[Cthulhu D]

kenberg, on 2014-January-20, 08:47, said:

The fault, dear Brutus, is not in our (movie) stars but in ourselves.

Oh yeah, it's funny watching the US situation vs the EU. Germany is trying to make Greece into Mississipi, but unlike Greece, Mississipi WANTS to be like Mississipi. I have no idea why. I think people have bought into prosperity gospel preachers or something.

[Trinidad]

Cthulhu D, on 2014-January-22, 22:10, said:

Oh yeah, it's funny watching the US situation vs the EU. Germany is trying to make Greece into Mississipi, but unlike Greece, Mississipi WANTS to be like Mississipi. I have no idea why. I think people have bought into prosperity gospel preachers or something.

I think the EU issue is a little more complicated than that.

Some would say that Greece is Mississippi, but over the past 30 years has been living like Colorado with money borrowed from Northern Europe that Greece cannot pay back. Neither characterization does justice to the situation.

Rik

[Zelandakh]

The Eurozone is worth an obscene amount of money to the German economy due to its heavy reliance on exports. This is the reason why Germany is willing to make such efforts to hold the thing together - the investment is definitely worth it once things stablise. For Germany not to take this route would be a political decision rather than a finincial one, since the average voter just sees the money going abroad rather than the returns on every export that the Greeks, et al end up buying down the line. I actually think Merkel has handled the whole situation rather well, not only on the macro-economic level but also politically. Most of the fallout from the bailout has fallen elsewhere and very little actually stuck to her.

[kenberg]

Winston started this thread back in early November, decrying inaccurate or incomplete reporting. I thought that be the end of 2013 things would be clearer. This was naive of me.

A story at http://www.cnbc.com/id/101356161 links to and comments on the recent government report at http://medicaid.gov/...ment-Report.pdf

From the article:

Quote

"The numbers are just so confusing," he said.

"I think it's safe to say that more people are getting Medicaid now than would have before the Affordable Care Act," Jost said. "I can't tell from these numbers how many new ernollees we have, hove many of these are new determinations, how many are under the Medicaid expansion, and how many are currently eligible."

I suppose we will eventually be able to evaluate all of this but not yet. What seems reasonably clear to me is that the expansion of Medicaid is far more successful than the expansion of purchased plans. Beyond that, who knows?



I would like to be able to comment on the recent notes about Greece, Germany, et al but I would only be displaying my ignorance.

[Vampyr]

kenberg, on 2014-January-23, 08:32, said:

I suppose we will eventually be able to evaluate all of this but not yet. What seems reasonably clear to me is that the expansion of Medicaid is far more successful than the expansion of purchased plans.

Perhaps the eventual expansion of Medicaid and/or Medicare to all is the answer.

[y66]

Greece is not like Mississippi

Medicare for all *is* the answer.

Angela Merkel is a political genius.

[HighLow21]

y66, on 2014-January-23, 10:48, said:

Greece is not like Mississippi

Medicare for all *is* the answer.

Angela Merkel is a political genius.

I posted the middle one on my facebook wall as a reminder. People quickly forget why all this started, as they become embroiled in the small (very small, in the grand scheme) problems with the Obamacare rollout.

[Winstonm]

Meanwhile, back at the ranch...

Quote

Now, 764,000 low-income adults in Florida will remain without insurance because of the coverage gap, according to the Henry J. Kaiser Family Foundation

These decisions by governors and legislators essentially consigned a huge swath of the very poor to a life of extreme insecurity.

"It's very frustrating," said Alphonse, who last worked as a security guard until being laid off 10 months ago. "It's kind of odd where an individual that has an opportunity to help millions of people in their own state, and they just totally refuse to do it."

Florida's legislature is poised to take up the Medicaid expansion again during this year's session, but the political dynamics don't appear to have changed much since last year. Meanwhile, one-quarter of Florida's population (under the age of 65) is without health insurance -- the second-highest of all the states behind Texas. In Miami-Dade County, where Alphonse lives, the uninsured rate was an astonishing 34 percent in 2011, the most recent year county-level data were available.

If one is not supplied with bootstraps, it becomes quite difficult to pull oneself up by that means.

[blackshoe]

Winstonm, on 2014-February-02, 08:32, said:

If one is not supplied with bootstraps, it becomes quite difficult to pull oneself up by that means.

The expression arose in an era when most men wore boots, and most boots had bootstraps. I suppose if you're so poor you can't afford any kind of footgear, you won't have bootstraps. If not, I suppose you're going to have to find some other way to pull yourself up.

