[DcifrThs]

[ QUOTE ]
[ QUOTE ]


you mean what is their sampling error?

if so then you'd need the standard deviation of the sample (which, i don't know). at lesat if you get it, then you can be comfortable with it since in these types of questions (jobs added), the sampling distribution can be treated comfortably as normal (i.e. one or a few massively outlying observations won't change the average that much).

further, what useful information would this sampling error give you?

and why does it matter?

i mean the # of jobs added every month is a bls estimate, sure, but i'd give them the benefit of the doubt that it is reasonably precise ...

Barron

EDIT: i just reread your question and i may not be answering it. was the 104k the jobs added/month #? because to relate that to the "labor force of 153,000,000" you'd also need the change in labor force (i.e. how much has unemployment changed on net from 1 month to the next).

typcially it is reported as a level rate rather than a change. i.e. 4.7% unemployment vs. a X% change in unemployment (since it would be a rate of change of a rate...like GDP). for labor statistics, i think most like to think in levels and rates rather than rates of change. i don't know though and i'm just trying to get at what you really what to know....

i'm just babbling at this point. can you rephrase your question in context so i know what you want to know and how you want to use it?

[/ QUOTE ]

Just asking as an extension to my confusion over why people give so much credence to this report.

The 104k is from their FAQ “An over-the-month employment change of 104,000 is statistically significant in the establishment survey”. Since there is rarely a 104K change from month to month I don’t understand why so much is made of numbers that according to the BLS aren’t even statistically significant.

It seemed to me that with statistical sampling based on a labor force of 153,000,000 and 104k being the point that the results became statistically significant, they are claiming a pretty high level of accuracy. Most other reports have an error rate in % I was just curious what this reports % error rate was to compare and contrast with others to see if they really are claiming an unusually high level of accuracy.

[/ QUOTE ]

i gotcha now.

the next question is "significant at what level? 90? 95? 99?" that makes a huge difference too.

judging from the way you relayed the quote though i'd guess 90% since it seems to cross the statistically signifcant barrier at 104k.

either way it's a good point, but in general i think despite the signifance level, it is how it comes in vs. the other (significant or not) expectations of bloomberg surveyed economists.

also, i don't htink compairing the figure of 104k to the 153,000,000 isn't appropriate. the comparison would be comparing 104k to the rate of change of the labor force.

that is apples to apples.

i guess though that would only make sense if both #s were significant.


Barron