I'm not a statistician but I've always been amazed at how amazingly accurate statistical analysis can be when given a large enough sample and enough information.
That led me to wonder what it would be like if you had
perfect statistical information about
everything!
What would perfect information consist of in the context of election fraud? You would need detailed information down to the precinct level. How would you get this information? Easy. If you 1) manage the system that registers voters; and 2) tabulate the election results.
Think of the information this would give you:
1) You would know the exact number of registered voters in each precinct and their party affiliation.
2) You would know every time each registered voter actually cast a vote in previous elections, enabling you to predict his/her likelihood of voting again
3) You would know the exact voting totals for each precinct going back in time
4) You would be able predict which precincts are most likely to be targets of hand re-counts given the election rules in effect
With this information alone you could probably predict the outcome within a small MOE. Make some adjustment a fews days before the election based on the latest polling data and voila! You know the results before a single ballot has been cast.
To give a candidate a 4% edge you would only have to "flip" 2% of the vote. Statistical analysis would enable you to spread your deception around in the smallest possible amounts to avoid detection. Remember, a state has thousands of precincts. You could entirely avoid precincts that have the highest probability of being recounted.
After a few cycles you would own the baseline and the expectations. Add a pinch or two of voter suppression and only the slightest amount of tweaking would be required to ensure your man victory every time.
BTW: did you catch the news that TRIAD provides BOTH voter registration software AND vote tabulation software for several counties in many states?