schematic overview of the experiment.
Vertical arrows indicate questionnaire sections.
3.3.2 Participants
WORK HARD, LAUGH HARDER
- 13 - Business, University of Groningen, but were
native Dutch speakers. This way, language would not become a barrier for fulfilling the creative tasks and idea generation task. The participants were recruited by means of emails, social media invitations and direct approach during the experiment weeks (Appendix C).
3.4 Procedure and Measurements
3.4.1 Pilot Study
Two distinct pilot studies were conducted, preceding the research experiment. During the first pilot study, 12 participants were exposed to 16 humorous film fragments with a total duration of 37 minutes. The participants were randomly selected with the condition that there was an equal distribution of the percentage of men and women (60%- 40%) among the 6 groups to take into account the gender differences regarding a possible preference for a type of humor. The film fragments were broadcasted on a big screen instead of on individual computer screens. In this manner, the participants watched the same fragments simultaneously in order to facilitate and create a collective humorous mood. Each fragment was separated by a break of 10 seconds during which the participants were asked to rate the preliminary fragment by means of answering the statement “I find this film fragment very funny”. Funniness was scored on a 5 point Likert scale (M=3,64). The fragments from the pilot test that were incorporated in the final humorous stimulus, were chosen based on the highest rating on funniness and also took into account the amount of negative ratings. Eventually, 7 fragments that scored the highest
on funniness and had the least negative ratings, were incorporated in the final stimulus, which led to a stimulus with a total duration of 10 minutes and 27 seconds (Appendix D).
The second pilot study identified the neutral stimulus and any ambiguities in the experiment’s instructions and scales. Based on existing literature, the neutral stimulus consisted of nature documentary fragments (Tice et al, 2007; Volkow et al, 2006; Vohs & Heatherton, 2000). To identify the final neutral stimulus, 10 participants were exposed to 4 nature documentaries with a total duration of 25 minutes. Between each fragment, there was a short break of 10 seconds during which the participants had to rate the previous fragment on a 5-point Likert scale based on three criteria: funniness, informative, and interesting. The fragments that were incorporated in the final neutral stimulus were selected on basis of a low funniness rating and a high rating on interesting. The criterion informative was merely used as a criterion to mislead the participants. Interesting was added as a criterion, because it is considered important that all participants focus on the stimulus while being exposed to create the desired mood. This eventually led to a final neutral stimulus consisting of 3 fragments with a total duration of 10 minutes and 3 seconds.
- 14 - 3.4.2 Measures
Humor Trait was measured by the short Trait form of the State-Trait Cheerfulness Inventory (Ruch, Kohler & van Thriel, 1996) and consisted of 30 statements on one’s character and mentality in general. The aim of the State-Trait-Cheerfulness-Inventory (STCI) is to provide a reliable, valid, and economical assessment of the three constructs of cheerfulness, seriousness, and bad mood both as states (STCI-S) and traits (STCI-T). Other scales for measuring humor trait are the Sense of humor Questionnaire (SHQ; Svebak,1974), The Multidimensional Sense of Humor Scale (MSHS; Thorson & Powell, 1993) and the Wit and Humor Appreciation Test (WHAT; O’Connell, 1960). These measurements do measure the humor trait, but one by one, they remain quite ambiguous and broad on the definition of the humor trait. Ruch and Köhler developed a temperament approach in which they identify cheerfulness, seriousness and bad mood as the traits that form the temperamental basis of humor, and therefore the humor trait. This led to the development of the State-Trait Cheerfulness Inventory (from now on: STCI), (Martin, 2013). This instrument has been validated in a variety of settings, including the study of the humor of teachers (Rißland, 2002), of depressed elderly (Hirsch et al. 2010), or the effects of nitrous oxide ( Ruch & Kolhler,1998). Participants rated each statement based on a 4-point Likert scale (1=strongly disagree and 4= strongly agree). The Trait form was completed by the participants before being exposed to the stimulus to prevent the results from being
influenced by the stimulus. The results were summed up using the Trait-scoring key (Appendix F) to form a score on the following dimensions: cheerfulness (CH),
seriousness(SE), and bad mood (BM).