You seem to be saying that "somebody" (government?) must pull these folks up, or at least help them to do so. While I'm not so sure that's a legitimate function of government, I'm not averse to other organizations doing so. I am averse to continuing to give people help in pulling themselves up long after they've demonstrated that they are either disinterested or incapable of doing so. Granted there aren't many of those, but there are some.

[hrothgar]

blackshoe, on 2014-February-02, 15:59, said:

The expression arose in an era when most men wore boots, and most boots had bootstraps. I suppose if you're so poor you can't afford any kind of footgear, you won't have bootstraps.

The expression was(supposedly) first used in Baron Munchausen.

It categorically refers to something that is impossible.
The prevalence of boots in the culture have nothing to do with one's ability to pull oneself up by your bootstraps.

[Winstonm]

blackshoe, on 2014-February-02, 15:59, said:

The expression arose in an era when most men wore boots, and most boots had bootstraps. I suppose if you're so poor you can't afford any kind of footgear, you won't have bootstraps. If not, I suppose you're going to have to find some other way to pull yourself up.

You seem to be saying that "somebody" (government?) must pull these folks up, or at least help them to do so. While I'm not so sure that's a legitimate function of government, I'm not averse to other organizations doing so. I am averse to continuing to give people help in pulling themselves up long after they've demonstrated that they are either disinterested or incapable of doing so. Granted there aren't many of those, but there are some.

It is not a matter of must but one of should....and in cases of large numbers of peoples spread across the country it should indeed be a federal intervention. The Reagan fantasy that the private sector can be both willing and capable of replacing the government's role in aiding the poor and weak has been so unsuccessful as to be considered fantasy rather than policy.

And personally, I don't care that a handful abuse the system if hundreds of thousands get the help they need. Find and prosecute the wrongdoers as you go, but don't relegate everyone to misery due to fear of being taken in by a petty thief.

[Zelandakh]

hrothgar, on 2014-February-02, 16:37, said:

The expression was(supposedly) first used in Baron Munchausen.

I think this is something of a myth. The Baron pulled himself (and his horse) out of a swamp by his hair (1781, translation 1785: The Surprising Adventures of Baron Munchausen) and there is no reference to bootstraps. More likely is an American origin in the early 1800s with the earliest reference in print that I know of being 1834: Workingman’s Advocate.

[kenberg]

One learns much by carefully by regularly reading the forum.

[mike777]

bootsrap has seeped in math......fwiw complicated math that many of my company use and that I find confusing.
bootstrapping was part of my CFA exams over the 3 years. It is very very common in Bond math, derivative, and other methods.

I suspect Ken and other posters understand it much better.

-----------------------------------------------------


fwiw to totally put this out are we talking about economic mobility at some point?

to put this in simple terms I understand....where poor people become richer and richer people become poorer?

[kenberg]

mike777, on 2014-February-02, 22:52, said:

bootsrap has seeped in math......fwiw complicated math that many of my company use and that I find confusing.
bootstrapping was part of my CFA exams over the 3 years. It is very very common in Bond math, derivative, and other methods.

I suspect Ken and other posters understand it much better.

-----------------------------------------------------


fwiw to totally put this out are we talking about economic mobility at some point?

to put this in simple terms I understand....where poor people become richer and richer people become poorer?

Bootstrap techniques in Statistics are one of the many things that I should know more about than I do. I see the Bootstrap metaphor as an assertion that you can sometimes do more than you might think in limited circumstances. Whether we are speaking of statistical analysis or overcoming poverty, there probably is something to that idea. Intelligence and an open mind can be useful in this regard. So can assistance.

[hrothgar]

Bootstrapping methods in statistics are something that I have a fair amount of experience with.
(I did a lot of work on this stuff back when I was at The MathWorks)

A bootstrap is an example of what's known as a resampling technique.

The basic idea is the following:

Suppose I want to generate some summary statistics for a population of interest.
For whatever reason, its not feasible for me to derive this summary statistic analytically.
Bootstrap methods draw multiple samples from the population of interest, and then use this set of samples to derive the summary statistic.

The prototypical example is generating confidence intervals for the median of a population.
There isn't a nice closed for solution to specify the 95% confidence bound for the median. However, the following works remarkably well:

Suppose that I have a data set with 100 data points.

Start by generating a new data set by sampling (with replacement) from the original data set 100 times.
Calculate the median of this new data set.

Repeat this process a large number of times (say 1,000 or so)

At the end of this process, you'll have a data set that consists of 1,000 medians of the various data sets that you created.

