If you have a bullshit measure that determines some phenomena (e.g. crime) to happen in some area, you will become biased to expect it in that area. It wrongly creates a spotlight effect by which other questionable measures are used to do the actual conviction (“Look! We found an em dash!”)