Humor State was determined by the short State form of the State-Trait Cheerfulness Inventory (Ruch, Kohler & van Thriel, 1997) and consisted of 18 statement which described one’s current mood and mental state (Appendix G). Each statement was rated using the same 4-point Likert scale and the results were summed up using the State-scoring key to form a score on CH, SE, and BM. Subsequently, these three subscales were assembled to compose a Humor State Component variable by means of dimension reduction and factor analysis, which was possible since the three subscales had strong correlations (Appendix H). The State form was completed after being exposed to the stimulus to optimally capture the participants’ temporary mental state.
WORK HARD, LAUGH HARDER
- 15 - were associated with a less creative
personality. The adjectives which were associated with a creative personality were scored by a +1, while the adjectives which were associated with a less creative personality were scored by a -1. Eventually, in line with Gough (1979), all scores were summed to form the CPS index, which had a potential reach from -12 to +18. A higher score on this CPS index indicates a higher creative personality.
Creativity State was recorded by association tasks in which the participants had to generate as much associations as possible with the given adjectives. These creative tasks had a 5 minute time limit and could be answered in both Dutch and English to prevent the participants from being limited in their creative abilities due to a language barrier. The first creative task was carried out before the video stimulus and contained the following Dutch adjectives: amusant (amusing), saai (boring),vermoeiend (tiresome), afleidend (distracting). The second creative task was executed after the participants were exposed to the video stimulus and consisted of the adjectives gemakkelijk (easy), moeilijk (difficult), monotoon (monotonous), gefocust (focusing) (Shibata & Suzuki, 2004). By executing the creative tasks, the creativity state can be analyzed. For the definition of Creativity Trait in the hypotheses, we will wield the first creativity task, since this task takes place before the stimulus and therefore, its scores are not influenced by the humor manipulation .
Individual Creative Performance was measured by an idea generation exercise. This section measured the individual creative performance. Since creativity is frequently seen as generating novel and useful ideas in the literature, this exercise captures the participant’s creative performance (Amabile, 1996; Mumford et al, 2002). The canteen facilities of a university campus were used as the subject to generate ideas for. The participants were instructed to generate ideas within a 5 minute time frame. The time frame for this exercise is based on the results from the pilot test. Each idea had to contain a short description of 10-15 words, to allow for serious evaluation by experts (Poetz & Schreier, 2012). The best idea was rewarded with a prize of €100,-. This way, the participants were motivated to be as creative as possible. The idea that eventually won the €100,- was “Introducing Small, Medium and Large servings”. Participants were allowed to complete the task in both English and Dutch, to prevent the language skills to be a barrier for the participants’ creativity. This exercise was executed after the exposure to the video stimulus to examine the manipulation effect of the participants’ humor mood on their creative performance.
- 16 - sorted and categorized by the researchers. This
way, the researchers succeeded in tracking and bundling similar ideas to reduce unnecessary overlap. This action led to the reduction of the 1219 ideas into 191 unique final ideas, which all were connected to single or multiple participant ideas. The final set of ideas was randomized before use in the idea rating (Appendix J). The experts rated the ideas using the Consensual Assessment Technique (Amabile, 1982). This technique assumes that a creative idea about a particular subject is best measured by the combined assessment of experts in that particular field. Based on previous literature, the ideas were rated on novelty, customer benefit, and feasibility (Piller & Walcher, 2006; Whittle et al, 2010; Poetz & Schreier, 2012) on a 7 point Likert scale. All ideas were rated separately by the experts on one dimension before continuing to the second and third dimension. Eventually, all ideas received a total score using a three-way interaction term (novelty x customer benefit x feasibility). The ideas were all presented in Dutch, to avoid a language barrier for the experts. These measurements led to the identification of a twofold measurement of creative performance. In this research, individual creative performance will be measured by quantity (number of ideas per participant) and by quality (best idea per participant) (Poetz & Schreier, 2012).
Control variables
Socio-demographic and experimental variables are incorporated as control variables to observe and control the influence of other independent
variables on the dependent variable. We included the control variables gender, age and frequency of canteen visits. The first two control variables are common control variables in the research on creativity (Amabile, 1996), the last control variable (frequency of canteen visits) is added because we expect that participants that are more often visiting the canteen, will be better informed and aware of the canteen facilities and therefore will be more capable of generation ideas for on this subject matter (Fat, 2000).
