Beyond satisfaction or extortion : effects of the gig economy on business gig workers : effects of boundary conditions of gigs on the career satisfaction and work life balance of business gig workers

(1)

1

Beyond satisfaction or extortion:

Effects of the gig economy on business gig workers

Effects of boundary conditions of gigs on the career satisfaction and work life balance of business gig workers

Thesis author: Rob Willemse (1152524) Thesis supervisor: dr. Nick van der Meulen Second reader: prof. dr. Peter van Baalen

Univeristy of Amsterdam - Amsterdam Business School Msc Business Administration, Digital Business Track Word count: 19.849

(2)

2

Statement of originality

This document is written by Rob Willemse who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Abstract

The gig economy is a growing group of workers in the labour market. These workers earn their living via performing (short-term) assignments rather than (long-term) labour contracts. A gap exists in the field of research on the gig economy. On the one hand, an opinion is formed that gig work is good for the satisfaction of workers. On the other hand, an opinion is formed that the gig economy is driven by corporate firms wanting to pay less for labour. Business gig workers are workers who engage in a labour relation with any type or

organisational form to provide one or many services via alternative contracts. This research aims to fill in a segment of this gap by quantifying effects and focussing specifically on the group of business gig workers in The Netherlands. A cross-sectional study is performed via a survey strategy in collaboration with ZZP Nederland (n = 393). Furthermore, this study aims to confirm whether or not business gig workers experience a higher career satisfaction and work-life balance compared to traditional workers. In addition, the effects of several boundary conditions, such as percentage of gig income, length of gig and digital platforms, are tested in relation to the career satisfaction and work-life balance via multiple regression models. Overall, no differences are found between the work-life balance and career satisfaction of business gig workers and traditional workers. However, one effect of gig work earnings in relation to career satisfaction is partially supported. Furthermore, several additional effects are found, such as the negative relationship between perceived need for income and career

satisfaction & work-life balance. Concluding, the results suggest that business gig workers and traditional workers within the Netherlands seem similar in regards of career satisfaction and work-life balance.

(4)

4

1. Introduction

The creation of full time jobs used to be one of the main drivers of the U.S. economic growth prior to the dotcom crash of 2000 and especially the economic crisis of 2008. Full time jobs used to add between 2-3% of full time jobs each year. A declining trend emerged when after these crashes the creation of new full time jobs dropped to less than 1% in 2015 (Katz & Krueger, 2016). This decrease in full time job creation was not the only trend which shifted after the crashes. A drastic decline in long-term employment also emerged. These two factors seem to have driven the growth of what we now call the gig economy (Friedman, 2014; Mulcahy, 2016b). The gig economy is the growing group of workers who no longer

exclusively work via long-term contracts with a company, but rather earn some or all of their income via rapid short periods of work, called gigs, under flexible conditions (Kuhn, 2016). These alternative forms of work, such as freelancing, are not a new phenomenon. However these ways of working are empowered by a growing set of (digital) platforms, which create large-scale efficient marketplaces. These digital platforms further fuel the growth of the gig workforce (McKinsey & Company, 2016).

So far there is no consensus on the overall nature of the gig economy. Is it good for satisfaction or could it be seen as extortion? According to Mulcahy (2016a), the gig economy is initiated by companies and workers who increasingly look for more flexible and

independent types of work out of their own desire. In addition, gig work can liberate people who feel that they need to stay with the same company because of high legacy pay and benefits, resulting in a potential higher fit between people and organisations (P-O fit)

(Friedman, 2014). This P-O fit can be defined as the congruence between the attributes of the organisation and the personality of the worker (Cable & Judge, 1996). Similarly, a factor which can persuade people to go for gig work is freedom. Some gig workers prefer gig work because they do not want the commitment to go to the same office every day but want to

(8)

8

decide for themselves when they work, possibly even at home in pyjama’s (Friedman, 2014). This offers them autonomy to structure their own career path and time. Making it possible to, for example, have a healthy work-life balance while for example paying the rent and starting up an own company.

There is a stream of research, in spite of these arguments, which believes that the motivation is driven solely by firms wanting to pay lower wages and saving costs on benefits during business down-turns (Friedman, 2014). They say that there is a high possibility of people being exploited and being exposed to too many risks by companies and platforms offering gig work (Aloisi, 2015; Friedman, 2014; De Stefano, 2015). Gig workers forfeit a large amount of certainty when choosing gig work over regular work as it increases their uncertainty of income and exposes them to economic risks. Workers will need to work extra hard during gigs to compensate for possible future down-times (Friedman, 2014). They also need to work extra hard for additional insurances or at least reserves since they are not applicable for social security which is designed for traditional jobs. The same goes for sick days and other legislations which differ per country (Kuhn, 2016). Concluding, the easiest way to receive a pension, employer-benefits, protection against workplace injuries, a social safety net and in some cases even health insurance is a traditional labour contract and not gig work (Mulcahy, 2016a; Leighton, 2016).

However, the distinctions between different groups within the gig economy are slowly becoming larger and clearer. Mainly the intention to join the gig economy yourself versus the forced entry due to financial need divides the workers into different groups (Mulcahy, 2016a; Rosenblat, 2016; McKinsey & Company, 2016). Furthermore, McKinsey & Company (2016) surveyed over 8,000 people in the U.S. and six countries of the EU-15 on the gig economy. Within this survey they measured the type of gig work performed by the gig workers. They concluded that most gig workers perform some type of services for consumers or companies.

(9)

9

From the 20-30% of the workforce who work via gigs, only 3% of them sell good and 1% rent out assets, leaving 16-26% who provide services.

Another aspect of interest is the length of the gig assignments. For example, most UBER rides will only last for minutes or hours maximum. Gig work based on pay per sale is not focussed on time duration but on results, gig workers can choose to quit every second. But service and business related gigs are not as homogenous, since the contracts usually include a timeframe of certain months, years or even a rolling notice. McKinsey & Company (2016) chose to use a cut-off point of 12 months in their dataset. As this is a consultancy paper their cut-off point is not scientifically proven. However using 12 months as a cut-off point does provide a practical and feasible boundary to exclude abnormal cases, for example disguised employees. Ideally, a scientifically proven grounded cut-off point should be established to further strengthen practical cases. Because what if the positive effects, such as satisfaction or a good work-life balance, of gig assignments already stop after 6 months or if they might even last for 18 months? This can lead to implications which could change the definition of gig workers as a whole. No research has been published on the length of gigs to date. This research will, amongst other things, further explore this train of thought.

