Music at work : how ones concentration can cause distraction for someone else

I

Bachelor Thesis - Business Studies

Music at work: How ones concentration

can cause distraction for someone else.

Abstract

In this research the perceived effects of listening to music at work will be investi-gated for the Dutch working population. This research builds on the research from Haake (2011) as it investigates the beneficial and detrimental effects that music at work can be perceived to have. In order to find these effects a questionnaire based survey was constructed and a sample was drawn (N = 123). The exploratory factor analysis revealed five factors that are perceived to be related to listening to music at work: productive, negative, affect, communication and masking. Also the influence of personal and task characteristics on these perceived are measured. The theoretic-cal and managerial implications along with the limitations and suggestions for future research are also given.

Table of Contents

Abstract 1

1. Introduction: 3

2. Literature Review: 6

2.1 Effects of music at work 6

2.2 Music Characteristics 10 2.3 Personal Characteristics 12 2.4 Task Characteristics 13 2.5 Conclusion 14 3. Methodology: 15 3.1 Research design 15

3.2 Sample and Data collection 16

3.3 Measure 17

4. Results 20

4.1 Descriptive statistics 20

4.2 Factor analysis 23

4.3 Inferential statistics 26

4.3.1 Influence on listening behaviour 26

4.3.2 Influence on perceived effects 27

5. Discussion 30

5.1 Theoretical implications 30

5.2 Managerial implication 33

5.3 Limitations and future research 34

6. Conclusion 35

Acknowledgements 36

Reference 37

Appendix A: Survey - Music at work 40

1. Introduction

Engagement with musical activities at work has not been a recent development. In the period before the industrialization music and work were mutually constituted to a significant degree, the work was constituted through music, and the music was constituted through work (Korcynzki, 2003). The music was generally termed ‘work songs’ and was sung by the people performing the work. Music in this period was considered as a tool setting a certain rhythm and pace, but was also an opportunity for the workers voice themselves expressing their own opinion although often coded (Colonial Williamsburg, 2003). Korcynzki (2003) speculates that the music distracted the workers from the repetitive labour undertaken day in day out and took the mind of the singer away from the place of action.

During the period of industrialization ‘the big split’ occurred, which indicated a sharp divide between music and work (Korcynzki, 2003). Machines controlled the pace in the industrialized labour process and thus music was not considered to be functional, but recreational, as it no longer set the pace. Management demanded rationalized and disciplined behaviour at work and were afraid that music could threaten managerial power. Therefore music became prohibited and was labelled a negative distraction, not suitable for the workplace. Music production and consump-tion also became separated during this period; this development was reinforced due to the development of the radio and music records.

The transition from the absence of music, towards the rise of background music at work is due to a shift in perception about the effects of music. Music allowed for distraction and transportation of the mind, which was a coping mechanism during monotonous and often routinized labour process. Its presence does not intrude upon the aural space, appropriated by management for the issuing of hierarchical instructions where necessary (Korcynzki, 2003). The distraction that music supplied came to be seen as beneficial for the work performance. Fox (1971) reviewed the use of music in industrial settings and his research indicates that music can have beneficial results as it can aid alertness. Especially in jobs where a certain amount of repetition is involved (Fox, 1971, p.70).

The effects of background music on task performance have been a topic of interest amongst applied psychologists, cognitive psychologists and personality theorists (Doyle & Furnham, 2011). Research into background music as with music in general shows inconsistent and sometimes even contradictory results. According to Dobbs, Furnham & McClelland (2010) noise and music both have equally distracting effects but cause quite different affective reactions. Someone could choose to be distracted by music (to take away from the repetitive nature of the task at hand) or to have a beneficial effect on mood. Kämpfe, Sedlmeier and Renkewitz (2011) made a comparison between studies that examined background music compared to no music and found mixed results. The research indicates that background music disturbs the reading process, has some small detrimental effects on memory, but has a positive impact on emotional reactions and improves achievements in sports. Cassidy and MacDonald (2007) compared the effects of high-, low arousal music and noise to silence on the performance of extraverts and introverts during five cognitive tasks. The performance on all five cognitive tasks decreased in the presence of background sound (music or noise) compared to silence. Introverts were more detrimentally affected than extraverts, in the presence of highly arousing music and noise.

The introduction of the personal stereo heralds the period of post-industrialism and allows for creating a personal soundtrack for his/her workday (Korcynzki, 2003). Bull (2005) argues that the new technology like iPods and other MP3 player gives users unprecedented power of control over their experience of time and space, through the creation of privatised auditory bubble. They do so by managing their mood and orientation to space through the micro-management of personalised music. The opportunities for experiencing music are now more diverse than at any other point in history (Greasley, 2011, p.40). Lesuik (2005) and Oldham, Cummings, Mischel, Schmidtke and Zhou (1995) both investigate the effects of listening to self-selected music at work. Self-self-selected music appears to influence mood, which in turn influences work responses. Oldham and colleagues (1995) found that the mood state of relaxation best explained the relationship between personal stereo use and performance. They also discussed that the use of headsets could reduce distraction and interruptions. Haake (2011) concludes that listening to music at work can fulfil a

wide variety of functions - affect management, engaging in/escaping from work activities, and environment/interruption management - and not only for simple routine tasks. Self-selected music provided employees with a sense of control over their surroundings and emotions and music could be considered to be conductive in one work situation and distracting in another. However she mainly focuses on the positive effects of listing to music at work.

The literature reviewed above showed inconsistent or even contradicting results, and the effects of music at work were either beneficial, detrimental or had no results on work responses and/or affect (emotion and mood). To contribute to this debate and to provide novel findings about the subject, the main question of this research is: Which functions do employees attribute to music while working? To answer this research question, the employee’s opinion about, and perception of these functions of music, will be researched. Also, possible mechanisms that could explain these functions will be explored.

The following chapter of this research consists of a literature review that will look into the possible effects of music on work responses and affect. This literature review will also provide mechanisms operationalized by previous studies that explain how these effects work; music characteristics, personality differences, type of tasks. After the literature review the research design for this research will be provided. Followed by the results from this research. The results will be discussed in the next section. Before ending with the conclusion of this research and here several limitation and opportunities for future research will be highlighted.

2. Literature Review

In this section, the existing literature about music at work will be discussed. First, the possible outcome effects that music can have during work will be summarized. These effects have been described to be detrimental, beneficial or non-exiting and are related to either work responses or affect. The following section will describe several mechanisms - music characteristics, personal characteristics, task character-istics - that influence the mentioned outcome effects. Finally, a summary will be given where important factors for this research will be emphasised.

