USA Brings Freedom of Religion to Iraq
"the Special Inspector General for Iraq Reconstruction ... Bowen and his deputy Ginger Cruz (a wiccan who has allegedly spooked staff with threats of hexes)"
I am absolutely sincerely delighted that a wiccan held a position of high responsibility in the US effort in Iraq.
That Hideous Strength
Back to Kirsch et al. A study which claimed that antidepressants provide clinically significant benefits only for the severely depressed has received a lot of attention
33 google news hits and lots of plain google hits.
Much commentary completely missed the "clinically significant" and, in fact claimed that Kirsh et al had shown that patients who received a placebo did "just as well" as patients who received antidepressants. This was a null hypothesis overwhelmingly rejected by Kirsch et al.
The huge amount of attention received by the paper is, I think, entirely due to the appeal to the concept of "clinical significance" as authoritatively defined by the National Institute for Health and Clinical Excellence (NICE) in the UK. Hmm where have I read that acronym before ?
NICE declared that to be clinically significant a benefit had to be 3 points on the Hamilton Rating Scale of Depression or at least one half of one standard deviation of the changes in the HRSD in the treated subsample.
The first definition is arbitrary and, I think, nonsensical (if the only change were from "I think life is not worth living" to "I think life is worth living" that would be one point on the HRSD). However the second definition is much much more absurd.
The standard deviation of changes is very important for testing whether an apparent benefit could be due to chance. It is useful for constructing a confidence interval around estimated benefits. It is not useful for determining clinical significance.
I think an example should be sufficient to prove this. Assume there are two huge controlled trials of drug A and Drug B. Each has a subsample of patients given a placebo. These people show improved depression with an average improvement of HRSD of 5 (in both trials). The standard deviation of changes in HRSD is 5 in both Placebo subsamples. Change over standard deviation is 5/5.
Patients in the trial of drug A who received drug A have an average improvement of 8 with a standard deviation of 5. change/standard deviation is 8/5 which is 0.6 higher than 5/5 so it is concluded that drug A provides a clinically significant benefit.
For each patient who receives drug A there is a patient who receives drug B (convenient coincidence). For 90 % of patients who receive drug A there is a patient who received drug B and who had exactly the same change in HRSD. For 10% of patients who got drug A (and who had the same average benefit as the 90%) there is a patient who received drug B whose improvement was greater by 20 HRSD points.
The mean change with drug B is 8 + 0.1*20 = 10.
Given the assumption in parentheses, the variance of changes for patients who received drug B is 25 + (400)(0.1)(0.9) = 61 so the standard deviation is 7.81
change/standard deviation is 1.28. 1.28-1 = 0.28 < 0.5 so NICE declares that drug B does Not have a clinically significant benefit.
But wait a minute, experience with drug B shows first order stochastic dominance over improvement with drug A. The problem with drug B is that a few patients had a wonderful experience. This added proportionally more to the standard deviation than to the mean (has to do with square root of 0.1 is much bigger than 0.1).
Using the mean divided by the sub sample standard deviation to assess clinical significance is utterly idiotic.
Back to Kirsch et al. A study which claimed that antidepressants provide clinically significant benefits only for the severely depressed has received a lot of attention
33 google news hits and lots of plain google hits.
Much commentary completely missed the "clinically significant" and, in fact claimed that Kirsh et al had shown that patients who received a placebo did "just as well" as patients who received antidepressants. This was a null hypothesis overwhelmingly rejected by Kirsch et al.
The huge amount of attention received by the paper is, I think, entirely due to the appeal to the concept of "clinical significance" as authoritatively defined by the National Institute for Health and Clinical Excellence (NICE) in the UK. Hmm where have I read that acronym before ?
NICE declared that to be clinically significant a benefit had to be 3 points on the Hamilton Rating Scale of Depression or at least one half of one standard deviation of the changes in the HRSD in the treated subsample.
The first definition is arbitrary and, I think, nonsensical (if the only change were from "I think life is not worth living" to "I think life is worth living" that would be one point on the HRSD). However the second definition is much much more absurd.
The standard deviation of changes is very important for testing whether an apparent benefit could be due to chance. It is useful for constructing a confidence interval around estimated benefits. It is not useful for determining clinical significance.
I think an example should be sufficient to prove this. Assume there are two huge controlled trials of drug A and Drug B. Each has a subsample of patients given a placebo. These people show improved depression with an average improvement of HRSD of 5 (in both trials). The standard deviation of changes in HRSD is 5 in both Placebo subsamples. Change over standard deviation is 5/5.
Patients in the trial of drug A who received drug A have an average improvement of 8 with a standard deviation of 5. change/standard deviation is 8/5 which is 0.6 higher than 5/5 so it is concluded that drug A provides a clinically significant benefit.
For each patient who receives drug A there is a patient who receives drug B (convenient coincidence). For 90 % of patients who receive drug A there is a patient who received drug B and who had exactly the same change in HRSD. For 10% of patients who got drug A (and who had the same average benefit as the 90%) there is a patient who received drug B whose improvement was greater by 20 HRSD points.
The mean change with drug B is 8 + 0.1*20 = 10.
Given the assumption in parentheses, the variance of changes for patients who received drug B is 25 + (400)(0.1)(0.9) = 61 so the standard deviation is 7.81
change/standard deviation is 1.28. 1.28-1 = 0.28 < 0.5 so NICE declares that drug B does Not have a clinically significant benefit.
