Figure 1 Overlap between measures of ill-treatment at work and ‘bullying’
In Part Two we explain the key quantitative results for different types of workplace troubles. Supporting data, for example the tables which summarise our multivariate analysis, are available at www.bloomsburyacademic.com/view/Trouble-At-Work/book-ba-9781849664677.xml. Unless we indicate otherwise, all of the data we discuss are drawn from our nationally representative survey of employees, the British Workplace Behaviour Survey (BWBS). The three chapters in Part Two are organised around themes which have been suggested by the factor analysis of our survey data which revealed three types of troubling experiences at work: unreasonable treatment (Chapter 2), incivility (or denigration) and disrespect (Chapter 3), and finally violence and injury Chapter 4). In each of the three chapters, we discuss
To illuminate the experiences and meanings behind the statistics, we illustrate each of the main results with interview material taken from the qualitative phase of our research (described in detail in Part Three). This introduction provides just a few important details about our survey methodology to illustrate some of the key choices we made when gathering quantitative data on trouble at work.
In Chapter 1, we explained our decision to frame our research as the study of ill-treatment, or trouble at work, rather than workplace bullying, but the question naturally arises of how much overlap exists between these concepts when they are put to work in surveys. A variety of (usually non-representative) studies have suggested that almost all behaviour that people regard as bullying can be described as ill-treatment. These studies have also suggested, however, that there is plenty of ill-treatment people would not think of as bullying. Researchers have discovered this by asking a standard battery of questions about ill-treatment such as the NAQ (Einarsen and Raknes 1997) along with a question asking respondents if they have been bullied. This technique has been used to define a threshold (particular frequency, number or seriousness of events) above which researchers can claim evidence of ‘bullying’, regardless of whether respondents themselves recognise they have been bullied.1 While we were not interested in imposing a definition of bullying in this way, we used a similar methodology to explore the overlap between ill-treatment and bullying in a smaller (representative) pilot survey for our BWBS.
In the pilot survey, face-to-face interviews were conducted with 1,083 employees during the spring of 2007. They were asked the questions in the NAQ battery and whether they had been bullied at work in the past two years. Only 76 respondents answered ‘yes’ to the bullying question, whereas affirmative responses on the individual NAQ items ranged from 87 to 392 respondents. As Figure 1 illustrates, most people who experienced some form of ill-treatment did not label it ‘bullying’.
Figure 1 shows that there was quite considerable variation in the proportion self-identifying as bullying depending on which type of ill-treatment was reported. For example, employees were not as likely to think they had been bullied if they had experienced an unmanageable workload or impossible deadlines (Item 21) as they were when they had been ignored, excluded or sent to Coventry (Item 6) or had allegations made against them (Item 17). Do such variations indicate that the label of bullying is applied consistently and predictably?
We followed up the pilot by undertaking some ‘cognitive testing’ to find out more about why people answered questions about bullying and ill-treatment as they did. In particular, we wanted to know anything that might help us to understand why ill-treatment was sometimes seen as bullying and sometimes not. Cognitive testing is a form of qualitative interviewing which allows researchers to examine how far questions are understood in similar ways by respondents, whether they have sufficient information to understand and answer the questions asked, and to develop suggestions for removing ambiguity in the wording of particular questions. Sixty of these cognitive interviews were conducted in the summer of 2007 (see Fevre, Robinson, Jones and Lewis 2010 for more detail about this process). The interviews focused in particular on respondents’ interpretations of different definitions of ‘bullying’ and the individual NAQ items. Respondents were asked about their experience of ‘bullying at work’, and then the interviewer initiated a discussion of what the respondent understood about the question, using probes such as ‘how would you describe “bullying”?’ and ‘what sorts of incidents are you thinking of?’.
