We are measuring the wrong things about the labor movement

Nick Driedger argues that most metrics by which unions are measured obscure what is really important.

Measuring the labor movement is a popular undertaking: leftist analysis is full of bigger picture stuff like strike rates per year since the 1930s, overall union density, and percentage of the public that is supportive of unions.

Awhile back, Dave Kamper wrote a provocative piece in The Forge reemphasizing the need for solid measurements as a foundation for rebuilding the labor movement. In “We Don’t Know What We’re Doing,” he compared the situation to the movie Moneyball, where a baseball team is turned around by an open-minded general manager and Harvard economics PhD who urged him to do things by the numbers. As Kamper put it:

Moneyball is the story of people who looked at this institution and realized that many of the truisms they had taken for granted had gone unexamined, unchallenged, and unproven. These people recognized that if you simply stopped and asked yourself what you really knew to be true — and what you didn’t know at all — you might find out that what you were doing was wrong all along.

Specifically, Kamper compared the work of Jane McAlevey and the work of Shaun Richman in an effort to lay out two poles for labor analysis — one anecdotal and values-driven and the other more focused on hard data. Among the objective comparisons that could be made, Kamper suggested certifications won and lost, contract language, and union density in various sectors and geographical areas. On the other side of that are things Kamper laments as being hard to measure, like the effectiveness of union organizer training, the effectiveness of organizing at all, and why some workers step up and others don’t.

What can be measured

Let’s start with what can be measured.

In most jurisdictions, contract language is easy to get ahold of. It also tends to follow a standard process and the language tends to be copied from agreement to agreement. Unions do compare this language — just watch a raid or jurisdictional dispute play out. They don’t do so publicly or systematically, partly because negative comparisons can be the kinds of things elected officials lose their jobs over and partly just because there is no upside for a union to do it.

Grievance procedures can be compared. There are places where that language is stronger and places where it is weaker, but the easiest thing to measure is who pays for it. Most language says the costs of arbitration are shared by both parties. The IWW’s Burgerville Workers Union contract says that the loser of the arbitration process pays. Some rare language says the employer pays, like this contract from a faculty union (AUFA) in Alberta, Canada. Employer-pays language usually exists in some kind of restricted sense — in the AUFA case, it is limited to discipline. Right there you have okay language, bad language, and very good language.

You can do a similar exercise with no-strike clauses, management’s rights clauses, and language particular to discipline in the agreement. Almost any language can be measured by looking at weasel words — does it say the employer “shall” give bereavement leave or that the employer “may”? Context of course is important, as Kamper points out. When you play the contract game you sometimes let weak language slide as a sort of opening for future rounds down the line. The problem though is that in many cases no language (and no contract!) is better than bad language or a weak contract — as long as you have the shop floor organized.

Grievance outcomes can be measured too. You can compare the remedies asked for in the grievance and look at how those grievances were settled, or whether they were dropped outright. Arbitration awards will generally directly reference remedies sought. So at least there you have a straight line from what the workers ostensibly “want” to the outcome of the process. Everything short of an arbitrator’s award isn’t documented, making the record keeping sparse and hard to track. And there are political costs to sharing that information publicly: there can also be trade-offs in settlements, which means there are winners and losers among the membership.

So, you can make comparisons on things like contract language and grievance settlements, but no one agrees on what a good settlement is because the question is inherently political. For that matter, there are people who are always going to think every deal is a total sellout — look at the Spartacists or the World Socialist Website. These people don’t think unions can win, they just see strikes as a good place to peddle their alternative to unions: their own political leadership. Then you are also always going to have conservatives who say that any deal is good, because the thought of a strike gives them hives — look at the major labor leaders who are involved with the Democrats in the US or NDP in Canada.

Disagreement is a healthy dynamic and exists in any movement. This does not mean there are no wrong answers but there is always going to be a tension between when to take a deal and when to keep pushing and these two poles are going to be there in even the most radical moments. The important thing is workers deciding when to hold them and when to fold them, because that is how they learn through struggle.

What is usually measured

Half of the point of Moneyball was that you could measure things — the other half is putting into question what you are measuring. There is a scene where the economist is sitting in a room full of conventional baseball experts and the economist just asks the same question over and over when analyzing a player: “does he get on base?” The economist finds a single data point he can reduce his analysis down to and then builds a plan around that. When you get people on base, according to his analysis of the data, you get more runs. You don’t look at the player.

We need to think about what “getting on base” is in union terms. The things that are currently measured in the labor movement are almost entirely legal benchmarks created by the laws and regulations around unions. The thing about labor law is that it gives you canned metrics that are sound legal categories but empty politically. Pretty much every step in the legal process exists to replace meaningful benchmarks in organizing with legal ones.

