Apples to Oranges.

How to understand the myriad jobless reports - Stocks & economy- msnbc.com:

"Q: What does the report on weekly jobless claims measure?
A: The report captures the number of people who file their first claim for unemployment insurance after being laid off.

This can be a good measure for how many jobs were cut the week before, but it doesn't capture the whole picture. For whatever reason, not everyone who loses their job ends up filing for unemployment insurance benefits, or at least they may not file right away. Still, the rough number can point to upward or downward trends in the labor market between monthly reports filed by the Bureau of Labor Statistics.

Q: What does the monthly unemployment report measure?
A: The monthly unemployment report is based on a Census Bureau survey of 60,000 U.S. households. The survey asks respondents several questions, including whether they have a job, are looking for a job, or are so discouraged that they have quit looking for work altogether.

The report lays out a whole range of unemployment rates. The most-cited rate only counts people actively looking for work; the latest unemployment report, released on May 8, calculated that rate as 8.9 percent. A much broader unemployment rate — 15.8 percent in the latest report — also includes everyone who has dropped out of the labor force or has been forced into part-time work."

This explains the conflicting recent news that the economy continued to lose jobs but the unemployment rate decreased. The two statistics aren't drawn from the same data. It's apple to oranges, sort of.

Drawing conclusions using those numbers provides conflicting conclusions. GIPS tries to prevent this by offering standards for performance presentation.