After the 2004 election, there was an uproar in some circles, among people who felt that, for a variety of reasons, there had been enough irregularities tilted towards President Bush, most notably in Ohio, but also in other "battleground states" that the election had been effectively "stolen". I don't know if it really was stolen, but I looked at the documentation and the noted and investigated irregularities, and I thought that it was enough to warrant an investigation.
There was no investigation. Instead, people calling for investigation were variously described, generously, as cyberspace conspiracy theorists, and less generously, as losers and nuts.
There was a large amount of collected data that was withheld from study in 2004, not by the government, but by news media subscribers to the NEP, the National Election Pool.
There were two prominent in depth looks at the possibility that the 2004 election was stolen in Ohio: "What Went Wrong in Ohio, the Conyers Report on the 2004 Presidential Election" and "was the 2004 presidential electin stolen? exit polls, election fraud, and the official count" by Freeman and Bleifuss. There you can find long lists of irregularities at the precinct layer, including long lines, manipulation of voter rolls, felon lists, alleged falsification of recounts and so on.
And you will find, if you look carefully in these or other sources, that there are two parts to election monitoring: watching for irregularities, which will be done this fall by complaint lines at 1-866-OUR-VOTE. The other part is done by comparing voting tallies to exit polls.
In 2004, Steven Freeman was able to get a screen grab of the unweighted exit polls from Edison/Mitofsky which does NEP, because CNN "accidently" published "preliminary data". Because of this, he had some evidence that there was a shift in the numbers from the preliminary data to the final exit poll results, done by weighting the data to correct for inaccuracies in the exit polling procedure. Except that the correction was done for reasons that would have to have been very skewed, and weren't supported by the underlying data on who was polled and who refused to answer.
When statisticians and survey analysts who thought there was merit to what Freeman was saying (not all did) asked to see the precinct level pre-weighted data, NEP clammed up and refused, citing intellectual property rights of its subscribers. Most importantly, they "promised" that a "mistake" like releasing the "preliminary data" would NEVER HAPPEN AGAIN.
It is easier and easier to manipulate elections in the technological age. Some of the manipulations are legal, for instance computer calculated redistricting lines that have the appearance of fairness when you look at a map, but entrench one party or the other permanently in a state or locality. Some are not, the one that shows up in the press, because it is hot, is the electronic voting machines and their flawed paper trails. A computer gives you the ability to know what you need to target, how to make it unobvious, and where you need to do it.
Like computer viruses and identity frauds, there is no way to compete except with computers analysing data to look for patterns of irregularities. If I go into your state, and flip votes sprinkled around the counties, more in counties that favor my candidate, less in the opposing one, and combine it with old fashioned methods that are "legal" like creating long lines and pushing for a candidate to concede before provisional and absentee ballots are counted, I can change your election, and you can't prove it without more data than you can currently lay your hands on, even assuming that you have the "ballots".
So it all comes down to the data.
I mentioned 1-866-OUR-VOTE. The exit polls are the other data needed to have a free and fair election. NEP owns those, in a strange consequence of the IP world and the free market. The only way to get that data, or get it analyzed, short of subpoenas, is to convince the owners, the subscribers to NEP, the news media, to make it publicly available, or provide the watchdogging and analysis themselves. The list of subscribers is on their website http://www.exit-poll.net . If you want a free and fair election, go find a subscriber, and tell them what you want BEFORE the election, when they need to hear it.