That means that what we need to measure is hiding behind what we can measure. Take membership. Before the Wagner Act in the USA (and its knock-off process in Canada), measuring union membership was pretty straightforward. People who paid dues to the union were members; those who did not were people the union was trying to win over. From that exercise you knew where the union stood in relation to the workers. So behind dues checkoff (or the Rand Formula in Canada) lies our first possible metric that has been completely hidden by the legal framework. Paying dues by hand, voluntarily, was something that could be measured in the past and now isn’t measured because we are measuring something else — the number of members paying dues, whether they want to or not, because of the contract.  

On top of that, unions ate what they killed when it came to servicing their members prior to the current contract system. Every concession was won largely through workers’ own actions. Now, while you can follow the chain of events through the grievance procedure from remedy to discussions around settlement to arbitration, every step is filtered through whether the issue is even a violation of the contract at all, and then judged within the norms of how similar language was judged before. The grievance procedure is easy to measure but also gives you information that tells you nothing about the strength of the union because it trades on the power and legitimacy of the labor board or arbitrators — not the workers.

Contract language is the most interesting example. Contracts reflect the power of the parties to the contract. If the workers have power they can bargain better language; it’s the one test of strength between the union and the employer that remains. So by measuring the outcome of a contract you are also measuring the strength of the union.

But you’re really only measuring its strength at a particular moment in time. The balance of power between the union and the employer can shift over time, especially as issues arise, etc. But the contract language is static. So what is going on “behind” the language (the level of organization on the floor) is actually more important.

The goal of a union, any union, is to improve the conditions of work and compensation of their members. How a union does that, and what they say about how they engage in that activity, is political. Contracts are where concessions often get written down but they aren’t the only place concessions happen and measuring the existence of a contract in the “win” column without looking at what that contract says gives you junk data. Same with measuring the existence of union recognition itself, without looking at what the union got them. It’s like measuring skulls to find out how smart someone is.

What we should be measuring

Ultimately Kamper is right that we need to strip away the folklore and anecdote from discussions about what works and what doesn’t in unions. But we also need to strip away the impulse to measure things as they appear to us at first glance and really think about what we are measuring. I am going to put forward a different set of metrics that unions should judge their effectiveness on.

The first things unions should measure are the things that indicate an active and engaged membership that is independent of the legal benchmarks. These are things like:

  • The proportion of the workforce that has voluntarily signed a union card.
  • The proportion of the workforce that attends union meetings.
  • Ratio of stewards to workers in a given workplace.
  • The distribution of grievances among the members. How many people are your frequent flyers on the process and how much more broad engagement to do you have?

The second broad category that should be measured is the amount of action taken around shared concerns among the workers:

  • How many campaigns around workplace issues have the members launched?
  • What concessions (changes in policy or direction) has management undertaken in response to those campaigns?
  • How many different actions were required to get management to cave?
  • How many different parts of the company (or industry) did the workers have to bring into the campaign to have sufficient power to win on their demand?
  • Was the demand conceded outright or was it bargained into a collective agreement?

The goal with measuring these things is to draw a straight line between what the workers want (demands) and outcomes (concessions from management).

The third broad category of things that can and should be measured are union bodies that contribute to organizing, and the actual act of organizing on the job itself:

  • How many workers do you have regularly working through a structured system of contacting their peers about shared concerns; in other words, how many one-on-ones are happening?
  • How many union members do you have that are experienced and capable (not just trained!) in how to do one-on-ones and engage in them regularly?
  • How long does it take for the committee to one-on-one the entire membership? For example, a workplace organizing committee of 20 people in a workplace of 400 is going to do roughly 20 calls each; more people means you can do it faster.
  • How many members do you have of a stewards council, network of workplace organizing committees, or job branches?

If you have a team of organizers on the job who can do these agitational conversations you are going to have an easier time getting the workplace mobilized and you are going to see greater concessions from the employer. You are also going to have a bigger and better pool of people who can reach out and organize similar workplaces to bring more pressure on groups of employers.

Concessions from the employer or groups of employers are the home runs we are looking for and the way we get on base by building a team that can make contact on the job as thoroughly and quickly as possible.

Scaling up the quality of the on-the-job organizers brings more pressure; building out your membership (across the entire enterprise and into competing firms) expands your power. The basic game here is building more organizers on the job of a higher quality than the ones you are losing through attrition, burnout, hiring into the ranks of union staff, and promotion into management.

The metrics outlined above (rather than typical labor relations benchmarks) are what let you build people up rather than building up formal legal structures. You can build these things in spite of the legal structures, but not through them. What metrics you choose lead to important political questions. For working people, the question is: what builds my power and control over the economic forces that dominate my life?

Nick Driedger is the Director of Labour Relations and Organizing for the Alberta Union of Provincial Employees and a frequent contributor to Organizing Work.