In his State of the Union Address late last month, President Barack Obama declared – to great applause – that there were two million Americans working now who would otherwise be unemployed if not for the American Recovery and Reinvestment Act, better known as the stimulus.

Three days later, the same night that Recovery.gov released a slew of new data on individual projects funded by the stimulus, the site Web ticker that tracks the total number of jobs reported by recipients dropped from 640,349 to 599,108.

Did we suddenly lose 40,000 jobs? And didnt the President say that the number of new jobs was more than three times that latest figure? What gives Recovery.gov?

For those not willing to slog through blogs, press releases, or congressional reports, there is an explanation for this seemingly contradictory information.

According to a White House blog post by Ed DeSeve, the data available on Recovery.gov, including the number of jobs created, represents only a small pool of the money spent–specifically only $50 billion, or about one-fifth of the total amount of spending and tax cuts the the Recovery Act has pumped into the economy through the end of 2009.

DeSeve’s blog post says the number of jobs created does not include Recovery Act spending on small business loans, tax credits and financial assistance for individuals and families. It also doesnt include indirect jobs that may be created as a result of Recovery funds, like jobs given to subcontractors of a government contract, or a waitress hired to handle a growing lunchtime rush due to a Recovery construction project being built nearby.

So when Obama announced that 2 million jobs were created, he didnt cite job figures from Recovery.gov, but instead used estimates calculated by the Presidents Council of Economic Advisers, which said the stimulus raised employment by anywhere from 1.5 million to 2 million jobs. He also used a report by the Congressional Budget Office that estimated the number of jobs created as anywhere from 600,000 to 1.6 million.

The White Houses DeSeve said that Recovery.gov only tracks 20 percent of the Recovery dollars because Congress only asked that these reports be completed by a portion of Recovery recipients, particularly in infrastructure projects and education spending.

As for the total job numbers reported on Recovery.gov, the two tallies count different things over unequal lengths of time. The roughly 640,000 jobs that were reported over a period beginning February 17 and ending September 30, 2009, while the roughly 600,000 jobs currently on Recovery.gov cover a time period from Oct. 1 to Dec. 31, 2009. The two numbers can’t be combined to reach a total number of jobs created or funded by the stimulus, because some jobs would be double counted, said Ed Pound, spokesman for the Recovery Accountability and Transparency Board, which oversees Recovery.gov.

The same job could be counted in both time periods, Pound said. So for the most precise number of how many jobs have been created by Recovery money, its always the amount listed at any one time on the Recovery.gov site that plus the other jobs created from the other 80 percent of Recovery money that the site does not capture.

The administration also changed the criteria for what constitutes a Recovery job. In the earlier period, only newly created jobs or existing jobs that would have been eliminated absent stimulus funds were supposed to be counted. For the later period, any job funded by stimulus dollars could be counted.

Further complicating the count was the question of hours: new jobs or retained jobs were expressed as full-time employees, but the period of the performance was not always explicitly stated.

A GAO report in November 2009 found that this data was being inconsistently reported and that there were differing interpretations of the rules on counting full-time employees. In California for example, two different colleges calculated full-time employees differently, one chose a 2-month period while the other chose a year period. GAO recommended that the White House clarify how to standardize the definition of measuring full-time employees.

Based on these recommendations, the White House adopted clearer language, telling recipients to calculate jobs created based on the employee hours paid for with Recovery funds during a single quarter. Heres some of the fine print from Recovery.gov:

…if three employees of a recipient worked on a Recovery-funded project for a total of 1,300 hours in a given quarter, and the recipient defines a full-time schedule as 520 hours per quarter (40 hours per week, for 13 weeks), the recipient would report that the full-time equivalent (FTE) of 2.5 jobs (1,300/520 = 2.5).

Of course, that type of fine print reporting only covers part of the $787 billion in spending and tax credits from the stimulus. When the new Recovery data was first reported late on Saturday, January 30, Vice President Joe Biden issued a statement that acknowledged that it only represented 20 percent of all Recovery funding. Because of this unprecedented transparency effort, the American people are getting a look at some of the ways Recovery dollars are benefiting their neighborhoods and communities directly from the recipients themselves, Biden said.

Since this is a partial survey based on reports filed by recipients, we know its not perfect or complete but it is providing a level of detail about a government program that has never before been made available to the public.