public health

[Martin Taylor 2007.07.15.12.45]

[From Rick Marken (2007.07.09.0950)]

Richard Kennaway writes:

>Rick Marken (2007.07.09.0930)]

Here's a remarkable quote that was in Paul Krugman's column today:

"We have always known that heedless self-interest was bad morals; we
know now that it is bad economics." F.D.R. in 1937

I guess "we" stopped knowing that at some point. I'd say in the early 1980s.

Argument from authority?

No. Just an interesting historical observation. I personally agree
with FDR's sentiments; heedless self-interest is immoral from my point
of view. I wish more people felt that way too but there is certainly
no way to convince people of that if they don't want to set that
reference on their own. I think there is also data that shows that
heedless self interest is, indeed, bad economics. But that depends on
what one considers bad economics. Krugman quoted FDR in the context of
the single payer health care issue. The data shows that single payer
systems (which Krugman and I see as oriented toward wise common rather
than heedless self interest) produce much better outcomes for far less
than a private insurance system. That seems like good economics to me.
But it might seem like bad economics to private insurers for whom good
economics is huge profits.

The following isn't PCT, but offers two bits of evidence on the so-called health care problem. (I say so-called, because the argument seems to have been an ideological one about who pays rather than a data-driven one about what leads to good public health). Marken [2007.07.05.1020] said:

If you have data on a system that works better than single payer (in
terms of the variables that matter to me: cost and outcomes) then
that would be great. I'd love to hear about it. I'm more interested
in implementing what works rather than what is _supposed_ to work
based on some theory or ideology.

1.There was an article in "American Scientist" that looked at the relation between health care costs, income disparity, and public health outcome (Clyde Hertzman, Health and Human Society, American Scientist v89, Nov-Dec 2001, pp 538-545). The study suggested that there is effectively zero correlation between cost and outcome (paying more neither helps nor hurts), but there's a fairly strong relationship between income disparity and public health -- more disparity, worse life expectancy and worse child mortality. It includes a fairly dramatic graph showing the relation between mortality rates and the share of income obtained by the poorest 50% of the population across American states and Canadian provinces. The graph suggests that the difference in public health performance in Canada and the US can be attributed almost entirely to the greater income disparity in much of the US, not to the manner of paying for health care. (Or, since correlation can't show causation, that there is a a common factor influencing both mortality and income disparity). There were also graphs showing some data for different countries, but that was less helpful.

2. After reading that article, I created a little spreadsheet from the CIA World Factbook, in which I related per capita income, GINI index of income disparity, and health care outcome (defined in terms of child mortality and expected life span). I wanted to see whether the study's relationships held over a wider range of societies.

My result was that over a rich enough range of per capita income (if I remember rightly, the cut-off was somewhere around 1/4 to 1/3 of US per captia income), average income had no relation to public health, but for lower incomes, it did (agreeing with Hertzman). At all income levels the Gini index did affect the outcome. Taking our both of those effects left a residual variation, which identified some countries as having better public health than one might expect from their income average and oncome disparity index.

I don't have the spreadsheet. It was a lot of work to create, and one of my attempts at using a "sort" command destroyed the inter-relationships among the columns in a way I could not easily recover. So I have to go from memory. What I do remember was that Canada was neither good nor bad; it lay more or less where one would expect from the two indices.

By a substantial margin, the best country for health after allowing for income level and income disparity was Cuba, and the worst were the former Soviet republics of central Asia. The US was not much better than those republics, so it's not a question of the political system. Neither do I think it's a question of the unreliability of data in the World Factbook.

What is clear is that if you don't want to pay much for good public health, you don't have to, provided you can arrange to avoid the kind of great income disparities prevalent in the US and in the former Soviet Union. Furthermore, even if a jurisdiction does arrange for (what I think to be) a morally defensible distribution of incomes, different ways of managing public health make a difference.

I don't think you (US psople) should look to Canada for an example of how to run a public health system. Canada is perhaps better than the US, but it's just neutral in its accomplishment. You should look to Cuba, which succeeds brilliantly despite its low income, and ask what it does right that other dictatorships don't.

There is a PCT point, to follow in another message if I have time.