The cognitive interviews soon established that there were major problems with providing a comprehensive yet clear definition of ‘workplace bullying’ that was interpreted in similar ways by different groups of respondents. For example, some respondents reported that they viewed bullying as primarily physical abuse, whereas others saw bullying as more about verbal activities, such as teasing. One respondent argued that bullying could well be negative treatment by the organisation as a whole rather than by one or some particular individuals. When asked to define bullying, a wide variety of examples were given, including
Many interviewees felt that all of the NAQ items might be considered ‘bullying’ depending on the circumstances in which they occurred. These referred to the job, the individual, how the behaviour was intended by the perpetrator and how it was received by the victim. The perception of an overlap between bullying and ill-treatment, therefore, not only varied between individuals but also changed for the same individual, depending on the circumstances.
Our pilot survey therefore suggested that there was far too much inconsistency in the application of the label of bullying for us to be able to construct a general explanation of why people regarded some, but not all, ill-treatment as bullying. Whatever limitations we, as sociologists, see in the use of the concept to further research, it seems that the bullying label is not sufficiently familiar to, and similarly understood by, British employees to allow a general explanation. One measure of this lack of familiarity and common understanding is given in Table 1, which contains three lists of typical bullying behaviours posted by organisations which consider it their job to inform employees on whether they have been bullied or not. Not only is there some disagreement between these lists, but there is sometimes little correspondence with the patterns observed in Figure 1. Most notably, whereas only 16 per cent of employees in the pilot survey who had been given ‘tasks with unreasonable or impossible targets or deadlines’ considered themselves bullied, this type of ill-treatment is mentioned in all three lists of bullying behaviour.
This kind of information is, presumably, intended to help British employees reach a common understanding of what workplace bullying is, but the evidence from the lists themselves, as much as our pilot survey, suggests we are some way short of such a common understanding at present. For this reason, we are unable to say very much that can help us to understand why some forms of ill-treatment qualify as bullying in employees’ minds and others do not.
The BWBS is the product of the extensive piloting and refinement just described. The BWBS is a structured survey that was administered to a representative sample of UK employees (or those with experience of employment in the previous two years) during the winter months of 2007–2008.2 The total weighted numbers responding to the BWBS were 3,979. Of these, 14.6 per cent were not employed but had experience of employment in the previous two years (the rest were currently employed to some level).
The BWBS gathered data on individual demographic factors, including age, income, ethnicity, gender and sexual orientation, along with data on job and workplace characteristics including occupation, industry, size of workplace, trade union membership, gender/ethnic/age composition of workplace and respondents’ views about their levels of control over the pace and nature of their work. We also sought to ascertain workers’ perceptions about those responsible for ill-treatment in the workplace and why it occurs.
The cognitive testing described above helped us to revise the NAQ for use as a battery of questions on ill-treatment (for further details on the way items were revised, see Fevre, Robinson, Jones and Lewis 2010). When they had answered all the NAQ questions, respondents were given an opportunity to confirm or deny the choices they made about which of the 21 items they had experienced; subsequently there were some small reductions to the incidence rates across the items (usually 1–2 per cent, never more than 5 per cent). Because these more conservative ‘confirmed’ estimates are more accurate, we use these data in all analyses presented in this book. We also asked respondents whether they had witnessed or perpetrated each of the 21 items. Finally, respondents were able to remark on any aspect of their experiences in an unrestricted way at a couple of points in the survey. So, although mostly quantitative, we do have some illuminating qualitative data.
A final section of the BWBS gathered information about who was responsible for perpetrating the various types of ill-treatment and why they might have done it. Details on perpetrators included their gender and ethnicity and whether they were ‘internal’ to the workplace (such as fellow co-workers, subordinates or employers) or ‘external’ (such as clients, customers or members of the public). Respondents were then able to offer their own judgement about the causes of the ill-treatment. This information was gathered by offering 20 potential reasons that respondents could select, falling into four broad categories: characteristics of the workplace (e.g. position in the organisation or feeling that ‘it’s just the way things are at work’), characteristics of other employees (e.g. members of a group or clique who exclude other employees from it), the respondent’s demographic characteristics (e.g. race, age and disability) or other characteristics (e.g. accent and trade union membership). These categories were derived from extant research and our cognitive testing process.
