The same web page as before is now
updated to split towns into big, small, and medium. This is a good page.

http://ronrox.com/paulstats.php?party=DEMOCRATS

Note that in large towns the machines DISfavor Clinton.
This would seem to show that the fraud hypothesis is bunk…
However, before you drop that hypothesis, I just want to point out that this big-towns result is based on only 1891 hand-counted-votes, which is
a small amount of data, and therefore leads to a large standard error, namely, +-2.3%. The anti-Clinton machines give Clinton -4.0% versus hand count in big towns.
So this big-town stat is NOT by itself statistically significant.
Similarly, their small-town Clinton stat is not significant either.
..
They have a similar page on republicans in NH:

http://ronrox.com/paulstats.php?party=REPUBLICANS

and NONE of the Romney discrep figures (for either big, small, or medium) are
significant by themselves.
===
This web page was pointed out on the rangevoting blog:

http://ronrox.com/paulstats.php?party=DEMOCRATS
http://ronrox.com/paulstats.php?party=REPUBLICANS

I am not responsible for and cannot vouch for their data.

But here is the thing. Clinton had what all said was a remarkable come-from-behind victory in New Hampshire, completely off what the pre-election polls had predicted, and supposedly “saving” her
campaign.

And golly. Gee willickers. According to this data, Obama would have beaten Clinton if we’d had hand counting. The machines gave it to her. Machines systematically gave Clinton more and Obama fewer votes than hand counts.

(Approximately 207K votes were counted by machine and 59K by hand in New Hampshire.)

Is this statistical proof of fraud?
Well, IF the two (hand & machine counted) voter subpopulations were identical in every way
except how they got counted, and if the voters were all independent, and if this data is correct, then yes, this is proof of fraud.

Because the chance of an anomaly this large (7.5 sigma)
is, I find using normal distribution tables
http://www.rangevoting.org/NormTble.html
about 3.2*10^(-14) = 0.000000000000032 .

HOWEVER, voters are not actually independent (e.g. husbands and wives) and conceiveably machines just happen to be located in pro-Clinton
anti-Obama regions. So, it might not be fraud.
(Meanwhile, machines also heavily favored Romney.)

Still, it makes you wonder…

The fact that they did not choose to do so would presumably validate Cameron’s objections.

There goes the neighborhood.

Election fraud is about election corruption when you have machines and electronic records which are easy to tamper with, being used in elections where, say, the overseeing officials are also the chairs of a candidate’s campaign. As you can guess, this scenario only worries lunatics and fringe folk and not democrats, republicans, nor the media.

And then there is voter suppression, which are tactics used to depress the turnout of unwanted voters. This involves creating lists of felons and using loose name matches to knock legitimate voters off the list; or sending voters who live in a certain precinct junk mail and then challenging their right to vote if they failed to respond to the junk mail.

Oh to go back to the simple days when it was enough to say “You cheated.” Dishonest people have been much too creative.

It is still a weird discrepancy, to say the very least.