But wait a minute, experience with drug B shows first order stochastic dominance over improvement with drug A. The problem with drug B is that a few patients had a wonderful experience. This added proportionally more to the standard deviation than to the mean (has to do with square root of 0.1 is much bigger than 0.1).
Using the mean divided by the sub sample standard deviation to assess clinical significance is utterly idiotic.
Sunday, March 30, 2008
Game On in Zimbabwe
It is clear that Morgan Tsvangirai has won the Zimbabwean Presidential election and almost equally clear that Robert Mugabe will try to steal it.
The key piece of evidence for both propositions is the extraordinary delay in the reporting of official results.
Also
Thus it is too late to steal the election in a way which will not be obvious to Zimbabweans and foreign observers.
The evidence that Mugabe will attempt to steal the election is not subtle either
I think we will know soon how determined the Zimbabwean people are. If they are as determined as they seem to be, it will depend on the army. Sad to say revolutions generally do.
Since I saw no hope for Zimbabwe as of yesterday, this post is wildly optimistic.
It is clear that Morgan Tsvangirai has won the Zimbabwean Presidential election and almost equally clear that Robert Mugabe will try to steal it.
The key piece of evidence for both propositions is the extraordinary delay in the reporting of official results.
Also
unofficial vote totals were posted at most polling stations soon after counts there were completed, and were spread through the rumor mill, passed along on cellphones from one part of the country to another. Independent election observers and journalists who saw some of the results confirmed what many ordinary people were reporting: that the opposition appeared to have won by wide margins at the polling sites they saw, including some places considered traditional strongholds of the government.
Thus it is too late to steal the election in a way which will not be obvious to Zimbabweans and foreign observers.
The evidence that Mugabe will attempt to steal the election is not subtle either
No results were officially announced by evening, 24 hours after the end of voting. Independent election monitors said riot police and other security forces were deployed in densely populated suburbs of the capital, Harare, a traditional base of opposition to President Robert Mugabe, 84. [snip]
In past elections, officials have released results as they were tallied by district, often within hours of the close of voting.
I think we will know soon how determined the Zimbabwean people are. If they are as determined as they seem to be, it will depend on the army. Sad to say revolutions generally do.
Since I saw no hope for Zimbabwe as of yesterday, this post is wildly optimistic.
I support S2743
having been convinced by Harold Pollack
Please Praise Senators Casey and Hatch who introduced the bill
By Mr. CASEY (for himself and Mr. HATCH):
S. 2743. A bill to amend the Internal Revenue Code of 1986 to provide for the establishment of financial security accounts for the care of family members with disabilities, and for other purposes; to the Committee on Finance.
Casey at the bat for the disabled.
Hatch. You have got to be kidding me, I'm praising Senator Orrin Hatch ?!?!
having been convinced by Harold Pollack
SEC. 3. TREATMENT OF FINANCIAL SECURITY ACCOUNTS FOR INDIVIDUALS WITH DISABILITIES UNDER CERTAIN FEDERAL PROGRAMS. (a) Treatment as a Medicaid Excepted Trust- Paragraph (4) of section 1917(d) of the Social Security Act (42 U.S.C. 1396p(d)(4)) is amended by adding at the end the following new subparagraph: `(D) A trust which is a financial security account for an individual with a disability described in section 530A(b)(1) of the Internal Revenue Code of 1986.'. (b) Account Funds Disregarded for Purposes of Certain Other Means-Tested Federal Programs- Notwithstanding any other provision of Federal law that requires consideration of 1 or more financial circumstances of an individual, for the purpose of determining eligibility to receive, or the amount of, any assistance or benefit authorized by such provision to be provided to or for the benefit of such individual, any amount (including earnings thereon) in any financial security account for an individual with a disability of such individual, and any distribution for qualified disability expenses (as defined in section 530A(b)(2)) shall be disregarded for such purpose with respect to any period during which such individual maintains, makes contributions to, or receives distributions from such financial security account.
In plain English: Families can set aside funds to help a disabled loved one without jeopardizing her access to services she needs, reducing the need for pointless, sometimes dishonest and dangerous shenanigans families are now prone to do.
Please Praise Senators Casey and Hatch who introduced the bill
By Mr. CASEY (for himself and Mr. HATCH):
S. 2743. A bill to amend the Internal Revenue Code of 1986 to provide for the establishment of financial security accounts for the care of family members with disabilities, and for other purposes; to the Committee on Finance.
Casey at the bat for the disabled.
Hatch. You have got to be kidding me, I'm praising Senator Orrin Hatch ?!?!
Pruning Sheers are OK but the Chainsaw was excessive
Recent News, the Bush administration wants to eliminate the Office of Thrift Supervision
Not so recent news. Said office went over the top in stupid press conference stunts.
Recent News, the Bush administration wants to eliminate the Office of Thrift Supervision
The administration proposal would also consolidate the current scheme of bank regulation by shutting down the Office of Thrift Supervision and transferring its functions to the Office of the Comptroller of the Currency, which regulates nationally chartered banks.
Not so recent news. Said office went over the top in stupid press conference stunts.
Consider the press conference held on June 3, 2003 — just about the time subprime lending was starting to go wild — to announce a new initiative aimed at reducing the regulatory burden on banks. Representatives of four of the five government agencies responsible for financial supervision used tree shears to attack a stack of paper representing bank regulations. The fifth representative, James Gilleran of the Office of Thrift Supervision, wielded a chainsaw.