Martin

[From Rick Marken (2007.07.15.1400)]

Martin Taylor (2007.07.15.12.45) --

1.There was an article in "American Scientist" that looked at the
relation between health care costs, income disparity, and public
health outcome (Clyde Hertzman, Health and Human Society, American
Scientist v89, Nov-Dec 2001, pp 538-545). The study suggested that
there is effectively zero correlation between cost and outcome
(paying more neither helps nor hurts),

I believe that. That is similar to the old RAND study that found no
change in outcomes with increased use (due to free insurance) of
health services.

but there's a fairly strong
relationship between income disparity and public health -- more
disparity, worse life expectancy and worse child mortality.

I just got some data on infant mortality (IM), GINI and per capita
income (PCI) for 121 countries. The correlation between GINI and IM is
.38, which is in the right direction (greater disparity -- GINI--
leads to greater IM) but not overwhelmingly large. The correlation
between IM and PCI is -.65, which also seem right; greater personal
income leads to lower infant morality. When you include both GINI and
PCI in a multiple regression on IM, the income disparity (GINI) effect
all but disappears; PCI is the only statistically significant
predictor of IM. The proportion of variance in IM predicted by PCI
alone is .42; adding GINI brings this up to only .45. The reason is
the correlation (-.38) between GINI and PCI. It turns out that the
greater the PCI of a country, the lower tends to be its GINI (income
disparity).

The
graph suggests that the difference in public health performance in
Canada and the US can be attributed almost entirely to the greater
income disparity in much of the US, not to the manner of paying for
health care.

Well, that seems to go against the international data. I'll take a
look at the paper.

My result was that over a rich enough range of per capita income (if
I remember rightly, the cut-off was somewhere around 1/4 to 1/3 of US
per captia income), average income had no relation to public health,
but for lower incomes, it did (agreeing with Hertzman).

I don't know what you did here. There may be differences in what you
get when you look at sub-range correlations. Let me know what you did
and I'll see what my data show. But the basic result for the data I
have (for 121 countries ranging from Malawi to Sweden) is that it is
per capita income (PCI) and not income disparity (GINI) that matters
in terms of infant mortality. And I would wager that that's because
most of the high PCI countries are the industrialized nations with
single payer healthcare.

I don't think you (US psople) should look to Canada for an example of
how to run a public health system.

I'm not. I'm looking to France! Did I mention that God gave me the
South of France? I'm glad the current residents are taking good care
of it because I'll be returning to reclaim it for me and my ancestors
soon (unless Kucinich or Gravel become President over here;-). I hope
the current residents don't mind moving to the slums of Marseilles;-)

Best

Rick

···

--
Richard S. Marken PhD
Lecturer in Psychology
UCLA
rsmarken@gmail.com

[Martin Taylor 2007.07.15.17.40]

[From Rick Marken (2007.07.15.1400)]

Martin Taylor (2007.07.15.12.45) --

I just got some data on infant mortality (IM), GINI and per capita
income (PCI) for 121 countries. The correlation between GINI and IM is
.38, which is in the right direction (greater disparity -- GINI--
leads to greater IM) but not overwhelmingly large. The correlation
between IM and PCI is -.65, which also seem right; greater personal
income leads to lower infant morality. When you include both GINI and
PCI in a multiple regression on IM, the income disparity (GINI) effect
all but disappears; PCI is the only statistically significant
predictor of IM. The proportion of variance in IM predicted by PCI
alone is .42; adding GINI brings this up to only .45. The reason is
the correlation (-.38) between GINI and PCI. It turns out that the
greater the PCI of a country, the lower tends to be its GINI (income
disparity).

That's an interesting observation in itself. If it holds up over different ranges of PCI, one might ask whether there's a causal relationship, and if so, whether it's between these two measures (and in which direction) or between some common factor and them both.

Where did you get your data? They must be formatted more conveniently than the CIA World Factbook data, which took a lot of regexp type of editing before I could make the spreadsheet (which is why I've always been discouraged from trying to redevelop what I did before).

The
graph suggests that the difference in public health performance in
Canada and the US can be attributed almost entirely to the greater
income disparity in much of the US, not to the manner of paying for
health care.

Well, that seems to go against the international data. I'll take a
look at the paper.

My result was that over a rich enough range of per capita income (if
I remember rightly, the cut-off was somewhere around 1/4 to 1/3 of US
per captia income), average income had no relation to public health,
but for lower incomes, it did (agreeing with Hertzman).

