MSc Business Administration
International Management
Master Thesis
T
HE EFFECT OF
B
IG
D
ATA USAGE
ON EMPLOYEE JOB SATISFACTION
Author: Kjell Verschuren
Student Number: 11397101
Date of submission:
21/01/2018
Supervisor: Dr. Markus Paukku
Statement of originality
This document is written by Student Kjell Verschuren 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.Acknowledgements
I would like to thank my advisor, Dr. Markus Paukku for the great advice, comments and pleasant conversations. I would also like to thank my family, friends and colleagues that have supported me throughout the last couple of months. Thank you for always challenging me.
Table of Contents
STATEMENT OF ORIGINALITY ... 2 ACKNOWLEDGEMENTS ... 2 TABLE OF CONTENTS ... 3 ABSTRACT ... 5 INTRODUCTION ... 6 LITERATURE REVIEW ... 8 BIG DATA ... 9EMPLOYEE JOB SATISFACTION ... 13
COMBINED ... 20
METHODOLOGY ... 22
RESEARCH DESIGN ... 22
SAMPLE AND DATA COLLECTION ... 24
Sample Characteristics ... 27 VARIABLES AND MEASURES ... 27 Independent Variables ... 28 Dependent Variables ... 29 STATISTICAL PROCEDURE ... 30 RESULTS ... 31 AT A GLANCE ... 31 A CLOSER LOOK ... 36 Three influencers ... 36
OTHER INDEPENDENT VARIABLES ... 38 CONCLUSION ... 39 DISCUSSION ... 40 RESULTS ... 40 LIMITATIONS ... 43 PRACTICAL IMPLICATIONS ... 44 PROPOSAL FOR FURTHER ANALYSIS ... 45 REFERENCES ... 47 APPENDIX A ... 51
Abstract
There has been no research into how Big Data usage inside a firm can have an impact on the employees. This study tries to answer that question: “What effect does Big Data usage inside a firm have on employees?” The study is founded on the research of different scholars from the social, economic and computer sciences, with regard to Big Data and employee job satisfaction, separately. The data was collected through social medias and gathered more than a thousand responses. Through a quantitative data analysis, it was concluded that there is a small, yet significant, increase in employee job satisfaction when a firm utilizes Big Data. The most notable finding was that the employees that work in companies that utilize Big Data were more positive towards their career advancement opportunities, their level of authority and felt more positive about their social status. The consequences of these results will be discussed at the end of this paper. Key Words: Big Data, employee job satisfaction, Minnesota job satisfaction Questionnaire, new technologies
Introduction
You are reading this paper in 2018 or later, this means that you have heard of the term “Big Data”. It does not matter if your field is academics, management, IT, marketing, sales and so on. This was not always the case, the term Big Data has only been around since the 1980’s, and only in the last decade it has become a mainstream social and academic topic. Many papers focus on how Big Data can provide a company with an advantage over the competition (Sagiroglu and Sinanc, 2013), or how Big Data solves many problems, but also has its own set of issues (Snijders et al., 2012). There have, however, not been any papers written on how Big Data affects the employees, and thus, the organization itself. Right now, there are companies that are rushing to implement Big Data into their workflow, without understanding what effect that will have on the organization itself. They can’t know, because there has been no research into this subject, yet. This paper tries to change that.Answering the question of how, or if, Big Data affects the employees of a firm is an enormous task. First, it has to be defined what influences employees, then how that can affect the organization. Finally, it has to be shown that Big Data does, in fact, influence the employees. It is possible that Big Data usage in a firm does not influence employees, this would be a helpful conclusion as well.
The next pages will feature a literature review which will discuss both literature on Big Data and employee job satisfaction, to give the reader a good overview of what is being discussed, and why it matters. Subsequently an explanation on the manner of research will be provided. The answers that this paper will provide come from a quantitative data collection and analysis. The reasoning behind the tool-selection will be provided and justified.
After this explanation, the results will be shown. The most important results will be highlighted, after which the additional findings will be presented. To close the paper, a conclusion will be written and subsequently a critical discussion of the literature, results and value of this paper will be presented.
Literature Review
Big Data is a relatively new concept in the academic world. While the subject off job satisfaction is decades old, the research into Big Data dates back to less than 15 years ago, around beginning of the new century, with most of the research coming in the last five years. A consequence of a young academic topic is that there’s no accepted definition of what Big Data is. Different people in different situations perceive Big Data as something different. An individual that does not have much knowledge on the subject depends on experts to define the definition for him/her (Labrinidis and Jagadish, 2012). This has led to many different definitions, as experts use a definition that fits their field or interest best (Labrinidis and Jagadish, 2012). Adding to this, it must also be stated that Big Data is also a popular subject, not only in the academic world but also in the world of business. Employee job satisfaction on the other hand, is a mature subject with decades on research behind it. Over the years, it has become accepted that there are certain elements that influence job satisfaction, like autonomy and creativity (Judge et al., 2001). It is also known that job satisfaction itself affects employee productivity, and thus company productivity and profitability (Bettencourt and Brown, 1997). So, to find out how Big Data and job satisfaction interact with one another, is not merely an academic question, it is a corporate question.
In the coming pages, a brief overview of past research into Big Data and employee job satisfaction will be provided, the chosen definition of Big Data will be explained and the most suitable manner to measure job satisfaction will be substantiated.
Big Data
Big Data is one of the most popular subjects in modern management literature. It is also one of the newest widely researched topics. The first mentions of the concept of Big Data start around the late 1980’s, and since the late 2000’s, Big Data really has become a mainstream academic topic. According to Google Scholar, between 1980 and 2000, there were three thousand papers published that mentioned Big Data, from 2000 till 2018, there were 95 thousand. The Data on the topic is thus, quite new.As stated before, there is no widely accepted definition of Big Data. There are articles that describe the flow of Big Data, there are articles that describe what Big Data entails and articles that describe the problems and challenges with Big Data, and then there’re articles to try to summarize all of this. It is in these articles that one can find the most logical and accepted definition of Big Data.