Even though the funding for our project was generous, we could not gather this more detailed information from every respondent due to financial constraints. Instead, only respondents who said they had experienced three or more of the 21 types of ill-treatment were asked these follow-up questions and, no matter how many types of ill-treatment they reported, we only asked follow-up questions about three of them. A methodology was employed to select the three types that we judged to be most serious (e.g. all of those who experienced ‘actual physical violence at work’ as one of three or more types of ill-treatment were routed into the follow-up section of the survey). Therefore the data on perpetrators of, and explanations for, ill-treatment at work reflect (a) respondents who experienced multiple forms of ill-treatment and (b) what we considered to be the more serious types of ill-treatment. It is for this reason that we consider them to be the ‘troubled minority’. Their experiences in particular enable a deeper understanding of trouble at work.
Having multiple measures of ‘trouble at work’ is a good thing. As noted in the description of cognitive testing above, getting more than one reliable measure of ill-treatment at work was no easy task. Using past research and the comments of ‘people on the street’ to guide our measurement of such a complex phenomenon, however, proved very useful. In Table 2 we present the final version of the 21 different items that were included in our survey. To facilitate our understanding of the items and how they potentially group together, we conducted a factor analysis which revealed three types of ill-treatment. This is a common technique in social research, including research on problems at work such as bullying. For example, using data collected in the late 1990s in the United Kingdom, Einarsen, Hoel and Notelaers (2009) found three types of bullying: personal, work related and intimidation. Researchers then use these types in order to refine their explanations of bullying (e.g. Hauge et al. 2007).
The three factors identified in our analysis are also presented in Table 2, along with some descriptive statistics indicating that they are very robust and reliable factors. In other words, we have at our disposal 21 specific items and also three broader factors with which to discuss our findings. As indicated earlier, the three factors provide the structure for the three empirical chapters that follow. It is important to note that the three factors of unreasonable treatment, incivility and disrespect, and violence and injury can be considered as distinct experiences. In other words, the data indicate that they are qualitatively different from each other and can be treated as separate entities. This is not to say, however, that some items clearly fell into only one category. Indeed, three of the items could have been placed in either the unreasonable treatment factor or the incivility and disrespect factor (‘people excluding you from their group’, ‘hints or signals that you should quit your job’ and ‘persistent criticism of your work or performance which is unfair’). So, for these items, the distinction between whether they were better suited as measures of unreasonable treatment or something more akin to denigration or disrespect was a bit more blurred. It is not surprising that some items have both a social and a work-related aspect. Likewise, the item ‘feeling threatened in any way while at work’ could have been considered to belong to either incivility and disrespect or violence and injury. We made our choices about which items belonged to which factors based on not only an empirical but also a conceptual framework.3 This is just worth bearing in mind, as sometimes social science cannot be as exact or tidy as we might wish.
The chapters in Part Two deal with each of these factors in substantial detail. Before considering them separately, however, it will be useful to show how they interrelate. As the Venn diagram presented in Figure 2 shows, most of our respondents experienced more than one type of ill-treatment at work. For example, 33 per cent experienced both unreasonable management and incivility and disrespect. A much smaller proportion (6 per cent) experienced violence, and nearly all of them (5 per cent) experienced the other two types as well. So violence was never experienced on its own, and only when combined with some other type of ill-treatment. The diagram is useful for getting a sense of the overlapping and interlocking nature of these different aspects of trouble at work.
Before drawing this introduction to a close, it is important to note that there is also a high degree of overlap between experiencing, witnessing and perpetrating trouble at work (see Figure 3). Our analyses of the data revealed that across all 21 items there were significant positive correlations (indicated by asterisks) between these three measures. In other words, people who experienced a particular type of ill-treatment were also more likely to report witnessing it and even to admit perpetrating it themselves. For example, the top three items in terms of highest incidence were very similar for experiencing, witnessing and perpetrating:
This is not to say that every ‘victim’ is also an ‘offender’ but rather that ill-treatment at work seems to affect people in a variety of ways – as victims, witnesses and perpetrators. This finding lends support to the main argument of our book, which is that it is the characteristics of troubled workplaces (i.e. ‘hotspots’) rather than of people (i.e. those especially prone to recognise, perpetrate or suffer workplace bullying) that will provide the best explanations of ill-treatment at work.
