Sunday, March 22, 2009

Social network analysis - a misapplication

I went to an interesting on social network analysis a few weeks ago. An analyst presented a visual graph of the numbers and types of communication between members of three geographically dispersed work units that were working on the same project. The network showed that peers working in the two groups in the US were communicating regularly. However, there was no communication between peers in the US and the overseas unit, where in fact managers had expected to see communication. The analysis had accurately uncovered a previously-unknown issue.

The analyst had stressed the importance of knowing the context in which the analysis was done. The graphs by themselves weren't of value unless you knew who and what was going on in the project. He wrapped numerous caveats around social network analysis, including some privacy issues.

Here's an unfortunate use of social network analysis being touted by a company that, unsurprisingly, sells data mining services that apparently include social network analysis. Dots and circles and lines on a chart represent employee "performance," as captured by this wholly inadequate tool. The employees with the dark colored circles? "On a relative scale, they don't add a hell of a lot," asserts the CEO of said data mining company. Really? Without knowing what the employee's skills are, their productivity level, their opportunities to contribute? HR departments should start cutting based on this company's little graphs of numbers of phone calls and emails sent?

I hope HR departments are smarter than that. Of course, I've complained about the use of personality testing as an HR tool for selection, but that seems to be making some inroads as well.

No comments: