Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
A confidence interval that spans one million numbers is unreliable, regardless of what the data concerns. That large of a gap would be the same as issuing a margin of error around ±40%. If we were talking about an election, say 50% of people would vote for Obama, with a ±30% margin of error. That means your actual figure could be between 20% and 80%, which makes the data completely useless. Same concept here.. If your measurement of how many jobs are saved or lost gives you that large of a difference, the resulting figure is worthless.
the resulting figure is worthless
This is just not so when you are talking about Obama sheep. They will believe any data, hmmm...... propaganda that is put out. They need/want to despite what reality is. It helps them cope with their decision last election.
You had better watch out, good ole Sangy is going to get you. You're going to get a lecture about garden variety statistics. I guess they could qualify if you are growing the right stuff in your garden.
Nobody should waste time trying to get you to understand basic math.
We know that.
You don't need to scream it from the rooftops constantly.
Statistics are all about measuring or describing something in an informative and unbiased manner. The concept of trying to rule out bias is foreign to many right-wingers I know, but the results are quite valuable to those of us who care more about what is actually going on than whether we are going to end up with something we can use to bash Obama. If that's your only objective, no, statistics aren't going to be of much use to you...
A confidence interval that spans one million numbers is unreliable, regardless of what the data concerns. That large of a gap would be the same as issuing a margin of error around ±40%. If we were talking about an election, say 50% of people would vote for Obama, with a ±30% margin of error. That means your actual figure could be between 20% and 80%, which makes the data completely useless. Same concept here.. If your measurement of how many jobs are saved or lost gives you that large of a difference, the resulting figure is worthless.
When you're doing a poll such as the one you describe, the range of possible outcomes is constrained across the interval of 0% to 100%. When you are doing a model or projection of employment for instance, the potential range is in at least the millions of individuals in either direction. You are going to get what appears to be a broader confidence interval as the result.
Polls and models differ from each other also in that for a poll or survey, you do power analysis and sample size and CI estimation as part of defining the poll or survey up front. These calculations tell you how many responses you will need to analyze in order to have a desired level of precision in estimating the effect of the variable or variables you are trying to measure. You can then simply choose to move that number up or down in order to either improve the level of precision or reduce the survey's cost.
In a model or projection, quality improvements come only from refining the nature of the equations that represent the relationships you have included in the model. You start with the obvious ones, then add others that are statistically significant. Each addition improves the potential precision of the model, but each addition also comes with its own built-in variability and hence contributes to the potential imprecision of the model as well. You want to end up including only those equations that appear to offer a net improvement in precision. Once those have been selected you can only tinker with them at the margins. Rather than being adjustable up front, the precision of a model can only be determined ex post. In other words, you make the model as statistically valid as you can, then see what sorts of confidence levels it produces. If this process didn't produce useful and informative results, people wouldn't invest in models and projections and a lot of statistical types would be out of a job. Instead, models and projections are run and relied upon all over the place in government, business, and academia. Somebody must be doing something right.
the resulting figure is worthless This is just not so when you are talking about Obama sheep. They will believe any data, hmmm...... propaganda that is put out. They need/want to despite what reality is. It helps them cope with their decision last election.
Something on an anti-intellectual approach. If it's too complicated for you personally to understand, it must be stupid or useless or propagabda. Kind of weak in a universal sort of way...
And I suppose that would be because even though that big fat plus representing your paycheck got deposited, you wrote so many checks that the entire plus was used up, and some of the balance that had been brought forward to boot. So it looks like a net negative can still include a substantial plus if substantial minuses are occurring at the same time.
Just in the month of September, there were more than 4 million new hires in the economy. Unfortunately, there were enough terminations and contract expirations that non-farm employment fell by 263,000 anyway. And I guess that because of that, you're going to try to tell me that NONE of those more than 4 million new hires could have had anything at all to do with ARRA. Is that about the size of it?
This is just great. "Progress" is measured by increased unemployment; that certainly is a sad state of affairs for the current administration. Most previous presidencies (at least the competent ones) would call rising unemployment a failure. But when you are a failure, one must look at things from a slightly different perspective. It is like the Special Olympics of politics.
Saggy, give us some more spin. It is very entertaining to hear you try to explain why rising unemployment is a good thing.
Last edited by hawkeye2009; 12-02-2009 at 06:29 AM..
Considering about $300B has been spent or is in the process, and given the CBO numbers being possibly correct, this equates to
$187,500 to $500,000 SPENT to save ONE job
I think its kinda ridiculous to argue about the numbers of jobs, when the cost per job is so much more "in your face" waste..
Please register to post and access all features of our very popular forum. It is free and quick. Over $68,000 in prizes has already been given out to active posters on our forum. Additional giveaways are planned.
Detailed information about all U.S. cities, counties, and zip codes on our site: City-data.com.