I don't know what you did here. There may be differences in what you
get when you look at sub-range correlations.

Yes, I think that's part of the point. PCI matters, but only for what we used to call "developing" or "Third World" countries. For "developed" countries, PCI doesn't seem to matter, but disparity does. For the lower-income countries, I seem to remember finding that they both matter.

Let me know what you did
and I'll see what my data show. But the basic result for the data I
have (for 121 countries ranging from Malawi to Sweden) is that it is
per capita income (PCI) and not income disparity (GINI) that matters
in terms of infant mortality. And I would wager that that's because
most of the high PCI countries are the industrialized nations with
single payer healthcare.

I'd guess the result is because of the 121 countries, far more of them are not in the high PCI range, but in the income range where income does matter.

Please do note that memory is fallible. I think I remember the pattern of my results pretty well, but I don't have the ability to go back and check, so it's possible that I may have misstated some detail (though not the main pattern, I think).

If you can point me to the dataset you found, I'd appreciate that, without promising to do anything with it soon. I've got a lot of catching up to do in the next two weeks, before I'm away again (for only a week this time).

Martin

[From Rick Marken (2007.07.15.1545)]

Martin Taylor (2007.07.15.17.40)

Where did you get your data? They must be formatted more conveniently
than the CIA World Factbook data,

Here's the spreadsheet, including data, regression analysis and cross
correlations. I got the data from several different places -- via
Wikipedia. I only kept the data for the 121 countries that over-lapped
in all three sets. Unfortunately, Cuba was not one of them. What I
would like to get now is a fourth column classifying each country in
terms of the type of health care system: 0 = Private Insurance; 1 =
Single Payer.

Ain't data fun!

Best

Rick

health.xls (24 KB)

···

--
Richard S. Marken PhD
Lecturer in Psychology
UCLA
rsmarken@gmail.com
Content-Type: application/vnd.ms-excel; name=health.xls
X-Attachment-Id: f_f4646b4g

[Martin Taylor 2007.07.15.18.55]

[From Rick Marken (2007.07.15.1545)]

Martin Taylor (2007.07.15.17.40)

Where did you get your data? They must be formatted more conveniently
than the CIA World Factbook data,

Here's the spreadsheet, including data, regression analysis and cross
correlations. .. What I
would like to get now is a fourth column classifying each country in
terms of the type of health care system: 0 = Private Insurance; 1 =
Single Payer.

Thanks for the spreadsheet. In your extension you would need more categories. Most of Europe has a mixed model, I believe, with basic single payer, but the option for patients and doctors to opt out and go private, or to add services. The big argument in Canada is whether to go that route, and allow private insurance and the existence of private clinics and hospitals that offer services that are already covered in the public system (so as to reduce waiting times for the rich). We already do have private insurance for services that are not covered by the public system (e.g. private hospital rooms instead of ward beds). I don't know whether any country has totally single-payer health services (Cuba, perhaps?).

Ain't data fun!

Fun, but like any good metaphor, the temptation is to follow it further than it warrants.

Martin

[From Bill Powers (2007.07.15.2111 MDT)]

Rick Marken (2007.07.15.1545)]

It would be interesting to see a scatter plot of infant mortality against
each of the independent variables, perhaps with the regression line
overlaid. That might give a clearer picture of the data than the various
statistical quantities, especially for those of us not to used to
statistics.

Best,

Bill P.

Cuba's and the U.S. Veterans' Administration's programs are fully socialized, as is the health plan enjoyed by all of our Congress critturs. In other words, all the personnel are government employees, as opposed to being contractors who bill the government for their services which is the way a single payer program such as Medicare and Medicaid and the Canadian system works.

These are the best programs in the world by any standard.

Ted

···

On Sun, 15 Jul 2007 19:02:14 -0400 Martin Taylor <mmt-csg@ROGERS.COM> wrote:

[Martin Taylor 2007.07.15.18.55]

[From Rick Marken (2007.07.15.1545)]

Martin Taylor (2007.07.15.17.40)

Where did you get your data? They must be formatted more conveniently
than the CIA World Factbook data,

Here's the spreadsheet, including data, regression analysis and cross
correlations. .. What I
would like to get now is a fourth column classifying each country in
terms of the type of health care system: 0 = Private Insurance; 1 =
Single Payer.

