Wednesday, April 26, 2023

Workplace safety versus worker privacy

Employers often struggle to balance their interest in improving workplace safety with workers’ right to privacy. For example, the history of workplace drug and alcohol testing often turns on the circumstances under which is it appropriate for an employer to require a worker to submit to testing (e.g., post incident, suspicion of impairment, randomly).

Employers often assert (and behave as if) workplace safety considerations trump workers’ privacy rights. This is good rhetorical terrain for employers to argue from because it frames opponents of testing regimes as being opposed to (or at least not prioritizing) safety.

When there is an absence of evidence to support the efficacy of initiatives like testing (which is often the case), employers can revert to some version of ”better safe than sorry” as a rationale to justify their position. This rationale runs contrary to the generally acceptable proposition that they who make a claim must substantiate it.

I recently read a 2018 arbitration decision about cognitive testing for Edmonton transit drivers that was quite interesting. You can find the full decision on under this reference:

Amalgamated Transit Union, Local No. 569 v Edmonton (City), 2018 CanLII 82319 (AB GAA)

The nub of the case (and I’m paraphrasing pretty liberally) is there had been two bus-related pedestrian fatalities and the government regulator required the city to implement a transit driver evaluation policy. The city’s response was to implement mandatory (1) road testing and (2) cognitive testing.

The cognitive testing included a computerized screening tool. If workers scored above a threshold on the tool, they were then suspended with pay and required to undergo medical evaluation. (There was no evidence that the two fatalities were related to cognitive impairment of the drivers.) The medical testing and release of information violated these workers’ privacy.

The grievance basically asserts that the city had no legal or factual basis for implementing (1) the mandatory screening and, for those who fail the screening, (2) the follow-on medical assessment. The union also argued the cognitive screening test, having been developed primarily to screen for cognition decay in older drivers, was not a valid test for an otherwise healthy population.

In the end, the arbitration panel ruled based upon the union’s argument around the testing being unreasonable and declined to address the (rather troubling) issue of the test’s validity and reliability. What makes this case interesting is that, while the matter awaited adjudication, the employer proceeded with the testing under the “work now, grieve later” principle and we actually have results about the efficacy of the testing.

The firm providing the testing predicted that, of the 1535 drivers tested, 1-2% would be suffering from cognitive impairment (so 15 to 31 drivers, roughly). At the time of the hearing, only one driver was confirmed as having cognitive impairment and a second driver’s status was undetermined (so the true rate of cognitive impairment was 0.12%, or one-tenth the rate the testing firm asserted). The screening tool sent 88 drivers for medical assessment, of whom the vast majority were false positives. (A small number of other drivers returned to work with modest work restrictions related to other medical conditions.)

This sort of outcome (where the proponents vastly over-state the true level of risk in order to push forward with testing) is not uncommon. Random drug testing is another example where, despite decades of effort, there is no good evidence that random testing reduces injuries. Certainly, we would expect a company that is selling testing to make claims that create the appearance that their product is valuable to potential clients. And, these kinds of circumstances are why, generally speaking, we expect those who make a claim to substantiate it.

It is also interesting to note the uneven application of the better safe than sorry principle by employers.
  • When it is employees who bear the cost of an OHS intervention (i.e., have their privacy invaded), employers are happy to play by better safe than sorry and not demand high levels of proof. 
  • When employers must bear the cost (e.g., face disrupted production or higher material costs) because workers have concerns about unsafe working conditions or materials, employers generally demand very high levels of proof before they will alter their processes. 
This existence of this double standard speaks to which (and whose) interests are prioritized in workplace regulation.

-- Bob Barnetson

Wednesday, April 19, 2023

Podcast: Vriend 25 Years On

The Well Endowed Podcast is publishing a series on the 25th Anniversary of the Vriend decision. While sexual orientation had been deemed an analogous ground under s.15 of the Charter, Alberta had refused to include sexual orientation as a prohibited ground in its human rights legislation. This permitted discrimination on the basis of sexual orientation by private actors. In Vriend, the Supreme Court found that this exclusion offended the Charter and should be read into human rights legislation.