In their paper on Big Data myths, (Labrinidis and Jagadish, 2012) state there are five steps that a process must take in order to be qualified as Big Data. First there is “Data acquisition”, then “information extraction and cleaning”, “data integration, aggregation and representation”, “modeling and analysis” and at last “interpretation”. The authors state that this is a track that Big Data follows, this is not the definition. It does however bring us closer to what a workable definition of Big Data is, a collection of data that has been cleaned, prepared for usage, analyzed and that subsequently has to be interpreted. Big data is thus, especially in the interpretation part still a fairy human intensive area (Labrinidis and Jagadish, 2012).
This knowledge adds a human element into the Big Data mix. Big Data provides insights that were not fathomable ten, or even five years ago. But this data is still delivered to a human, who has to interpret the data, and do something with it. It is all interconnected, Big Data collection, analyzation and implementation (McAfee and Brynjolfsson, 2012). Management literature, especially, emphasizes that the collection, analyzation and management of Big Data are connected and cannot be seen as independent of one another (Ward and Barker, 2013). In their paper on the management revolution caused by Big Data, (McAfee and Brynjolfsson, 2012) state five requirements for managers if they want Big Data to be properly used by their firm. First off, the leadership should use Big Data as a supplement to human insight. Big Data offers many benefits, it does not show what you can do with the data. Second, the firm should hire more and more talented data analysts that know how to navigate the web of data. Third, the firm must be up-to-date with the latest technology to both manage the Data and to keep up with competing firms. Fourth, the decision making should occur across different departments in the firm, every department is affected by the decisions that another department takes. And fifth, the company culture should support a move away from relying on “hunches” and instinct, and focus more on a data reliance.
The usage of Big Data inside firms is most notable to employees as data-driven decision making (Brynjolfsson et al., 2011). Big Data has provided information, analysis and predictions, and then it is up to an employee to decide what to do with it (Chen et al., 2012). Big data is required for and can lead to data driven decision making, according to (Brynjolfsson et al., 2011). They then state three definitions of data driven decision making:
1. The usage of data for the creation of a new product or service 2. The usage of data for business decision making in the entire company 3. The existence of data for decision making in the entire company We now know that the hard requirement for data driven decision making is Big Data usage or ownership. Yet we still do not have a clear-cut definition of Big Data. In his paper on the “pathologies of Big Data” (Jacobs, 2009) starts off by asking: “What is Big Data?”, then the content of the paper circles around the problems with data files, to conclude with the following definition: “data whose size forces us to look beyond the tried-and-true methods that are prevalent at that time.” This is a conclusion that suits the narrative of the paper, that Big Data evolves and that new tools are required to keep utilizing it. A popular definition is that Big Data is the point where data volume, velocity and variety intersect (Russom, 2011). Meaning that a lot of data by itself is not valuable for a firm. For data to be of value, it needs to come in different forms, structured, semi structured and unstructured. In addition to this, the data needs to be generated, delivered and processed frequently (Jee and Kim, 2013). Most papers that have been studied for this project used a definition of Big Data that suited the narrative that the researcher wanted. The closest definition of Big Data that is available is where volume, velocity and variety meet. The observation that there is no clear-cut definition of Big Data and that researchers can bend the definition to suit their narrative creates a problem: How do you measure its usage? Asking potential respondents if their company uses data that has both volume, velocity and variety is not a recipe for success. What we, however, have learned so far is that Big Data can lead to data driven decision making, and that the entire company is affected by
it. Asking respondents if their company or they themselves utilize data driven decision making automatically answers the question if the firm uses Big Data, since it is a requirement for data driven decision making. This is the definition that will be followed during this project, data driven decision making.
Big Data usage does not affect a small part of the company, it affects the entire company, and thus, its employees (Brynjolfsson et al., 2011). Yet, there is no popular academic literature on the organizational effects of Big Data usage or data driven decision making. The knowledge that exists is on managers and how to manage Big Data. No knowledge exists on how employees are affected, on a personal or professional level, by data driven decision making. However, as stated before, Big Data is a relatively new occurrence or technology in organizations. On this subject, there has been interesting research performed. The article by (Barley, 1986) describes the social change that occurred after a new technology was implemented that was best understood by a previously lower social group. The group that understood the new technology rose in the social order while displacing groups that did not understand or work with set technology. The same effect can occur with Big Data, social change due to a change in the way the business operates. A change in social order influences employee job satisfaction, since job satisfaction can be considered the mean of many influencers. This will be explained on the next pages.
Employee Job Satisfaction
Employee job satisfaction is a much more mature subject than Big Data. The first popular research into this subject dates back to the 1930’s. The subject of employee job satisfaction has thus been researched for almost 90 more years at the moment of writing. Longevity of a research topic has many benefits for researchers, the saying “standing on the shoulders of giants” does not come out of nowhere. If there has been previous, peer-reviewed research into a topic, it is possible to build upon that research. This allows for more in-depth and faster research, as many data collection methods are already pre-made and most importantly, more accurate research, since the mistakes of the past are known, and are thus preventable.Since there is not one widely-accepted definition of Big Data, and research on its organizational effects are limited, it is difficult to assess how Big Data can influence employee job satisfaction. However, a start can be made by comparing the effects of Big Data and what makes employees happy. If these two outcomes interact, there is an effect that follows. Here, Big Data usage functions as a moderator for employee job satisfaction.
Employee job satisfaction has always been difficult to measure. The academic community has been refining, adding and removing elements that influence job satisfaction for more than 70 years. (Hoppock, 1935) discussed that there is a relation between job satisfaction and emotional adjustment, religion, social status, interest, age, fatigue and size of community. As one of the earliest work on the subject, this is already an impressive list. Showing that employee job satisfaction is linked to much more than salary only paved the way for further research into employee job satisfaction, which can clearly be seen by the popularity of the paper. The paper written by (Hoppock, 1935) and other popular papers
written in the 30’s caused a rapid growth in the research efforts on the subject. This interest, as noted by (Locke, 1969) has however not lead to an equal growth in the knowledge on the subject. Whereas the research has increased, the knowledge has stagnated. For example, after thirty years of additional research, researchers are still divided on what causes job satisfaction. There are recent papers published on both intrinsic job satisfaction and subjective job satisfaction, but no papers combine these two. However, authors of these papers do suggest that job satisfaction is not solely dependent on any one of the two views (Judge et al., 2010). The main cause of job satisfaction is difficult to pinpoint, (Aziri, 2011) describes three different models that try to explain what factors influence job satisfaction. (Christen et al., 2006) state that job performance, job factors and problems with role perceptions influence job satisfaction, this however is still quite broad. All models in (Aziri, 2011) however concur that the there are many facets to job satisfaction. Other papers focus more on independent, smaller, influencers of job satisfaction. (Rue and Byars, 2003), for example created a model with 8 so called “influencers”. Manager’s concern, job design, compensation, working conditions, social relationships, opportunities, aspiration and need for achievement all influence job satisfaction. Although it is not specifically stated, these are both intrinsic and subjective influencers to job satisfaction.