Saturday, March 29, 2008
Nick Barrowman has a clear summary of the work of Pyjamas in Bananas and me.
PJ writes well himself, but I sure don't.
PJ writes well himself, but I sure don't.
Matt Taibbi channels Wittgenstein
Exactly how Wittgenstein explained how language works. Also justice "I can't define it but I know it when I see it" White.
ahhh now I get it, a Forz'idiota (sad to say over here they are, at the moment, in the majority and without any terror at all).
Ooops fell off the bicycle. I knew that they were creating the New Man, but I mean, you know, biology. I doubt many managed true Sovoknitzi before the age of 30 (it's hard to be indifferent to sex under 30).
That's just the way we are, and maybe it's time to wonder why that is. In Russia they have a word, sovok, which described the craven, chickenshit mindset that over the course of decades became hard-wired into the increasingly silly brains of Soviet subjects. It's a hard word to define, but once you get it — and all Russians get it — it's like riding a bicycle, you've got it.
Exactly how Wittgenstein explained how language works. Also justice "I can't define it but I know it when I see it" White.
Sovok is the word that described a society where for decades silence and a thoughtful demeanor might be construed as evidence of a dangerous dissidence lurking underneath; the sovok therefore protected himself from suspicion by babbling meaningless nonsense at all times, so that no one would accuse him of harboring smart ideas. A sovok talked tough, and cheered Khruschev for banging a shoe at America, but at the same time a sovok would have sold his own children for a pair of American jeans.
ahhh now I get it, a Forz'idiota (sad to say over here they are, at the moment, in the majority and without any terror at all).
The sovok talked like a romantic and lavished women with compliments, but preferred long fishing trips and nights spend in the garage tinkering with his shitty car to actual sex.
Ooops fell off the bicycle. I knew that they were creating the New Man, but I mean, you know, biology. I doubt many managed true Sovoknitzi before the age of 30 (it's hard to be indifferent to sex under 30).
It's hard to explain, but over there, they know what the word means. More than anything, sovok described a society that spent seventy years in mortal terror of new ideas, and tended to drape itself in a paper-thin patriotism whenever it felt threatened, and worshipped mediocrities as a matter of course, elevating to positions of responsibility only those who showed an utter absence not only of objectionable qualities, but any qualities at all.
Kuttner and Krugman on Obama
Paul Krugman had two problems with Barack Obama's speech on the economy
Also he
Robert Kuttner was very favorably impressed
Kuttner stresses that he has agreed with Krugman's criticisms of Obama in the past.
I agree with both of them on mandates and strongly disagree on Social Security. Obama's social security proposal was to raise the payroll tax ceiling and thus to increase taxes paid by individuals with labor income over $102,000. These are not poor people. A fellow populist claims that they are about the top 6% of earners. Why is Kuttner opposed to raising their taxes ?
In his column Krugman makes an excellent point "The moral is that it’s important to take a hard look at what candidates say about policy." Krugman means listen what they propose to do, not their tone, their rhetoric or their manner. In the case of the proposal to raise the FICA ceiling, he, like Kuttner absolutely refuses to follow this advice. Neither discusses whether the proposed reform would make America better or worse. Both stop when Obama says that there is a problem which should be addressed -- the actuarial balance of the Social Security Administration. They argue, convincingly, that this is a small problem compared to all other sectors of the US federal budget. So ? If they have no problem with the policy proposal, why do they object to the way it is presented. Note Krugman's excellent point.
Now how about the middle class tax cuts which Krugman doesn't like.
Provide Middle Class Americans Tax Relief
Obama will cut income taxes by $1,000 for working families to offset the payroll tax they pay.
Now it's pretty clear how Obama plans to pay for this. He plans to raise the FICA ceiling. His plan clearly is to make FICA less regressive by cutting it by a fixed dollar amount and raising the ceiling. He doesn't want to be accused of being a class warrior (which he clearly is) so he presents the increase as shoring up social security and the cut as a cut from the general budget. The shift from the general fund to the SSA trust fund is only as good as the lock on the lockbox, which can be picked by any idiot, even a Republican. However, it does no harm.
Krugman is convinced that Obama is not solidly progressive because Obama wants to finance a (roughly) lump sum transfer to workers by raising payroll taxes on the richest 6% of wage earners.
Paul Krugman had two problems with Barack Obama's speech on the economy
his proposals for aid to the victims of the current crisis, though significant, are less sweeping than Mrs. Clinton’s
[snip]
Mr. Obama also continues to make permanent tax cuts — middle-class tax cuts, to be sure — a centerpiece of his economic plan. It’s not clear how he would pay both for these tax cuts and for initiatives like health care reform, so his tax-cut promises raise questions about how determined he really is to pursue a strongly progressive agenda.
Also he
Robert Kuttner was very favorably impressed
Barack Obama's speech on the financial crisis was a remarkable breakthrough.
Kuttner stresses that he has agreed with Krugman's criticisms of Obama in the past.
I actually shared Krugman's critique of Obama's health insurance individual mandate and his proposal to tax the upper middle class to pay for a much exaggerated Social Security shortfall that is more like a rounding error.