3.5 Quality of Research
To verify the internal and external validity, reliability, consistency and objectivity of this research paper, we conducted a number of analyses.
3.5.1 Inter-rater reliability
WORK HARD, LAUGH HARDER
- 17 - concert and both experts got the possibility to
discuss each other’s ratings. This way, the rating behaviour will be aligned as much as possible. If the ratings were far apart, the previous steps would be repeated until consensus was reached. This process was executed for ten example ideas. When final consensus was reached, the experts individually started rating the 191 ideas on laptops. During this assignment, they were not allowed to discuss their scores until they had finished all ideas. The complete set of unique ideas was rated on one dimension, before continuing to the next dimension. Thirdly, the previous steps were applied to the second and third dimension; respectively customer benefit and feasibility . Expert1 scored an average score of 3,75 and these scores had a standard deviation of 2,39, while Expert 2’s average mean was 5,24 and consisted of scores that had a standard deviation of 2,078. The mean of the average score of both experts was 4,49. On average, Expert 1 was scoring 1,49 point lower than Expert 2.
3.5.2 Intraclass Correlation
To confirm inter-rater reliability, the intraclass correlation method was exerted. Since the experts were not drawn randomly from a larger population of raters, but were deliberately choosen beforehand, a two-way mixed measure is used. Consistency had an average measure of .79 with a lower bound of .78 and an upper bound of .81. The Cronbach’s alpha for the inter-rater reliability is .80 (Appendix K). Therefore we can confirm that it meets the reliability standards.
3.5.3 Factor-analyses State and Trait Scales The State and Trait scales of the STCI Questionnaire were purified by performing a factor analysis using principal component analysis with varimax rotation and a cut-off value of .50 in SPSS. The analysis was performed on the State and Trait-forms of the STCI. The State-form showed minor problematic results. The CH subscale had a Cronbach’s alpha of .85, which is above the threshold of .70 (Cortina,1993). Item 3 and 17 on the CH subscale had fairly high cross loadings, but will be maintained, since the items load on the right factor and do not influence the Cronbach’s alpha tremendously. The BM and SE subscales had Cronbach’s alphas of .89 and .85 respectively. Therefore, the complete STCI State scale will be applied for this research.
- 18 - remaining items capture the essence of each
subscale (Ruch, 1998) (Appendix L).
3.5.4 Reliability CPS
The reliability of the Creative Personality Scale was calculated via a weighted composite technique (Oldham & Cummings, 1996). First, the Cronbach’s alpha of each subscale was calculated. The positive and negative subscales had alpha’s of .70 and .35 respectively and a correlation of .27. By removing the items individualistic and snobbish, the highest possible alpha (.72) for the positive subscale was reached. Deleting artificial, conservative, conventional, dissatisfied, suspicious, narrow interest, and submissive from the negative subscale created the highest possible alpha for this particular subscale (.43) which led to a correlation of .30 with the positive subscale. Including submissive in the negative subscale, however, slightly decreased the Cronbach’s alpha for this particular subscale to .42, but increased the correlation with the positive subscale to .31. Eventually, including submissive did not lead to a lower overall CPS reliability score and therefore was not deleted. The total alpha of the CPS index was .68. Therefore we can confirm that the CPS meets the reliability standards.
3.5.5 Manipulation check stimulus
In the experiment, the participants were exposed to different stimuli; a neutral film stimulus and a humorous film stimulus. To check for the significant effect of this manipulation, we conducted a check by means of the participant’s self-report Humor State,
WORK HARD, LAUGH HARDER
- 19 - 3.6 Data analysis overview
Table 3 provides an overview of the different analyses and the corresponding hypotheses that will be applied to determine the antecedents of creative performance.
- 20 - 4. RESULTS
The next chapter presents the descriptive analyses (section 4.1), sample characteristics (section 4.2) and the bivariate correlations (section 4.3). Supplementary, we will elaborate on the results of the one-way ANCOVA and Regression analysis. We strive for finding tentative answers by means of assertions. Based on the outcomes of the experimental data and its analyses, the hypotheses will be confirmed or rejected(section 4.4).