An answer for the stagnation of knowledge is provided by (Locke, 1969), who in his own researched encountered significant correlations between outside factors and job satisfaction. The problem with these correlations is that in each independent research between 1965 and 1969 the magnitude of the correlation was highly irregular between the studies. Even more important was that the author realized that he did not explain the
correlation or the difference therein between the studies. There was something wrong with the studies toward employee job satisfaction, none of the researchers pinpointed the reason for found correlations. (Locke, 1969) states that the concept of “causation without correlation” has been the cause of this lack of understanding. (Skinner, 1953) promotes the idea that scientist can never provide causality, only correlation. For example, there have been many studies correlating the performance ratings with test scores, an entire industry has been built upon this research. Yet, only a few papers have been published that try to answer why these correlations exist, or why sometimes, they are not there. It is more important that these correlations exist, than why they exist. (Locke, 1969) argues that it is impossible to predict an event in the future, if you cannot explain the causes of a phenomenon.
After providing criticism of previous-research into employee job satisfaction, and offering a new view on future research efforts. Locke goes on to explain what according to him, constitutes job satisfaction. He states that “Job satisfaction is the pleasurable emotional state resulting from the appraisal of one's job as achieving or facilitating the achievement of one's job values.” (Locke, 1969). This definition is not far off from (Hoppock, 1935), stating that “any combination of psychological, physiological and environmental circumstances that cause a person truthfully to say I am satisfied with my job”. Another way of describing job satisfaction is “affective orientations on the part of individuals toward work roles which they are presently occupying” (Vroom, 1964). In other words, job satisfaction is the level on which the reality exceeds the expectations. This consists three distinct elements: the perception of some aspect of the job, an implicit or explicit value standard and conscious or sub- conscious judgment of the relationship between one's perception(s) and one's value(s) (Locke, 1969).
These elements are indistinguishable for the employee and together form the level of job satisfaction. A logical way to find the degree of job satisfaction is to subtract the value standard from the perceived aspect of a job. If this number is positive then the employee is satisfied with his/her job (Porter, 1963).This is according to (Locke, 1969). neglects a crucial element in the job satisfaction equation: intensity. Intensity refers to how much an employee wants something. The concept of artistic freedom, for example, can be much more important for person A, than for person B. If the artistic freedom is then limited, person A will suffer more with regard to job satisfaction. Just as the perceived aspect of a job gets a numerical value for the equation, so should the explicit value standard.
The thinking that Locke started, that there is more to job satisfaction than the equation “Reality - Expectations” has since been expanded upon. When in 1971, the paper from (Hackman and Lawler, 1971) was published, it stated that in addition to the four core dimensions that they found: Variety, Autonomy, Task identity and Feedback, there also interpersonal dimensions and worker characteristics involved in the job satisfaction equation. The researchers based their research on the notion that for employees to be satisfied, the four core dimensions need to be fulfilled. When you try to put these core dimensions into a sentence, it would look like this: An employee needs to be able to figure out that he/she personally accomplishes something meaningful, when he/she performs well. In this paper, the researchers add the interpersonal dimension, as an addition to the already existing core dimension. The interpersonal dimension exists of dealing with others and friendship opportunities. These dimensions, according to the authors are highly dependent on the characteristics of the workers. For example, an employee in an urban area has different
expectations than an employee in a rural area (Hackman and Lawler, 1971). The authors state a solution for the uncertainty of employee characteristics is to let the employees describe the job characteristics themselves. Only the employees know what tasks they perceive, and can thus have an influence on them. However, using the data from the employees causes the loss of important methodological and conceptual advantages that come with independently designed variables. The researchers state that the right job characteristics and tasks can be found when the researchers question both the employees and managers, while also researching the characteristics themselves.(Turner and Lawrence, 1965). show how to describe a job in operational, measurable, terms. However (Hackman and Lawler, 1971) do put a higher emphasis on the core dimensions than on either the interpersonal dimension or the worker characteristics. The difference between research that takes place around the same time within the field of employee job satisfaction is staggering. The Job Descriptive Index (JDI) is to this day a popular way to measure job satisfaction. The JDI, combined with the later to be discussed Minnesota Job Satisfaction Questionnaire (MSQ) has the highest validity between job satisfaction measurement tools (Van Saane et al., 2003). The first version of the JDI, designed in 1969, by (Smith, 1969) proved highly popular within the scientific community, amassing more than 6000 references on the original paper alone. The Survey is based on five aspects of jobs: Supervision, co-workers, work, pay and promotion. The sum of the five is used as a measure of overall job satisfaction. The popular paper, that has already been discussed, by (Hackman and Lawler, 1971) was written with the knowledge of the JDI in mind, but completely ignored it, and instead used the core dimensions. This difference between
research work, that still exists today, makes collecting data on job satisfaction difficult. Processing this data into a valid result, is even harder.
The main issue with academic literature on the subject of employee job satisfaction is that it is ever growing. After almost a hundred years of research, there is still no widely accepted definition of job satisfaction, nor is there a tried and tested, fool proof, method of measuring it. New research often creates more questions than it answers. This can be seen in the paper on the relation between job satisfaction and turnover by (Mobley, 1977). During the research into this topic, the author mentions that the level of perceived job satisfaction increases when it is difficult to obtain a job at a different firm. This is caused by a feedback loop, where the level of perceived job satisfaction is re-evaluated by elements, both inside and outside the employee’s work-environment. This knowledge adds to the validity of future research into job satisfaction, yet it also causes the measurement of it to be that much more difficult. All previous methods of data collection and subsequent analyzation will have to be remade, and revalidated. The difficulty to find a proper job satisfaction questionnaire is visible in even the most popular research. In the paper on the correlation between job satisfaction and performance by (Ostroff, 1992) a high correlation between performance and job satisfaction is found. However, the job satisfaction survey included items like student discipline, school curriculum, parental support and supervision. The survey was designed specifically for this research and partially derived from previous research. The combination of previous research into one survey, and the way it was specifically designed for this survey into the job satisfaction of teachers, causes a survey that is both unreliable and has a low validity. It is possible, of course,
that this survey accurately measures job satisfaction in the given situation, but this has not been proven and the results are equal to previous research.