I agree with both of them on mandates and strongly disagree on Social Security. Obama's social security proposal was to raise the payroll tax ceiling and thus to increase taxes paid by individuals with labor income over $102,000. These are not poor people. A fellow populist claims that they are about the top 6% of earners. Why is Kuttner opposed to raising their taxes ?
In his column Krugman makes an excellent point "The moral is that it’s important to take a hard look at what candidates say about policy." Krugman means listen what they propose to do, not their tone, their rhetoric or their manner. In the case of the proposal to raise the FICA ceiling, he, like Kuttner absolutely refuses to follow this advice. Neither discusses whether the proposed reform would make America better or worse. Both stop when Obama says that there is a problem which should be addressed -- the actuarial balance of the Social Security Administration. They argue, convincingly, that this is a small problem compared to all other sectors of the US federal budget. So ? If they have no problem with the policy proposal, why do they object to the way it is presented. Note Krugman's excellent point.
Now how about the middle class tax cuts which Krugman doesn't like.
Provide Middle Class Americans Tax Relief
Obama will cut income taxes by $1,000 for working families to offset the payroll tax they pay.
* Provide a Tax Cut for Working Families: Obama will restore fairness to the tax code and provide 150 million workers the tax relief they need. Obama will create a new "Making Work Pay" tax credit of up to $500 per person, or $1,000 per working family. The "Making Work Pay" tax credit will completely eliminate income taxes for 10 million Americans.
Now it's pretty clear how Obama plans to pay for this. He plans to raise the FICA ceiling. His plan clearly is to make FICA less regressive by cutting it by a fixed dollar amount and raising the ceiling. He doesn't want to be accused of being a class warrior (which he clearly is) so he presents the increase as shoring up social security and the cut as a cut from the general budget. The shift from the general fund to the SSA trust fund is only as good as the lock on the lockbox, which can be picked by any idiot, even a Republican. However, it does no harm.
Krugman is convinced that Obama is not solidly progressive because Obama wants to finance a (roughly) lump sum transfer to workers by raising payroll taxes on the richest 6% of wage earners.
Friday, March 28, 2008
Fearfully No !
Calling Them Rude Names Will Not Help
Just because it is in a colon is no reason to call a robot "shit".
A little diplomacy is in order while we might still find a robot willing to say
Calling Them Rude Names Will Not Help
It’s like we’re begging the robots to kill us
Posted at 3:03 by Brad
This shit just has to stop:
As if the idea of colonoscopies didn’t sound uncomfortable enough, now researchers are developing self-propelling probes that crawl inside the colon and grip its sides with the aid of sticky films.
Still, these slithery devices could lead to better, safer, more comfortable colonoscopies to help uncover cancerous polyps.
Just because it is in a colon is no reason to call a robot "shit".
A little diplomacy is in order while we might still find a robot willing to say
Can’t we just talk to the humans?
A little understanding could make things better.
Can’t we talk to the humans and work together now?
Iraq Update
U.S. Armor Forces Join Offensive In Baghdad Against Sadr Militia
Things are not going well for Prime Minister Maliki.
By Sudarsan Raghavan and Sholnn Freeman
Washington Post Foreign Service
Friday, March 28, 2008; Page A01
By JAMES GLANZ and STEVEN LEE MYERS
Published: March 28, 2008
Journalists are practicing actual journalism. In addition to "a Washington Post correspondent" an "Iraqi employee of The New York Times, driving on the main road between Basra and Nasiriya, observed numerous civilian cars with coffins strapped to the roofs, apparently heading to Shiite cemeteries to the north."
They are also directly saying that Bush is full of it. From the Post
Ouch.
To me the interesting news is that un-named administration officials are very eager to argue that this was all al Maliki's idea
again the Post
I wonder if they are thinking of dumping al Maliki. It would be a way to restore the cease fire, and it's not like he has such great accomplishments so far. Recall he became prime minister because of the support of the Sadrists who opposed Adel Abdul-Mahdi of then SCIRI (now ISCI) who they considered to be too pro-Iranian.
My personal guess is that Bush is too mentally rigid to make such a shift. Also al Maliki is not likely to resign quietly and a vote of no confidence could get out of hand (wouldn't want to have to send Strykers into the parliament building would we).
U.S. Armor Forces Join Offensive In Baghdad Against Sadr Militia
Things are not going well for Prime Minister Maliki.
By Sudarsan Raghavan and Sholnn Freeman
Washington Post Foreign Service
Friday, March 28, 2008; Page A01
BAGHDAD, March 27 -- U.S. forces in armored vehicles battled Mahdi Army fighters Thursday in Sadr City, the vast Shiite stronghold in eastern Baghdad, as an offensive to quell party-backed militias entered its third day. Iraqi army and police units appeared to be largely holding to the outskirts of the area as American troops took the lead in the fighting.
By JAMES GLANZ and STEVEN LEE MYERS
Published: March 28, 2008
BAGHDAD — American-trained Iraqi security forces failed for a third straight day to oust Shiite militias from the southern city of Basra on Thursday, even as President Bush hailed the operation as a sign of the growing strength of Iraq’s federal government.
Four U.S. Stryker armored vehicles were seen in Sadr City by a Washington Post correspondent, one of them engaging Mahdi Army militiamen with heavy fire. The din of American weapons, along with the Mahdi Army's AK-47s and rocket-propelled grenades, was heard through much of the day. U.S. helicopters and drones buzzed overhead.
