Wednesday, August 14, 2013


Data and Debate

Bonnie Stewart made a statement on Twitter yesterday that made me sit up.  She wrote that big data inherently undermines shared governance, placing more power in the hands of administration.  

If that’s true -- and I don’t believe that it has to be -- then higher education has a serious problem.

The statement stopped me short because it’s easy enough to describe the conditions under which that happens.  Statistics show that program A works and program B doesn’t.  But program B has a loud and aggressive constituency, and program A doesn’t.  There’s enough money to fund either, but not both.  Who wins?

In that scenario -- any experienced administrator has been through it -- you have a choice.  Do you go with the politically easy choice, or the one with truth on its side?  How much political pushback are you willing or able to endure to do what needs to be done?

In the abstract, it’s easy to criticize that scenario as oversimplified.  What do you mean by “works”?  Are the goals themselves legitimate?  What about when the data are flawed, or ambiguous, or cherry-picked?  What about the things the data can’t (or don’t) capture?

But in actual cases, those larger issues are often less relevant than one might expect.  Defining the conflicts away through epistemological caveats may score debating points, but it doesn’t lead to making an actual decision.  It leads to postponing, possibly forever.  Sometimes, postponing really isn’t an option.  Or it amounts to making a decision in itself, favoring the currently favored over alternatives.  This is the flip side of the idea that data is a cudgel for administrators to use.  

(One of the habits of mind I had to learn, when I moved from faculty to administration, was accepting the reality of limited or conflicting information and making decisions anyway.  If you wait for the dust to settle, you will wait forever.  That’s especially true in settings in which some people kick up dust deliberately, precisely to prevent decisions from happening.  You’ll even see that in job ads for administrative positions -- language like “must be capable of working with ambiguity” pops up for a reason.)

If we replace the word “data” with the word “facts,” the problem is clear.  Does shared governance require a fact-free environment to thrive?  If it does, what does that imply about shared governance?  (Or, as Stephen Colbert puts it in a different context, “reality has a liberal bias.”)

Yes, people can use inside information to get what they want.  They also use bluster, and emotion, and force, and all sorts of other things.  If we ban information, but leave bluster, emotion, and force intact, I don’t see the decisions getting any better.

The issues aren’t confined to data, or to information generally.  State and Federal governments are taking much more active -- some would say intrusive -- roles in academic decisions than they have in the recent past.  Those mandates don’t come with the option of local nullification.  In those contexts, decisions that were once well within the purview of shared governance are now shared with the public at large, with the public having the final say.  

My sense of it is that the meaning of “shared governance” is much murkier than it used to be.  How does it work when a faculty is majority adjunct?  How does it work when a state hands down a mandate?  How much sense does it make to divorce shared governance from budgeting, when so many curricular and policy decisions have budgetary impacts?  (For those keeping score at home, I had the same question about the political thought of Hannah Arendt.  If you separate “the economic” from “the political,” you’re left with a pretty thin concept of “the political.”) nThe external constraints on colleges are far greater than they’ve been, and increasing, and often not really subject to internal veto.  

Last Spring I saw a presentation by a community college from Michigan about “Data Day,” during which they take over a classroom and put up posters with key data points on every wall, along with sheets for comments.  The idea is to bring a common base of information for discussion.  Yes, the selection and framing of the data is non-neutral; that’s inevitable.  But I like the idea of educating the discussion before getting to the conflictual part.  Here’s the mandate, here are the facts as we know them; now what should we do?  Skipping those first two steps isn’t a sign of respect for shared governance; it’s a passive-aggressive way of destroying it.

I can understand the concern, because in most places the faculty do not have the ability to access Big Data and ask it their own questions. For example, I have a question that everyone agrees is a good one but the ones who control access to Big Data will not ask it. Don't get me wrong, because there is no us-vs-them conspiracy here. The only people allowed to approach the altar are too busy to deal with faculty requests.

Shared governance requires shared access to Big Data.

One positive move at my college is that the faculty may have that access sooner rather than later. If that happens, the set of facts being considered will be broader. You might get better buy-in on that example you gave about what "works" if the vocal defenders were expected to test their ideas against Big Data rather than just listen to yours.

I see this as important because those mandates you mention leave the identification of outcomes and their assessment to the collective us at present. Define the outcome first (whether for a class or the staff roles you mentioned a few days ago), including the operational definition set by how you agree to assess it. Each answer will generate new questions that is at the heart and soul of shared governance.

I love that idea of Data Day, but it would be even better if faculty could identify gaps and have them filled before a reprise session a day or a month or a semester later.
Bonnie Stewart has it exactly backwards: the use of big data is a bulwark against autocratic administrators who make decisions based on "truthiness". Using data to make decisions is a fundamental component of meaningful shared governance.
As Dean Dad mentions, Big Data cannot always be trusted to give an accurate picture of the success or failure of any given program. For one, the accuracy of any picture that is generated by Big Data depends greatly on who gathered the data in the first place. Are the people who gathered the data truly objective and unbiased, or do they have an agenda? Are they tilting their data toward showing either that the program under review is a great success or that it is an unmitigated failure? The numbers can probably be manipulated and massaged to show either alternative.

So a natural question to ask when confronted with a pile of data is to be somewhat skeptical. Who gathered this data? Do they have some sort of agenda?

In addition, is there some motive for the people who are supplying the data to lie to the people who are gathering the data? If I report the “wrong” numbers to the reviewers, can I end up looking bad? Can I be dinged by the administration if I don’t meet some set of prior-designated numbers? Maybe I should make sure that I always report numbers that please the powers that be.

It might be sort of like the Five Year Plans in the old Soviet Union, where everyone lied to everyone else all the way up the chain, so much so that the Politburo made decisions based on numbers that almost everyone knew were entirely fake.

Shared access to data is certainly not something that favors administration. But, a few things:
1) As others have said, access to data is not always egalitarian. The gatekeepers of the data answer to people, and I might have a story or two about that...

2) Limited data doesn't necessarily paint a complete picture. I find it fascinating that people can be so easily impressed by some small data set that actually tells us very little.

3) If data is incomplete, it's OK to have a Bayesian prior. I'm open to having my mind changed with data, but until I see adequate data I'm also OK with defaulting to common sense and personal experience. Lack of data should not be a recipe for paralysis, it should just be grounds for caution before making expensive or irreversible decisions.

4) Requests for data can be turned into power plays. I was once in a committee meeting where an administrator wanted us to approve a certain project. We knew that similar projects had been started and abandoned at other campuses in the same state system. So I asked her if she knew why these projects had failed to take off at peer institutions. Her response was that she had no idea why, and we need to look at data before we assume that her remarkably similar proposal would fail.

Um, no. The default should be that when something has failed repeatedly, the data-collection burden should be on the person who wants to try it one more time, not on the person who says "Hey, wait, this has failed every time it's been tried!" However, this administrator is far more slick than me, so somehow her response is "I don't know the answer to your questions, so the burden is on you to answer it."
Something I'd like to see you write more about:

"How does [shared governance] work when a faculty is majority adjunct?"

First, how do you count adjuncts? Is it by pure numbers regardless of load, or is it by sections taught? At my CC, we are close to 50% of sections taught, but there are many more adjuncts than full-time faculty. The fraction also depends on whether and how you count lab classes.

(If you counted all the grad students, my grad school was majority adjunct 40 years ago.)

Second, how do you do governance? Is it a representative body or does it consist of every faculty member? Are adjuncts members?

(When I was in grad school, graduate students had voting representation on all major university shared governance bodies. Always a minority, but always a voice.)
Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?