2.1 Effects of music at work

Listening to music at work has often been described as either being a form of, or directly causing distraction (Avila, Furnham & McClelland, 2011; Dobbs, Furnham & McClelland, 2010; Dibben & Williamson, 2005; Doyle & Furnham, 2011; Furnham & Bradley, 1997; Furnham & Strbac, 2002; Frith, 2002; Goethem & Sloboda, 2011; Haake, 2011; Jett & George, 2003; Oldham, Cummings, Mischel, Schmidtke & Zhou, 1995; Scheufele, 2000; Speier, Vesey & Vlacich, 2003; Stein, 2012). According to previous research music can also mask-off other distractions (Bull, 2004; Loewen & Seudfield). This shows that findings about the effect of music at work have been inconsistent and sometimes even contradictory. To find out what the effects of listening to music at work are on work responses and affect, this section will give: a definition of what constitutes a distraction; the possible effects of distraction; the effect of music (negative and positive) on both work responses and affect.

Speier, Vesey and Vlacich (2003) and Jett and George (2003) have investigated the potential consequences of interruptions, for the person being interrupted. Speier et al. (2003) defined interruptions as uncontrollable, unpredictable stressors that requi-re additional decision-maker effort. An interruption brequi-reaks the attention away from the primary task - if only temporarily. Distractions and interruptions are similar in that they can both occur while a decision maker is performing a primary task. However, the manner in which distractions and interruptions are detected by sensory channels differs: distractions are detected by a different sensory channel from those of the primary task and may be ignored or processed concurrently with a

primary; interruptions, on the other hand, use the same sensory channel for both the interruption and the primary task. Thus, decision makers cannot choose to ignore interruption cues, resulting in both capacity and structural interference. Capacity interference occurs when the number of incoming cues is greater than a decision maker can process. Structural interference occurs when a decision maker must attend to two inputs that require the same physiological mechanisms (e.g., attending to two different auditory signals - listening to a conversation and music). Distraction conflict theory (DCT) describes a research stream investigating the influence of distractions (e.g., industrial noise or background music) on decision performance. This theory suggests that distractions facilitate performance on simple tasks and inhibit performance on more complex tasks (Baron, 1986). The underlying premise behind DCT is that as distractions occur during simple tasks, stress increases and attention narrows, resulting in the possible dismissal or exclusion of irrelevant information cues, thus facilitating decision performance. However, as the number of information cues (i.e., complexity) increases, a decision maker’s excess cognitive capacity decreases. The consequent narrowing of attention likely results in a decision maker processing fewer information cues, some of which may be relevant to completing the task successfully, which results in deteriorating performance. Finally, because interruptions use the same sensory channels as the primary task, interruptions may also result in the loss of working memory contents or confusion between cues in memory, which further inhibits decision performance.

Jett and George (2003) distinguish four types of work interruptions - intrusions, breaks, distractions and discrepancies - and propose conditions under which each type is likely to have negative and positive consequences for the person whose work is being interrupted. An intrusion is defined as an unexpected encounter initiated by another person that interrupts the flow and continuity of an individuals work and brings that work to a temporary halt. Breaks are planned or spontaneous recesses from work on a task that also interrupts the task’s flow and continuity. Distractions are psychological reactions triggered by external stimuli or secondary activity (music) that interrupt focussed concentration on a primary task. Discrepancies are perceived inconstancies between one’s knowledge/expectations and one’s immediate observa-tions that are relevant to the task and personal well-being. “The potential negative

consequences of intrusions are often recognized, whereas the potential positive con-sequences are often overlooked’ (Jett & George,2003, p. 496). Listening to music is a form of distraction according to this definition of interruption types and so here the negative and positive consequences of distractions will be highlighted. Negative consequences can occur when available time to work on a task is scarce. Additional negative effects related to time pressure may include heightened feelings of stress and anxiety, as the person is being distracted. Distractions can also lead to mediocre performance when the person's work is complex, demanding, and requires learning and one's full attention and/or when the person has certain personality traits. But distractions can also have positive effects as they lead to enhanced performance when other environmental stimuli are filtered out and/or increases stimulation levels on routine tasks.

Thus whether a person experiences negative or positive consequences from distractions, depends on the characteristics of both the person and the task being performed (Jett & George, 2003). In this research these influences, but also the influence of the characteristics of music and of how the music is presented will be listed as well later on. But first several other effects of music on work responses and/or affect will be mentioned. Starting with the negative effects followed by the positive effects of music at work.

The most extreme perspective is given by Cloonan and Johnson (2002), who have researched the ‘darker-side’ of popular music. They investigated how popular music might have been used as a source of pain and/or to accompany the inflicting of pain. A broad division exist between pain, which is incidental to the intended function of the music (e.g. loud music played in an adjacent room or office) and occasions when music is used deliberately for inflicting pain. Since sound is an ancient marker of a person’s physical and psychic territorial identity. Thus music, a form of sound, can be seen as an invasion to the personal acoustic space, which in turn can lead to acoustic conflicts. Frith (2002) support this notion as he notes that because music is now used to mark out private territory, it can also invade it; music has become part of us and thus it can be misused; and because music is nowadays used as an emotional tool, this misuse is genuinely upsetting. Marti (1997) calls music that is we are compelled to hear without wanting to listen: psychological torture. He investigated the most

annoying sounds on the street of Barcelona through mapping complaints in a local newspaper, and found that street musicians, neighbours practising instruments, and recorded music in public spaces were perceived as the most irritating. He suggested that the sounds were annoying, because the events were out of the listener’s control and invaded their own private spheres. In fact noise has been known to contribute to: deafness, tinnitus, strokes, migraines, peptic ulcers, colitis and hypertension as well, of course, as stress. It can even be, a deadly problem (Coonan & Johnson, 2002, p.33). Factors other than volume, such as repetitiveness and the projection of nuis-ance noise can also cause pain. Frith (2002) states that all uninvited noise raises the blood pressure and depresses the immune system.

On the other hand several other studies showed that music could have beneficial effects for either affect or work responses. People consciously and unconsciously use music to change, create, maintain or enhance their mood (affect) on a daily basis for their personal benefit (Goethem & Sloboda, 2011). This is known as mood or . Most regulation aims to change negative effects such as stress and sad-ness and feel more positive, other people aim to become nostalgic and melancholic. Lesuik (2008) and Lai and Li (2011) have proved that music effectively reduces stress amongst air traffic controllers and nurses respectively. Several other studies confirm that music can be used to regulate emotions and mood. Bull (2004) researched in which manner iPod users manage their time and space through the use of music and the creation of an auditory bubble. Music allowed them to control thoughts, feelings and observations, the following statement confirms this: “I can’t overestimate the importance of having all my music available all the time. It gives me an unprecedented level of emotional control over my life (Terry 3, Bull, 2004)”. People also create an auditory bubble when travelling and use the music for enjoyment, passing time and enhancing emotional states (Heye & Lamont, 2010). Scheufele (2000) compared the effects of listening to music and progressive relaxation with two control conditions - attention control and silence - to see how attention, relaxation, and stress responses differ for healthy, male human subjects. Results showed that all four of the groups completed a number cancellation task faster following the relaxation or control conditions than following stress manipulation. Listening to music appeared to distract listeners from the experimental stressor and resulted in beneficial