So, given the knowledge on the difficulties in researching and measuring job satisfaction, what do we know? We know that there is a correlation of .30 between job satisfaction and job performance (Judge and Piccolo, 2004). This number has been derived from the combination and review of 301 studies into the subject. This correlation is higher when related to more complex jobs (Saari and Judge, 2004). Job satisfaction is predictive of future job performance, this leads to the statement that employees that enjoy a high level of job satisfaction, are more productive, and thus, more valuable to employers. It has been stated before, and it is confirmed again, 40 years later, that different personalities enjoy different levels of job satisfaction. This finding puts a limit on the highest achievable level of correlation, since the results will be different per person, team, location, culture and more influencers of personalities. The last, and potentially most important piece of information that comes out of the literature review on employee job satisfaction is that there are two popular measurement tools used in academic literature: The JDI and the MSQ. To measure employee job satisfaction these measurement tools ask the respondent how they feel about a small aspect of their job, like “independence” or “co-workers”. The MSQ is based around twenty job-aspects (Weiss et al., 1967), whereas the JDI is based around five (Kinicki et al., 2002). Both measurement tools come with a manual that helps the researcher to give a value to “mean job satisfaction score”. After reviewing the literature on the subject, it is concluded that accepting the outcome of either the JDI or MSQ is the most reliable way of measuring job satisfaction. Other tools, as
discussed, are not applicable situations outside the one they were originally developed for. The right tool for this specific project will be decided on in the methodology part.
Combined
We now know that in order to measure employee job satisfaction, either the JDI or the MSQ will be used. These two measurements provide a total job satisfaction score after receiving responses from participants. These participants answer questions that can be categorized in twenty different subjects for the MSQ, and five for the JDI (Aziri, 2011). If Big Data usage inside a firm has an influence on employee job satisfaction, that influence must be on one of the influencers/categories, as shown in model 1. Model 1: How Big Data can affect employee job satisfaction. Possible Big Data influence is shown in red Advancement Independence Recognition Employee job satisfaction
In the model, it is shown that if Big Data has an impact on employee job satisfaction, it happens on the influencers/categories of job satisfaction. In the model, a higher level of advancement opportunities, leads to a higher job satisfaction score. If Big Data has a negative influence on advancement opportunities, the overall employee job satisfaction score will be lower because of it.
Methodology
This chapter discusses both the research approach and the research design that is used. The entire research process from design to analysis will be explained in detail in the coming pages.Research Design
Designing a valid, reliable, accurate and -not unimportantly- feasible data collection method is a difficult task. When researching the potential impact of Big Data on job satisfaction, it is important to have data from firms/employees that utilize Big Data, but also from firms/employees that do not utilize Big Data. Furthermore, the research question is not “what is the impact of Big Data usage on the job satisfaction of managers in the transport industry”, that would be too specific. This research is the first to attempt to uncover the potential effects of Big Data on job satisfaction, paving the way for more specific studies.The data collection happens through a survey. Different theoretical models were originally created, including a longitudinal study and a model that included more than five independent variables. However, due to the limited time frame that is available for this research project, performing a longitudinal study is not feasible. The data will be collected at one moment, a cross sectional study, and will be based on one independent and one dependent variable (Saunders et al., 2009).
The research is founded in a positive philosophy. This means that the research tries to answer, and predict in the future, the question “Does Big Data usage have an effect on employee job satisfaction” (Easterby-Smith et al., 2012). This prediction is done purely based on gained scientific knowledge, not opinions or previous research.
The two variables that are being tested inside this study are Big Data and employee job satisfaction. Job satisfaction, as stated before has been the subject of academic research for decades. As stated before, there are many methods of determining employee job satisfaction. Most of these methods are specifically created to answer the job satisfaction question in a specific setting, these methods are not usable in a different research project. The two most widely used methods of measurement that are applicable to more than one study and setting, are the Minnesota Job Satisfaction Questionnaire (MSQ) and the Job Descriptive Index (JDI). Both the MSQ and the JDI have an internal consistency of 0.81, which is by the highest of all heterogeneous surveys into employee job satisfaction (Van Saane et al., 2003). The revised Job Descriptive Index (JDI-Revised) is a more accurate version of the original JDI, bringing the internal consistency to 0.88. The main difference between the JDI(-revised) and the MSQ questionnaires is that the latter offers a simplified version, going into less detail while keeping the validity. Both the JDI and the regular MSQ were created as long surveys that cover a hundred or more questions. Since then, the authors have designed shortened versions of the measurements, while keeping the internal validity equal to the longer version (Weiss et al., 1967, Kinicki et al., 2002). Due to the necessity of having many respondents, it was decided that it would not be beneficial to let respondents fill in a very extensive survey, it was hypothesized that this would endanger the data collection. The shortened MSQ contains twenty questions, while the shortened JDI contains 40 questions. Both surveys were analyzed by Qualtrics, an online data collection tool. After simulating ten thousand responses, it was shown that the MSQ took less than five minutes to fill in. The JDI took 40% longer to complete than the MSQ. As stated before, it was hypothesized that a short
fill-in time would lead to a higher completion percentage, and thus more complete responses. Since there is no difference in internal validity between the shortened MSQ and shortened JDI, it was decided to utilize the MSQ for this project
Sample and Data Collection
I was originally planning for the survey to be distributed by HR managers towards the employees. This was to be done without a pre-selection on the basis of location or predicted Big Data usage of the firm. This plan was carried out but did not lead to positive results. Of the 20 emails that were send out, two answers were received, with both a kind, yet negative response. The manner of data collection had to be adapted to the situation. The sampling method changed from purposive or expert sampling towards convenience sampling (Saunders et al., 2009)It has to be stated that the original sampling method also contained a “convenience element”. HR managers were specifically targeted non-only because of their direct relation with job satisfaction and thus, possible interest in the outcomes of this study, but also due to their position inside a firm, with the power to ask employees to fill in the survey. Yet, the selected HR managers were chosen due to the perceived probability of an answer. The step from purposive sampling towards convenience sampling is thus not as big as it seems. The decision to use convenience sampling brought the issue of control with it. It is important to have a sample of people that are both managers and non-managers, work in different industries and different countries. If this is not the case the data will be biased towards either industry, country or managerial position. Since the first method, of using HR managers to distribute the survey to employees failed, it became clear that it was not viable
to be too selective. For this project, there was not enough time to gather respondents from one specific industry, country or managerial position. One of the way to solve this problem is to have a large sample size (Saunders et al., 2009)A large sample size is more likely to provide responses from different countries and industries then a small sample size would be. It is possible that the data collection method is not perfect, providing many more responses from industry A, than from industry B. this however is a more faithful outcome then a few from A and none from B. in addition to this, more likely to find flaws in a data collection, if there are many respondents (Saunders et al., 2009). The new sampling method directly targeted employees that are currently working in an office environment and speak English. Selection was not based on country of employment, managerial status, Big Data usage, time of employment or Industry. However, these variables were included in the survey, for possible use in the analysis. The survey, hosted by Qualtrics, was distributed on social media. Respondents were received from scientific communities on Reddit and Facebook. The survey was specifically not distributed to communities that centered around Big Data, as it was hypothesized that this would have skewed the results.