Also important is recent media footage, including the Wall Street Journal, claiming that the growth of the gig economy is hyped and that there is in fact no gig economy. Possibly leading to a non-existing field of research (Zumbrun & Sussman, 2015). Kuhn (2016)

acknowledged these articles and has showed strong arguments to prove these news articles incorrect. She provided insights that the filing of tax forms on additional income has

structurally risen since the dotcom- and economic crashes. Furthermore people with regular jobs, who earn on the side via online platforms or via other alternative forms of work, do not necessarily perceive themselves as working on more than one job. This could raise the debate that labour statistics classifying people as either traditional- or alternative labour relation are

(10)

10

missing much data and insights on the actual size of the gig economy (Katz & Krueger, 2016; Kuhn, 2016; McKinsey & Company, 2016; Donovan, Bradley, & Shimabukuru, 2016).

This is further supported by the data provided by the Bureau of Labour Statistics in the article of Friedman (2014). The graphs in his paper only describes two factions, traditional arrangements and alternative arrangements. It is clear that even within this limited data the alternative arrangements are increasing, from 10.7% in 2005 towards 13.4% in 2013. But what about workers doing both arrangements? The gig economy could possibly be bigger than expected if this group is included. Basing arguments purely on existing labour statistics, which classify workers as either traditional labour or alternative labour, should thus not be accepted as reality anymore. The reality is more complex. Recent estimates which include both arrangements and combinations have reported that the size of the gig economy is between 20% and 30% over the U.S. and EU-15 in 2016 (Katz & Krueger, 2016; McKinsey & Company, 2016). If this trend continues it could well be that the gig economy could become the new standard for work in the future. Perhaps going as far as degrading traditional work to a minority.

Concluding, the growth of the gig economy, in importance, practical numbers (shown in the increasing size) and academic contribution in the form of publications (Google Scholar found 86 hits in 2014, 173 in 2015 to 605 in 2016), is rapidly growing and gaining

momentum while traditional labour is stagnating (De Stefano, 2015). Unfortunately, the gig economy has downsides as well. These mainly consist out of concerns on extortion and increased financial risks. However, it is possible for some groups to benefit from this development, possibly via an improved work-life balance and more career satisfaction. In addition, the largest segment of gig workers is working via services. This research will

therefore focus on the gig workers in this segment working for businesses. They will be called business gig workers from here on. In addition, this research will investigate different effects

(11)

11

which are mentioned on this group in comparison to traditional labour. The following research question is identified:

“What effects do boundary conditions of gigs have on the work life balance and career satisfaction of business gig workers?”

This research will contribute to science as well as practice in multiple ways. Firstly, the results will shine light on which factors can improve the quality of life of business gig workers, by explaining a part of the variance of career satisfaction and work life balance. Secondly, from an organisational point of view the results will show patterns of flexible and satisfied business gig workers. These criteria could help in selecting those business gig workers who benefit from working in the gig economy, creating a win-win situation. Thirdly, comparing business gig workers with traditional employees could also aid in practical

researching on deciding if organisations would be better off with gig workers in place of traditional employees or vice versa. Fourthly, this research will contribute to literature on the gig economy by confirming if business gig workers actually are more satisfied and balanced then regular employees. Finally, it will analyse the lengths of the gigs possibly resulting in a grounded cut-off point of effective gig length for business gig workers in place of the arbitrary use of 12 months.

The data used for this research are collected via a survey and analysed quantitatively. The survey is distributed under the associated business gig workers of ZZP Nederland. This is an association of independent workers, under which business gig workers, in the Netherlands. The comparison group of traditional employees is collected via the LinkedIn and Facebook platforms. The effects of the business gig workers are tested via multiple hierarchical regressions and other statistical tests. All comparisons between groups are done via T-tests and ANOVA where possible.

(12)

12

Next, the theoretical constructs will be discussed and linked to hypothesis in a

literature review which will result in a conceptual model and hypothesis overview. Following, all steps taken in the data collection and statistics are put apart. Afterwards the results of the statistical tests and a discussions section are located where the results are intertwined with the theoretical insights. Finally a conclusion was written at the end of this paper.

(13)

13

2. Literature review

This chapter elaborates and defines the constructs which are relevant to the research question. First, the gig economy as a whole and business gig workers are defined. Afterwards, the different constructs, work-life balance, career satisfaction, length of gigs, income related constructs and digital platforms, are set forth developing hypothesis which are directly

formulated in the text. Finally, a conceptual model and an overview table of all hypothesis are shown at the end of this chapter.

2.1. Defining the gig economy

The gig economy in general is the disaggregation of work which is offered via alternative arrangements. Different researchers define these alternative arrangements differently, for example consulting projects, freelance assignments, contracted opportunities, independent contractors, online task execution, ride hailing and many more (Rosenblat, 2016; Mulcahy, 2016b; Kuhn, 2016; Katz & Krueger, 2016). Within this research the definition of the gig economy will include all arrangements which are not traditional labour arrangements conform national labour contracts.

These gig workers are different to traditional workers in regards to their highly

specific tasks which often are planned within a shorter time period together with a narrow and finite set of obligations (Lemmon, Wilson, Posig, & Glibkowski, 2016; Kuhn, 2016).

Furthermore, there is no clear consensus on the boundaries of the gig economy per definition and it is therefore used in combination with a variety of other keywords such as crowdwork, work-on-demand, on-demand economy, digital labour platforms and the sharing economy (Aloisi, 2015; De Stefano, 2015; Huws & Joyce, 2016; McKinsey & Company, 2016).

(14)

14

2.2. Defining business gig workers

The introduction chapter mentions that the majority of gig work is done via providing services and that this research would focus specifically on business gig workers (McKinsey &

Company, 2016). However, there is no published research which focussed solely on the business gig workers yet. Therefore they have not been defined in peer-reviewed articles. The definition below is built up out of different existing definitions to properly describe the unit which is the business gig worker.

Important to note is the distinction between peer-to-peer gig workers and business gig workers. Peer-to-peer platforms, such as UBER, connect one person with another person (Aloisi, 2015). They can provide a wide variety of services for example offering cleaning services, putting together an IKEA closet, mowing the lawn or filling in a tax return form. Essential is the short term relation between two people, peers, where one person pays the other for a gig. The interactions can take place via a digital platform. In contrast business gig workers engage in short term alternative work arrangements in which they provide their services to businesses only. Whereas a business and a person are not the same and thus not equal this cannot be a peer-to-peer relationship. However, there are digital platforms which connect workers and businesses for business services such as Upwork or Freelancer (Aloisi, 2015; Donovan et al., 2016). Thus, business gig work can be facilitated via digital platforms but it excludes peer-to-peer agreements.