physio-logical effects (i.e., lower heart rates). Music caused a distraction, but also relaxed and reduced stress. Several other studies supported the notion that music can mask disturbing noise, which could then lead to enhanced cognitive performance (Loewen & Seudfield, 1992, Oldham et al., 1996; Schlitmeier & Hellbrück, 2009). Haake (2011) notes that previous research has focused on positive mood and negative effects of distraction on task performance, but she identified additional beneficial functions for music in listening in the workplace: inspiration, concentration, positive distraction, stress relief and managing personal space. She explored the music-listening practices and experiences in office settings in the UK. Performance, turnover intentions, mood and turnover intentions were also found to improve, due to listening to self-selected music on personal stereos (Oldham et al., 1995). According to Oldham et al. (1996) music boosts enthusiasm, increases relaxation and lessens nervousness, and these elevated mood states contribute to higher productivity.

Finally, there are some studies that seem to suggest that listening to music at work have no effect at all. Kämpfe, Seflmeier and Renkewitz (2011) performed a meta-analysis in an attempt to summarize the impact of background music. A global analysis showed a null effect, but a detailed examination of the studies revealed that this null effect is most probably due to averaging out specific effects. Furnham, Trew and Sneade (1999) indicated that whilst music improved mood and induced positive attitudes towards work, it had no effect on productivity or performance.

The above-mentioned effects of music are either negative or positive in nature, other research pointed out that the effects of music are rather ambiguous and the effects often depend on other factors (Avila et al., 2011; Dobbs et al, 2010; Doyle & Furnham, 2011; Furnham et al., 1999; Furnham & Bradley, 1997; Furnham & Strbac, 2002; Jett et al., 2003; Kämpfe et al., 2011; Loewen & Seudfeld, 1992; Ransdell & Gilroy; 2001; Stein, 2012). These factors will be highlighted in the following sections. 2.2 Music Characteristics

Several characteristics of music have showed to influence the effects music has on work responses and affect. These influences have been inconsistent and research has struggled to find significant results. In his review about the influence of music on human behaviour, Fox (1971) also looked at the type of music and in the optimum

length of music listening. He states that there is allot of confusion and contradiction as to what is the best type of music and optimal length of programme within the industrial context. He states that music should not be played continuously as it would lose the stimulating value. Newman, Hunt and Rhodes (1956) looked at the effects of four types of music - dance, show, folk and popular - versus no music on the quantity and quality of production and the attitude of workers engaged in the routine task of assembling and packing skateboards. The results showed that neither the type of music nor music versus no music had effect on quantity and quality of worker output. However, the attitudes of employees toward the music were highly favourable. Tucker and Bushman (1991) played rock and roll music at 80 dB and found that it decreased the performance of undergraduates on mathematical and verbal tasks, but not their performance on a reading comprehension test. Huang and Shih (2011) contradict these findings and claims that not the type of music, but the fondness for the music influenced the attention of the listener.

Other factors besides the type of music and the fondness for the music include, the tempo of the piece or whether the songs include lyrics or not. Avila et al. (2011) investigated the effect of familiar musical distractors on the cognitive performance of introverts and extraverts. Participants completed a verbal, numerical and logic test in three conditions: vocal music, instrumental music and silence. Familiarity and vocals were expected to have a detrimental effect on performance for introverts and might be a beneficial influence for extraverts. The results showed patterns in the predicted direction but failed to reach significance. Furnham et al. (1999) struggled with finding significant results as well, but also concluded that this trend existed. Shih et al. (2012) did however manage to find that background music with lyrics had significant negative effects on concentration and attention. With a laboratory study Schlittmeier and Hellbrück (2009) tried to determine whether background music could mask the detrimental impact of background sound on cognitive performance in the same fashion as continues noise. They tested cognitive performance using verbal serial recall, using staccato music, legato music or continuous noise superim-posed on office noise. The results showed continues noise repressed the detrimental effects of office noise on cognitive performance, although participants preferred listening to legato music subjectively.

2.3 Personal Characteristics

In general cognitive task can benefit from moderate levels of aurousal, but it can also impair performance at extreme levels. Researchers have agreed that music can have negative effects on task performance if the music produces very high levels of arousal (although there are individual differences in terms of what constitutes very high levels), as this can cause problems narrowing and controlling attention (Haake, 2011). This conclusion is based on the psychobiological framework of Yerkes and Dodson (1908), which stipulated that there is an inverted U-relationship between performance quality and arousal.

Eysenk’s (1967) cortical arousal theory holds in account individual variation in optimal cortical arousal levels. Introverted individuals have a smaller amount of available resources for working memory capacity and thus require much less arousal before reaching their optimum level. Extraverts require a higher level of arousal to reach optimum levels. Introverts experience inhibition in task performance when their neurological threshold of arousal has been exceeded, while extraverts need extra arousal to reach their optimum functioning level. This type of individual difference in arousal has been supported in laboratory studies into the effects of music on task performance, extraverts have performed better than introverts in the following tasks: memory recall (Furnham & Bradley, 1997) coding tasks (Furnham et al., 1999), and reading comprehension tasks (Furnham & Bradley, 1997). Avila et al. (2011) failed to find a significant interaction between music and personality. Cassidy & MacDonald (2007) studied the effects of music on the cognitive task performance of introverts and extraverts, and manipulated music’s arousal potential and affect, and also compared the effects of music conditions with effects of everyday noise. Music with high arousal and negative affect appeared to be most detrimental for performance of recall tasks and the Stroop test, and that introverts were more negatively affected than extraverts. Stein (2012) proved the moderation effect of extraversion such that extraverts were facilitated with different forms of distractions (of social distractions (co-actor and evaluator) and non-social distractions (low and high complexity music)) while introverts were impaired. Extraverts thus need more outside stimulation to achieve performance facilitation while introverts tend to become over-stimulated with too much outside stimulation.

Individual differences other than personality traits have also been studied. For example, a study of the effects of background music on word processed writing by Ransdell & Gilroy (2001) showed that musically trained participants, or those with a high working memory span wrote better essays with longer sentences (compared to untrained participants, or those with a lower working memory span), but generally music disrupts word fluency, as music listening influences working memory. Doyle and Furnham (2011) predicted an interaction, such that musical distraction would have a greater negative effect on the performance of non-creative individuals com-pared to creative individuals. No significant interactions were found although trends indicated that creative individuals performed better in the music condition. Using experience sampling methodology Greasley and Lamont (2011) found three different types of listeners: those identified as having low, moderate, or high engagement with music. People who listen to music for fewer hours a week were labelled as less engaged, and were less likely to listen to self-selected music, and used music to pass the time, out of habit or to help them feel less alone; and the highly engaged, who listened for a greater number of hours few week, were found to listen to more self-selected music, and were more likely to use music to evoke specific moods, create an atmosphere, or enhance an activity. Chamorro-Premuzic and Furnham (2007) show that open and intellectually engaged individuals, and those with higher IQ scores, tended to use music in a more rational/cognitive way, while neurotic, non-conscientious and introverted individuals were all more likely to use music for emotional regulation (e.g. change or enhance moods).