The following chart shows the questions that were asked in the survey:
Question Question Scale
1 Do you work in an office environment? Yes/No 2 In what industry is your company located? 20 options 3 In what department do you work? 13 options 4 Are you a manager within your department/team? Yes/No 5 Of how many people does your department/team exist? 5 scales between 1-5 and 51+ 6 How long have you been employed by your current employer? 5 scales between -6 months and 5+ years 7 How long have you been inside your current position? 5 scales between -6 months and 5+ years 8 In what country are you employed? List 9 What is the name of the company you work for? (Optional) Text 10 MSQ questions 20 Questions on a 5pt scale. From Not satisfied till Extremely Satisfied 11 Does your employer use data driven decision making / Big Data? Yes/No 12 Do you personally use data driven decision making / Big Data in your job? Yes/No 13 Are you personally affected by data driven decision making / Big Data? Yes / No 14 If you have any comments or additions, you can write them here: Text Chart 1: List of questions in the survey
Sample Characteristics
After distributing the survey for two weeks, the total number of participants was n=1063. After removing incomplete and duplicate answers, the total of unique completed surveys ended up at n=934. Out of these 934 respondents, 854, or 91,4% work in an office environment. It is possible that employees in different work environments perceive Big Data differently. If the number of non-office-based respondents would have been equal to office-based respondents, this could have been researched. Since the percentages differ significantly, the respondents that do not work in an office environment will not be included in the analysis. The reasoning behind this decision is that the 8.6% is too small to compare, but large enough to influence the analysis. In This removal brings the total to n=854.
No industry is overrepresented, both “Computer and Electronics” and “Education” score high, with 108 respondents each. Three departments score high; Operations, Customer Services and Information Technology, with around 130 each. Managers account for 47.5% of all respondents. The United States takes 53.4 percent of all respondents, India has 239. With regard to Big Data usage, 67.9% of the respondent’s state that their employer uses Big Data, while 54.9% used Big Data themselves. The last question of the Survey asks if the respondent is personally affected by data driven decision making and/or Big Data, 51% states that they are, while 49% states that they are not personally affected.
Variables and Measures
The research is the first of its kind, which makes designing a reliable and valid survey difficult, since there is limited work to build upon. But to find out what the influence of Big data is on
employee job satisfaction, the survey, at the minimum has to measure the level of job satisfaction and if the respondent uses Big Data.
Independent Variables
The independent variable in this study is the Big Data usage inside a firm. The idea is that the usage of the Big Data has an effect on employee job satisfaction, either negative or positive. It has been stated before in this paper that it is difficult to assess Big Data usage, since there is no clear-cut definition of what Big Data actually is. For some it might be quarterly results that influence current decisions, but for others it can be customer usage data. Since there is no clear definition and it would be unethical to “just make one up”, it was decided to ask three simple questions inside the survey: 1. DOES YOUR EMPLOYER USE DATA DRIVEN DECISION MAKING / BIG DATA? 2. DO YOU PERSONALLY USE DATA DRIVEN DECISION MAKING / BIG DATA IN YOUR JOB?3. ARE YOU PERSONALLY AFFECTED BY DATA DRIVEN DECISION MAKING / BIG DATA? 3A. HOW ARE YOU AFFECTED BY DATA DRIVEN DECISION MAKING / BIG DATA?
The first three questions are yes or no, while the fourth offers the option to elaborate, and will only appear if the third one is entered as yes.
These straightforward questions remove the need to define what Big Data is, by using the Big Data definition that the respondents themselves have. This removes the need to explain a certain definition of Big Data, which can lead to incorrect results due to confused respondents. By letting respondents use their own definition the concept of Big Data is still maintained. Until there is a clear definition of what Big Data is, letting respondents use their
own definition is the only manner in which you can do research into the effects that Big Data has on employees.
Dependent Variables
Employee job satisfaction is the dependent variable of this study. It is envisioned in the conceptual model that Big Data has some kind of influence over employee job satisfaction.
As explained before, the Minnesota Job Satisfaction Questionnaire (MSQ) was chosen to measure the level of job satisfaction of the respondents. The MSQ aims to determine the level of total job satisfaction by finding answers to twenty influencers of job satisfaction. These influencers are: Ability utilization, achievement, activity, advancement, authority, company policies and practices, compensation, co-workers, creativity, independence, moral values, recognition, responsibility, security, social service, social status, supervision-human relations, supervision-technical, variety and finally working conditions. These twenty influencers are categorized in three different scales: Intrinsic satisfaction, extrinsic satisfaction and general satisfaction (Weiss et al., 1967). In appendix A, there is a list of the questions that were asked to the respondents, and what influencer of job satisfaction that question measured.
Every response will get a raw score, which will be converted towards a percentile score on the MSQ-developed “employed-disabled” norms. With these percentile scores, the influence of Big Data can be found. For this, the “Big Data users” will have to score significantly different than “Non-Big Data users”. In addition to the percentile score, a more detailed look can be aimed towards the twenty individual influencers, it is possible that Big Data does not have a significant influence on employee job satisfaction, but does have a significant effect
on “recognition”, for example. All the influencers and the three scales will be thoroughly analyzed in the results part.