Journalists are practicing actual journalism. In addition to "a Washington Post correspondent" an "Iraqi employee of The New York Times, driving on the main road between Basra and Nasiriya, observed numerous civilian cars with coffins strapped to the roofs, apparently heading to Shiite cemeteries to the north."
They are also directly saying that Bush is full of it. From the Post
As President Bush told an Ohio audience that Iraq was returning to "normalcy," administration officials in Washington held meetings to assess what appeared to be a rapidly deteriorating security situation in many parts of the country.
Ouch.
To me the interesting news is that un-named administration officials are very eager to argue that this was all al Maliki's idea
again the Post
Maliki decided to launch the offensive without consulting his U.S. allies, according to administration officials. With little U.S. presence in the south, and British forces in Basra confined to an air base outside the city, one administration official said that "we can't quite decipher" what is going on. It's a question, he said, of "who's got the best conspiracy" theory about why Maliki decided to act now.
In Basra, three rival Shiite groups have been trying to position themselves, sometimes through force of arms, to dominate recently approved provincial elections.
The U.S. officials, who were not authorized to speak on the record, said that they believe Iran has provided assistance in the past to all three groups -- the Mahdi Army; the Badr Organization of the Islamic Supreme Council of Iraq, Iraq's largest Shiite party; and forces loyal to the Fadhila Party, which holds the Basra governor's seat. But the officials see the current conflict as a purely internal Iraqi dispute.
Some officials have concluded that Maliki himself is firing "the first salvo in upcoming elections," the administration official said.
"His dog in that fight is that he is basically allied with the Badr Corps" against forces loyal to Sadr, the official said. "It's not a pretty picture."
I wonder if they are thinking of dumping al Maliki. It would be a way to restore the cease fire, and it's not like he has such great accomplishments so far. Recall he became prime minister because of the support of the Sadrists who opposed Adel Abdul-Mahdi of then SCIRI (now ISCI) who they considered to be too pro-Iranian.
My personal guess is that Bush is too mentally rigid to make such a shift. Also al Maliki is not likely to resign quietly and a vote of no confidence could get out of hand (wouldn't want to have to send Strykers into the parliament building would we).
Wednesday, March 26, 2008
Waldmann Smackdown Watch
Someone who is not as enthusiastic about Barack Obama as I am.
We had an e-mail debate about Obama's willingness to attempt to kill Osama in Pakistan
with or without Pakistani permission.
He wrote to me (among other things) "As he [Obama] put it, he would not hesitate to bomb people.
It's always wise to think it over before you start WW III.
I thought Huh wah ?!?!?!? and replied (quoted in full)
don't recall Obama every saying he wouldn't hesitate befor bombing.
If I am aware of the statement by Obama which you are paraphrasing, it
has since been done, and didn't start WWIII. I bet you didn't even
notice. He said he would send missiles after Osama even without
Pakistani permission. Since then, Bush has done so twice against
lower al Qaeda figures. As far as I know, Pakistan didn't even
protest. I would like a quote supporting your accusation that "As he
put it, he would not hesitate to bomb people." I expect to find the
word "hesitate" in it.
Which is clearly throwing down the googling gauntlet.
He came back with a quote from the Washington Post
This week saw the unusual spectacle of a foreign government criticizing a U.S. presidential candidate. The government is Pakistan and the candidate is Sen. Barack Obama -- and while such criticism is rare, Obama's remarks were pretty strange, too.
Clicking on "remarks" took me to proof that Obama did, indeed, say "I will not hesitate to take out terrorists who pose a direct threat to America"
He wins. I lose.
Now I happen to agree with Obama about not hesitating if there is actionable intelligence about the location of Osama bin Laden, but I sure wish I had hesitated before starting a googling war.
Someone who is not as enthusiastic about Barack Obama as I am.
We had an e-mail debate about Obama's willingness to attempt to kill Osama in Pakistan
with or without Pakistani permission.
He wrote to me (among other things) "As he [Obama] put it, he would not hesitate to bomb people.
It's always wise to think it over before you start WW III.
I thought Huh wah ?!?!?!? and replied (quoted in full)
don't recall Obama every saying he wouldn't hesitate befor bombing.
If I am aware of the statement by Obama which you are paraphrasing, it
has since been done, and didn't start WWIII. I bet you didn't even
notice. He said he would send missiles after Osama even without
Pakistani permission. Since then, Bush has done so twice against
lower al Qaeda figures. As far as I know, Pakistan didn't even
protest. I would like a quote supporting your accusation that "As he
put it, he would not hesitate to bomb people." I expect to find the
word "hesitate" in it.
Which is clearly throwing down the googling gauntlet.
He came back with a quote from the Washington Post
This week saw the unusual spectacle of a foreign government criticizing a U.S. presidential candidate. The government is Pakistan and the candidate is Sen. Barack Obama -- and while such criticism is rare, Obama's remarks were pretty strange, too.
Clicking on "remarks" took me to proof that Obama did, indeed, say "I will not hesitate to take out terrorists who pose a direct threat to America"
He wins. I lose.
Now I happen to agree with Obama about not hesitating if there is actionable intelligence about the location of Osama bin Laden, but I sure wish I had hesitated before starting a googling war.
