Tackling over-generous ratings in managers

Posted by Anne Hamill

I recently saw a talent management question on LinkedIn – “Do you use a basic performance curve to force distribution of performance grades? If so what are the pros/cons?” This refers to the practice of asking managers to plot their team against a performance curve, so that they are only allowed to put 10% of names in the top category, and must put 10% of the names in the bottom category, to achieve a normal distribution of ratings. In some organisations, the bottom 10% are routinely managed out of the organisation.

Some of the best points made in the discussion included the fact that if an organisation has a lot of poor talent to deal with and leaders who are ‘too polite’ to confront this, a forced performance curve could help to winnow out the poor performers. It’s likely to demotivate any good teams you have left, however! And if you continue using the same system, over time you will end up forcing strong performers into the bottom 10%.

There is evidence that at employers like Microsoft where this has been done, performance appraisals deteriorate into discussions of how a staff member can increase their visibility with senior managers and claim credit for results – in order not to be forced into the bottom 10%. Is that really what we want to be discussing with good people at appraisal?

Taking a strategic step back, the real question is, ‘what are you trying to do, by creating a forced normal distribution curve?’

Usually the answer is that we are trying to force line managers to use the bottom ratings as well as the top ones. This is a priority in a situation where there is a culture of being over-generous, too ‘polite’, and evading tough conversations with poor performers. Or when a lack of consistency in confronting poor performers is causing employee disengagement.

A much more sensible approach to tackling generosity bias is to make managers accountable for their ratings. But how?

Why don’t we put some IT resource into ensuring all our managers become consistent and accurate ratings? For example, via the approach below.

Educate managers about rating bias before they do appraisals. You can do this via a small automated self-test, where you ask them to rate 5 staff or colleagues on 4 or 5 elements of effectiveness and then feed back the distribution of all their ratings, showing how these relate to an average distribution. Provide pointers such as “you might be showing generosity bias” or “you may be showing halo effect”, with explanation of these biases. This makes managers aware of their possible biases, and the need to justify apparent generosity with out of the ordinary team performance
Every year at appraisal, feed back the distribution of all the ratings managers made on their staff. This can include their ratings on subscales, if available. The feedback should display the distribution of their ratings against the average distribution of all managers for all staff. This will immediately show any possible bias. Again, explain what each manager needs to watch out for, in making ratings – generosity, toughness, central tendency, halo effect.
Make managers accountable through a managed discussion. Based on this feedback, senior managers and HR can hold discussions with apparently ‘generous’ managers, asking them for evidence that their team as a whole is really a high performing team based on outputs and results. If it is – no problem! If not, the possibility of rating bias can be raised, the unfairness of this explained, and the issue tackled.

We took this approach with one client and found that over a 2 year period (i.e. by the time managers had had the set up exercise and 2 sets of feedback) appraisal ratings shifted to a much more balanced distribution of ratings, with less bias and halo effect. What is a ‘balanced’ distribution? Well, the natural performance curve of a healthy organisation that is doing a good job of managing performance, is likely to be skewed – with most people in the Satisfactory and Good levels, and a few in the unsatisfactory and outstanding ends.

Take Away
Giving line managers evidence on their possible bias, and personal feedback on how to adjust the way they rate, coupled with making them accountable for improvement is far better than hitting them over the head with a statistical club!