Statistical Procedure
The original dataset, provided by the Qualtrics survey platform contained many variables that were helpful in the cleaning process. Both the “time” and “completion percentage” variables were analyzed in order to remove outliers. Only complete submissions where accepted, while no responses under a minute were accepted, since this was deemed impossible. As already stated before, only respondents with an office job were accepted. These filtration steps brought the total number of valid respondents to 854, after being open between October 1stand November 15th.
To calculate the mean employee job satisfaction score, the averages of all twenty influencers were combined and then divided by twenty. The five-point scale includes; not satisfied, somewhat satisfied, satisfied, very satisfied and extremely satisfied. According to the MSQ manual, a final employee job satisfaction score between 3 and 3.5, or “satisfied” is expected (Weiss et al., 1967).
Results
In this chapter, the results will be presented. First the most important results will be presented; does Big Data influence employee job satisfaction? Later, there is a closer look into results that are not as obvious as the research question.At a glance
The goal of this study is to find out if there is a correlation between Big Data usage inside a firm and employee job satisfaction. The analyzed results indicate that employees who are employed in a firm that use Big Data come in at a 3,53. Employees who work in a firm that does not use Big Data come in at 3,19. If translated to human terms, both employee groups come in between “satisfied” and “very satisfied”. Answers Answer explained N Job Satisfaction scoreBig Data usage The company uses Big Data
603 3,53
No Big Data usage
The company does not use Big Data 245 3,19 Mean Both 848 3,43 Chart 2: Summarized Mean of Big Data and MSQ These results can be considered valid, the mean of a job satisfaction survey usually comes in around 3 – 4, with the interesting finds occurring in the sub-variables (Weiss et al., 1967). The standard deviation for the total is .716. There are only nine instances where the results are more than one standard deviation apart from the mean results, in a later section these instances will be discussed.
Since Big Data is not a clearly defined concept – as stated before – it was decided to ask three questions: 1. DOES YOUR EMPLOYER USE DATA DRIVEN DECISION MAKING / BIG DATA? 2. DO YOU PERSONALLY USE DATA DRIVEN DECISION MAKING / BIG DATA IN YOUR JOB? 3. ARE YOU PERSONALLY AFFECTED BY DATA DRIVEN DECISION MAKING / BIG DATA?
The answers to these questions revealed that there are some significant deviations between the answer options:
Answers Answer explained N Job Satisfaction score
YYY The company uses Big Data, I use it myself too, and I am affected by it
349 3,59
YYN The company uses Big Data, I do too, but I am not affected by it
101 3,54
YNY The company uses Big Data, I do not but I am affected by it
68 3,28
YNN The company uses Big Data, I do not and I am not affected by it
85 3,48
NYY The Company does not use Big Data, but I do use it and I am affected by it
22 3,20
NYN Company does not use Big Data, I am not affected by it, but I myself do use it
20 3,02
NNY The company does not use Big Data, I don’t either, but I am affected by it
11 3,01
NNN The company does not use Big Data, I don't either and I am not affected by it
192 3,22
First of all, the results show that employees who are employed in firms that utilize Big Data and/or data driven decision making, are generally happier in their job than employees whose employer does not utilize Big Data. All of the job satisfaction scores come in lower than the counterparts that do utilize Big Data. The one outlier is “YNY, The company uses Big Data, I do not, but I am affected by it”. This score of 3,28 is almost half a standard deviation lower than the “YYY” and “YYN”. That score brings it closer to a score that is expected when the company does not utilize Big Data. This is a trend that is visible in the data, if the employee does not utilize Big Data, but does feel affected by it, the job satisfaction score drops to the lowest point in both categories. However, trends that are visible with the naked-eye, and observations made from looking at raw data output cannot be the basis of an academic conclusion to this research project. To answer the question if Big Data influences employee job satisfaction, the coefficients model, and specifically the “beta value” must be utilized. The beta value shows the degree by which an independent variable(s) influences a dependent variable(s). In this research project, there is one dependent variable that is constructed of twenty sub-variables, or influencers. Presenting the coefficients table for twenty variables would result in an incredibly large table. The interesting data, like significant changes in “authority” for example, will be rendered invisible by all the data that does not present any interesting finds. To prevent this, a closer look was taken at how the three independent variables influenced the total MSQ score of participants. Especially interesting finds are discussed in a later part of the project.
Chart 4: Coefficients model, Big Data and MSQ What becomes immediately obvious about these results is that both company and personal Big Data usage has a positive effect on employee job satisfaction, representing a 18,1 and 9,3 percent increase respectively. These results are both significant. The initial observation about personally being affected by Big Data was deemed true, having a negative impact by -2,7%. The limit for significance is .05, the significance of the last question is 0,523, about ten times too high to be considered valid. This means that no definitive conclusions can be drawn from the data on if employees feel personally affected by Big Data. There is however room to take a closer look, the only negative relation between Big Data and job satisfaction can provide noteworthy insights into why this negative relation exists.
Coefficients Model Unstandardized
Coefficients Standardized Coefficients t Sig. B Std. Error Beta Constant 3,938 0,088 44,997 0 Does your employer use data driven decision making / Big Data? 0,286 0,066 0,181 -4,356 0 Do you personally use data driven decision making / Big Data in your job? 0,135 0,061 0,093 -2,201 0,028 Are you personally affected by data driven decision making / Big Data? -0,038 0,06 -0,027 0,639 0,523
To summarize:
• Big Data usage inside a firm has a positive impact on employee job satisfaction, employees are 18,1% more satisfied at work when the employer utilizes Big Data. • Personal Big Data usage also has a positive impact on employee job satisfaction,
employees are 9,3% more satisfied at work when they utilize Big Data themselves. • Personally being affected by Big Data usage has a small negative impact on employee
job satisfaction, by -2,7%. This result is however, not significant and no hard conclusions should be made from this.