This is Good News
Gallup reports that 28% of Democrats who support Clinton claim that they would vote for McCain if Obama were the nominee.
good news ?!?!?! yes, because we know about overall support for McCain vs Obama (they are roughly tied). If Gallup is to be believed, the support for McCain in national polls currently includes 28% of Democrats who support Clinton. In the actual election, it is extremely unlikely that so many of them will actually vote for McCain.
Thus the news from the Gallup poll is not that Obama has a problem (we know he has a problem he isn't ahead of McCain in the polls) but that, it appears, that a significant part of his problem is something that is likely to get worse before it gets better, but get better before election day.
McCain is not likely to get 28% of the votes of Democrats who support Clinton, therefore he is not likely to get as many votes as suggested by current national polls.
via TPM.
Gallup reports that 28% of Democrats who support Clinton claim that they would vote for McCain if Obama were the nominee.
A sizable proportion of Democrats would vote for John McCain next November if he is matched against the candidate they do not support for the Democratic nomination. This is particularly true for Hillary Clinton supporters, more than a quarter of whom currently say they would vote for McCain if Barack Obama is the Democratic nominee.
good news ?!?!?! yes, because we know about overall support for McCain vs Obama (they are roughly tied). If Gallup is to be believed, the support for McCain in national polls currently includes 28% of Democrats who support Clinton. In the actual election, it is extremely unlikely that so many of them will actually vote for McCain.
Thus the news from the Gallup poll is not that Obama has a problem (we know he has a problem he isn't ahead of McCain in the polls) but that, it appears, that a significant part of his problem is something that is likely to get worse before it gets better, but get better before election day.
McCain is not likely to get 28% of the votes of Democrats who support Clinton, therefore he is not likely to get as many votes as suggested by current national polls.
via TPM.
HIV Vaccine
Below I argue that the failure of recent HIV vaccine trials suggest that it is necessary to put even more effort into finding vaccines which will cause us to make antibodies which block HIV infection of cells, that is neutralizing antibodies. I now know more about just how frustrating this effort has been.
I am relying on a fairly recent review from 2004
My version of the story. Long long ago a monoclonal, called B12, which blocks HIV infection was discovered. It binds to the part of the HIV surface protein gp-120 which binds to CD-4. Unfortunately, other antibodies to this protein only block HIV which has been grown in standard cell line not HIV isolated from patients. The problem appears to be that the glycosylation (sugar coating) of gp-120 is different if the HIV grew in different types of cells.
Other neutralizing antibodies (X5 17b) bind to another part of gp-120 which seems to be required for HIV binding to another protein on the target cells surface either CCR5 or CXCR4. They aren't as famous as CD-4 but HIV has to bind to them. The problem is that this part of the gp-120 is only exposed after gp-120 sticks to CD-4.
There is a very strange mutant neutralizing antibody which would not normally be made in response to a vaccine so it doesn't count.
Finally there are some antibodies which bind to parts of HIV which are not conserved and so neutralize some strains of HIV but not others.
Finally the known neutralizing antibodies which block the most different strains of HIV 4F5 and 2E10 bind to a different HIV protein gp-41. They don't block binding to target cells but seem to block a later stage of infection.
For all of the broadly effective neutralizing antibodies other than B12 bind to parts of HIV which are only briefly exposed during the infection of cells. Thus they aren't made normally in HIV infected people, although they could be made in response to a vaccine in which these parts (epitopes) are exposed.
I think that it is not known why production of antibodies like B12 is so rare. One important issue is that after a B-Cell is activated, the antibody gene mutates. It is possible that there is no gene for any antibody which works like B12 in resting B-Cells and that several different mutations were required to produce B12. It is also possible that the epitope to which B12 binds is normally destroyed during antigen processing and they lucky patient had a lucky antigen presenting cell which didn't do so.
There are some avenues for research which were not discussed in the brief review and which I would like to discuss.
First, notice I was guessing about the closest analogue to B12 in resting B-cells. I think it would be interesting to clone the genes analogous to B12 from resting B-cells. For one thing, a vaccine would have to bind to their IG-d product in order to stimulate mutation to a B12 like antibody.
I think that it would be possible to come up with other candidate neutralizing monoclonal antibodies by screening for antibodies which compete with the currently known neutralizing antibodies for binding sites. Also to make modified proteins which are partly like gp-120 and gp41, screen for those which bind the known neutralizing antibodies and then select antibodies which bind to those proteins and not the others (I said select which means I am still into phage display).
That is, the next step isn't interesting until there are a fair bunch of known useful neutralizing antibodies.
OK then the problem becomes not we want to design a vaccine so that it will induce antibodies which neutralize HIV but rather we want to design a vaccine which will induce the following n antibodies. This is, it seems to me, definitely doable.
Basically to induce an antibody first an antigen has to stick to it. It is very easy to get proteins which stick to something - make antibodies. An antibody which specifically sticks to another antibody is called an anti-idiotype antibody. An anti-idiotype antibody would not work at all as a vaccine, but I'm not so sure about an anti-idiotype Fab.
Furthermore, the other aspects a protein must have to be an antigen are comprehensible. It must also bind to the receptor of a helper t-cell and it must be processed and presented by an antigen presenting cell. I think that fiddling with a protein to make it antigenic should be feasible. I think it would be done often but it hasn't been necessary.
Below I argue that the failure of recent HIV vaccine trials suggest that it is necessary to put even more effort into finding vaccines which will cause us to make antibodies which block HIV infection of cells, that is neutralizing antibodies. I now know more about just how frustrating this effort has been.
