Most accountability processes change what people say, not what they do

When a leader tells us their team has an accountability problem, it is worth asking what they actually want. The answer is almost never “better decisions”. It is usually closer to: when something does not happen, I want someone to have to answer for it.

Those are different objectives, and the research suggests they may pull in opposite directions.

Accountability does not do one thing

Jennifer Lerner and Philip Tetlock reviewed the accumulated research on accountability for Psychological Bulletin, the American Psychological Association’s review journal. Their subject was what accountability actually does to people’s judgement and their choices.^[1] The headline is that there is no headline. Accountability sometimes improves thinking, sometimes does nothing at all, and sometimes makes cognitive bias worse rather than better.

Which of those you get depends on the arrangement. Who you answer to, when, for what, and whether you know in advance what they want to hear. Their review is explicitly organised around those conditions, because the conditions are the whole story.

Their synthesis is unusually specific about when accountability produces real thinking. It happens when people learn, before they have formed a view, that they will answer to someone whose opinion they do not already know, who cares about accuracy, who is interested in how the judgement was reached rather than only the result, who is reasonably well informed, and who has a legitimate reason to ask.^[1]

Change any one of those and the coping strategy changes with it. If the audience’s view is known in advance, people shift towards it. If it can be guessed, they stop working and guess. And if the account is only required after the decision, people defend what they already chose, with commitment to a failing course of action tending to increase rather than fall.^[1]

This is inconvenient for anyone hoping accountability is a lever you can simply pull harder.

The question that matters most

One of the four questions Lerner and Tetlock set out to answer is the one worth sitting with: does accountability change how people think, or only what people say they think?^[1]

Put that question to the accountability processes most organisations actually run, and the answer is uncomfortable.

The monthly review where someone explains why the number was missed. The project meeting where each workstream reports green. The performance conversation held after the fact, with a person whose views are already well known, about a decision that was made weeks ago and cannot now be unmade.

Every one of those is structured to produce an account. None of them is structured to produce a different decision, because by the time the process runs, the decision has already happened.

If nobody has ever changed their mind in one of these meetings, it is not a decision process. It is a reporting process.

What happens over time

People are good at what they practise. A team that runs a process rewarding well-constructed explanations gets better at explanations.

You can see this happening, and it looks like progress. The updates get more articulate. The reasons become more sophisticated. The person presenting has anticipated the questions and prepared for them. That is a real skill, and it is not the skill anyone intended to build.

Meanwhile the behaviour that produced the problem is untouched, because nothing in the process ever reached it.

This is also why increasing the pressure tends not to work. A harder version of a process that rewards accounts produces more defensive accounts, delivered by people who have now learned that the meeting is a thing to survive. Which is roughly the point at which the leader concludes the team lacks accountability, and the loop closes.

Where it would have to land instead

If the aim is different decisions rather than better explanations, the process has to reach the point where the decision is still open.

That means before the decision is committed to rather than after it has been made. It means a genuine question rather than one where everyone in the room knows the expected answer. And it usually means someone in the conversation whose view is not already predictable, because explaining yourself to a known audience mostly teaches you what that audience wants.

None of this requires a new meeting. It mostly requires moving an existing one earlier and changing what gets asked in it.

A more useful question than “who is accountable”

Teams described as lacking accountability are rarely full of people avoiding responsibility. Far more often they are full of people who have correctly worked out what the process rewards and are supplying it.

Which makes the more useful question not who should be held to account, but what our accountability arrangements are currently training people to be good at.

About the author