A closer look
There are some less noticeable finds that have come out of the data analysis. Insights that can potentially shed more light on, and answer not only the “if” but also the “why”.Three influencers
There are three influencers that show a large difference in results between Big Data usage and no Big Data usage. These influencers are “Advancement”, “Authority”, and “Social Status”. All three show lower job satisfaction values when Big Data is not usedChart 5: MSQ 14 Advancement on Big Data
It is shown that Employees who personally use Big Data are more satisfied with their advancement/growth opportunities inside the firm. These differences are not significant, but they shed some light on why employees are more satisfied when the firm uses Big Data. Employees who do not use Big Data, either in the firm or by themselves rate their job satisfaction as “somewhat satisfied”, as opposed to “satisfied”.
Employer use Big Data / DDDM MNQ 14 - Advancement Personally use Big Data / DDDM MNQ 14 - Advancement Personally affected by Big Data / DDDM MNQ 14 - Advancement
Yes N 606 Yes N 494 Yes N 453
Mean 3,33 Mean 3,41 Mean 3,33
No N 247 No N 357 No N 399
Mean 2,78 Mean 2,83 Mean 2,98
Total N 853 Total N 851 Total N 852
Employer use Big Data
/ DDDM
MNQ 10 – Authority
Personally use Big Data
/ DDDM MNQ 10 – Authority Personally affected by Big Data / DDDM MNQ 10 – Authority
Yes N 606 Yes N 494 Yes N 453
Mean 3,29 Mean 3,34 Mean 3,26
No N 247 No N 357 No N 399
Mean 2,86 Mean 2,92 Mean 3,06
Total N 853 Total N 851 Total N 852
Mean 3,17 Mean 3,16 Mean 3,17
Chart 6: MSQ 10 Authority on Big Data
The difference in satisfaction scores for “Authority” is slightly smaller than the effect of “Advancement”. Yet it offers an interesting insight, in this study, employees are more positive about their positioning the firm. The complete questions went as follows: “the chance to tell people what to do”. So, it turns out that employees that utilize Big Data and or are affected by it, feel more satisfied with the level of authority they have over other employees. Chart 7: MSQ 4 Social Status on Big Data The last peculiar influencer is “social status”. Employees who use Big Data or who work in a firm that uses Big Data rate their themselves to be 18% happier with their current social status.
Employer use Big Data
/ DDDM
MNQ 4 - Social Status
Personally use Big Data
/ DDDM MNQ 4 - Social Status Personally affected by Big Data / DDDM MNQ 4 - Social Status
Yes N 606 Yes N 494 Yes N 453
Mean 3,42 Mean 3,48 Mean 3,42
No N 247 No N 357 No N 399
Mean 2,89 Mean 2,97 Mean 3,09
Total N 853 Total N 851 Total N 852
Other independent variables
In the survey, a number of independent variables were tested that were not subject to the main analysis of the research project. The survey was designed to be as accessible to an as large as possible group, because of the lack of previous research that could be built upon. The additional independent variables were implemented in the survey in the hope that they would provide more specific insights.After the analysis had been performed, it became clear that participants that hold a managerial position inside a firm that uses Big Data are 11,6% more satisfied with their job, when compared to non-managerial employees. Another interesting find is that industry has no impact on employee job satisfaction. There is no industry that differs significantly from the others in employee job satisfaction. The same can be said for departments. There is no significant difference in level of job satisfaction between different departments in a firm. The three largest department samples; operations, customer services and information technology (IT) all have more than a hundred respondents and differ by less that .1 in average job satisfaction.
The last two independent variables, “time employed by firm” and “time inside position” also don’t provide significant differences in employee job satisfaction. For time employed, the results all fall between 3,3 and 3,4, a minor difference with no significance. The same can be said for “time in position”, there is a slight tendency towards a higher job satisfaction the longer an employee works inside the same position, but this is less than 0,2.
Conclusion
The conclusion of this research project is not one that can be properly written down in one sentence, but this is the closest: Big Data usage inside a firm has a positive impact on employee job satisfaction, however, feeling affected by Big Data has a negative impact on job satisfaction The results of this research project fall in line with most research work into employee job satisfaction, most respondents are moderately to very satisfied with their jobs. The fact however, that in this relatively small study, it has been shown that Big Data can have an influence on employee job satisfaction paves the way for future –more elaborate- studies into this area. The power of this conclusion does not lay in the mean-job satisfaction score or the percentage of job satisfaction that Big Data usage is responsible for. The power lies in the influence that Big Data usage has on the twenty influencers of employee job satisfaction. Being able to pinpoint where employees feel most affected by Big Data usage can not only guide future research, it is also immediately usable by HR managers in organizations. Especially the influencers of social status, authority and advancement are affected by Big Data usage. In total, there are eight influencers that differ more than 0,45 between Big Data usage or not.
Discussion
In this part, a closer look will be taken at the results. How does the outcome of this study tie back into the previously discussed literature? Does this project have a practical implication on business that utilize Big Data, or do not yet? Last but not least, the limitations of this project will be discussed and future research recommendations will be shown.Results
The results show that Big Data usage inside a firm leads to a higher employee job satisfaction. The specific result was 3,53 out of 5 for Big Data usage, and 3,19 for non-Big Data usage. These results are in line on what the manual for the MSQ describes, it is stated that a result between 3 and 4, or between satisfied and very satisfied is expected (Weiss et al., 1967). This result, is thus not surprising, but interesting nonetheless. It shows that changes have occurred within the twenty influencers, who together create the mean employee job satisfaction value (Van Saane et al., 2003). If we take a closer look and split “Big Data usage” and “No Big Data Usage” into four different variables each, we see that there are more interesting finds. If the company and the employee both use Big Data, and the employee is affected by Big Data the highest job satisfaction grade is achieved. This can for example indicate that companies that work top-to-bottom with Big Data have happier employees, it can also mean that employees who want to work with Big Data, and thus go to a company that does so, are happier. This cannot be confirmed unless qualitative data collection through interviews that provide more detailed insight into what makes employees satisfied at work. This however, is not possible on theThe lowest score when the firm does use Big Data is when the employee does not use Big Data but does feel affected by it. The score drops from 3,59 to 3,28. It is only a drop of 0,3, but it is a 9,45% decrease in job satisfaction. This drop could potentially be explained by social change that occurs. In his paper on social change that can occur when technologies change, Barley noticed a change in social between nurses, doctors and technicians (Barley, 1986). The doctors, that did not know how to operate the new machinery dropped down in hierarchical position, in favor of the technicians, who were the ones that knew how to operate the new systems. The same can occur now, with the influx of new technologies and ideas like Big Data. Employees that work in a company that started to adopt Big Data, but do not know how to use it, or simply in their job do not utilize it, can feel left out. On the other hand, the engineers that do know how it works, and potentially get paid more because of it, can see a rise in hierarchical position and, with it being one of the twenty influencers, a rise in employee job satisfaction. This is however hypothetical and can once again only be confirmed with interviews. It is an interesting find that this study could potentially confirm the work by (Barley, 1986) more than twenty years later, and over a wide selection of industries.