2.4 Task Characteristics

There is great variation from one study to another in how task complexity is un-derstood and operationalised (Byström, 1999). Task complexity has generally been approached from two different perspectives. In the first perspective, complexity is treated as an interaction between the task and the person performing the task, i.e., perceived task complexity. According to this view, task complexity is determined on the basis of the characteristics of both task performer and task. The task performer's knowledge and skills play a significant role in determination of task complexity. In the second perspective, task complexity is viewed as a function of objective task

characteristics. Contrary to the first perspective, task complexity is determined independently of the task performer. Relevant characteristics in determining task complexity are the number of alternative actions, multiple and/or conflicting goals, uncertainty of actions and goals, etc. This approach allows for evaluation of task performer’s action based on objective characteristics.

In the research of Speier, Vesey and Valacich (2003) an interruption framework is proposed, were not only complexity but also whether the task is novel or automatic is considered as characteristics of the primary task. Results from their experimental study indicate that interruptions facilitate performance on simple tasks, but inhibit the performance on more complex tasks. In general findings support this notion, but the University College of London (2007) found that people were less distracted performing a difficult task. This was attributed to the notion that the brain was filled with information that was relevant for the task, and therefore there was no extra brain capacity for processing distracting information and so people will focus all their attention on the task at hand.

2.5 Conclusion

In the studies reviewed above several functions and effects of listening to music at work have been listed. It illustrates a picture where music can be experienced as being beneficial (for concentration and affect) but also detrimental (distracting and annoying). Several mechanisms that could influence these effects have also been mentioned, such as the selected music (vocal/instrumental), the task (complexity) and personal characteristics (introvert/extravert).

This research will build upon the research from Haake (2011) to determine which functions people attribute to music at work. In her research she concentrated on individuals who choose to listen to music at work and therefore found only positive effects, this can be considered to be a form of participant bias (Saunders, Lewis & Thornhill, 2009, p. 156). This research will look into both the beneficial and detri-mental effects of music to see how the effects music is perceived in the Netherlands. Several of the factors listed above are incorporated into this research; these factors will be highlighted in the methodology.

3. Methodology

In the previous sections the existing literature about the effects of music at work were discussed, and the main question, which this research attempts to answer was posed. Below the research design, exploratory research using a questionnaire-based survey, and sample will be described and explained. Before describing how the data from this sample will be collected. Finally, the measures will be discussed to clarify how the main question will be answered through the research design.

3.1 Research design

This research will form an extension to the research of Haake (2011) and will also be exploratory in nature. In her research she described the music-listening practices and experiences in office setting in the United Kingdom. But because she filtered her responses to people who choose to listen to music at work, her responses contained positive bias. In this research the music listening practices and experiences of the en-tire Dutch working population will be examined. To find out whether music at work is considered to be a positive or a negative experience and to discover the self-perc-eived effects of the music. This research can be classified as an explanatory study, which is a study that establishes causal relationships between variables (Saunders, Lewis & Thornhill, 2009, p.140). Haake (2011) notes that the experimental approach taken in many of the previous studies has been valuable for gathering quantitative proof of the effects of music in a work situation, and for testing and validating theo-ries about individual differences in arousal potential in music-listening situations. For this research a survey strategy is used because with this strategy it is easy to explain relationships between variables and to understand those relationships. This research will use a questionnaire-based survey, because this allows for the answers of a large amount of people to be compared (Saunders et al., 2009, p. 144). This design allows for the questions and statements to be standardised and consistent, so that every person gets the same questions, in order to compare the answers of different people and to make sure this comparison is reliable (Saunders et al., 2009, p. 373).

Besides these advantages, there are some limitations to the survey design. One limitation is the number of questions that can be included, since people are

general-ly not willing to fill in a large questionnaire (Saunders et al., 2009, pp. 144-145). This meant that the number of questions was kept to a minimum to ensure that the larg-est amount of respondents was collected in the limited time available. Another pote-ntial drawback of the questionnaire-based design is that there is only one chance to collect the data, therefore it must be done properly from the beginning (Saunders et al., 2009, p. 366). This was kept in mind while designing the questionnaire.

The questionnaire will be self-administered, because this will reduce subject and participant bias (Saunders et al., 2009, p. 156) since the researcher will not be stand-ing next to the participant to check the answers or write them down for them and this will in turn improve the reliability of the data. Furthermore self-administration saves a lot of time in collecting the data and it will guarantee anonymity.

3.2 Sample and data collection

For this research a sample will be taken from the Dutch working population. This research will make no further distinction between sector, industry or firm, because it is interested in the general opinion of the entire Dutch working population about the self-perceived effects of listening to music at work. In order to be able to generalize the results, a reliable sample has to be drawn from the population (Saunders et al., 2009, p. 217). To test whether the sample is reliable some general demographics will be compared, the three demographics that will be used are age, gender and level of education. Of the Dutch population, 49.5 percent is male, and 50.5 percent is female (CBS, bevolking; kerncijfers, 2013). In term of age distribution, 23.1 percent is below 20 years old, 24,6 percent is 20 to 40 years, 35.5 percent is between 40 and 65 years, 12,6 percent is between 65 and 80 years and 4,2 percent has an age above 80 years (CBS, bevolking; kerncijfers, 2013). Furthermore, the highest education completed by people between 15 and 65 years is divided as follows (from low to high): 8.4% primary school, 23% LBO, vmbo, mbo-1 and AVO onderbouw, 40.1% MBO 2, 3, and 4, HAVO or VWO, 18.2% HBO or WO Bachelor and 9.4% WO Masters or Doctor (CBS, beroepsbevolking; behaalde onderwijs, 2011).

The size sample was strived to be as large as possible, because a large sample size allows for better generalization of the results (Saunders et al., 2009, pp. 217-218). But due to time and money constraints, a sample size as is used in previous studies

(Haake, 2011) could not be achieved. To collect the largest sample size possible with-in the limited time available, snowball samplwith-ing was used. The snowball samplwith-ing te-chnique involves distributing the questionnaire through one’s own personal network and asking existing subjects to recruit future subjects among their acquaintances. It allows for easy access to participants, but this non-random sampling strategy does include bias, since it gives people with more social connections a higher chance of selection (Berg, 2006).

