Wednesday, May 19, 2010

Performance Reviews

We're coming up on performance review time for the administrative and staff ranks. That means I have to have my annual internal debate about performance reviews. (Apparently, it's "ambivalence week" here at Dean Dad HQ.)

As many folks have pointed out, performance reviews are deeply flawed in the best of times, and often just destructive. There's no end of reasons for that.

- They only occur once per year. That leads to predictable temporal distortions -- the most recent stuff outweighs the older stuff -- and some unavoidable discontinuity.

- They reflect only what the manager sees, and/or how the manager sees it. Depending on the manager, this can be a real problem.

- They're "one size fits all," even though jobs are very different. In many cases, there's only one person in an entire job category, so any basis for comparison will necessarily be arbitrary.

- Clean measures of performance are often absent. Managing creative people is not like selling cars.

- They're undeniable sources of tension.

And yet, we keep doing them. Why would otherwise intelligent people keep using such a flawed system?

Because performance evaluations aren't just for the benefit of the employee. They're also indispensable elements of the paper trail you'll need if you need to rank people, whether for promotion or for termination. They give the organization protective legal cover.

That's why the oft-heard suggestion of "just replace them with regular informal coaching," which sounds nice, rarely gets picked up. Informal discussions don't show up in personnel files. That means that if you have to lower the boom on someone, there's nothing in the file to defend a claim of arbitrary and capricious (or discriminatory) decisionmaking.

As with outcomes assessment and student grades, there's a tension between the formative and the evaluative functions. Yes, it's great when an evaluation serves as a teaching moment and results in better performance. I actually had that happen to me at my last job once. My boss made a negative comment that struck me as random and mean at the time, but as I stewed for a few days, I gradually figured out what was behind it. Subsequently I made some adjustments, and we both noticed the improvement. So yes, it can happen.

But when that doesn't work, and someone just isn't working out, you'd better have a paper trail.

The usual rebuttal to that involves documenting incidents as they occur, but that presumes that individual incidents are the issue. Sometimes they are, but frequently the issue is a longstanding pattern. Showing up late once means nothing; it happens to everyone from time to time. Showing up late every day is something else altogether. A single awkward or hostile interaction could be the result of crossed wires, but a pattern of them usually indicates something more fundamental. By definition, patterns only show up over time. A document specifically intended to cover a span of time can capture that in a way that a single-incident document just can't.

So these awful, draining, semi-accurate documents survive. They're not my favorite task -- not by a long shot -- but I know that if they don't get done, there'll be hell to pay. They're almost as bad as not doing them.
We're coming up on performance review time for the administrative and staff ranks. That means I have to have my annual internal debate about performance reviews. (Apparently, it's "ambivalence week" here at Dean Dad HQ.)

As many folks have pointed out, performance reviews are deeply flawed in the best of times, and often just destructive. There's no end of reasons for that.

- They only occur once per year. That leads to predictable temporal distortions -- the most recent stuff outweighs the older stuff -- and some unavoidable discontinuity.

- They reflect only what the manager sees, and/or how the manager sees it. Depending on the manager, this can be a real problem.

- They're "one size fits all," even though jobs are very different. In many cases, there's only one person in an entire job category, so any basis for comparison will necessarily be arbitrary.

- Clean measures of performance are often absent. Managing creative people is not like selling cars.

- They're undeniable sources of tension.

And yet, we keep doing them. Why would otherwise intelligent people keep using such a flawed system?

Because performance evaluations aren't just for the benefit of the employee. They're also indispensable elements of the paper trail you'll need if you need to rank people, whether for promotion or for termination. They give the organization protective legal cover.

That's why the oft-heard suggestion of "just replace them with regular informal coaching," which sounds nice, rarely gets picked up. Informal discussions don't show up in personnel files. That means that if you have to lower the boom on someone, there's nothing in the file to defend a claim of arbitrary and capricious (or discriminatory) decisionmaking.

As with outcomes assessment and student grades, there's a tension between the formative and the evaluative functions. Yes, it's great when an evaluation serves as a teaching moment and results in better performance. I actually had that happen to me at my last job once. My boss made a negative comment that struck me as random and mean at the time, but as I stewed for a few days, I gradually figured out what was behind it. Subsequently I made some adjustments, and we both noticed the improvement. So yes, it can happen.

But when that doesn't work, and someone just isn't working out, you'd better have a paper trail.

The usual rebuttal to that involves documenting incidents as they occur, but that presumes that individual incidents are the issue. Sometimes they are, but frequently the issue is a longstanding pattern. Showing up late once means nothing; it happens to everyone from time to time. Showing up late every day is something else altogether. A single awkward or hostile interaction could be the result of crossed wires, but a pattern of them usually indicates something more fundamental. By definition, patterns only show up over time. A document specifically intended to cover a span of time can capture that in a way that a single-incident document just can't.

So these awful, draining, semi-accurate documents survive. They're not my favorite task -- not by a long shot -- but I know that if they don't get done, there'll be hell to pay. They're almost as bad as not doing them.