Vriend was ground-breaking litigation and this multi-part podcast begins by examining how Canada and Alberta treated members of the LGBTQ2+ community in the decades leading up to 1991 (when Vriend was fire by an Alberta college because of his sexual orientation). The degree of discrimination faced by the LGBTQ2+ detailed in the first episode is, frankly, shocking.

This decision has had significant impacts for labour relations, including the Charter, human rights, immigration, and sex work.

-- Bob Barnetson

Tuesday, April 11, 2023

Complaint over “Mafia-esque” union Xmas cards resolved

An unfair labour practice complaint, alleging Christmas cards sent by a union to the employer’s bargaining team amounted to “Mafia-esque” intimidation, provides insight into the unexpected impact that Alberta’s restrictive picketing laws may have on union pressure tactics during bargaining.

Alberta’s picketing laws

In 2019, the United Conservative Party (UCP) formed government in Alberta. In the summer of 2020, the UCP passed Bill 32: Restoring Balance in Alberta’s Workplaces Act (2020). This act substantially restricted picketing activities by:
  • rendering it illegal to obstruct or impede someone from crossing a picket line,
  • requiring a union to seek Labour Board permission to engage in secondary picketing, and
  • allowing the Labour Board to determine the conditions of any secondary picketing.
These changes effectively rendered legal picketing ineffective and effective picketing illegal. This, in turn, reduced the ability of workers and unions to exert pressure on the employer to move at the bargaining table (which was the intent of the legislation).

Christmas card “intimidation”

Athabasca University Faculty Association (AUFA) served notice to bargain in the spring 2020. By the late autumn of 2021, the employer had not yet provided its monetary proposal and bargaining was stalled. The union began applying pressure in order to generate movement. For example, it filed a bargaining in bad faith complaint with the Labour Board. This proved predictably ineffective due to delay in getting the matter to hearing in a timely way.

The union also began experimenting with the alternative strike tactics that it had developed, in part, because of Alberta’s restrictions on effective picketing. These tactics included choking-off revenue by applying reputational pressure. The first effort was a 12 Days of Christmas meme campaign based on the song “All I want for Christmas is my two front teeth.” Members tweeted these memes at the employer and its bargaining team.

At the end of the online campaign, the most popular meme was then made into a Christmas card. Copies of the card were mailed to homes of the university president and bargaining team co-chairs. In January of 2022, the employer filed an unfair labour practices complaint, alleging the cards were intended to be intimidating, an implicit threat to the safety of the employer’s representatives and their families, and were a “Mafia-esque” tactic.

In April of 2023, the union and the employer settled the unfair. In this settlement, the union agreed, in future, to collect personal information in accordance with Alberta’s privacy legislation (which it is legally bound to do in any case). The union also “acknowledged that those who received the Christmas Cards and members of their families felt that they had been intimidated and harassed.” This settlement is, I think, best read as saving the union the financial cost of the hearing and saving the employer the political cost of losing.


The UCP’s changes to Alberta’s labour laws were intended make it more difficult for unions to exert meaningful pressure on employers via picketing that disrupts operations. The desired effect was to attenuate unions’ abilities to make meaningful contract gains.

These changes do not, however, eliminate the need for workers and unions to exert pressure on employers during bargaining. That doesn’t mean these picketing restrictions have no effect on union power. Rather, they just push unions to (1) develop alternative tactics and/or (2) ignore the law and take whatever punishment that entails.

On the surface, mailing Christmas cards to the boss was a very mild alternative pressure tactic. Yet, it triggered a very strong response from the employer. This reaction may have been an effort by the employer to generate some pearl-clutching and internal dissent within the union membership by equating the union with the mob. Or it may have been designed to generate litigation to trade away against the union’s bad faith bargaining complaint.

The tenor of the employer’s complaint, though, suggests real outrage. (I recognize these explanations are not mutually exclusive.) The memes and cards may have driven home for the recipients that collective bargaining can have real world consequences for bosses (just like it always does for workers). It may also have highlighted that the government restricting traditional picketing activities increases the likelihood that unions will expand their tactics to include applying pressure directly on bosses.