Concluding, a business gig worker is a worker who engages in a labour relation with any type of organisational form to provide one or many services via a short term alternative work arrangement with a narrow and finite set of obligations, possible via both non-digital and digital platforms. This definition further excludes the small portion of the gig economy which sells goods or rents out assets (McKinsey & Company, 2016).

(15)

15

2.3. Work-Life Balance

Both scientific work and regular press releases have noted that the gig-economy provides workers with an unprecedented degree of freedom and flexibility (Friedman, 2014; Kuhn, 2016; Mulcahy, 2016a; Rosenblat, 2016). This often mentioned flexibility led to the inclusion of the construct of work-life balance (WLB) within the research question. A definition of WLB that has reached relative consensus is “an individual assessment of how well her or his multiple life roles are balanced” (Haar, Russo, Suñe, & Ollier-Malaterre, 2014, p. 361). In essence, the WLB is dependent on the subjective perception of someone on him- or herself. Therefore, WLB does not consist out of an objective ideal outcome. The ideal WLB is different for everyone and varies between people in general and life-stages in specific (Chandra, 2012).

Süss & Sayah (2013) qualitatively explored the WLB of contract workers in Germany, which are closely related to business gig workers due to their limited and short-term

obligations towards their employer. The differentiating factors between traditional employees and contractors, which they encountered in their interviews, are; long working hours, much travelling and flexibility requirements of place and time. They noted that for most contract workers the boundary between work- and personal life is slowly dissolving. However, this does not necessarily have to have a negative influence. These factors can also be seen as enablers via which they can positively influence their own work and private schedule to create a better WLB. These possibilities converge with the diverse possibilities of the business gig workers as the gigs offer more control over the planning of time than a traditional jobs do.

Many gig workers seek this flexibility and find it appealing to work outside the standard 9-to-5 schedule (Lobel, 2016). They can set their own agenda when to work while earning a stable living (Burtch, Carnahan, & Greenwood, 2016). This implies that business gig workers can exploit this flexibility by planning their work in such a manner that they free

(16)

16

up time and resources for other activities, such as other assignments or private life, without directly sacrificing their income. In addition, a recent survey of Huws & Joyce (2016) (n = 2,234) collected data on the reasons why people want to engage in gig work. A majority of the gig workers who are looking for new gigs (88%) are looking for gig work which can be done at home or wherever they prefer. Again, offering them more control over their work and personal lives.

Concluding, working as a business gig worker offers them more options to influence their own WLB, which traditional employees do not have or have less, be it working breaks between gigs, alternate assignments between full-time and part-time gigs or working at home.

H1a There is a significant difference between the WLB of business gig workers and traditional employees, where business gig workers have a higher WLB.

H1b There is a positive relationship between participation (% of working hours that is gig work) as business gig worker and WLB, where more participation as business gig worker leads to a higher WLB.

2.4. Career Satisfaction

More than 65% of the Americans are structurally not engaged in their work-life as employee (Gallup, 2017). This engagement issue might be one of the reasons why the business gig workers choose gig work over secure, regulated, labour relationships. Traditional work arrangements do not seem to work equally well for all people and can thus lead to a general dissatisfaction and disengagement of work (Mulcahy, 2016a). As there is no research exploring the relation between business gig workers and career satisfaction yet, research on self-employment and career satisfaction is further used as a proxy to argue for this

(17)

17

Benz & Frey (2008) concluded that self-employed experience a higher degree of satisfaction from their work than the compared group of traditional workers. They explain their results by describing the concept of procedural utility, as the self-employed have more self-determination and freedom offering them utility to make their own decisions, similar to that of the business gig worker. Andersson (2008) concluded a very similar result in her Sweden based research. She concludes that self-employed workers have an overall increased life satisfaction in addition to higher levels of work related satisfactions. Both Andersson and Benz & Frey agree that the self-employed might work more hours than an average wage-earner but that the empowerment of deciding their own jobs, working hours and being a boss over themselves explains their higher levels of career satisfaction. This so called ‘free worker’ who consciously chooses to work via self-employment thrives on the independence and will develop themselves faster (Guest, 2004). However, a note can be placed by this image in which it looks like everyone should become self-employed. To benefit from these effects Guest (2004) explains that the self-employed needs to actually prefer the temporary contract above the permanent contract or else the effects are not as strong. This phenomenon has similarities with another topic and will be further explored within the topic of perceived need for income.

On the other hand, there are traditional workers who do engage in deviations from the regular traditional fulltime employment standards to tailor work to their own needs. However, deviating from the standard full time traditional labour contract often results in workers accepting sacrifices on pay and career perspectives (Leighton, 2016; Omar, 2013). These deviation are seen as, for example, part-time working, homeworking or job sharing. Usually the practices on pay and career progression are not designed for these deviations of the 40+ hours workweek. As the policies focus on traditional full-time career paths. Leighton (2016) notes that in these cases workers are often better served in independence than compromise

(18)

18

within their traditional arrangements. The choice for independence results in a sacrifice of employee status and security rather than the earlier mentioned sacrifice in pay and career perspective. Choosing this independent status, for example business gig worker, means that these workers are not full-time committed to companies on the long term anymore. This frees them up to seek the working conditions which they require to be successful, be it part-time, full-time or anything else. Or they can set up a business for themselves on the side. Finally, Friedman (2014) interviewed multiple of these gig workers and noted the following:

“…perhaps most surprisingly, many of them love it” (Friedman, 2014, p. 173)

Concluding, traditional work does not fit everyone’s needs. The autonomy which business gig workers have is similar to that of self-employment and as such the effects might be comparable. Therefore business gig workers might experience higher levels of career satisfaction in general. Especially workers who do not fit the corporate 40+ hours workweek will most likely experience a positive effect in their career satisfaction when engaging in business gig work.

H2a There is a significant difference between the career satisfaction of business gig workers and traditional employees, where business gig workers have a higher career satisfaction.

H2b There is a positive relationship between participation (% of working hours that is gig work) as business gig worker and career satisfaction, where more participation as business gig worker leads to a higher career satisfaction.