I am relying on a fairly recent review from 2004
My version of the story. Long long ago a monoclonal, called B12, which blocks HIV infection was discovered. It binds to the part of the HIV surface protein gp-120 which binds to CD-4. Unfortunately, other antibodies to this protein only block HIV which has been grown in standard cell line not HIV isolated from patients. The problem appears to be that the glycosylation (sugar coating) of gp-120 is different if the HIV grew in different types of cells.
Other neutralizing antibodies (X5 17b) bind to another part of gp-120 which seems to be required for HIV binding to another protein on the target cells surface either CCR5 or CXCR4. They aren't as famous as CD-4 but HIV has to bind to them. The problem is that this part of the gp-120 is only exposed after gp-120 sticks to CD-4.
There is a very strange mutant neutralizing antibody which would not normally be made in response to a vaccine so it doesn't count.
Finally there are some antibodies which bind to parts of HIV which are not conserved and so neutralize some strains of HIV but not others.
Finally the known neutralizing antibodies which block the most different strains of HIV 4F5 and 2E10 bind to a different HIV protein gp-41. They don't block binding to target cells but seem to block a later stage of infection.
For all of the broadly effective neutralizing antibodies other than B12 bind to parts of HIV which are only briefly exposed during the infection of cells. Thus they aren't made normally in HIV infected people, although they could be made in response to a vaccine in which these parts (epitopes) are exposed.
I think that it is not known why production of antibodies like B12 is so rare. One important issue is that after a B-Cell is activated, the antibody gene mutates. It is possible that there is no gene for any antibody which works like B12 in resting B-Cells and that several different mutations were required to produce B12. It is also possible that the epitope to which B12 binds is normally destroyed during antigen processing and they lucky patient had a lucky antigen presenting cell which didn't do so.
There are some avenues for research which were not discussed in the brief review and which I would like to discuss.
First, notice I was guessing about the closest analogue to B12 in resting B-cells. I think it would be interesting to clone the genes analogous to B12 from resting B-cells. For one thing, a vaccine would have to bind to their IG-d product in order to stimulate mutation to a B12 like antibody.
I think that it would be possible to come up with other candidate neutralizing monoclonal antibodies by screening for antibodies which compete with the currently known neutralizing antibodies for binding sites. Also to make modified proteins which are partly like gp-120 and gp41, screen for those which bind the known neutralizing antibodies and then select antibodies which bind to those proteins and not the others (I said select which means I am still into phage display).
That is, the next step isn't interesting until there are a fair bunch of known useful neutralizing antibodies.
OK then the problem becomes not we want to design a vaccine so that it will induce antibodies which neutralize HIV but rather we want to design a vaccine which will induce the following n antibodies. This is, it seems to me, definitely doable.
Basically to induce an antibody first an antigen has to stick to it. It is very easy to get proteins which stick to something - make antibodies. An antibody which specifically sticks to another antibody is called an anti-idiotype antibody. An anti-idiotype antibody would not work at all as a vaccine, but I'm not so sure about an anti-idiotype Fab.
Furthermore, the other aspects a protein must have to be an antigen are comprehensible. It must also bind to the receptor of a helper t-cell and it must be processed and presented by an antigen presenting cell. I think that fiddling with a protein to make it antigenic should be feasible. I think it would be done often but it hasn't been necessary.
Battle of Basra
As far as I can tell, the al-Malaki government and the Islamic Supreme Council of Iraq (the supreme council formerly known as SCIRI) which are the clients of both the USA and Iran are trying (again) to take control of the port from the anti Iranian British client Fadhila.
Michael Kamber and James Glanz write in The New York Times
Fadilla is definitely outspokenly anti-Iranian. In The Guardian, Ghaith Abdul-Ahad wrote
"Fadhila, widely seen as backed by the British, split from the main Shia alliance in Baghdad after accusing it of having a sectarian agenda."
The risk is that the Irano-American assault on UK clients might trigger armed conflict between the Iran and US supported government and ISCI and the Iran liking US hating Mahdi army. So far, Moqtada al Sadr is calling for non violent resistance.
Hmm so is the US involved in a proxy war allied with Iran and against the UK ?
I'll ask Juan Cole.
He just links to Reuters where Aref Mohammed presents the fighting as mostly Iraqi government vs Mahdi army.
Update: McClatchy too describes the fighting as the philo-Iranian supported Iraqi central government against the Philo-Iranian Mahdi army. Leila Fadel and Ali al Basri at McClatchy have plenty of quotes from prominent figures on both sides.
They suggest that the battle is an electoral campaigning conducted by other means "The operation was portrayed by Sadrists and some analysts as an effort by Maliki to consolidate power in the south for the Islamic Supreme Council of Iraq before the election in October."
As far as I can tell, the al-Malaki government and the Islamic Supreme Council of Iraq (the supreme council formerly known as SCIRI) which are the clients of both the USA and Iran are trying (again) to take control of the port from the anti Iranian British client Fadhila.
Michael Kamber and James Glanz write in The New York Times
In the weeks leading up to the operation, Iraqi officials indicated that part of the operation would be aimed at the Fadhila groups, who are widely believed to be in control of Basra’s lucrative port operations and other parts of the city. The ports have been plagued by corruption, draining revenue that could flow to the central and local governments. But the operation also threatens the Mahdi Army’s strongholds in Basra.