While the overall effectiveness of this sort of pressure tactic remains unclear, the employer’s over-reaction to the Christmas card complaint certainly suggests that bosses intensely dislike even the mildest personal pressure and are surprisingly easy, according to their own complaint, to intimidate. This, in turn, tells unions that they should continue to explore this space.

There is significant room to escalate these forms of personally targeted pressure while still staying within the bounds of legal leafletting activity. And the nothing-burger settlement of the employer’s unfair suggests the cost to the union of using these tactics is low.

-- Bob Barnetson

Wednesday, April 5, 2023

Research: Some thoughts on an arbitrator-selection project

A widely held belief by labour-relations practitioners is that a union or employer can influence its likelihood of success at grievance arbitration via arbitrator selection. That is to say, practitioners believe that arbitrators have observable, stable, and significant decision tendencies that affect their chances of winning an arbitration. That’s not to say that facts, evidence, and jurisprudence don’t matter. But a party rarely has much control over these factors (i.e., you have the case you have). What is within a party’s control (more or less) is who hears the case.

“Who you get affects what you get” is an interesting conjecture that, if true, could suggests strategies for parties to adopt to increase their odds of success. Taking this conjecture as starting point, a colleague and I have been working to test whether or not arbitrators have observable, stable, and significant decision tendencies that can be used to predict future decisions. This is a conceptually and methodologically challenging project and I thought thinking aloud about this project might be of interest to SOSC 366 and IDRL 316 students (as well as a useful process of metacognition for me).

Our underlying approach is rooted in social constructivism. This theory, loosely speaking, asserts that there is infinite stimuli in the world. What stimuli we pay attention to and how we interpret those stimuli is shaped by our experiences, values, and beliefs. In this way, we socially construct our world. For example, bosses often frame conflict with workers as rooted in a communication or attitude problem rather than as an expression of conflicting interests.

Social constructivism is, I think, a reasonable starting point for analyzing arbitration decisions. Arbitrators are normally tasked with making complex decisions, often by sifting and weighing evidence and arguments and applying principles and precepts to come to decisions about what has happened and what ought to happen. This kind of work entails exercising significant judgment about what information is important and what it means. While arbitrators carefully apply many useful conventions and tests when making these decisions (e.g., around witness credibility), social constructivism assumes that arbitrators are ultimately relying upon their experiences, values, and beliefs (more on this below) when exercising their judgment.

This approach suggests, to the degree that arbitrators have, among themselves, different experiences, values, and beliefs, they might come to different conclusions in a case when faced with the same information. (There is some research that concludes (1) arbitrators are consistent in the factors they consider their decisions over time, and (2) different arbitrators can come to very different conclusions when deciding identical cases. This research broadly accords with the social constructivist approach we've adopted.)

The nature of grievance arbitration makes it hard to test the “who you get affects what you get” conjecture. Facts, evidence, and jurisprudence clearly affect individual decisions in important ways. The unique nature of each case impedes direct comparisons of decisions rendered by different arbitrators. And, to the degree that social constructivism occurs outside of our awareness, its operation may be difficult to see in arbitration decisions (although I do acknowledge how arbitrators carefully walk readers through the facts and arguments, and their analyses).

A different approach to testing this conjecture (and the one we've settled on) is to look at patterns in arbitrator decision-making over a large number of cases. The idea here is that the unique facts of each case (which will sometimes favour the employer and sometimes the union) will “average out” over a large enough number of cases (>1000 at this point) to create a baseline of wins and losses. Once the dataset is coded, we can then assess:
  1. whether there are significant differences in the win-loss ratios (i.e., decision tendencies) among arbitrators and compared to the baseline,
  2. how stable these decision tendencies are over time, and
  3. the degree to which these tendencies are usefully predictive of future decisions.
I’ve read and coded over 700 awards in the past year or so. The patterns we saw early in the analysis (such as those reported here) continue to hold true. The raw win-loss ratio data is quite stark and, at times, eyebrow raising. This suggests that the “who you get affects what you get” conjecture may have some merit in the sense that some arbitrators seem to have clear tendencies which may make them, from an outcomes perspective, more (or less) desirable as adjudicators for specific cases.

