I just want to point out that the data is coming from a BLS report
link:
http://www.bls.gov/news.release/laus.nr0.htm
The article can't even copy data correctly.
BLS:
In August 2014, nonfarm payroll employment increased in 35 states and decreased in 15 states and the District of Columbia. The largest over-the-month increases in employment occurred in California (+44,200), Florida (+22,700), and Texas (+20,100).
yahoo:
The biggest job gains were in Texas, which added 46,600 jobs, followed by California, with 27,700 and Michigan, with 17,900.
It is mathematically impossible for the number of jobs to increase and have unemployment go up unless the denominator (labor force) increased. The article mentions that in most states the number of jobs increased.
If you look at the Tables 3 through 6, on both a seasonal adjusted and not seasonal adjusted basis, for Georgia, the labor force shrank, number of jobs went up and number of unemployed went up. The impossible happened... 4,755,200 in the labor force, 393,200 unemployed, and 4,132,900 employed... wait those don't add up, there are 229,100 people missing in the non seasonally adjusted data for Georgia.
You are forgetting that the jobs data comes from a different survey(CES) than the labor force and unemployment data(CPS). There are many difference between these two surveys. For example, a person working two different jobs would be counted twice in CES but once in CPS. The Self-employed are counted in CPS, but not CES. There are many more differences and if you feel like reading more:
http://www.bls.gov/bls/fesacp2101703.pdf