Thanks for the spreadsheet. In your extension you would need more categories. Most of Europe has a mixed model, I believe, with basic single payer, but the option for patients and doctors to opt out and go private, or to add services. The big argument in Canada is whether to go that route, and allow private insurance and the existence of private clinics and hospitals that offer services that are already covered in the public system (so as to reduce waiting times for the rich). We already do have private insurance for services that are not covered by the public system (e.g. private hospital rooms instead of ward beds). I don't know whether any country has totally single-payer health services (Cuba, perhaps?).

Ain't data fun!

Fun, but like any good metaphor, the temptation is to follow it further than it warrants.

Martin

In a message dated 7/16/2007 12:26:39 P.M. Eastern Daylight Time, tcloak@UNM.EDU writes:

Cuba’s and the U.S. Veterans’ Administration’s programs
are fully socialized, as is the health plan enjoyed by all
of our Congress critturs. In other words, all the
personnel are government employees, as opposed to being
contractors who bill the government for their services
which is the way a single payer program such as Medicare
and Medicaid and the Canadian system works.

These are the best programs in the world by any standard.

Ted

Are you not aware of the Vet. Adm. health care scandal earlier this year?

Kenny

···

Get a sneak peak of the all-new AOL.com.

[From Rick Marken (2007.07.16.0955)]

Bill Powers (2007.07.15.2111 MDT)]

Rick Marken (2007.07.15.1545)]

It would be interesting to see a scatter plot of infant mortality against
each of the independent variables, perhaps with the regression line
overlaid. That might give a clearer picture of the data than the various
statistical quantities, especially for those of us not to used to
statistics.

Here you go. The graphs are in the "Graphs" tab. I agree that it is
always a good idea to look at the data and not just the summary stats.
The graphs I made reveal a couple of interesting things: First is that
the relationship between per capita income and infant mortality is
logarithmic. I have redone the regression using log income and income
and Gini together now pck up 70% of the variance in infant mortality
rates across countries. Second, the relationship between Gini and per
capita income is kind of odd. It looks like there is an approximately
equal number of low and high per capita income countries with a low
Gini (equitable income distribution). But it is only low per capita
income countries that have high Gini (>50, the range of Gini is from 0
to 100).

Best

Rick

health2.xls (48.5 KB)

···

--
Richard S. Marken PhD
Lecturer in Psychology
UCLA
rsmarken@gmail.com
Content-Type: application/vnd.ms-excel; name=health2.xls
X-Attachment-Id: f_f4774f19

[From Bryan Thalhammer (2007.07.16.1155 CDT)]

Kenny,

Your point about the VA and the way it has been managed since the 80s? After the
Vietnam War, vets who had severe mental distress were turned out on the streets
by Ronny Raygun. Agent Orange was denied by the Reagan admin, and vets had to
fight back to get medical coverage. Then in the Gulf War, another WMD used by
the US was harming Gulf War vets and their families. Senator Webb has pointed
out in abundance that Iraq Vets are not being cared for since the VA has been
cut to ribbons by THIS Bush Admin. During the Republican Congress since 1992,
they have virtually ignored the VA.

This is blaming the victim. Design a good system, fund it appropriately, use it
appropriately, and let it be managed and evaluated by people who are qualified
(not Republican political appointees!)...

--Bry

···

In a message dated 7/16/2007 12:26:39 P.M. Eastern Daylight Time,
tcloak@UNM.EDU writes:

Cuba's and the U.S. Veterans' Administration's programs
are fully socialized, as is the health plan enjoyed by all
of our Congress critturs. In other words, all the
personnel are government employees, as opposed to being
contractors who bill the government for their services
which is the way a single payer program such as Medicare
and Medicaid and the Canadian system works.

These are the best programs in the world by any standard.

Ted

Are you not aware of the Vet. Adm. health care scandal earlier this year?

Kenny

************************************** Get a sneak peak of the all-new AOL at
http://discover.aol.com/memed/aolcom30tour

[From Rick Marken (2007.07.16.1010)]

···

On 7/16/07, Kenneth Kitzke Value Creation Systems <KJKitzke@aol.com> wrote:

Are you not aware of the Vet. Adm. health care scandal earlier this year?

Kenny, do things in the US seem to you to be better now than they did
in 1999? Does it seem to you like we (the US) are moving in the right
direction?