2.5. Length of gigs

An arbitrary point within the gig economy is the length of the assignments of business gig workers. Business gig workers are classified, among others, due to the legal aspects of their work arrangement. However, what if a business gig worker has a full-time and long-term

(19)

19

alternative arrangement with only one business? This worker could legally be working in the gig economy. Hence, it could be that this worker is working as a disguised employee (De Stefano, 2015; Leighton, 2016). If a business gig worker is practically working as a traditional employee under an alternative arrangement the aforementioned hypothesis concerning the increased WLB and career satisfaction might be invalid for this worker. As this worker is trying to be a traditional employee. This would result in a curve like relation between the length of the gig assignments and the WLB & career satisfaction, where the positive effect would start to decrease after a certain length of gig. In the meanwhile this worker will still have the increased risks of working in the gig economy. This could result in a highly

undesired situation. Ideally, the gig worker should work via a traditional arrangement in this case. However, some authors choose not to focus on the length of the assignment but choose to focus on the legal framework surrounding the assignment as the disguised employee is a shadowy topic without clear boundaries (De Stefano, 2015; Aloisi, 2015). A consultancy paper used a hard cut-off point of maximum 12 months to exclude gig workers who seemingly still work as long-term workers out of their data sets (McKinsey & Company, 2016). This cut-off point could of course practically eliminate most of the disguised

employees out of the data set, however why is it not 11 or 13 months? There is no scientific argumentation on the length of gig work yet.

Concluding, it is unknown what effect the lengths of the gig work has on the outcome variables WLB and career satisfaction so far, it might be possible that people who feel forced to employ themselves in gig work try to recreate a long-term traditional relationship with a single business while legally being a gig worker, creating a disguised employee with no increase in WLB and career satisfaction.

(20)

20

H3a There is a curvilinear relationship between length of gigs and WLB, where the WLB initially increases with the length of the gigs but after an optimal point it eventually starts to decrease, creating an inverted u-shape relation.

H3b There is a curvilinear relationship between length of gigs and career satisfaction, where the career satisfaction initially increases with the length of the gigs but after an osssptimal point it eventually starts to decrease, creating an inverted u-shape relation.

2.6. Percentage of income via gigs

A large scale survey (N=2,235) performed in the UK led to the finding that the largest group of gig workers do so for supplemental income (49%). They defined supplemental income as income from gig work between 1% and 50% of the total income. Furthermore, a limited amount of gig workers earn 100% of their income (5%) via gigs. Of the survey, 23% did not know or did not want to share how much they earned and the remaining 23% earned between half and all of their income via gig work. They concluded that around one third of the gig workers earn most of their income via gig work (Huws & Joyce, 2016). These results correspondence with the earlier mentioned survey of McKinsey & Company (2016) which notes the supplemental income group as the biggest (54-61%). The variance in percentages in this consultancy paper is explained by different sizes in different countries. However, this survey excluded entries which did not know or did not want to share information concerning their income, leading to bigger shares in the other segments. They conclude that around 28-46% of the workers within the gig economy earn their primary income via gig work

(McKinsey & Company, 2016). However, primary source of income and more than 50% of income are slightly different definitions. This might be one of the reasons why the

percentages differ slightly. Nevertheless, it does illustrate that most workers earn income via a combination of income streams which cannot exclusively be classified as only gig work or only traditional work, in essence becoming hybrid workers.

(21)

21

In addition to McKinsey & Company (2016) (54-61%) and Huws & Joyce (2016) (49%) Berg (2015) also notes in the results of his survey on Amazon Mechanical Turk that 45% of the gig workers do so to supplement income from different jobs. Earning

supplemental income, in addition to the before mentioned group earning primary income, can possibly be seen as a separate segment within the gig economy. Supplemental gig work differs mostly in that it can enable the utilization of underused assets or time to create an efficient supplement on the existing, possibly traditional, income (De Stefano, 2015). An example could be a business professor offering consultancy or guest speaker services during evening hours and weekends.

Concluding, the results of the before mentioned surveys all report different patterns of income earned by gig workers. The main groups consist out of earning primary income and supplemental income. The classification of gig workers from McKinsey & Company (2016) will be set forth in the subsections below. These classifications will structure the link between the income related factors & classifications and the outcome variables WLB & career

satisfaction. The hypothesis concerning the percentage of income will follow after the classifications.

2.6.1. Casual earners and free agents

One of the most prominent promises of the gig economy is the flexibility to work in spare time and the autonomy to decide how much hours to work. This is a main motivational driver for people, such as retirees, professionals or empty-nesters, who start to do gigs for

supplemental income, in addition to their jobs (Rosenblat, 2016). This group which does gig work by their own choice for supplemental income is called “casual earners” (McKinsey & Company, 2016).

In addition to casual earners there is a group of gig workers called “free agents” (McKinsey & Company, 2016). This group of gig workers also values the freedom and the

(22)

22

independent character of gig work. However they prefer to earn their primary income via alternative working arrangements in place of only a supplement. This group participates out of their own choice as well. They are the people who feel liberated by leaving their full-time job and who benefit the most of all from the gig economy (Mulcahy, 2016a). The results of the survey of McKinsey & Company match with this statement noting “Free agents across both EU-15 and the United States report the highest levels of satisfaction with their work lives of any other group in the survey” (McKinsey & Company, 2016, p. 65). This satisfaction was measured on 14 different dimensions including flexibility of schedule, recognition, income level and opportunities to learn, grow and develop. These dimensions include aspects both of WLB and career satisfaction which are computed into a single satisfaction variable in their analysis. Concluding, the effects of the gig economy should be strongest in the group of free agents.

H4a There is a positive relationship between percentage of income via gigs and WLB, where a higher percentage of income via gig leads to a higher WLB.

H4b There is a positive relationship between percentage of income via gigs and career satisfaction, where a higher percentage of income via gigs leads to a higher career satisfaction.

2.6.2. Perceived need for income: financially strapped and reluctants

Furthermore, gigs often have less requirements then full-time positions. Companies are not liable for these workers and thus feel less necessity to maintain high standards and costly recruiting procedures. This offers new working opportunities for people who lack education, who are underemployed or who have criminal records (Donovan et al., 2016; Rosenblat, 2016). The groups of gig workers, within the classification of McKinsey & Company (2016), who benefits from these low standard are called “financially strapped” for supplemental income and “reluctants” for primary income. They participate in gig work out of necessity for

(23)

23

primary or supplemental income and not by their own choice due to intrinsic motivation. They would rather not participate in the gig economy and have a traditional working arrangement to make ends meet, but somehow they are not able to. Of the gig economy, 30% consists out of financially strapped and reluctants. While this is a minority within the gig economy, the size of this problem is still strikingly large (McKinsey & Company, 2016). In other words, reluctants and financially strapped experience a high perceived need for the income which they earn with gig jobs. As they would rather not participate they feel obliged to do gig work for the income. The following quote, while positively formulated, expresses the discrepancy between free agents/casual earners and financially strapped/reluctants well: “While gig work is a necessity for some, it is a luxury for others” (Rosenblat, 2016, p. 1.). Further results on these two groups describe that financially strapped and reluctants have an overall lower satisfaction. Strong effects of their dissatisfaction are measured in their income, hours worked and overall benefits (McKinsey & Company, 2016). Therefore different effects on people’s lives in relation to gig work could be contingent on the perceived need for income.