We’ll need to wait until I finish coding all of the awards/decisions before we move onto testing the degree to which past decision tendencies can predict future decisions. That will involve (I think) segregating decision data into two groups for each arbitrator (maybe two thirds as a predictor pool and one third as a test pool) and assessing the degree to which we can, knowing an arbitrator’s win-loss ratio in the predictor pool, predict the outcomes of arbitrations in the test pool. (Obviously, there are many complexities to control for in the analyses, such as differences on which side bears the initial onus of proof and such).

Perhaps the biggest potential critique of this research is the premise that arbitrators’ values, beliefs, and decisions play a meaningful enough role in the outcome of decisions to warrant paying attention to them. You could very reasonably take the position that the impact (if any) of social constructivism would be rendered irrelevant because of the importance of the facts of each case plus the careful decision-making process that arbitrators routinely exhibit.

This view accords with the widely (but not universally) held belief that adjudicators (e.g., judges) are for the most part, neutral actors who are unlikely to be systematically biased in one direction or another. If we set aside, for the moment, that labour-relations practitioners, who have extensive experience with adjudicators, don’t believe this to be true, it is fair to ask if there any evidence that arbitrator bias operates in important ways? (We’re also setting aside the broader literature on bias in other forms of adjudication).

I’ve been keeping a diary of observations during coding. One of the striking things is how few of the 700+ decisions I’ve read where I got to the end and went “yeah, the arbitrator totally blew that call”. I’ve only run across (I think) one case so far where I’ve thought the decision was just clearly wrongheaded. In all of the other cases, the decision (based on the analysis presented by the arbitrator) was plausible (even if, maybe, I might have made a different decision). I don’t know if that pattern reflects that (1) arbitrators are good at getting to a sensible decision, (2) arbitrators are good at writing reasons that justify the decision they’ve reached, or (3) both.

I went back and forth about whether to link to the “you blew it” decision and, in the end, decided not to. There is no reason to dog pile on an arbitrator who made a decision in a complicated case with some ambiguous facts and whose other decisions seem generally fair minded. I mention the case only because it illustrates how an arbitrator’s values, beliefs, and expectations can shape the decision. It is a bit hard to explain this without going into identifying details of the case so you'll have to trust me a bit here.

The case revolved around an assault in the workplace and the worker defending themselves. There was clear and uncontested evidence that the worker had to and had cause to defend themselves: they were assaulted, put in a headlock, and feared for their safety. The arbitrator discounted this evidence and instead blamed the worker for triggering the assault because the worker was inattentive to the assailant’s needs. The arbitrator used the circumstances of the assault (e.g., a vulnerable assailant, moderately ambiguous and inconsistent evidence) to conclude that the worker’s termination was justified. The subsequent dissent by the union representative on the panel was, at the risk of understatement, pretty sharp.

An award that is not plausible is a pretty surprising outcome. Awards are designed to require arbitrators to clearly justify their decisions. This structural feature of awards ought to preclude decisions that clearly reveal arbitrator bias in interpretation of the evidence and drawing conclusions. Basically, the decision ought to be at least plausible on the face of it (and the vast majority are). That we have an example where the arbitrator’s reasoning is just not plausible runs contrary to the purpose of the system, which is to preclude both the fact and appearance of bias.

This is, certainly, just a single case and, Lord Vader knows, I’ve had off days myself. I think the point of it is that this decision is evidence, independent of the views of labour-relations practitioners, that arbitrator bias can occur. (I suspect most often it is harder to detect because the impact of an arbitrator’s values, beliefs, and experiences are less obvious, perhaps being less stark and profound or better obscured by the procedural narrative.

In any event, this case combined with the widespread belief of labour practitioners supports, however tentatively, inquiring into whether arbitrator’s values, beliefs, and decisions play a meaningful enough role in the outcome of decisions using a social constructivist frame. Basically, there’s some smoke here; let's see if there is also fire.

Anyhow, I hope that was an interesting view into how a researcher thinks about research questions before and during the process of collecting and coding data.

-- Bob Barnetson