Best

Rick
--
Richard S. Marken PhD
Lecturer in Psychology
UCLA
rsmarken@gmail.com

[From Kenny Kitzke (2007.07.16)]

<Rick Marken (2007.07.16.1010)>

<Kenny, do things in the US seem to you to be better now than they did
in 1999? Does it seem to you like we (the US) are moving in the right
direction?

Best

What a question! Some things are better and other things are worse as I perceive them.

If I had to generalize, I would say it seems to be getting worse year by year. I have no confidence that government can move the USA in the “right” direction…whatever that may be.

BTW, prophecy in the Bible predicts that Jerusalem will be a cup of trembling in the end times for the whole world and that evil deeds will increasingly predominate the affairs of men and nations. Does it seem that way to you? It surely does to me.

···

Get a sneak peak of the all-new AOL.com.

[From Kenny Kitzke (2007.07.16)]

<Bryan Thalhammer (2007.07.16.1155 CDT)>

<Senator Webb has pointed
out in abundance that Iraq Vets are not being cared for since the VA has been
cut to ribbons by THIS Bush Admin. During the Republican Congress since 1992,
they have virtually ignored the VA.

This is blaming the victim. Design a good system, fund it appropriately, use it
appropriately, and let it be managed and evaluated by people who are qualified
(not Republican political appointees!)…

–Bry>

You sound like a broken record. I guess the floods and hurricanes and my leaking sink drain are the fault of Reagan and the Republicans too?

My recollection is that funding of the VA was at record levels under Bush. Do you have any facts to the contrary?

If the Congress did ignore the VA , then the problems in the VA health care system are all the more related to its leaders. BTW wasn’t the top VA man who resigned in disgrace appointed by President Clinton?

···

Get a sneak peak of the all-new AOL.com.

[From Bill Powers (2007.07.16.0800 MDT)]

Rick Marken (2007.07.15.1545), Martin Taylor (2007.07.15.17.40)

···

It occurs to me that it might be worthwhile mining the public health data
bank for other relationships. For example, what is the correlation
between infant mortality and other public-health measures on the one
hand, and number of people with no health insurance on the other hand?
There must be some big obvious correlations in there somewhere.

Also, it’s often said that medicine in the US is highly advanced, partly
because so much money is spent on it. We probably consume more new drugs
per year per capita than anyone else. So how do our cure rates for
different diseases compare with those in other countries?

We have read about some really big studies that have been done on things
like cholesterol and heart disease. Unfortunately, big studies involving
tens or hundreds of thousands of people will find significant effects
when the percentage of positive results hardly differs from chance. Also,
drugs are often pronounced effective when they appear to help only a
minority of the population. So we have a situation in which a person can
be persuaded to take an expensive drug which reduces the incidence of
some condition by 20%, meaning that at least 4 out of 5 people taking the
drug will have wasted their money and worsened their conditions since
they will experience only the side-effects. Would it be possible to get
some real numbers relating to this thought-example?

I think a truer story about any statistical relationship is obtained by
calculating the chances that a decision made about an individual on the
basis of population statistics is incorrect. The present discussion is
about a set of individual countries and their health-care systems. What
are the chances that the relative ranking of, say, France and the United
States on a measure like infant mortality will correctly predict that a
newborn baby in one of those countries will outlive a newborn baby in the
other country? This number should be directly obtainable from the
actuarial tables without going through correlation calculations. Will it
be greater than chance, and how much greater? Will the chances be 50.1%?
or 80%? The meaning of the ranking depends very much on those
numbers.

Statistics seems to be used in medicine and psychology mainly as a way of
salvaging some slight positive effect out of treatments that, for any
individual, are more often ineffective than effective. Remember the
“coefficient of uselessness” that I once proposed? It is a
measure of the likelihood that a statistically-established relationship
will NOT hold true in any individual case (I’m not sure that is the right
way to put it). It is, as I remember, the square root of (one minus the
correlation squared). That’s the square root of the “coefficient of
alienation”. For a correlation of 0.36, mentioned by Rick as
“being in the right direction”, the coefficient of uselessness
would be 0.933 (the maximum possible is 1). We can call this coefficient
the “anticorrelation” to avoid pejorative terms.

