I looked into this a little bit a number of years ago. My recollection is that on average, for a given position (IF vs OF) errors cost just slightly more outs (obviously), but - plays cost slightly more in terms of bases. The differences in bases may not have been statistically significant. Given that outs matter more than bases, I think it's fair to say that errors are more costly than - plays, but I think the gap is pretty insignificant. I generally just add them together if I'm looking at a performance history, i.e. + plays - (- plays + errors). I do suspect that one could gain a small edge on the field by properly weighting all of those groups of plays, but I'm not sure it would be a big enough edge to be worth the effort.