Except that every study seems to show that actual voter fraud is minimal
I always hear that claim made, yet every so called 'study' I examine has gaping holes in it's methods as more often then not there is simply not enough data to make an accurate measure as to what level of fraud exists because of the lack of ID requirements.
If I show up to your polling place and claim to be a recently deceased person, someone I know to be on vacation, or not likely to vote... unless the election worker actually knows the person I am pretending to be... I'm almost certainly going to be able to cast a vote illegally and the chances of being caught are slim, ditto for me ever showing up on the radar of someone investigating fraud.
Let me fix that for you:
Voter ID laws are (taken at face value) an attempt to reduce a unknown number of people from voting illegitimately while keeping a larger number of people from voting legitimately.
Somehow one needs a photo ID to fly on a plane, buy a six-pack of beer or open a bank account in most places... but only with regards to voting is it suddenly a horrible thing, even when free IDs are usually available.
To go back to my earlier example, a person doing so would only be casting a single extra vote which on it's own isn't much, however it's not too hard to imagine an individual or group being a little more dedicated and spending the morning of election day doing this over and over again.
How many fraudulent votes would they need? Lets look at some recent close races which come to mind:
in Florida in 2000 George Bush beat Al Gore with a margin of 537 votes out of 5,962,657 cast (0.00901%).
In South Dakota in 2002 Tim Johnson beat John Thune by a margin of 524 votes out of 33,4438 cast (.001567%).
In Washington in 2004 Christine Gregoire beat Dino Rossi by a margin of 133 votes out of 2,810,058 cast (0.00473%).
In Minnesota in 2008 Al Franken beat Norm Colman by a margin of 312 votes out of 2,887,646 cast (0.0108%).
We are talking margins not only within the range of error (accidental over votes, identifying marks, etc), but very possibly within the margin of fraud, something the current system makes very difficult to detect.
Most of us don't wait to put locks on or home doors until after we are burgled, we do so because we recognize a potential threat and prefer to take some actions to reduce the risk, why not here as well?