I do not feel comfortable to draw any conclusions from the two lowest scores, respectively 3,02 and 3,01. These low scores occur with the answer option “the company does not use Big Data and I am not affected by it, but I personally do use it.” And “The company does not use Big Data, I don’t either but I am affected by it”. The number of respondents that caused these low ratings are 20 and 11. These numbers are too low to draw a valid conclusion from. In addition to this, and as will be explained later, it is unsure what the respondents define as “affected by it” and “I myself use it”. The respondent can misinterpret the question
and say that he is affected by Facebook’s or Amazon’s Big Data. Which is of course, not what we are after in this research. For that reason, these results cannot be accepted for more than an interesting observation.
As stated many times before, employee job satisfaction is constructed of twenty influencers (Aziri, 2011). After the data analysis, it showed that there were three influencers that deviated more than one standard deviation from the mean, these are “Advancement”, “Authority” and “Social Status”. These are the three influencers that are the biggest cause of change between the employees whose company uses Big Data, and whose company does not. Again, there are many different reasons why specifically these three influencers differ more than one standard deviation from the mean, yet it again brings it back to the article by (Barley, 1986). All three influencers seem like they confirm the research, employees that do use Big Data feel better about their career advancement opportunities, about the level of authority they have over co-workers, and about their own social status at work. These all seem to indicate that people that are not using Big Data are lagging behind and are aware of the change that is occurring. Seeing this change though, does not mean that social change is the reason behind these ratings. Once again, it is also just as likely that employees who work with Big Data do this in firms that are more pleasant to work for, or the employees are younger and are happier at work. The change we can see, observe and discuss, we cannot stick a reason to it. It does provide an opportunity for future researchers to perform a qualitative study into the same research question as this study. Understanding the “why” behind the numbers can help in creating a theory. This theory can confirm what Barley said more than twenty years ago, or it can provide a whole different reason as to why employees
that work in companies that use Big Data are more satisfied with their job. The last option, of course, is that new research will have to state that this research had flaws and that the results are incorrect, which is always a possibility with academic research.
Limitations
There are a number of limitations that will have to be kept in mind when reviewing this study. First, there is the issue of the definitions. There is no widely accepted definition of Big Data. The definitions of Big Data that have been written are academic and not applicable in the professional world. Asking a respondent if their company uses data that has a high variety, velocity and volume is not something that would benefit the data collection. So, it was decided that the respondents could self-define what Big Data means. I do believe that data driven decision making removes some of the confusion, but it is possible that the respondent(s) misinterpreted the question. The question: “are you affected by Big Data usage?”, is possible to be misinterpreted. Some examples are: “I am able to email any of my 50.000 colleagues, I think that is Big Data.”, Amazon knows what I like to buy, that is Big Data and I am affected by this. In future research, this question will have to be written in such a way that it less likely to be misinterpreted. However, for this to be possible, there has to be an easier way to describe Big Data, or the specific type of Big Data that you are looking for. Another limitation of the study is that it utilized the Minnesota Job Satisfaction Questionnaire. This questionnaire was first designed in the 1960’s and has since not been updated more than once. It is possible that in the years since, employees’ needs and expectations have changed. In the 1960’s there was no internet, there was no social media, there was no mobile PC even. It is possible that since then, many influencers of employee job
satisfaction have changed. It was not found in the literature review, however that can also mean that no one thought about it yet.
The scope and the time of the project did not allow for a critical selection of respondents. The study would have had more potential impact if it would, for example, have measured “How is Big Data affecting employee job satisfaction in the entertainment industry in the Netherlands.” That, together with many more potential combinations would not have been possible to answer. I believe, that the more than thousand responses from all over the world were the most that I could collect within this time frame and scope. In the future, a more specific sample would benefit not only the impact of the conclusion, but also the practical implications. This research, if performed by a professor or PhD candidate carries more weight and might persuade companies to work with the university.
Practical Implications
The practical implications of this study are what companies or researchers decide that they are. This has been the first research project into the effect that Big Data usage can have on employees. The results show an increase in job satisfaction when companies do utilize Big Data. I believe that it would have been most interesting for companies if the research would have shown that there are one or two influencers of job satisfaction that are negatively affected by Big Data. It would have been possible to further investigate these influencers, especially for HR managers, this would have been a very interesting find that can help them in their own jobs. Also, the find that Big Data usage in a firm improves employee job satisfaction does not mean that the work in itself is better. It is also possible that the type ofcompany that utilizes Big Data attracts a certain type of employee that is happier at their job or reacts extremely positively to new technologies or Big Data usage. The practical advice that organizations can take from this study is that they have to be mindful of the potential effects that Big Data usage can have on their employees’ happiness at their job, and thus as stated before, organizational wide productivity. More, in-depth, practical implications or advice cannot be given with 100% accuracy, and thus should not be provided.
Proposal for further analysis
I would describe the research project as successful. In a short period of time, it has been shown that Big Data has a positive impact on employee job satisfaction. As stated in the pages above, no hard conclusions or practical implications can be drawn from this study yet. The scope and most notably, sample selection, is not specific enough. However, there is ample opportunity for future scientific research into the topic of Big Data and employee job satisfaction. If first, one company, or one industry in one country is researched, further research could build upon this, this research could actually provide firms, and especially (HR) managers with information with which they can improve the working conditions of the employees. As stated before, an in-depth, qualitative, continuation of this research project can potentially confirm the work by Barley, or expand upon it for new technologies in the 21st century. More research will also start to show if the MSQ or another job satisfaction measurement tool like the JDI is still valid. There was no time to analyze the validity of both tools, the MSQ was chosen due to availability, applicability and time. Future research will beone of the first to show the effects that Big Data usage can have on a firm, to show the effects that technological advancement can have on the workforce.