4.1 Descriptive statistics
Table 4 displays the descriptive statistics of
the variables of our research.. The average age of the research participants was 22.68 years, with a population that consisted of 62% male and 38% female participants. This percentage is consistent with the overall population of the Faculty of Economics and Business, of which 66% is male and 34% is female. Furthermore, 53% of the participants was enrolled in a Master study, while 47% was enrolled in a Bachelor study at the Faculty of Economics and Business. Of the participants, 13,25% visited the Kapteynborg Canteen Less than once a Month, 9,25% Once a Month, 23,84% 2-3 times a Month, 15,89% Once a Week, 29,14% 2-3 Times a Week and 8,61% on a
WORK HARD, LAUGH HARDER
- 21 - Daily basis (Appendix N).
4.2 Sample characteristics
A Shapiro-Wilk’s test (p>.05) (Shapiro & Wilk, 1965) , the calculated Z-values and a visual inspection of their histograms, normal Q-Q plots and box plots showed that the major part of our data were approximately normally distributed (Appendix O) . The variables Creative Performance-Quality, Creativity Trait Age, Trait Cheerfulness, Trait Seriousness, Trait Bad Mood and State Cheerfulness are a little skewed and kurtotic, but it does not differ significantly from normality. We can therefore assume that these variables are normally distributed in terms of skewness and kurtosis. As well, in terms of the Shapiro-Wilk test, we can assume that our data are approximately normally distributed.
Ideally, all data is normally distributed, but this is seldom true for all variables. For the variables Creative Performance-Quantity, Creativity State, Frequency of visits, Humor State Seriousness and Humor State Bad Mood, we found Z-values that slightly differed from the required range of -1.96<Z>1.96. A Shapiro-Wilk test for these variables exposed that they are significant for non-normal distribution. To exterminate these extremes, we transformed the relevant variables with log and reflect. For Creativity State, this led to a normal distribution according the Shapiro-Wilk test. For the remainder variables, creating a Log did not change the significance of the normality tests. Nonetheless, for the variables Idea Quantity, and Trait Bad Mood , the Log variant is preferred for further use since this
provides us with less skewed and kurtotic values. For executing the analyses correctly, variables with less extreme values are strongly desired (LaLonde, 2005). Therefore LogIdea Quantity, LogTrait Bad Mood, and LogCreativity State will be exerted in the hypotheses testing.
4.3 Bivariate correlations
A correlation matrix (table 5) is drafted to represent the relations of each variable to the other variables. As well Humor State as Humor Trait consist of three items: Cheerfulness, Seriousness and Bad Mood. By conducting a bivariate correlation analysis, we are able to observe the correlations between de subscales of these measurements of humor and the remaining experiment variables.
- 22 - Creative Performance Quantity (.190*).
For Creative Performance Quality, there is a significant correlation with Trait Seriousness (.174*), which means that participants that score high on Trait Seriousness as well score high on Creative Performance Quality. Trait Cheerfulness correlates significantly negative with Trait Bad Mood (-.568**) and State Bad Mood (-.315**). As well, Trait Cheerfulness and State Cheerfulness are significantly correlated (.326**), which is consistent with the theory that the Humor Trait measurement is a predictor for the Humor State (Ruch, 1998) An equivalent correlation is manifest for Trait Bad Mood, which has a significant correlation with State Cheerfulness (-.343**) , State Seriousness (.247**)and State Bad Mood(.554**). Furthermore, we observe a significant negative correlation for State Cheerfulness with State Seriousness (-.569**) and State Bad Mood (-.715**) and a significant correlation for State Seriousness and State Bad Mood (.445**). Moreover, participants that score high on State Seriousness score
WORK HARD, LAUGH HARDER
- 23 -
- 24 - 4.4 ANCOVA analyses , regressions and
hypotheses
To test the hypotheses and the conceptual model, ANCOVAs are conducted. In section 5.4.1, we will elaborate on the assumptions for ANCOVA. In section 5.4.2, the ANCOVA’s will be performed. In section 5.4.3, the regression analysis for Humor State will be conducted.