Concluding, necessity and a perceived need for income seem to drive 30% of the gig workers to participate in gig work. As they do not prefer alternative working arrangements it seems logical that they are not as satisfied as the hypothesized free agents and casual earners. Furthermore, claims made on the increased satisfaction of the gig economy as a whole feel rather counter intuitive when 30% of the gig workers would rather not participate in the gig economy at all and who would, logically reasoning, be more satisfied within a traditional job. It therefore looks to be that the perceived need for income moderates the aforementioned relation between percentage if income and the outcome variables, where a high perceived need for income leads to lower scores on WLB and career satisfaction.

H5a The positive relationship between percentage of income via gigs and WLB is moderated by perceived need for income, so that the relationship stays positive with a

(24)

24

low perceived need for income and that a high perceived need for income leads to a negative relationship between percentage of income via gigs and WLB.

H5b The positive relationship between percentage of income via gigs and career satisfaction is moderated by perceived need for income, so that the relationship stays positive with a low perceived need for income and that a high perceived need for income leads to a negative relationship between percentage of income via gigs and career satisfaction.

2.7. Digital platforms

Online gig platforms, such as UBER, Upwork or Amazon Mechnical Turk, are nearly known to all people as platforms for gig work (Aloisi, 2015). The main functions of these platforms is creating an infrastructure in which they can connect supply and demand. Furthermore, these platforms enable interactions in which the supply and the demand can communicate. In short, they can be seen as the brokers of gigs who create a three-sided arrangement (Donovan et al., 2016; Aloisi, 2015). These three sides consist of the platform, requester (business or peer) and gig worker. This three-sided arrangement can be fundamentally different from non-digital platform gig work. The third party, the platform or intermediary, is optional offline and can be included only if necessary. Whereas within the three-sided arrangement via digital platforms this third party is required, as it facilitates communication, payments and structures the work. The digital platform requires commissions in return for these services (Horton, Kerr, &

Stanton, 2016; Donovan et al., 2016). The platforms facilitate a crucial role in online gig work and they can therefore not be excluded from this way or working in the current situation.

The digital platforms have created a powerful position for themselves. A consequences is that the commissions can be high, for example UBER claims between 20-30% of the

revenue created by the drivers (Aloisi, 2015). Furthermore, limited compensation of expenses should be included in the consideration as well, in addition to possibility of high

(25)

25

commissions. UBER has been sued multiple times on compensational issues as they do not provide mileage- or toll costs expenses for example (De Stefano, 2015). Another example can found in the Amazon Mechincal Turk platform. 90% of all the tasks which are offered on this platform pay less than $0.10 per task. Both the low compensation of costs necessary to

complete a task or a low pay per task make it difficult to gain a high hourly wage if doing gigs in the lower skill segments of these digital platforms (Webster, 2016). However, only 8% of all the people in the working population have earned income via such a platform (Rosenblat, 2016). Within the gig economy 15% of all gig workers frequently use online platforms for their gig work (McKinsey & Company, 2016). These statistics both suggest that gig work without the usage of an online platform is still, by far, the dominant standard in the gig economy.

Concluding, the involved parties within gig work via digital platforms seem to be structurally different from the offline types of gig work. This different set-up leads mainly to lower compensations and the permanent powerful role of the digital platforms provider, limiting the freedom of the gig worker. Therefore negative effects are to be expected when using digital platforms.

H6a There is a negative relation between business gig work via digital platforms and WLB, where working via digital platforms leads to a lower WLB.

H6b There is a negative relation between business gig work via digital platforms and career satisfaction, where working via digital platforms leads to a lower career satisfaction.

(26)

26

2.8.Conceptual model

The conceptual model is displayed below as figure 1. Note that the conceptual model includes all 12 hypothesis out of the literature review (H1a until H6b). However, the order of the hypothesis is altered in order to create a visually clear model. The control variables are not included in the conceptual model as they are further discussed in chapter 3 concerning data, measures and methods.

(27)

27

2.9. Overview of hypothesis

H1a There is a significant difference between the WLB of business gig workers and traditional employees, where business gig workers have a higher WLB.

H1b There is a positive relationship between participation (% of working hours that is gig work) as business gig worker and WLB, where more participation as business gig worker leads to a higher WLB.

H2a There is a significant difference between the career satisfaction of business gig workers and traditional employees, where business gig workers have a higher career satisfaction.

H2b There is a positive relationship between participation (% of working hours that is gig work) as business gig worker and career satisfaction, where more participation as business gig worker leads to a higher career satisfaction.

H3a There is a curvilinear relationship between length of gigs and WLB, where the WLB initially increases with the length of the gigs but after an optimal point it eventually starts to decrease, creating an inverted u-shape relation.

H3b There is a curvilinear relationship between length of gigs and career satisfaction, where the career satisfaction initially increases with the length of the gigs but after an optimal point it eventually starts to decrease, creating an inverted u-shape relation. H4a There is a positive relationship between percentage of gig income and WLB, where a

higher percentage of gig income leads to a higher WLB.

H4b There is a positive relationship between percentage of gig income and career satisfaction, where a higher percentage of gig income leads to a higher career satisfaction.

(28)

28

H5a The positive relationship between percentage of income via gigs and WLB is

moderated by perceived need for income, so that the relationship stays positive with a low perceived need for income and that a high perceived need for income leads to a negative relationship between percentage of income via gigs and WLB.

H5b The positive relationship between percentage of income via gigs and career

satisfaction is moderated by perceived need for income, so that the relationship stays positive with a low perceived need for income and that a high perceived need for income leads to a negative relationship between percentage of income via gigs and career satisfaction.

H6a There is a negative relation between business gig work via digital platforms and WLB, where working via digital platforms leads to a lower WLB.

H6b There is a negative relation between business gig work via digital platforms and career satisfaction, where working via digital platforms leads to a lower career satisfaction.

(29)

29

3. Methods & measures

The methods & measures of this research are described in this chapter. Furthermore, this chapter will elaborate on the research design and research strategy. After which the survey related topics, measures, populations and control variables are put forth. The chapter ends with the data collection and the sample characteristics.

3.1. Research design

This research is designed with a mind-set based on interpretivism. This is the belief that reality is complex and that therefore not every phenomena can be developed into definitive “law” (Saunders, Lewis, & Thornhill, 2009). This can for example be seen in the construct of WLB and perceived need for income where the outcome depends on the subjective perception of oneself. Furthermore, the research adopted a deductive approach in order to investigate the effects and relations between the relevant variables. The hypothesized relations are found in a combination of different fields of research as described in the literature review.