See:

[
http://www.utdallas.edu/~herve/Abdi-Correlation2007-pretty.pdf

](http://www.utdallas.edu/~herve/Abdi-Correlation2007-pretty.pdf)The “anticorrelation” concept is supported by thinking of
the correlation as the cosine of the angle between two vectors (as Martin
finally convinced me). If a correlation between two vectors is 0.36, the
angle between the vectors is 68.9 degrees. A correlation of zero would
mean an angle of 90 degrees, so the difference between the actual angle
and 90 degrees is 21.1 degrees. The cosine of 21.1 degrees is 0.933, the
same anticorrelation calculated above. If correlation is the degree of
relatedness, the anticorrelation is the degree of unrelatedness in the
same units. So if we interpret a correlation of 0.36 as saying that two
vectors that are 36% related, they are 93% unrelated in the same terms.
This shows the bias of correlations in exaggerating relatedness. I’m sure
that bias explains why correlations are used in sciences that find it
difficult to see any regularities in their subject-matter. A 30% degree
of relatedness looks a lot better in print than a 95% degree of
unrelatedness, though both numbers describe the same result.

It seems to me that what we really need is a “predictive value”
for using statistical relationships to predict one variable from knowing
another. I propose using the ratio of the correlation to the
anticorrelation. This ratio is 1.0 for a correlation of 0.7071, meaning
that the measures are equally related and unrelated. I assume that would
come out to be equal chances of a yes or a no. We could subtract 1 to
make this predictive value 0 or neutral. So I’ll define Powers’
Predictive Value as

PPV = correlation/anticorrelation - 1.

For a correlation of 0.36, and an anticorrelation of 0.93, the predictive
value is 0.36/0.93 - 1 or -0.61, so the decision should not be made
(or maybe the decision should go the other way).

The point where the correlation and anticorrelation are equal is at r =
0.7071. The angle is 45 degrees. By my proposed measure, that is the
correlation at which predictions have a predictive value of 0. In other
words, a prediction based on a correlation of 0.7 has equal chances of
being correct and incorrect, the same as using a coin-toss. The following
table is interesting: it seems to fit my contention that correlations
less than 0.9 are not worth a lot, Those less than 0.7 have a greater
chance of predicting wrong than right, and so have a negative predictive
value.

Corr
Anti-corr PPV

0.050
0.999 -0.9

0.100
0.995 -0.9

0.150
0.989 -0.8

0.200
0.980 -0.8

0.250
0.968 -0.7

0.300
0.954 -0.7

0.350
0.937 -0.6

0.400
0.917 -0.6

0.450
0.893 -0.5

0.500
0.866 -0.4

0.550
0.835 -0.3

0.600
0.800 -0.3

0.650
0.760 -0.1

0.700
0.714 0.0

0.750
0.661 0.1

0.800
0.600 0.3

0.850
0.527 0.6

0.900
0.436 1.1

0.950
0.312 2.0

0.960
0.280 2.4 (
note finer intervals)

0.970
0.243 3.0

0.980
0.199 3.9

0.990
0.141 6.0

Richard K, I assume you’re going to be all over this like an
octopus.

Best,

Bill P.

[From Bryan Thalhammer (2007.07.16.1320)]

* The VA was found a Billion bucks short in 2005. Republicans were not pleased
with Jim Nicholson, VA Secretary.

* NPR reports the fight over the VA shortfall.
http://www.npr.org/templates/story/story.php?storyId=4722633

Gee, all you have to do, Kenny is search Google for the generically neutral "VA
Funding" and the first things that come up are not funding satisfactory, or
funding beyond the needs of vets, but instead, funding shortfalls. What about
Walter Reed do you not understand regarding funding shortfalls? My father used
the VA and finally had to go to the healthy Medicare System, which Bush has
repeatedly raided to finance his dirty war.

When the government sends military to Iraq and other places, ploying them with
$50K tax free reenlistment bonuses, are you telling me that you believe in your
heart of hearts that there should not be equivalent funding for taking care of
them when they are hurt or dead? Why one and not the other? Are you so cruel
that you believe that $50k on the front-end is going to compensate for 60+ years
of poverty and misery?

You no-tax, no-govt types are all alike. You want to keep your money, but then
you don't mind riding on the roadways provided in the Eisenhower Interstate
system, and using other resources that responsible govt. provides. You expect
individuals to take care of their government-sponsored injuries. That is real
tripe.

