Pitfalls of measurement

In my last post, I talked about some ways people try to influence behavior by measurement, and some of the disastrous consequences.

The more I think about it, the more convinced I am that these are not simply exceptional cases. These are the rule, not the exception.

Any time you measure performance hoping to achieve an outcome, you are more likely than not to be disappointed.

There are reasons for this, and some possible answers. I'll talk about solutions in another post, but first it's important to understand why measurement is so difficult.

The biggest problem is that people are inherently competitive. The moment you start measuring something, people are extremely good at making sure they "measure up" - whatever that means. Sometimes this shows itself as explicit and conscious behavior. Sometimes the effect is subconscious or more subtle. Sometimes it is indirect, such as through pressures created within the organizations that people belong to, as a result of organizational measures such as profit.

Whatever it is, measurement influences behavior, and often not in the way you want.

Why is that?

Here are a few reasons...

The measure is not a good indication of the outcome

Many of the things we want to achieve are difficult to measure directly, so we end up with lots of indirect measures. Some of these are reasonable, many are not.

For example, we would like to think we are being led by people most fit to govern.

The theory of democratic voting might work well in small groups, where people truly know each other, and assuming that people have the integrity to vote for those they genuinely believe would be the most suitable.

The trouble is that modern democracy is little more than a measure of popularity, and in a large world where voters have no prospect whatsoever of actually knowing the candidates, popularity has in turn become a measure of all sorts of undesirable traits such as the ability to lie convincingly, or to most effectively trash your opponent.

These traits have nothing whatever to do with fitness to govern.

The measure is appropriate, but is not the whole story

People will strive to achieve whatever is being measured. Sometimes this comes at the expense of something else important, but the other important thing suffers because there is little motivation to pay attention to it. It is not being measured.

For example, a focus on saving lives, and on life expectancy, has led to people being kept alive at all costs - whether they want to or not. In focusing exclusively on one (albeit important) measure, other things like quality of life tend to suffer. Also, health services are under pressure to spend disproportionate amounts on expensive procedures to prolong the lives of a few individuals, at the expense of less glamorous treatments that would ease the suffering of millions.

The measure can be cheated or subverted

One of the biggest problems with many measurements - it is often seen as easier to cheat than to comply. Often, the only thing stopping people is honesty - which is another measure that people respond to in varying degrees.

The point of cheating is to improve the measure by some means other than that which was intended. This varies from outright dishonesty - ballot boxes stuffed, records falsified - to more subtle tactics that fall into a grey zone, such as pressuring marginal students into not sitting exams in order to boost pass rates.

And the scariest part is...

Most of the ways in which measurements misfire are not down to dishonesty.

People act as a result of myriad - often conflicting - motivations. Measuring some aspect of performance alters that web of motivation, and it's only natural that people most feel the pressure of the measurement itself rather than the underlying intent. As a result, people tend to follow the line of least resistance, even if the results are clearly undesirable.

And, given how difficult it is to devise a truly good measure in the first place, bad things happen more often than any of us would dare to imagine.