Friday, July 2, 2010

Do your social networks influence your performance appraisals?

Here is a question: when it comes to the performance ratings you receive does it, or should it, matter who you are connected to at work? Intuitively we would expect that certain powerful friendships could influence how you are rated, but surprisingly, there has been very little actual research that tries to uncover the impact of social networks on performance evaluations. Oh, and by social networks I mean your personal connections in the workplace, and not whether you are an avid user of Facebook/Linked-in/Twitter etc!

This is a really important issue - folk theory would suggest that social connections can significantly influence a lot of things managers care about: fair pay and recognition, promotions and terminations, training and career opportunities. It is important to try and understand, are such connections at work always beneficial, or can they also be detrimental? My colleague Gianluca Carnabuci and I thought we would investigate this issue and see how mapping employee social networks might shine some new light on the problem of the accuracy of performance appraisals.

Although it is truly difficult to define job performance in a way which everyone can agree upon, the accuracy of subjective ratings of job performance may be captured by four key measures:

1. leniency - are the ratings given higher than they should be?
2. central tendency/range restriction - does a supervisor tend to give the same ratings to all of their direct reports?
3. halo effects - does a supervisor tend to rate each direct report similarly across objectively different dimensions of performance (e.g., timeliness, quality, quantity of output) regardless of real differences?
4. inter-rater disagreement - when different supervisors are asked to evaluate the same employees, do they agree?

There are several reasons to expect that social ties may influence the accuracy of performance evaluation, both positively and negatively. The most obvious social connection to consider when looking at accuracy of performance ratings would be whether there is a strong friendship between manager and direct-report. These kind of relationships might be expected to reduce the ability or willingness of the supervisor to provide an accurate evaluation, for fear of damaging the relationship. However, strong relationships also imply frequent and open interaction, and this can lead to a supervisor having a lot more accurate information about the true performance of the person they are rating. Which of these effects is stronger is not at all obvious.

In addition to such friendships, we were also curious about another aspect of social networks, which we refer to as 'social embeddedness'. This concept refers to the extent to which any individual is connected to many others through strong and mutual exchanges of trust, friendship, advice and support. The idea here is that being more deeply connected within a work community may exert some additional pressures on the performance rating process.

What happens when a manager or supervisor is 'networked' in the organization? Social embeddedness would increase their ability to access information about performance of a particular direct report, simply by being connected on the 'grapevine'. This should increase the ability of managers to perform accurate ratings.

What about the embeddedness of employees (direct reports) in social networks? Consider the popular employee, connected to many others in strong relationships. It is likely to be much easier to give that person a positive rating than a negative one. After all, if you upset 'popular Pat' with a negative evaluation, then he or she may complain about you to his/her friends. If you already perceive it to be costly to give a negative evaluation, because of the emotional upset and potential loss of cooperation, then the threat could be magnified in the case of a popular employee. Not only might you lose the cooperation of your direct report, but possibly their friends as well. At the extreme, perhaps you would be perceived more negatively even if you are being quite fair. Supervisors are rational human beings, and recognize when the cost of giving an accurate rating outweigh the benefits. The social embeddedness of direct reports could alter the balance of this judgment.

We were interested in seeing whether we could measure these effects, and which of them was strongest with respect to the four measures of accuracy. We studied a network of employees in a large manufacturing firm, by gathering data on their social connections in the workplace, and by separately obtaining performance ratings across five dimensions of performance. In summary, what we have found is that:

1. When there is mutual trust between supervisor and direct report we see greater, and not lower accuracy in ratings (although the average ratings increase, we also see lower levels of halo effect and higher inter-rater agreement).

2. The extent to which supervisors are embedded in workplace social networks does seem to moderately enhance accuracy - they are better at differentiating across performance dimensions (reduced halo effects).

3. The social embeddedness of direct reports has the opposite effect. Consistent with the idea that it is harder to accurately rate popular people, we see greater leniency, more halo effect and some evidence of greater disagreement across raters.

What do we learn from this? Relative to folk theory, the reality is quite complex: social connections definitely matter; but whose social connections we consider is also important. For supervisors, being connected may help improve accuracy, while for direct reports broad social connections can lead to less accurate feedback. Quite surprising, and contrary to popular opinion, strong relationships between supervisor and direct report can improve accuracy rather than diminish it.

So what does this mean for managing? Different organizations have their own patterns of social relationships. Bureaucratic organizations tend to stifle informal social relations while clan-like organizations are built upon them. In bureaucratic organizations the network embeddedness effects may be more limited. However, if you are trying to build an organic, clan-like organization, perhaps because you are interested in being innovative and entrepreneurial, then the effects of social networks on the performance management process are likely to pose a significant challenge. In these conditions, promoting the benefits of embeddedness for managers, while preparing managers for the biasing effects of subordinate networks should help with the effectiveness of performance management.

No comments: