By popular demand...
I made an initial stab at a web page about this.
The first item (turnout effects) is pretty much not computed
yet, however. Anybody want to try to compute what fraction
of US election winners would change if we had universal suffrage and
universal voting? Or how much the US vote would shift?

There are turnout biases correlating
with income, education-level, race, gender, and criminality. These also
are correlated with politics. You have to estimate all these effects
and their magnitudes and signs, to get an estimate of how much the
vote would shift. When we have that number we can compare with
other things like the "gender gap" estimating how important women's
suffrage is, etc.

Also the fraud item has a large error bar...

http://rangevoting.org/RelImport.html