--Bryan

···

[Kenny Kitzke (2007.07.16)]

<Bryan Thalhammer (2007.07.16.1155 CDT)>

<Senator Webb has pointed
out in abundance that Iraq Vets are not being cared for since the VA has
been
cut to ribbons by THIS Bush Admin. During the Republican Congress since
1992,
they have virtually ignored the VA.

This is blaming the victim. Design a good system, fund it appropriately, use
it
appropriately, and let it be managed and evaluated by people who are
qualified
(not Republican political appointees!)...

--Bry>

You sound like a broken record. I guess the floods and hurricanes and my
leaking sink drain are the fault of Reagan and the Republicans too?

My recollection is that funding of the VA was at record levels under Bush.
Do you have any facts to the contrary?

If the Congress did ignore the VA , then the problems in the VA health care
system are all the more related to its leaders. BTW wasn't the top VA man
who resigned in disgrace appointed by President Clinton?

************************************** Get a sneak peak of the all-new AOL at
http://discover.aol.com/memed/aolcom30tour

[From Rick Marken (2007.07.16.1125)]

Kenny Kitzke (2007.07.16)--

<Rick Marken (2007.07.16.1010)>

<Kenny, do things in the US seem to you to be better now than they did
in 1999? Does it seem to you like we (the US) are moving in the right
direction?

What a question! Some things are better and other things are worse as I
perceive them.

If I had to generalize, I would say it seems to be getting worse year by
year. I have no confidence that government can move the USA in the "right"
direction...whatever that may be.

I see it the same way. But wasn't it clear to you by 2004 (it was
clear to me by 2002) that this government was moving the USA in the
wrong direction? If not, what changed your mind recently? And if you
did know things were going south before 2004, did you vote to change
the government then? If not, why?

BTW, prophecy in the Bible predicts that Jerusalem will be a cup of
trembling in the end times for the whole world and that evil deeds will
increasingly predominate the affairs of men and nations. Does it seem that
way to you? It surely does to me.

I don't know about Jerusalem (why did God only make prophecies about
Jerusalem and not Athens, Paris or Beijing? He should get out more;-)
but I think evil deeds are running about at par. I agree that the evil
done by the US government has increased manifestly since Bush has come
into office. But overall I think it would be hard to beat the evil
deeds done in the mid 1900s by Germany and the Soviet Union.

Bill Powers once told me, when I suggested that someone would "never
get PCT", that prophecies are reference signals. I didn't want to
believe him then but I agree with him now. And I think it explains a
lot about why the President we have does the things he does. I think
that people who believe in the "end times" when "evil deeds will
increasingly predominate the affairs of men and nations" have
basically set a reference for seeing this occur. I've heard that Bush
believes this crap and it appears that he is doing what he can to
bring the state of the world to a reference state where "evil deeds
will increasingly predominate the affairs of men and nations".
Wouldn't you really rather have a leader who believed that the world
is destined to become better and who, therefore, acted in ways that
tended to bring this about even if that leader lies once or twice
about having sex?

Best

Rick

···

--
Richard S. Marken PhD
Lecturer in Psychology
UCLA
rsmarken@gmail.com

[From Richard Kennaway (2007.07.16.1901 BST)]

Have people seen Hans Rosling's work on trends in international health?

http://roslingsblogger.blogspot.com/

···

--
Richard Kennaway, jrk@cmp.uea.ac.uk, http://www.cmp.uea.ac.uk/~jrk/
School of Computing Sciences,
University of East Anglia, Norwich NR4 7TJ, U.K.

[From Bill Powers (2007.07.16.0800 MDT)]
Richard K, I assume you're going to be all over this like an octopus.

Ah, quite. One of the things that occurred to me from Richard Marken's spreadsheet (apart from "Runkel! Thou should'st be living at this hour") was to look at the correlations among the three variables: income, infant mortality, and GINI. The figures in the spreadsheet are:
     Income:IM -0.66
     Income:GINI -0.38
     IM:GINI: 0.39

Abstractly, let there be three variables A, B, and C, with three correlations rAB, rAC, and rBC. Suppose that we hypothesize a causal model, according to which there are four independently acting causal factors W, X, Y, and Z, which give rise to the observables A, B, and C like this:

A = W
B = W + X + Y
C = W + X + Z

That is, B and C are each caused by A and X, plus a separate causal factor for each. Clearly, B and C will each correlate positively with A, and with each other. If we replace income by negative income in Ricks table, then all his correlations become positive.