The generally accepted standard of 95% certainty was handled in all statistical testing (Field, 2013). Only the literature is available to argue for cases of causality, due to the limited scope and resources of the research it was not possible to add experimental- or a longitudinal elements (Saunders et al., 2009). Accordingly, this research can be classified as an

explanatory correlational research. The timeframe of this research project, 5 months, in combination with the available resources, only personal contributions of the researcher, make a longitudinal design undesirable. The choice was made to adopt a cross-sectional data collection in which data from one time period is collected and analysed.

3.2. Research Strategy

The primary data were collected via a survey strategy, using a mono method design. This choice was made with the knowledge that surveys do have strong and weak characteristics.

(30)

30

On the one hand the use of a survey will reduce observer bias and observer errors (Saunders et al., 2009). As there are no different observers they cannot ask questions in different ways or interpret the findings differently. Otherwise, surveys require clear formulated and validated questions in combination with random sampling in order to infer generalizable results from the dataset (Field, 2013). The measures which require validation are all adopted from validation research papers in order to ensure internal validity. The measures can be found further in this chapter.

In addition, random sampling can be difficult when the sampling frame, a list of the whole population, is not available when selecting a sample (Saunders et al., 2009). This research contains two different populations. The first population is all traditional workers, employees with labour contracts, within the Netherlands. This population was mainly used for comparisons with the second population. The second population is all business gig workers within the Netherlands. This includes all people working via alternative contracts for

businesses within the Netherlands, as explained in the literature review. This population was used to compare with the first and to test different hypothesis within this population.

However, these population are not mutually exclusive. There is a segment which fits in both populations, doing both types of work. This group was included in the business gig worker selection, as this group is the focal point of this study. The conceptual model includes

variables via which distinctions are made for this segment, percentage of income via gigs and percentage of work hours that is gig work. However, there are no sampling frames available of these populations. The lack of a sampling frame often leads to convenience sampling where there is no equal chance for all the units in the population to be selected. Generalizability is not guaranteed, but possible, when executing non-probability sampling (Saunders et al., 2009). Therefore this research deploys self-selection sampling techniques in order to avoid

(31)

31

convenience sampling and improve the chances of generalizability. These techniques are discussed further within this chapter.

Furthermore, the survey was performed as an online questionnaire, due to the obvious practical benefits such as language selection, automatic data collection and optimized

question paths for the two populations. All items are scored on a 5-point Likert scale, in correspondence with their validation papers. The data resulting from the Likert scales are treated as interval level data in this research in accordance to the finding of Norman (2010). He concluded the following: “Parametric statistics can be used from Likert data … with no fear of coming to the wrong conclusion” (Norman, 2010, p. 7). Factual items such as

percentage of income via gigs or percentage of hours that is gig work are not adopted due to their straightforward and descriptive nature.

Multiple steps were taken to guarantee the anonymity of the respondents. Firstly, all data generated by Qualtrics which conflicts with anonymity were disabled. The disabled data includes geo location, automatic name generation and IP address. The researcher wished to provide the respondents with an incentive for participation and to offer an option to send the final paper to the respondents. A second survey was built in order to provide these

possibilities without linking the e-mail addresses to the data. The end of the first survey thanked the respondents for participation and reminded them of the incentive in the second survey. If they click next, the first survey ended and an automatic redirect sent them towards the second survey. Via this method a second data set consisting of the e-mail address for the incentive and final paper are gathered. 65% of the respondents left their e-mail address for this incentive when the data collection was completed.

Finally, within the aforementioned survey strategy and research design this research aims to enrich the design with a second dependent variable to add robustness to the results and conclusions. This conceptual design will compensate for the decreased robustness of the

(32)

32

cross-sectional data collection. When two dependent variables have relationships with the multiple independent variables, it implicates a bigger and robust overall trend within the sample.

3.3. Measures

All measures of the variables are put forth in this section. As noted before, all multiple-item constructs using statement questions were measured using Likert scales. These scales will consist out of a 5-point Likert scale, ranging from 1 = “strongly disagree” to 5 = “strongly agree”. However, a number of more descriptive variables do not make use of the Likert scale. These items asked for number of hours, days, months or year and another for a percentage.

Work-Life Balance was measured using Brough, Timms, O'Driscoll, Kalliath, Siu, Sit and Lo's (2014) four-item measure, including “I currently have a good balance between the time I spend at work and the time I have available for non-work activities” (Cronbach’s α = .93).

Career satisfaction was measured using Greenhaus, Parasuraman, & Wormley's (1990) five-item measure, including “I am satisfied with the success I have achieved in my career” (Cronbach’s α = .88). This measure was recently validated a second time by Spurk, Abele, & Volmer (2011) (Cronbach’s α = .84).

Percentage of hours gig work was measured by asking two descriptive questions, “Could you please indicate for how many hours per week you are currently employed via labour contract(s)? (This concerns the contractually agreed upon number of hours, excluding overtime.)”and “Could you please indicate for how many hours per week you are currently employed via alternative contract(s)? (This concerns the contractually agreed upon number of hours, excluding overtime.)”. After which the percentage is computed and added as a new

(33)

33

variable in the dataset. In addition, the actual worked hours are also included as questions for further analysis.

Length of gigs was measured by asking one descriptive question, “(Choose the best fitting option and fill in the average) Usually my work assignments (no labour contracts) last…” giving the opportunity to choose hours, days, months or years. A text box added the option to fill in how many hours/days/months/years. The answers will be standardized into months for all questions for the further usage.

Digital platform was measured by asking one descriptive question, “Do you acquire the majority of your work assignments via digital platforms?”. An additional open question is added, exploring which platform they prefer.

Percentage of gig income was measured by asking one descriptive question, “What percent of your income, on a yearly basis, do you earn via alternative work assignments (f.e. freelancing, independent contractor, consultancy project, etc.) which are no labour

contracts?”. The respondent could then answer by moving a slider between 0 and 100,

representing gig income. This approach was chosen, in contrast to asking the factual earnings and computing a ratio, as providing data on income is a sensitive topic. Asking the percentage will invite participants to answer without hesitation. A drawback is that the estimation of the percentage will most likely be less accurate then computing it from absolutes.

Perceived need for income was measured using the revised four-item scale of Van Hooft & Crossley (2008) on perceived financial need, including “It is difficult for me to live on my current income right now” (Cronbach’s α = .83). In one question the word “money” was replaced by “income” to fully adapt the measure to this research. No further adjustments have been done on the measure.