The mean of the this data set should be (virtually) identical to the median of the original data set.
You can also define upper and lower confidence bounds that are

1. Centered on the median of the original data set
2. Contain 95% of all of the medians that you generated.

Really handy stuff now that computing resources are cheap....

[kenberg]

I suppose this is a serious hijack but I am not quite seeing this. You say you have an original data set of 100 points. You generate 1ooo new data sets, each with 100 points. You calculate the median of each of these 1000 data sets and you average the result, and you can expect this average to be close to the median of the original set. Fair enough, but it seems like a lot of work. Why not just calculate the median of the original set?

I thought it was more like this: Your original set is maybe 100,000 points, too large to handle efficiently. So you proceed as you indicate with 1000 samples of 100 points each. Under some circumstances tis average of the medians from the 1000 samples gives a good estimate of the median of the original but large sample.

I'm not trying to catch you, rather what I just said above is about my entire knowledge of bootstrapping, and maybe even that is wrong. So I am interested and I am not following what you said.

OK, I see we can also get confidence intervals. We are speaking here of confidence that this average sample median will be close to the actual median of the original sample, is that right? Say the original sample, of 100,000 in my formulation, comes from a population of 100,000,000. A confidence interval that refers back to this 100,000,000 population would be really great. But it seems like we would need some strobg assumptions to get this. Or not?But presumably the median for the 100,000 is of some decent resemblance to the median for the whole population.

[hrothgar]

kenberg, on 2014-February-03, 12:39, said:

OK, I see we can also get confidence intervals. We are speaking here of confidence that this average sample median will be close to the actual median of the original sample, is that right? Say the original sample, of 100,000 in my formulation, comes from a population of 100,000,000. A confidence interval that refers back to this 100,000,000 population would be really great. But it seems like we would need some strobg assumptions to get this. Or not?But presumably the median for the 100,000 is of some decent resemblance to the median for the whole population.

I don't have MATLAB up and running right now so I can't run a controlled experiment. However, if I used a decent size for the bootstrap (say 1K samples for simple stuff), the difference between the mean of the bootstrap medians and the population median was usually "epsilon" (too small to worry about). The confidence intervals should be as accurate.

With respect to your question about sample size:

In general, if you're doing statistical sampling you get to assume asymptoptic normality.

A sample size of 100,000 is lovely, but is probably severe overkill unless you expect something weird with the original population.

[kenberg]

hrothgar, on 2014-February-03, 13:24, said:

In general, if you're doing statistical sampling you get to assume asymptoptic normality.

A sample size of 100,000 is lovely, but is probably severe overkill unless you expect something weird with the original population.

Asymptotic normality of what distribution? The original population? (Of course the original population is probably discrete but the cdf can still strongly resemble the cdf of some normal population. But that's a pretty strong assumption.

Right, if you wait for 100,000 you wait a long time. What I really meant was that in your original post you had a sample of n=100 and you were doing trial samples from it of size m=100. I was under the impression this was used for n>>m. I probably overdid the >> part.


As an aside, Brad Effron is either the inventor of the Bootstrap or one of the inventors, I think there is some dispute and I am not in a position to say. In my brush with the possibly rich and famous, he and I were in the Philosophers Club when we were adolescents. He went to St. Paul Central, I went to St. Paul Monroe, the club was something that several of us put together on our own. I generally think back on that experience as the first time that I met someone who I was quite sure was smarter than I was. Probably there were many others before that, but this one I recognized.

[hrothgar]

kenberg, on 2014-February-03, 14:31, said:

As an aside, Brad Effron is either the inventor of the Bootstrap or one of the inventors, I think there is some dispute and I am not in a position to say. In my brush with the possibly rich and famous, he and I were in the Philosophers Club when we were adolescents. He went to St. Paul Central, I went to St. Paul Monroe, the club was something that several of us put together on our own. I generally think back on that experience as the first time that I met someone who I was quite sure was smarter than I was. Probably there were many others before that, but this one I recognized.

One of the last projects I did at MATLAB was a program that I called "FitIt". (FitIt is a application for data driving modelling. It used cross validation to derive an optimal spanning parameter for a localized regression model and then used a bootstrap to derive confidence bounds for the resulting surface)

I was proudly showing it off to Cleve Mohler (the co-founder of MathWorks) who said it was a really clever little toy and asked my how I came up with it. I explained that I had read a paper called "An Introduction to the Bootstrap" by Brad Efron and generalized his example. Cleve smiled and said that Brad had been his room mate back in undergrad and he knew the example well. Never have I been so glad to give credit whwn credit was due.

FWIW, I had the same experience with Cleve that you did with Brad. You work with him for a while and you recognize that he is operating at a whole different level.