The selection of ANCOVA analyses was based on the design of the experiment and the conceptual model. The conventional analysis of covariance (ANCOVA) is administered to test the main and interaction effects of categorical variables on a continuous dependent variable controlling for the effects of selected other continuous variables, which co-vary with the dependent variable. In this set-up, the control variables of our research will take the role of covariates and the independent variables will be applied as interaction factors. ANCOVA is very frequently used in experimental designs, to reveal the effects of manipulation and to control for factors which cannot be randomized but which can be measured on an interval scale. (Keselman et al., 1998, Mayers, 2013) With the ANCOVA analyses, we strive for exploring the potential effect of the independent variables in combination with the effect of the stimulus, which is considered to be the main effect. Scilicet; we identify the extent to which a variable is interacting on the relationship between type of stimulus and creative performance quantity and quality.
4.4.1 Prior tests for assumptions and restrictions for ANCOVA
Before the ANCOVAs are conducted, multiple initial tests need to be considered to ensure that none of the assumptions of the ANCOVA test are being violated. (Brace et al., 2006; Doncaster et al., 2007; Tabachnick & Fidell, 2007)
Correlations between covariates and dependent variable
A prerequisite for conducting the ANCOVA is reasonable correlation between the covariates and the dependent variable. In a twofold analysis (both for creative performance quantity and creative performance quality) of the correlation matrix (table 4), we observe a correlation of all of the covariates with the dependent variable. This correlation is not significant in every case with p,0.05, still, the prerequisite is met.
Independence of covariates
WORK HARD, LAUGH HARDER
- 25 -
Table 6: Independence of covariates
Normal distribution
Third, the ANCOVA analysis requires a normal distribution. By means of the Kolmogorov-Smirnov and Shapiro-Wilk analyses, we have already analyzed and corrected these distributions (Appendix O).
Homogeneity of regression slopes
Fourth, it is required to assess whether the correlation (or the regression slopes) differs significantly between the groups. This is executed by means of building a custom model to check for the interaction effect of the covariates, as displayed in table 7 and table 8. From this analysis, we can affirm that there is
no significant interaction between the Type of Group and the covariates Gender, Age and Frequency of visits since p > 0.05 for Idea Quality. For Idea Quantity, we find a significant interaction effect of Frequency of visits (p= .009). For the other covariates, there is no significant interaction since p> 0.05. Therefore, we will exclude Frequency of visits as a covariate from the main ANCOVA analyses and instead add an extra analysis for this variable to discover its specific effect on creative performance. Nonetheless, by
- 26 - 4.4.2 Main analyses ANCOVA
In this section, we will explore the effect of the covariates and interaction variables.
By means of two one-way ANOVAs, it becomes apparent that participants in the treatment condition generated more ideas (M=8.577) than did participants in the control condition (M=7.713). In like manner, the quality of the best idea per participant was higher for participants in the treatment group (M=190,298) than for the participants in the control group (M=184,875) (Appendix P).
Subsequently, a one-way ANCOVA is conducted for the relationship between the type of stimulus and creative performance quantity and quality, controlled for the covariates gender and age. The covariates could be an explanation for the mean differences between the treatment and stimulus group. Nonetheless, the results of the one-way ANCOVA disclose that there is no significant effect for the covariates gender and age in both the quantity (p=.216; .915) and quality ( p=.332; p=.534)
Table 7: Homogeneity of regression slopes idea quantity
WORK HARD, LAUGH HARDER
- 27 - dimension of creative performance (Appendix
Q).
Therefore, a custom model is built to test for the interaction effect of Humor Trait, Creativity Trait, Creativity State and Frequency of visits. (Appendix Q). For these models, the relationship between type of stimulus and creative performance is controlled for gender and age (covariates). These ANCOVAs were conducted on the quality and the quantity dimension of creative performance.
The ANCOVAs for humor stimulus versus neutral stimulus on creative performance quantity were found to have statistically significant interaction effects for Humor Trait Seriousness F(1,78) = 1.891, p<0.05 and Creativity State F(1, 83) = 1.489, p<0.05. Furthermore, the ANCOVA for humor stimulus versus neutral stimulus on creative performance quality was found to have a statistically significant interaction effect for Humor Trait Seriousness, F(1, 78) = 2.2 11, p<0.01.