The question is then, can random variables W, X, Y, and Z be found which result in the observed correlations rAB, rAC, and rBC?

The answer is yes, given certain conditions on the correlations.

Let the standard deviations of W, X, Y, and Z be w, x, y, and z.
Write Cov for covariance, Var for variance, and Std for standard deviation.
Define p = Std(B) = sqrt(w^2+x^2+y^2)
        q = Std(C) = sqrt(w^2+x^2+z^2)
        r = sqrt(w^2+x^2)

rAB = Cov(A,B)/Std(A)*Std(B)
     = w^2/(w*p)
     = w/p
Similarly, rAC = w/q
rBC = r^2/(p*q)

The correlations are unaffected by scaling everything equally, so take w = 1.
Then we have

     p = 1/rAB
     q = 1/rAC
     r = sqrt(rBC/(rAB*rAC))

From these we want to get back to x, y, and z.

From the definition of r and the last equation we get
     x^2 = r^2 - 1 = rBC/(rAB*rAC) - 1 = (rBC - rAB*rAC)/(rAB*rAC)
From the definition of p we have
     y^2 = p^2 - r^2 = (rAC - rBC*rAB)/(rAB^2*rAC)
and similarly
     z^2 = (rAB - rBC*rAC)/(rAC^2*rAB)

Since these must be non-negative, it is necessary and sufficient for a solution that
     rBC >= rAB * rAC
     rAC >= rBC * rAB
     rAB >= rBC * rAC
That is, the product of any two is not greater than the third. These all hold for Rick's data. (In fact, I believe these inequalities are true for any trivariate distribution, although I haven't proved it.)

Notice that these conditions are symmetrical in A, B, and C. That means that if there is a causal model consistent with the data, that says that B and C are caused by A plus other factors, then there are also causal models saying that B causes A and C, or that C causes A and B. And if my suspicion about the inequalities is true, there is always such a model.

Or as we all know, "correlation does not imply causality".

···

--
Richard Kennaway, jrk@cmp.uea.ac.uk, http://www.cmp.uea.ac.uk/~jrk/
School of Computing Sciences,
University of East Anglia, Norwich NR4 7TJ, U.K.

Where do you get your information on the Cuban medical system? You know, given that the country is a jail and owning a computer is a crime there. Do you think that Michael Moore saw anything but what the Cuban regime decided he would see, and that he wasn't happy to be their useful idiot?

http://www.civita.no/civ.php?mod=news&id=299

···

At 09:25 -0700 16/7/07, Frank T Cloak Jr wrote:

Cuba's and the U.S. Veterans' Administration's programs are fully socialized, as is the health plan enjoyed by all of our Congress critturs. In other words, all the personnel are government employees, as opposed to being contractors who bill the government for their services which is the way a single payer program such as Medicare and Medicaid and the Canadian system works.

These are the best programs in the world by any standard.

--
Richard Kennaway, jrk@cmp.uea.ac.uk, http://www.cmp.uea.ac.uk/~jrk/
School of Computing Sciences,
University of East Anglia, Norwich NR4 7TJ, U.K.

rBC >= rAB *
rAC

rAC >= rBC * rAB

rAB >= rBC * rAC

That is, the product of any two is not greater than the third.
These all hold for Rick’s data. (In fact, I believe these
inequalities are true for any trivariate distribution, although I haven’t
proved it.)

Notice that these conditions are symmetrical in A, B, and C. That
means that if there is a causal model consistent with the data, that says
that B and C are caused by A plus other factors, then there are also
causal models saying that B causes A and C, or that C causes A and
B. And if my suspicion about the inequalities is true, there is
always such a model.

Or as we all know, “correlation does not imply
causality”.
[From Bill Powers (20-07.07.16.1352 MDT)]

Richard Kennaway (2007.07.16) –

It’s fun to watch a mathematician at play, even if I get
lost rather easily. Could it be that the situation as you show it at the
end implies a closed loop of causality?

Anyway, what do you think of that nonlinear relationship I’m proposing
between correlation and anticorrelation? I wouldn’t be surprised to see
I’m inventing another wheel, but it’s intriguing to think we can show
that a correlation of 0.7071 is no better than a coin-toss for making
predictions. Could that really be true?

Best.

Bill P.