Interviewing with HAL

In a recent post, The New Teacher Project (TNTP) asks an interesting question. And by "interesting," I mean "dumb."

Can Better Questions Lead to Better Teachers?

By "better," TNTP means "very specific multiple choice questions asked by a computer." As it turns out, I'm pretty sure I know the answer to this question, so let's end the suspense right away. Jessica Cardullo, contributing writer to the TNTP blog, here is the answer to your question:

No.

Or possibly, "for the love of God, no." I was also tempted to go with "No, dummy," by I do not know Ms. Cardullo and for all I know, she's a highly intelligent person who was assigned to write about a dumb question.

TNTP is the big-boys-and-girls version of TFA; they take people who have an actual history of working for a living in some non-teaching job and try turning those folks into teachers. It's actually a far less ridiculous concept than TFA, and if TNTP didn't have the same corporate weenie connections as TFA, I might be inclined to take it seriously. But instead, they are run by the kind of people who come up with ideas like this one.

First we have to figure out what "better" means when applied to teachers. Cardullo starts out by contrasting broad questions such as "Why do you want to be a teacher" with focused questions such as "Which of these statements best describes you" followed by multiple choice answers from which to select. Or they could be asked agree-or-disagree statements such as “All students should be held to the same grade level standards,” and “teachers in high-need schools must be given equal resources to teachers in wealthier schools if they are going to be evaluated by the same standards.” And I'm starting to see their problem, because by "better," maybe we mean "fully aligned with the biases of this organization."

Cardullo assures us that lots of standardized test questions such as these will elicit a picture filled with "nuance" and "texture." Because I think we can all agree that when we really want a nuanced and textured picture of a person, we hand that person a set of standardized test questions.

But TNTP is really looking at this "interview pointillism."

One interesting possibility: using computer-based interviews, at least at the outset of someone’s application to one of our training programs, to ask these sorts of questions. It’s an interview model that sounds inhuman and even a little scary, like the quest for machine-based efficiency gone too far. But we’re looking into computer interviews not because we’re trying to skimp on time, but because we think they might actually predict future teacher performance better than our old model, a daylong in-person interview.

Well, yes. If by "interesting" you mean "inhuman." Now, it's not that they never believed in rigorous interviews and screening; they've got training screening and interview screening and even phone interviews. It's just that "when we took a hard look at whether performance during the screening process was connected to classroom performance, we found very little relationship."

Now, I can definitely draw a conclusion from that, and my conclusion is that TNTP's screening people are very, very bad at their jobs. Either they don't know what to look for in prospective fellows, or they make it really easy to game their system, or they just don't know how to draw impressions about the character and ability of other carbon-based life forms. I might send them for training. I might reassign them. I might fire them. But the brain trust at TNTP has a better idea.

Multiple choice questions.

Seriously.

Rather than relying on whatever meaning interviewers ascribed to an open-ended conversation, these “forced-choice” computer survey questions could give candidates—and us—a clear, specific and common set of terms and ideas to review regarding their skills, experience, aspirations and expectations for teaching as a career. They could replace part of our existing application screen to allow us to drill down on the handful of particular skills we know we’re looking for, like professionalism, critical thinking and receptivity to feedback, which our experience tells us that most effective teachers possess.

You know, my ordinary approach is to exaggerate outrageously in order to make my subject look foolish, but today I am stumped. And by "stumped" I mean "saddened and astonished." In addition to the virtues of drilling down and attempting to determine critical thinking with a bubble test, Cardullo goes on to tout the virtues of computer assessment because it can't be biased (only the person who writes the assessment questions can be biased-- seriously, what is this magical belief that when a person says something or writes something it has human bias, but as soon as you type his words into a computer, magical elves dance out and suck all the personal bias away, leaving nothing but sweet, sweet perfect objectivity. And by "sweet" I mean "imaginary"). (And yes, I know we're practically describing the Praxis test here, but I'll save curmudgeoning about that for another day.) There are so many things wrong here. Sooooooo many. Let's just pick two.

1) You know what the best way to gauge someone's ability to interact with other human beings? Have them interact with other human beings. Yes, all the human beings involved in that process will bring their own personal biases, ideas, personalities, human foibles, hair styles and histories into the process. So will all the students who walk into a classroom. This is how human beings are. Stop trying to create systems that don't allow for humans to be human! If your humans in charge of talking to the humans can't handle talking to humans, you need to hire different humans.

2) You know what's really easy to lie to? A bubble test. Your questions will be biased and you will be testing your candidate's ability to figure out what you want him to say. What you won't have is the benefit of seeing him roll his eyes and flip the bird at the computer screen. Your standardized test will tell you exactly nothing except how good he is at navigating bureaucratic baloney. That is, of course, excluding the candidates who will decide that any organization that tries to screen humans for a job that's all about working with humans without involving any humans-- well, that organization is to be avoided.

Cardullo acknowledges this last problem. It is possible that this approach might shrink the talent pool. "The idea of a computer predicting which candidates have the potential to become effective teachers might sound a little crazy." If by "a little crazy" you mean "unbelievably stupid," then yes, you are correct. But in reform land, they have fallen so far down the rabbit hole, they cannot see how far they've fallen and how little sense they make. And by "rabbit," I mean "stupid."