To develop and distribute the questionnaire the website www.qualtrics.com was used, because it provides a lot of options and is easy to use for both responded and researcher. This method saved a lot of time, since the data did not had to be entered manually (Saunders et al., 2009, p. 365). Furthermore it allows us to reach people in different geographical locations and it can reach a lot of respondents in a relatively short time. However, collecting data through the internet can lead to a systematic bias because people engage in self-selection, they choose whether they want to participate or ignore the invitation. However, because to achieve to sample size required in the limited time for this research this could not be avoided.

The questionnaire is written in the Dutch language, because it is focussed on the Dutch working population. The layout is made attractive to make the questionnaire easier to process for the respondents and thus might increase the response rate (Saunders et al., 2009, p. 387). The questionnaire includes an introduction letter, which explains the purpose of the survey and why the respondent’s opinion is important. It will also state that anonymity is guaranteed. The survey is closed with a short text in which the respondent is thanked for their contribution and provided an opportunity to contact me with if they have any additional questions or comments about the survey, as is recommended in Saunders et al. (2009, p. 393). The measures used will be listed in the following section.

3.3 Measures

In the survey attribute variables and opinion variables (Saunders et al., 2009, p. 268) are used to measure the perceived effects of listening to music at work and to collect some personal values and demographics. This section describes the measures used for these variables in the order in which they appear in the survey: starting with

demographic variables and additional variables followed perceived influences of music. For a copy of the final survey (in Dutch), see appendix A.

The questionnaire starts of with some general demographics, concerning gender, age, level of education and occupational sector. The question about gender provides two answer options, either male or female. Age was asked in a open ended question. Since the educational system in the Netherlands has changed over the past decades, several terms refer to the same educational level. The division the Centraal Bureau voor de Statistiek maintains, was used to find out the highest level of completed education: 1) Primary school, 2) lbo, mavo, vmbo, mbo-1, avo-onderbouw (First 3 years of havo en vwo), 3) havo, vwo, mbo-2-4, 4) hbo, wo 5) other. The division of the occupational sector was borrowed from the Vrije Universiteit van Amsterdam (VU, 2013).

The next section of the questionnaire contains questions that are concerned with either the subject’s personal opinion or more general information. This part starts of with an open-ended question regarding to the number of colleagues that share the same work environment. To next question is to determine how often the subject or his/her colleagues listens to music at work using a visual scale item, which requires the participant to describe how often they themselves (or their colleagues) listen to music at work on a scale from 0 to 100. The following question determines how the music was consumed during work. The participant was asked to divide 100 points, amongst 5 scenarios that described whether the subject uses headphones (or not), if there were colleagues around (or not) and whether the music was desirable (or not). The final question in this section is related to the notion of perceived control over the music and the kind of music. Three different, but interrelated questions accessed whether the participant felt like he/she could influence or control the music. These questions are obtained from Oldham et al. (1995).

The final part of the questionnaire that is important for this research is based on the research from Haake (2011), this part will replicate some of the questions that she posed in her research but will also form a extension to her research. She posed the question: When you listen to music at work, what functions do you think it has for you? The participants had to indicate whether they agreed with a particular statement using a 5-point Likert scale. However, the statements she propositioned

were all positive in nature and therefore included a bias. In this research these question will be replicated and in addition several other questions that are drawn from previous literature are added. These questions are about both positive and negative influences music at work can be perceived to have. The questions from Haake (2011) are translated into the Dutch language and adapted in such a way that they still pose the same principal. Also some questions about distraction, annoyance and several negative affect outcomes are added.

To examine whether complexity would influence the perceptions of the functions of music, these questions were duplicated for two different scenarios. For the first scenario the participant is asked to picture him/herself performing a relatively easy task. In this research perceived task complexity is used, because we want to test the perceived effects of music for the entire Dutch working population and not just for one specific task. This means that we could not base complexity on objective task characteristics and that complexity had to be based on both the task performer and the task (Byström, 1999). In the second scenario the participant is asked to picture him/herself performing a relatively difficult task. Oldham et al. (1995) also posed a question about the self-perceived complexity of their job and this validates our decision.

4. Results

The research design and measurements have been discussed in previously and in this section the results will be disclosed. First some general descriptive statistics will be reported, before conducting an exploratory factor analysis to find latent variables from the data. Then several inferential statistics will be described.

4.1 Descriptive statistics

In total 222 people started answering the questionnaire, however 99 people have not completed the questionnaire and are therefore excluded from the sample. This means that the final sample consist of 123 respondents. As indicated in the method section, the sample will be compared with the Dutch working population in order to test if a reliable sample has been drawn. The questionnaire has been completed by 66 female respondents (53,7%) and 57 male respondents (46,3%). The sample was compared with the population, of which 50,5% is female, using a one-sample T test and this shows that the gender distribution of the sample is just like the Dutch pop-ulation (t(122)= -.700, p=.485).

The age of the respondents ranged from 17 to 63 years old, which is comparable to the ages that are considered to be part of the Dutch working population, this are the ages 15 to 65. However, the average age of the sample 34,9 was compared with the average age of the population, which is 41,4, using a one-sample T test and the results show that the sample is different from the population in terms of age (t(122) =-4.917, p=.000). In table 1a the age of the Dutch population is compared with the sample and this additionally describes the difference between the sample and the population. The boxplot of the age depicted in graph 1 shows that the distribution amongst the younger and older half of the sample is less skewed. The 25th percentile is 21, the median is 29 and the 75th percentile is 50.

In table 1b the education of the Dutch population is compared with the education of the sample. The responded to the questionnaire consists of predominately highly educated individuals 35.8%, in contrast to the 17% that were lower educated. The different education levels allow us to compare the responses of lower, average and higher educated individuals.

Graph 1: Age Boxplot

Occupational categories include Consultancy (N = 5), Business services (N = 12), Financial services (N = 13), Information and communication (N = 11), Teaching and research (N = 19), Politics (N = 1), Health and social welfare (N = 17), Logistics (N = 7), Industrial (N = 5), Hospitality industry (N = 16) and miscellaneous (N = 18). In total 8 respondents worked without colleagues in their direct vicinity, 51 of the respondents shared an office with 1 to 3 colleagues, 41 respondents worked in an office with 4 to 8 colleagues and 23 respondents worked together with 9 or more colleagues in the same office.