(34)

34

3.4. Control variables

Four control variables and one pre-test variable were formulated on the basis of logical reasoning. These control variables can lead to potential causes of interference or incorrect relations and effect sizes within the conceptual model. The goal is to keep the control variables constant, to avoid influence of these variables on the relationship between the independent and dependent variables (Saunders et al., 2009). Practically, the language of the survey will be pre-tested (0 = English, 1 = Dutch) before the analysis. Although the content of the surveys were translated back and forth in order to discover discrepancies of the

translation, a difference in language is a potential threat to interpretation of questions. The English and Dutch surveys are analysed as one dataset after this pre-test if no mentionable deviations are found. Furthermore, four control variables are included. Firstly, the educational level of the respondents is added to investigate if the effects of the gig economy are constant over all levels. Secondly, work experience is included for the same reason. Finally, gender and age are added as commonly used demographic factors. They will be controlled for in the analysis.

3.5. Survey design

The before mentioned measures were built into an online survey using the University of Amsterdam licence of Qualtrics. Qualtrics is an online survey platform which is compatible with many different types of devices. This ensures that if someone is willing to participate in the survey, that they will be able to do so, optimizing the response rate and sample size. Furthermore, the survey included optimized question paths so that questions disappear if they are not applicable, creating an as short as possible and optimal experience. The survey was provided in Dutch and in English via the translate survey option of Qualtrics. As it is not 100% certain that all the people in the sample are capable of filling this survey in in English. Therefore the survey language is included as a pre-test to check for the effect of the language.

(35)

35

The language used in the survey was initially English, as the validated measures are English. Two independent students from the Msc Business Administration program then translated the survey back and forth from English to Dutch and from Dutch back to English. The researcher investigated the differences between the translated and the original English survey together with another independent student from the Msc Business Administration program in order to detect deviations in the translations. A number of deviations were found and a total of seven corrections were made to the Dutch survey. The corrections can be found in appendix E. Appendix E consists out of the original English, translated Dutch, translated English and corrected Dutch sentences of the deviations. Four out of the seven deviations were identified in the items of career satisfaction due to the interpretation of one word. Lastly, the corrected surveys were confirmed via a final check by the supervising assistant professor.

Before launching a survey it can be useful to do a pilot test in order to minimize small overlooked errors (Saunders et al., 2009). A small scale pilot test was performed to identify these overlooked errors by the direct family and friends of the researcher (n = 9). The remarks provided from the pilot test aided in creating a clear and concise survey. On the survey-level the need for a back button and progress bar were voiced by four testers. On the question-level a number of style and spelling errors were identified, such as switches from formal to

informal writing. Furthermore, a manual link was first added in the end of the first survey to link to the second, incentive, survey. The use of a manual link resulted in responses remaining “in progress” in the first survey. An automatic redirect was added in order to ensure that the first survey will be 100% finished before starting on the second survey. The end result of the surveys can be found in the appendix A & B for the English surveys and C & D for the Dutch surveys.

(36)

36

3.6. Populations

Two different populations were required to fill in the survey in order to compare the business gig workers and the traditional workers. There was no sampling frame available of all the business gig workers nor the traditional workers. Unfortunately, without a sampling frame it is not possible to collect a probability sample, resulting in a limitation for this study (Saunders et al., 2009). Therefore non-probability sampling techniques are used to gather data of the populations. Furthermore, as business gig workers and traditional workers populations are different profiles it was necessary to engage the populations via multiple channels.

On the one hand traditional employees working via labour contracts were asked to participate. This group was not be hard to reach, as it is the dominant method of working in the Netherlands at the moment. The survey was distributed in the LinkedIn and Facebook network of the researcher and in multiple related groups on LinkedIn, resulting in a self-selection sample technique (Saunders et al., 2009). As with almost all non-probability sampling techniques, the chance that the sample will become representative is limited.

The second group was the section of business gig workers. It was expected that this group would be harder to reach, as the population is smaller according to the data of the literature review. Multiple channels have been identified via which it is possible to reach this group. Firstly, the researcher made contact with ZZP Nederland, which is an association for independent workers. They have over 40.000 independent contractors who are members. However, not all members are business gig workers. ZZP Nederland agreed to cooperate and distribute the survey under about 2.000 business gig workers. In addition, different groups of gig workers were identified on LinkedIn and Facebook. The researcher joined these groups and asked for participation to further increase the sample size. Thus, self-selection sampling strategies are used for both populations. An advantage of this strategy is that it is useful when access is difficult, which corresponds with these groups (Saunders et al., 2009).

(37)

37

3.7. Data collection

The data was collected via Qualtrics only. The data collection has taken place during four weeks starting on April 1st 2017 until April 25th 2017. Posting of messages in the LinkedIn and Facebook groups was done between April 1st_{and April 15}th_{.The newsletter of ZZP}

Nederland containing the survey was sent on April 18th. The researcher waited until no new responses were added for a period of 24 hours. The data collection was closed on April 25th.

A Google URL shortener was used to distribute the survey. The link was clicked a total of 1,396 times. The analytical data from google shows that most references came in via Mobile view of Facebook (m.facebook.com, 57.7%) and via email (classified as unknown, 34.9%). Smaller channels were LinkedIn (linkedin.com, 1.4%) and non-mobile Facebook (facebook.com, 2.9%). In correspondence with the referrers, the most used platform was Android (37.6%) followed closely by iPhone (26.2%). The desktop based platforms,

Windows (15.7%) & Macintosh (17%), were smaller then mobile equivalents. In addition, the most used browsers were Chrome (50.6%) and Safari (20.9%). Most clicks (1085, 77.7%) came from Dutch IP addresses. A few other countries have contributed some clicks, which says little about the actual responses, which are the United States (13), Austria (8), Germany (7), Thailand (5), France (4), Indonesia (4), United Kingdom (4) and Italy (3). No country was identified for 266 clicks (19.1%) The data from the URL shortener can in no way be linked to the data of the respondents.

In total 458 records have been collected before starting any form of data cleaning and preparation. Thus, 32.8% of the clicks resulted in useable entries of the data set. Qualtrics has recorded all partial and empty responses as well. These needed to be cleaned out before the data could be processed in the results. All respondents who have not, at least, answered both dependent variables were deleted. All partial responses which at least answered the dependent

(38)

38

variables were kept in the dataset. This resulted in a cleaned dataset of 393 responses. This means that 28.2% of all clicks, 85.8% of all recorded entries, in the dataset are useable.

