Office noise levels

A couple of months ago I sat a decibel meter on my desk for a few days, and recorded the noise levels in my office. Although I feel my office is very noisy, the figures I got didn't seem too bad compared to typical levels I found on-line. But one thing I did spot was that the most disruptive noises - people talking in the corridor - were also not particularly loud.

At the weekend I came across a paper, "Defining the Acoustic Environment of (semi-)open Plan Offices Acoustic Measurements leading to Activity Based Design for retrofit Buildings", which included average noise levels for various types of office. The authors of that paper found that the average noise levels in the modern (open-plan) offices they looked at were fairly consistent at about 51dB, but there was variation in the standard deviation from ~3 for an office filled with programmers to ~5 in an office with mixed tasks.

So, I went back to my records, picking one day look at, and plotted average sound level, standard deviation and lowest 2 second average in 5 minute segments from 7am to 6pm.

The minimum sound level line gives an impression of the background noise - the higher levels for a while in the afternoon are due to a graduation, with the new graduates and their families congregating in the quad outside my window, so this is an unusual event that can be ignored.

I suspect that the standard deviation line at the bottom (and the large variation in noise levels through the day) show why I feel my office is too loud, even though actual average sound levels are OK, the number of disruptive noise events is high, and this is what makes if feel like a noisy and unpleasant environment to work in.

It would be interesting to get similar records from other offices along with information on their occupants impression of how noisy they are to compare with this - maybe a nice wee research project for someone. I've not had much success in finding research on the impact of noise on knowledge work, other than the well known work DeMarco and Lister covered in PeopleWare, but that research should make this a major concern to managers.