In the next section of the questionnaire the participants were asked to estimate how much percent of the time the participant and his/her colleagues spend listening to music while at work. On average the participants listen to music more then half of the time while at work (53,9%) and this was similar to the time they estimated their colleagues spend listening to music at work (52,3%). The standard deviation of music listening behavioor of both the participant themselves (SD = 37,2) and their colleagues (SD = 35,9) is rather large though. This deviation is rather obvious as 7 participants indicated that they did not listen to music at all while at work and 25% of the participants spend less then a fifth of their time at work listening to music (18%). While 21 other participants indicate that they listen to music all the time

Table 1b: Education Dutch population compared to education sample

Primary school VMBO, MBO1, AVO

onderbouw MBO 2-4, HAVO, VWO HBO, WO

Dutch Population 8.4% 23% 40.1% 27.6%

Sample (N=123) 1.6% 15.4% 47.2% 35.8%

Table 1a: Age Dutch population compared to age sample

<20 20-40 40-65 65-80 >80

Dutch Population 23.1% 24,6% 35.5% 12.6% 4,2%

while at work. A pai-red-sample T test was conducted that showed there was no significant difference between the music-listening behavior of the respondents and colleagues (p>0.05). The reports showed that the music-listening behavior of the respondents and their colleagues did have a significant high positive correlation (r=.788, p<0.01) i.e. the music listening behavior is perceived to be rather similar. In graph 2 the boxplots of the music listening behavior of the participants and their colleagues is displayed. The participant’s 25th percentile is located at 18% of the time, the median is 60% of the time and the 75th percentile is 91%. For the colleagues of the participants the 25th _{percentile is located at 18%, the median is}

located at 51% for them and the 75th _{per-centile 90% of the time.}

Graph 2: Listening patterns of participants and their colleagues

How the participants listen to music at work formed the basis of the next quest-ion. The participants had to divide 100 points amongst five pre-described scenarios, indicating how they consumed the music while working. On average the participants used headphones in a private setting for 6,6% of the time, in a shared office setting music was consumed with headphones 11,1% of the time. The no headset scenario, refers to situation were music is consumed through e.g. a radio. Music without head-sets was consumed in a private setting 14,7% of the time, in a shared office setting the music was desirable 55,2% of the time and was not-desirable 12,4% of the time. The perceived degree of control was accessed using three different, but interrelated questions. Whether the participant could decide if he/she listened to music or not, he/she could decide what music was heard or whether they felt that they could ask others to turn off their music when this was perceived as annoying or disturbing. To test whether these questions captured a single dimension, the internal validity was measured using Cronbach α. The Cronbach’s alpha of the three questions measured

α = 0,781, and since the α > 0,7 the items could be considered to measure the same dimension. In general the participants are mildly positive (M= 3,53, SD= 1,19), about the perceived control over the music at work.

4.2 Factor analysis

In order to test whether an exploratory factor analysis can be used to discover latent variables from the items in the questionnaire, several preliminary analysis were conducted. First, the R-Matrix was produced (see Appendix B), which shows that the statements from the questionnaire are correlated with one another. All of the statements have correlations that lie in between .3 and .9, which suggests that these statements could be measuring aspects of the same underlying dimension i.e. latent variables (Field, 2009, p 628). By reducing interrelated variables to a smaller set of factors, factor analysis explains the maximum amount of common variance in a correlation matrix using the smallest number of explanatory constructs. The Kaiser-Meyer-Olkin measure verified the sampling adequacy for the analysis (KMO = .894), which is great according to Field (2009, p.659). Moreover the anti-image correlation matrix (Appendix B) shows that all individual KMO-values exceed the minimum of .5 (KMO > .717). Barlett’s test of sphericity X2 _{(253) = 2080.582, p < .001, indicates that}

correlations between statements are sufficiently large for an exploratory factor anal-ysis.

To determine how many factors should be included in the factor analyses, a factor extraction process will be used. The first part of the factor extraction process is to determine the linear components within the data set by calculating the Eigen values of the R-Matrix (see Appendix B). The analysis shows five factors that would be retai-ned using the Kaiser criterion of Eigen values greater then 1. However, the criterion is accurate when there are less than 30 variables and communalities after extraction are greater then 0.7 for sample sizes below 250. The average of the communalities (see Appendix B) is lower then this norm (0.658). The points of inflexion on the scree plot (see Appendix B), however does indicate that there are either two or five factors that can be extracted. In the final analysis five factors were retained. The five factors with Eigen values over the Kaiser’s criterion of 1 explain 65.8% of the variance.

The exploratory factor analysis is conducted using the average of both the simple and the complex task to find the perceived effects of music at work. The statements from Haake (2011) were replicated and several statements were added in order to investigate the positive and negative aspect of music at work (for more information see the methodology). Principal Axis factoring with oblique rotation (Oblimin with Kaiser normalization) was used to find five latent variables. Oblique rotation was used, because correlations between factors were expected. The correlation between the factors will be discussed later on; first the pattern matrix (table 2) will be explai-ned and discussed.

Table 2: Pattern Matrix

Factor:

Productive Negative Affect Communica-tion

Masking Biedt mij een ander

perspectief 0,873 Stimuleert mijn creativiteit 0,775 Inspireert en/of stimuleert mij 0,630 Verbetert mijn concentratie 0,423 Creëert een passende sfeer Maakt me onrustig 0,803 Maakt me chagerijnig 0,803 Irriteert me 0,802

Leidt me af van mijn werk 0,768 Blokkeert denkproces 0,732 Vermindert mijn concentratie 0,650 Maakt me passief 0,441

Verbetert mijn stemming 0,889 Helpt tegen verveling 0,764 Maakt me gelukkiger 0,731 Helpt moeheid tegen te gaan 0,643 Maakt me rustig Blokkeert communicatie -0,716 Creëert een gesloten werksfeer -0,687 Stimuleert gesprekken met collega’s 0,405 Vermindert niet-werk gerelateerde gedachten 0,668 Blokkeert omgevingsgeluiden -0,482 0,509 Helpt me doorwerken Eigen-value 9.14 3.47 1.44 1.35 1.22 % of variance 39,7 13,7 4,8 4,4 3,2 α ,903 ,916 ,887 ,454 ,555

The first component consisting of five statements is labelled ‘productive’ and ac-counts for 39,7% of the variance, the second component consist of seven statements is labelled ‘negative’ and accounts for 13,7% of the variance, the third component is labelled ‘affect and consist of four statements that together account for 4,8% of the variance, the fourth component is labelled ‘communication’ and accounts for 4,4% of the variance, the fifth and final component is labelled ‘masking’ and accounts for 3,2% of the total variance. For all components the reliability was checked using Cron-bach’s alpha and for the first three components the α > .7, which is considered to be

good (Field, 2009). However the components ‘communication’ and ‘masking’ both failed to reach this threshold. The reliability of ‘communications’ component can be increased to .732 if the statement ‘Stimuleert gesprekken met collega’s’ is dropped. Since the ‘masking’ component only consist of two statements this was not possible. 4.3 Inferential statistics

In this section several inferential statistics will be disclosed, which investigate how specific variables influence the music listening behaviour and/or the perceived effect of music at work, for the participants. The variables that will be investigated are age, gender, education, perceived influence and task complexity.

4.3.1 Influence on listening behaviour

To test whether males and females differ in the amount of time that they spend listening to music while at work, an independent-samples T test was conducted. The Levene’s test for equality of variance shows that the music listening behaviour of the males (M = 54,6, SD = 36,5) is not significantly different from that of the females (M = 53,4, SD = 38,1) F (121) = ,926, p > 0,05. The same test was conducted for age and the participants were divided in half using the median. The young age group consists of participants younger then 29 (N = 62, M = 56,1, SD = 36,5) and the old age group consists of participants older then 30 (N=61, M = 51,7, SD = 38,1). Levene’s test of equality of variance shows once again that this difference is not significant F (121) = .383, p > 0.05. The interaction between gender and age was also controlled, using the univariate analysis of variance and this shows that the interaction between age and gender has a small significant effect on the amount of time listing to music at work F (1, 119) = 2.786, p < 0.1. To test

The correlation between the time spend listening to music at work and age was tested using the Pearson correlation test and this revealed that the relationship be-tween these to variables was negative but not significant r = -.130, p > 0.05. However when females were excluded, age appears to be negatively related to the time spent listening to music r = -.277, p < 0.05. The scatter plot (see Appendix B) depicts this relationship.

The level of education showed to have no influence on the time spend listening to music at work F (3, 119) = 1.67 p > 0.05. The self-perceived control over the music al-so appeared to have no influence on listening behaviour F (12, 110)= .560 p> 0.05. The research design did no allow for analysing the influence of task complexity on the listening behaviour of the participants.

4.3.2 Influence on perceived effects

The influence of task complexity on the perceived effects of listening to music at work will be revealed, before disclosing the influences of several demographics on the opinion about music. In order to test whether task complexity had an influence on the perceived effects of listening to music while working, a paired-sample T test was conducted (see Appendix B). The test revealed that the responses on almost all the statements were significantly different, when the simple and complex scenarios are compared. Only two statements ‘muziek biedt mij een ander perspectief’ and ‘muziek vermindert niet werk gerelateerde gedachten’ proved not to differ between the simple and complex scenario. In table 3 the mean and standard deviation of both the simple and the complex task are showed.

Table 3 Mean and Std. Deviation of the simple and complex task

(1 = Volledig mee oneens, 5 = Volledig mee eens) Statements N=123

Muziek..

Mean:

(Simple task) Std. Deviation: (Simple task) (Complex task) Mean: Std. Detiation: (Complex task) ..Inspireert en/of

stimuleert mij 3,55 1,196 2,91 1,268

..Stimuleert mijn

creativiteit 3,23 1,234 2,76 1,208

..Biedt mij een

ander perspectief 2,66 1,222 2,64 1,117 ..Helpt me doorwerken 3,87 1,123 2,97 1,274 ..Creëert een passende sfeer 3,88 1,013 2,95 1,186 ..Helpt moeheid tegen te gaan 3,57 1,229 3,12 1,252 ..Vermindert mijn concentratie 2,38 1,098 3,00 1,379 ..Maakt me passief 1,77 0,857 2,19 1,027

In order to test whether the overall opinions about music differed in one situation compared to another, several statements are recoded so they all reflected a positive influence. The statements that needed to be recoded were statements: 7, 8, 14, 15, 16, 19, 20, 21 and 22. The Cronbach’s alpha revealed that the regular and recoded variables formed a reliable measure for the positive influence of music during both the simple task (α = 0,905) and the complex task (α = 0,931). This measure allows us to discover in which scenario music is perceived to be most beneficial, using a paired sample T test. The average response of both the simple (M = 3.63, SD = .625) and the

..Verbetert mijn stemming 4,06 0,969 3,46 1,147 ..Maakt me rustig 3,43 1,109 2,97 1,201 ..Maakt me gelukkiger 3,70 1,063 3,20 1,259 ..Helpt tegen verveling 3,76 1,174 3,09 1,318 ..Verbetert mijn concentratie 3,12 1,068 2,49 1,155 ..Irriteert me 1,73 1,049 2,28 1,400 ..Maakt me chagerijnig 1,60 0,981 1,97 1,254 ..Maakt me onrustig 1,86 0,986 2,37 1,363 ..Blokkeert omgevingsgeluiden 2,72 1,290 2,93 1,269 ..Vermindert niet werk gerelateerde gedachten 2,70 1,187 2,56 1,174 ..Leidt me af van mijn werk 2,03 1,024 2,98 1,385 ..Blokkeert denkproces 2,08 1,098 2,74 1,366 ..Blokkeert communicatie 2,31 1,253 2,59 1,220 ..Creëert een gesloten werksfeer 1,98 1,109 2,37 1,155 ..Stimuleert gesprekken met collega's 2,91 1,064 2,72 1,068

complex (M = 3.14, SD = .782) scenario were compared, which revealed that the opinion about music is significantly different due to the complexity of the task (t (122) = 7.418, p < .001).

Further paired-sample T test analysis show that women perceive music at work to be more beneficial (M = 3.73, SD = .595), than men (M = 3.51, SD = .643) during the simple task scenario and this difference proved to be significant (t (122) = -2.021, p < 0.05). Similar results were discovered for the complex task, women described music to be more beneficial (M = 3.17, SD = .769), than men (M = 3.10, SD = .801), but here the difference showed to be of no significance (t (122) = -.500, p > 0.05).

This test was also conducted for age, once again dividing the participants into two age groups. The division was based on the median and the young age group consists of participants under the age of 29 and the old age group consists of participants ag-ed 30 or up. The results were comparable with the ones found for gender. The young age group perceived music to be more beneficial (M = 3.77, SD = .425), than the old age group (M = 3.48, SD = .754) during the simple task. This difference proved to be significant (t (122) = -2.585, p < 0.05). While for the complex task the difference be-tween young (M = 3.21, SD =.674) and old (M = 3.07, SD = .878) showed to be not significant (t (122) = -.971, p > 0.05).

ANOVA was conducted to compare the effect of complexity on the opinion about music, amongst different levels of education. The ANOVA results show that there is a significant difference in both the simple (F (3, 119) = 5.695, p < 0.05) and complex (F (3, 119) = 6.840, p < 0.05) scenario. To reveal were this difference occurs Turkey HSD post hoc test was conducted. The post hoc test shows that for the simple task the participants with average education perceive listening to music to be more beneficial then the highly educated participants (Mean difference =.451, p < 0.01). However, for the complex task this effect is the other way around (Mean difference = -.753, p < 0.01).

Finally, a Pearson correlation test was used to check whether the amount of time listening to music at work was related to the beneficial perceived effect of music at work. The time spend listening to music at work seems to be positively correlated with the perceived beneficial effects of music as it shows a significant result of r = .327, p < 0.01.

