Corps de l’article
Large-scale, government-conducted workplace surveys can prove a major boon to academics, providing data that enables them to address a whole host of questions and issues with a level of sophistication otherwise not possible. Yet there are a number of potential down sides to these surveys, not the least of which because of the risk of inappropriate government involvement in the production of knowledge, but also because the measures they generate can be poor, negatively affecting the quality of research. There is also reason to worry that they can distort research, away from competing methodologies, and away from topics not covered by the survey in question. As discussed in my article (and elsewhere), these and other possible limitations mean that, while government surveys can potentially represent a “new dawn,” they can also potentially represent a “bad moon rising.” The main purpose of my article was to assess the extent to which three such surveys appeared to fit either of these characterizations: the 1995 AWIRS (Australia), the 1998 WERS (the U.K.), and the 1999 WES (Canada). The WES, and its implications for IR, were of particular concern.
I defined IR as “the relations between labour (union and nonunion) and management and the context within which they interact” (p. 5). This definition included the organization of work and HRM as well as conventional union-management relations topics. One may quarrel with it, as do Professors Chaykowski and Slotsve (C&S), but it seemed to me that any survey that did not do a proper job of addressing these topics would not be of much use to mainstream IR scholars, and hence at best would not represent a “new dawn” for the field.
Based largely on a content analysis (table 2: 12), I found that each of the three surveys had various strengths and weaknesses. With regard to the WES, I concluded that it appeared to be technically sound and would likely yield high quality data on labour market and productivity issues. But I also concluded that it was not an IR survey. In marked contrast to its U.K. and Australian counterparts, it was weak in mainstream IR content and, in particular, contained very little on either labour unions and collective bargaining or worker attitudes, beliefs, and social-psychological outcomes (e.g., stress). I also commented that it appeared to be driven by a “managerialist” policy agenda and ideology, one that marginalizes mainstream IR topics and institutions. Finally, I expressed concern about control of the data. At the time my article went to press, the intention was for the government to select a small group of individuals to write papers analyzing the data, and to severely restrict access by outsiders.
An underlying concern was that the government was engaging not only in the production of knowledge, but also in its control, and that it was doing so in a way that could serve to promote a particular ideology. I was hesitant to state this in strong terms, because I have had a high regard for Statistics Canada and did not think this was by intention. I was not even sure that this concern was well-founded, and in fact hoped that it was not. But it was something that worried me considerably in view of trends I have noticed within the federal government, and I believed that it was, at the very least, worth voicing. Coupled with my other concerns, it led me to conclude that the WES may represent a bad moon rising for Canadian IR.
In reading the C&S comment, I had hoped to find something that would allay my concerns. Unhappily, I did not. After quibbling with my metaphors, C&S respond to the possible down sides I identified for government-sponsored workplace surveys, seemingly unaware that my purpose was not to prejudge the three surveys but rather to establish potential problems and hence criteria against which these surveys could be evaluated (see p. 10). C&S then essentially argue that the WES was not an IR survey and agree that it reflected a managerialist perspective, appearing to defend this perspective and its marginalization of labour unions. In the final part of their comment, they basically argue that the WES is great for labour market analysis. In the main, these arguments only substantiate and reinforce the concerns raised in my article.
I am especially perplexed as to how a survey with such weak coverage of core labour and employee relations topics can be considered conducive to meaningful analysis of (in the words of C & S) “institutional arrangements... including the determination of pay policies, workplace practices, and employment arrangements, etc.” (p. 387) or, more generally, of “workplace change and emerging institutional arrangements” (p. 387). The underlying issue here is not how one defines IR, nor is it what type of research one prefers, nor is it even what the WES was intended to accomplish. Rather, it is how one views workers and their representatives and whether one thinks that they matter.
To be fair, C&S do a reasonable job of establishing some of the research possibilities the WES holds for labour market analysis (if it were not for access problems). They also illustrate that those involved in the design of the WES have, from a technical standpoint, done an impressive job. I am in general agreement on both points, as should be apparent from my article (see especially pp. 25–26). But there is little in the C&S comment that would cause me to change my assessment of the WES and its implications for Canadian IR. Indeed, this comment has convinced me even more of the need either for changes to the WES or for a separate survey. Sadly, neither seems likely at the present point in time.
At the time, outsiders could only ask Statistics Canada to run the data on their behalf, providing them with a pre-specified model from a dummy data set. This remains the case for the employer data. In contrast, the employee data is now available through a handful of research centres across the country. But, as C&S suggest (see p. 387), it is the collection of employer data, and the ability to link them to employee data, that represent the main potential contributions of the WES. Moreover, access to the employee data still requires that one first write a proposal, have it approved by Statistics Canada, and submit to a security check. These restrictions may or may not be justified on confidentiality grounds, but they seem at odds with any notion of free academic inquiry, and I am unaware of any such procedures elsewhere. In the U.K., not only is the WERS data set freely available, there has been a significant effort to encourage its use by academics.
If I am to do penance, it should be for the bottom paragraph on page 18. To my chagrin, I realized shortly after the article went to press that I had misread a table (5–2) in the WERS report. Readers should disregard the second and third sentences of that paragraph