Creativity Trait
According to Hypothesis 1, creativity trait should relate stronger to the quality and quantity of creative performance in the humorous treatment conditions than in the control condition, leading to different mean scores on creative performance quality and quantity. When we address the results of the ANCOVA, the adjusted mean of Creativity Trait for the treatment group on the creative performance quantity dimension displays an
increase from .8744 (unadjusted) to .892, while the mean of the control group displays only a small increase; from .8527 (unadjusted) to .864. For the creative performance quality dimension, a mean decrease is found for the treatment group; from 190.2975 (unadjusted) to 184.221, while the mean of the control group only slightly decreases from 184.8750 (unadjusted) to 184.128. Therefore we can conclude that the results are partly in line with our prior expectations, since the effect of the stimulus on the quantity of creative performance is more positive for the treatment group than for the control group. Since this analysis lacks significant results, H1 cannot be confirmed.
Creativity State
- 28 - (unadjusted) to 191.241, while the mean of the
control group decreases from 184.538(unadjusted) to 177.105. Therefore Hypotheses 2 can be confirmed for the quantity dimension of creative performance.
Humor Trait
According to Hypothesis 3, humor trait should have been stronger related to the quality and quantity of creative performance in the humorous treatment conditions than in the control condition, leading to different mean scores on creative performance quality and quantity. In our research, we applied the three subscales of Humor Trait; Cheerfulness, Seriousness and Bad Mood. For the adjusted means that that take into account this moderating effect, the one-way ANCOVA broadcast a change in means due to interaction effect for both quantity and quality. The adjusted mean of Humor Trait for the treatment group on the creative performance quantity dimension increases from .8744 (unadjusted) to .875 while the mean of the control group decreases from .8527 (unadjusted) to .835.
For the creative performance quality dimension, we see a similar change; the mean of the treatment group increases from 190.2975 (unadjusted) to 190.642 while the mean of the control group decreases from 184.8750(unadjusted) to 180,688.
This effect is caused by the significant interaction effect of Humor Trait Seriousness. Therefore, we can partly confirm Hypotheses 3; for both creative performance quantity and quality, Humor Trait Seriousness is
significantly (p<0.05) higher for the humor treatment group.
Frequency of visits
Since the control variable Frequency of visits violated the prerequisites to be applied as covariate, an ANCOVA was conducted for the interaction effect of Frequency of visits. Since p>0.05, we can confirm that there is no significant interaction effect for frequency of visits on the relation between type of stimulus and creative performance for both the quantity and the quality dimension (Appendix P).
4.4.3. Regression Analysis Humor State
The relationship between humor state and creative performance quantity and quantity requires a different approach than the other variables of our conceptual model. Humor State is the only variable that is measured after the manipulation in the experiment. Therefore, it is not desirable to, as an analysis for the relationship between humor state and creative performance, explore the relationship between type of group and creative performance for the interaction effect of humor state, as was being carried out for the other hypotheses by means of ANCOVA. To explore Hypotheses 4, we conducted an OLS regression analysis for the separate treatment and control groups for the quality and quantity dimension of creative performance to be able to explore the in-between subject effects. We do this by means of the following hypotheses:
WORK HARD, LAUGH HARDER
- 29 - In this regression analysis, we corrected as
well for the control variables of our conceptual model; gender and age . In table 9 (Appendix R), the outcomes of these regressions are represented. The outcomes show us that there is a clear difference between the results on the relationship of humor state and creative performance for the treatment group and for the control group. For as well the
quantity as the quality condition, we observe a higher β and for the creative performance quantity, we see that the regression displays a significant effect (p<0.05) for the treatment group. Therefore, we reject H0 for creative performance quantity. Consequently, H4 is supported for creative performance quantity.
5. CONCLUSIONS AND IMPLICATIONS In this section, we will elaborate on our findings, the lessons learnt, limitations, managerial implications and suggestions for future research.
5.1 Discussion
Does exposure to a humoristic stimulus change the individual’s mood, and if so, does this change of mood affect the individual’s creative performance? From an existing psychological research perspective, we found that humor trait is a predictor of creative performance (Morreall, 1991) and that a humorous state can facilitate and enhance creative performance (Ruch, 1996). Likewise, creativity trait and state are considered to be positively related to creative performance (Amabile, 1996; Shibata & Suzuki, 2004).