5. Discussion

5.1 Theoretical implications

The results show that the sample is comparable to the Dutch working population in terms of gender ratio, but at first glance it seems to differ in terms of age and level of education. The average age of the sample (M = 34,9) proved to be different than that of the Dutch working population (M = 41,4). However, the age range of the sam-ple, which is 17 to 63, is comparable to the age range that is considered to be part of the Dutch working population, this is 15 to 65. The sample consist of relatively more higher educated then lower educated people compared to the Dutch population and the number of highly educated individuals can be considered to lay even higher. This is due to the formulation of the question that measured this variable and the age of the participants. The participants are requested to fill in the highest level of achieved education, this means that the younger participants who might still be studying filled in a lower level of education then they will probably have in the future. However the overall spread does indicate that it is similar to the Dutch population. In general the sample is considered to be comparable to the Dutch working population, since the gender ratio, age range and educational spread is deemed comparable. In addition, the participants are uniformly distributed amongst the occupational categories, thus depicting the opinion of the Dutch working population in general. Due to the money and time constraints that bound this research achieving a more accurate sample of the Dutch working population was not considered to be possible.

On average participants indicated that they listened to music more then half of he time while at work (53.9). Compared to Haake (2011) who investigated practices and experiences of listening to music in office setting in the United Kingdom, our findings suggest that the Dutch working population spend almost double the amount of time listening to music. The British employees indicated that they spend 36% of the week listening to music. The non-academic research that has been conducted on behalf of one of the largest employment agencies in the Netherlands, Randstad, support the notion that listening to music is considered to be important for the Dutch working population. The research (Randstad, 2012) suggests that 72% of the Dutch workforce requires both coffee and music to start of the day. However, the amount of music

that is consumed differs amongst participants (SD = 37,2) indicating that participants have different music listening behaviours. Given that 25 percent of the participants (N = 31) stated that less then one-fifth of the working day is spend listening to music (18%), however on the other hand the upper 25 percent indicate that they listen to music for more then 91% of the working week and 21 participants stated that they listen to music all the time. This supports the findings of Greasley and Lamont (2011) who identified two broad types of listeners: the less engaged and the high engaged.

The direct colleagues were perceived to have similar listening patterns to that of the participants themselves. In contrast to findings from Haake (2011) headphones were not the pre-dominant form of listening to music. The most amount of music was consumed without headsets e.g. radio, in shared office environments and was perceived to be desirable for 55,2% of the time or not in 12,4% of the time. Listening to music without headsets in private office settings were also more popular 14,7% of the time compared to with headsets in private setting (6,6%) and to shared settings (11,1%). This is however comparable with the findings of Randstad (2012) two-thirds of the week radio is selected and listened with colleagues. These findings seemed to suggest that listening to background music is not considered to be particularly anno-ying or disturbing (for the majority of the people). These might be because it masks off other office noises like Bull (2004), Loewen and Seudfield (1992) and Schlitmeier and Hellbrück (2009) suggest. Instead of music causing pain or discomfort as Coonan and Johnson (2002) imply, although this might still be the case for participants who indicated that the music was non-desirable. The perceived control of music might be of influence to this, as Kjellberg et al. (1996) suggest. In their research the perceived control over the sound (e.g. music) and the necessity of that sound was found to be related to the amount of annoyance and distraction that the sound caused. Since our participants indicated that in general they felt in control of the music being played in the workplace (M= 3,53, SD= 1,19), these negative effects might not have been of great importance to them.

In the next section the findings from the factor analysis will be discussed and the descriptions will be justified. The factor analysis extracted five different factors using the oblique rotation method, Oblimin with Kaiser normalization. The factors combi-ned explain 65,8% of the total variance. The first factor alone accounted for 39,7% of

the variance and has been labelled productive. The factor has a strong correlation with four statements that are all related to the self-reported performance of the participants. Our findings validate the findings of Haake (2011) and Oldham et al. (1995) that music is perceived to have a positive effect on work performance and help them to concentrate or be more creative. The second factor correlated with seven statements and accounted for 13,7% of the variance. The seven statements related to either negative performance or affect effects and has thus been labelled negative. The negative effects of music at work have been found in the researches from amongst others Huang and Shih (2011), Jett and George (2003), Ransdell and Gilroy (2001) and Speier et al. (2003). The third factor consists of four statements and accounted for 4,8% of the variance. This factor was labelled affect as the state-ments related to positive affect regulation (e.g. makes me happier or feel less tired). This corresponds with the finding of Bull (2004), Heye and Lamont (2010), Goethem and Sloboda (2011), Haake (2011) and Oldham et al. (1995). The fourth factor corre-lated with four statements and was labelled communication, since the statements are all related to working with colleagues and being approachable. Communications accounted for 4,4% of the variance. Haake (2011) discovered that people adjusted their listening behaviour in order to not disturb colleagues or appear unprofessional in front of clients and this might also be the case here and might also lead to more conversations amongst colleagues. The fifth and final extracted factor was labelled masking, consisted of two statements and accounted for 3,2% of the variance. The two statements refer to decreasing the non-work related thoughts and blocking off background noise. These masking effects of music have also been found by Loewen and Seudfield (1992), Oldham et al. (1996) and Schlitmeier and Hellbrück (2009). The first three factors all have a Cronbach’s alpha value above the 0,7 threshold meaning that the combined measurements seem to measure the same thing. The other two factors have not reached the threshold and should be investigated further in future research.

This research also investigated the influence of certain variables on the music lis-tening behaviour and the perceived effects. Gender appeared to have no influence to have no influence on listening behaviour and for age a negative trend was found but this failed to reach significance. However, when females were excluded age had

showed to have a negative correlation with the amount of time listening to music at work. This thus suggests that males listen to less music at work, as they get older and this influence has not appeared amongst women. Haake (2011) found no significant effects of age on listening time, however Bull (2004) found that younger people used personal listening devices (e.g. iPods) more then older aged individuals. The level of education and perceived control over the music had no influence on the listening be-haviour of the participants.

The complexity of the task was found to have significant influence on the opinion about music at work, except for two statements. In condition that were perceived to be simple for the person performing the task, music was rated to be more beneficial the compared to performing a complex task. This thus means that the complexity of the task influences the way music is perceived, confirming the findings of Speier et al. (2007). Women perceived music to be more beneficial then man did in the simple task but not in the complex. Similar results were found for age, younger participants indicated that they found music more beneficial then older participant in simple task but not the complex task. These finding might explain the mentioned correlation be-tween age and music listening behaviour for men. For the simple task participants with average education, perceived listening to music to be more beneficial then the highly educated participants. However, for the complex task this effect is the other way around. This could mean that participants with a higher IQ levels use music in a more rational and cognitive way as Chamorro-Premuzic and Furnham (2007) found in their research. Finally, the more participants listened to music, to more beneficial the music was perceived to be. This indicates that a learning curve exists for listening to music at work.

5.2 Managerial implications.

The findings of this research can also be useful applicable for managers as how to deal with music at work. People tend to differ in their opinion about the effects of music during work and this also influenced by other factors like task complexity. In general the employee should be free to decide to what extend music is incorporated in their work. Some participants want to work with their own music or shared music, but other prefer to have no music at all. Management could take this in account and