The last study that came out on this showed that engineers were significantly overestimating their own productivity gains.

If a stat like that is not accurately measured, it's useless.