The exploration of these theoretical relations was approached by the execution of a scientific experiment; we examined the effects of a mood manipulation in the form of a film stimulus on the relationship between different forms of humor and creativity and individual creative performance. The findings of this research underline that participants in the treatment group that were exposed to a humorous stimulus, on average generated more ideas and as well scored higher on the quality of their best idea than the participants that were in the control group and were exposed to a neutral stimulus. We found that the effect of a 10 minute humor stimulus was sufficient to significantly change the individual’s mood to a substantially more humorous mood, which is in line with existing literature. This mood, which we designate as
- 30 - humor state, positively relates to the amount of
ideas; only for participants that were brought into the humorous mood, there was a significant relationship between their humorous mood and the amount of ideas that they generated. Another very interesting finding, is the fact that an individual’s creativity state is a great predictor for creative performance, but only in combination with a humorous stimulus; only for the treatment group, there was a significant positive effect of creativity state on the amount of ideas for creative performance (quantity). In other words; from these results we can tell that a humorous mood enhances and increases the relationship between creativity state and creative performance.
Evidence for relationships with our other main variables was mixed and not completely in line with existing theories in this research. For humor trait, we found a significant relationship for the subscale humor trait seriousness for as well creative performance quality as quantity. This means that the positive effect of the stimulus is stronger for people that are characterized by a more serious nature. For the creativity trait, we found no significant differences for the treatment and control group. For both trait variables, we found only small differences between groups. This is not in line with the existing literature that establishes the relationship between these concepts.
Here it is important to point out that the aim of this study was not to investigate the specific relationship between the different independent variables and the dependent variable, but to find out if bringing individuals in a humorous
mood would lead to an increased innovative and creative performance. The scientific experiment helped increase our understanding of the mood influencing process and of how participants are brought into a different state, and its effect on subsequent actions. Since we have achieved to significantly change an individual’s mood and its subsequent actions, we consider this a great contribution, a great step towards bridging the literature gap and a great learning experience.
5.2 Managerial implications
WORK HARD, LAUGH HARDER
- 31 - innovativeness can be enhanced. Lastly, as we
argued at the introduction of our study, humor is a largely underexposed topic in business. It is often seen as something that is, to a greater or lesser extent, naturally present at the workplace. Instead of accepting this a fixed given, we hope managers will start to see humor as a useful influential and managerial tool. We believe more awareness of humor as a business device could have great benefits in terms of efficiency, creativity and quality of the innovation process.
5.3 Limitations
Although this research has contributed significantly to the existing knowledge field, there were several limitations. First of all, we can identify various methodological limitations of this study. An experimental design is pre-eminently prone to (unwanted) external influences. Although we have strived for ensuring the internal and external validity by means of several steps through the process, one has to take into account other factors (which were not addressed as control variables) that might have affected the outcomes. Secondly, although having 151 participants participate is considered a high performance for this type of experimental research, the relative small sample still entails sample problem; because of this, we experienced normality problems, which might have negatively affected our hypotheses testing. Furthermore, with a small sample, it is harder to generalize the findings for larger populations. Thirdly, to a large extent, we applied self-report scales. Although evidence is found for the reliability of these
type of scales, the type of data that it generates is limited by the fact that it rarely can be independently verified. This same limitation also applies for the determination of the creative performance quality scores; these scores that are assigned to the participants ideas are based on the (subjective) opinion of two experts and cannot be verified. This might be an explanation for the stronger significance for creative performance quantity (which was free from any subjective scoring) than for creative performance quality in this research. Additionally, although we strived to correct for the Hawthorne effect (Wickström, 2000) by means of our research design, we can still question the meaning of the humorous stimulus. Did this specific stimulus really induce the causal relations with creative performance, or is this due to the effect of being paid attention to and would a different type of stimulus have caused a similar effect? Finally, our experimental sample consisted of a rather homogeneous group in terms of age, educational background and nationality. The effect of a manipulative stimulus might differ for other groups, which could lead to different results.
5.4 Suggestions for future research
