I found this example particularly interesting, in particular because of the media attention heaped upon the twin positive readings of the ISM and construction spending numbers which were released yesterday morning.
Construction spending was expected to decline -0.2%, but instead rose 0.8%. By now you can guess the rest: the previous month, which had been reported as an increase of 0.8%, was revised down to -0.1%. Growth was certainly not explosive despite reports to that effect; rather than beating expectations by a massive 1%, the final index level is just 0.1% better than anticipated over the two months.
With revisions of this magnitude, one can either side with the conspiracy theorists - which might be a little extreme - or simply doubt the sincerity of any of these numbers as they swing wildly from month to month and have margins of errors larger than themselves.