3.8. Sample

The cleaned sample consists out of 393 entries. Taking all 393 respondents together (Mage =

43.79, SDage = 14.10, Range 17-72) 59% were male. The sample included multiple

educational backgrounds. The majority of the sample completed an educational programme on the Bachelor level (170, 47.8%). 111 (31.2%) followed an educational programme on the Master level. The remaining educational programmes are vocational degree/MBO (52, 14.6%), high school (22, 6.2%) and primary school (1, 0.3%). The whole samples had, on average, 21.07 years work experience (SD = 13.76, Range 17-52).

The sample contains traditional workers (124, 31.6%), business gig workers (174, 44.3%), hybrid workers (60, 15.3%) and respondents currently without working hours (35, 8.9%). The group without working hours contains people who do work, as this research focusses on two working populations, but who do not have a number of working hours described in their contracts as the survey question mentions “how many hours per week you are currently employed via labour contract(s)?”. Forms of labour that are part of this category are specified labour contracts such as 0-hours contracts, on-call workers and flex workers. It could be argued that these workers are part of the gig economy, but this research defined the business gig workers as working via alternative contracts. Concluding, this group does not consist out of unemployed as this survey was focussed on working populations. Hence, the currently without working hours group is a standalone group within this survey who work without a predetermined amount of contracted hours.

Demographics on the separate groups are shown in table 1 (p. 40). Of potential interest could be the differences in age, where the business gig workers (M = 50.93) seem to be older than hybrid (M = 41.89)- and traditional workers (M = 33.87). The average working hours of

(39)

39

traditional workers (M = 31.89) seem almost equal to that of hybrid workers (M = 30.08). While the hybrid workers on average also work an additional 21.02 hours a week in gig work. It seems that the overall workload of all working hours is highest within the group of hybrid workers. Furthermore, the education levels are reported in table 2 (p. 40). The group of business gig workers seem to have a higher educational level as the segment with a Master or PhD is higher (36.0%) then traditional workers (28.6%) but vocational degree (MBO) is lower for business gig workers (9.1%) then for traditional workers (21.1%).

(40)

40

Table 1: Descriptive statistics of sample populations

Business Gig

Workers Hybrid Workers Traditional Workers All respondents

Variables n M SD n M SD n M SD N M SD

1. Age _{164 50.93 10.05} 57 41.89 _13.47 112 33.87 _12.77 356 43.79 14.10

2. Gender (0=female, 1=male) ₁₆₇ 0.61 0.56 57 0.60 0.50 112 0.54 0.50 360 0.59 0.526 3. Years work experience _{167 26.96 10.57} 56 19.91 _13.50 112 12.31 _12.88 358 21.07 13.76 4. Labour contracted hours ₁₇₄ 0.00 0.00 60 30.08 _19.15 124 31.89 _10.73 393 14.66 18.34 5. Alternative contracted hours 174 31.97 13.46 60 21.02 15.57 124 0.00 0.00 393 17.97 18.32 Note: the totals do not add up, as the group currently without any working hours (n = 35) is not included as a separate group but are included in the "all respondents" columns.

Table 2: Cross tabulation of educational levels

Business Gig

Workers Hybrid Workers Traditional Workers All respondents

Variables n % n % n % n %

1. Master or PhD 59 36.0% 17 29.8% 32 28.6% 111 31.2%

2. Bachelor 84 51.2% 27 47.4% 47 42.0% 170 47.8%

3. Vocational degree (MBO) 15 9.1% 12 21.1% 18 16.1% 52 14.6%

4. High School 6 3.7% 1 1.8% 14 12.5% 22 6.2%

5. Primary School 0 0.0% 0 0.0% 1 0.9% 1 0.3%

(41)

41

3.9. Language pre-test

Of all respondents, 38 (10.6%) used the English version of the survey. A majority of 322 (89.4%) respondents used the Dutch version of the survey. Independent sample T-tests were used to pre-test if any variable has significant differences with grouping by the survey language. The two dependent and three out of five independent variables had no significant differences in regard to the used survey language. Two significant differences in means were found for independent variables and three for the control variables. Firstly, gig working hours was significant (t = -4.45, p < .01). Meaning that the English survey consisted out of

significantly less gig working hours than the Dutch version. Secondly, the percentage of income from gig work is also significantly different (t = -2.38, p < .05). The income of the English respondents have a smaller portion which is generated through gig work. This seems to be in line with the lower number of gig working hours. Thirdly, education (t = -2.40, p < .05) has a significant difference. The English respondents have a significant higher education, as a lower score indicates a higher educational level. Fourthly, the years of work experience has a significant difference (t= -8,354, p < .01). The English respondents have significantly less work experience. Finally, age has a significant difference (t = -8.289, p < .01) such that the English respondents have a significant lower age then the Dutch respondents.

Concluding, no significant differences were found within the multi-item constructs, such as WLB, career satisfaction (CS) and perceived need for income. The English

respondents do seem to have different demographics and some differences in variables. However, the absence of differences within the multi-item constructs indicate no translated or interpretation errors, which is the primary concern of this pre-test. The data can therefore be used together in further analysis.

(42)

42

4. Results

The chapter will describe the results of the analysis on the gathered data. Firstly, the internal reliability is analysed and constructs are computed. Secondly, descriptive statistics,

distributions and a correlation table are put forth. The empirical part will follow afterwards testing the different models and each hypothesis. The data is processed in the software program Statistical Package for Social Sciences (SPSS).

4.1. Reliability analysis

Firstly, all counter indicative items were recoded. The research contains three multi-item constructs for which the internal reliability must be tested before computing the means. The results of the internal reliability analysis will also be visible in the correlational matrix (table 3, p. 46). The norms used in the reporting of the analysis are adopted from Field (2013). Field (2013) states that a Chronbach’s Alpha of .8 indicates a high reliability. Furthermore, all items within the analysis should have a corrected item-total correlation of at least .3 to indicate that all items have a good correlation with the overall scale.

The work-life balance scale has a high reliability, with Cronbach’s Alpha = .862. The corrected item-total correlations were all above .30. One item had a slightly lower score then the rest. This deviation possibly originates from the recoding of the item due to the counter indicative nature. Removing this item would increase the Cronbach’s Alpha by .029. This increase is not substantial, therefore the item is kept constant in the scale despite the suboptimal score.

The career satisfaction scale has a high reliability, with a Cronbach’s Alpha =.817. The corrected item-total correlation are also sufficient. All five items are kept in the scale. The perceived need for income scale has a high reliability, with a Cronbach’s Alpha =.866. The corrected item-total correlations are also sufficient. The scale cannot be improved

Beyond satisfaction or extortion : effects of the gig economy on business gig workers : effects of boundary conditions of gigs on the career satisfaction and work life balance of business gig workers