Fadilla is definitely outspokenly anti-Iranian. In The Guardian, Ghaith Abdul-Ahad wrote
"Fadhila, widely seen as backed by the British, split from the main Shia alliance in Baghdad after accusing it of having a sectarian agenda."
The risk is that the Irano-American assault on UK clients might trigger armed conflict between the Iran and US supported government and ISCI and the Iran liking US hating Mahdi army. So far, Moqtada al Sadr is calling for non violent resistance.
Hmm so is the US involved in a proxy war allied with Iran and against the UK ?
I'll ask Juan Cole.
He just links to Reuters where Aref Mohammed presents the fighting as mostly Iraqi government vs Mahdi army.
Update: McClatchy too describes the fighting as the philo-Iranian supported Iraqi central government against the Philo-Iranian Mahdi army. Leila Fadel and Ali al Basri at McClatchy have plenty of quotes from prominent figures on both sides.
They suggest that the battle is an electoral campaigning conducted by other means "The operation was portrayed by Sadrists and some analysts as an effort by Maliki to consolidate power in the south for the Islamic Supreme Council of Iraq before the election in October."
Tuesday, March 25, 2008
HIV Vaccine
Experimental HIV vaccines have failed. Oddly, when I looked for a recent Washington Post story, the Washington Post sent me to this old ABC news story
The evidence suggests that, if anything*, the vaccine increases the risk of infection and speed progression of the disease. The result is horrible, but it is not astonishing (OK I was shocked when I learned about it back then but I am an irrational optimist). It is known that stimulating the immune system speeds the progression from HIV infection to AIDS. I think this is understood at the molecular level by now. So it seems, on balance, for infection.
Also, before the vaccine trials, it was widely noted that successful vaccines cause the production of antibodies which block virus binding to host cells. Of course, it is well known that anti-HIV antibodies don't do this (they are the first symptom of infection). It had been hoped that vaccines would work by stimulating Killer T-cells to kill the first virus infected cells. This would have been a first if it had worked.
So why don't we make antibodies which block HIV infection of cells ? One possibility is that there is no such antibody in our repertoire. That would mean we are out of luck. Another, however, is that we have B-cells which would make the antibody but don't have helper T-cells which would stimulate them. In that case, it might be possible to make a vaccine which will cause us to make an HIV blocking antibody by linking the key bit of the HIV surface protein (gp120) to another protein. A hapten. Such an approach is the basis of the H influenza vaccine which actually works (here it is H influenza polysaccharide linked to a protein).
Now it makes sense that helper T-cells which would help make exactly the right antibodies are missing. The reason is that we have proteins which bind to CD-4 just as HIV does. Therefore, it is possible that the helper T-Cells needed to make antibodies which bind to the HIV binding site are deleted, because they would help us make antibodies to our own proteins.
In fact, blocking antibodies exist in the human repertoire.
A fairly recent review is here.
Permanently promising perhaps, but not hopeless.
Previous googling sent me to the original now very old articles.
First Chimpanzees definitely do make antibodies which protect Monkeys from SIV (monkey HIV).
This article is 9 years old !
This seems much more promising and it is 13 years old !
B-12 a human antibody cloned from a long time HIV infected survivor blocks infection of human cells ! There must be something wrong.
They were still publishing about B-12 in 2003 and getting published in top journals.
HIV binding and infection has been understood in gruesome detail for over a decade.
OOh this is interesting. It is about how HIV evades blocking antibodies. The claim is that gp-120 is glycosylated (sugar stuck on it) when it reproduces in monocytes (not the helper t-cells which it kills but other cells in which it grows). If the virus made from helper T-Cells and monocytes are different, then an antibody which blocks one won't block the other. The authors don't say if the protein is glycoslyated when HIV grows in T-cells, but there must be some reason people were convinced that the antibody blocks infection.
I would say that
known and boring
In fact the evidence tends to suggest that the Merck experimental vaccine was worse than worthless
First what would the probability that as many as 49 or 82 infected men were vaccinated if the probability of infection were the same with and without the vaccine ?
The expected number would be 914/(914+922)= 40.821351
the variance 82*914*922/(914+922)^2 = 20.499611
the standard deviation (82*914*922)^0.5/(914+922)= 4.5276496
z score = (number - expected number)/standard deviation = 1.8063786
This should be distributed very very nearly like a unit normal under the null so the probability of a z score that high or higher is 0.035429612
Typically one considers a two tailed test, that is the probability that the absolute value of the z-score is that high that is 0.070859225.
I think that, during the interview with ABC, Dr. Gottesdiener was quite relieved that the standard confidence level for statistical significance is 5%.
To me that settles it.
Experimental HIV vaccines have failed. Oddly, when I looked for a recent Washington Post story, the Washington Post sent me to this old ABC news story
The evidence suggests that, if anything*, the vaccine increases the risk of infection and speed progression of the disease. The result is horrible, but it is not astonishing (OK I was shocked when I learned about it back then but I am an irrational optimist). It is known that stimulating the immune system speeds the progression from HIV infection to AIDS. I think this is understood at the molecular level by now. So it seems, on balance, for infection.
Also, before the vaccine trials, it was widely noted that successful vaccines cause the production of antibodies which block virus binding to host cells. Of course, it is well known that anti-HIV antibodies don't do this (they are the first symptom of infection). It had been hoped that vaccines would work by stimulating Killer T-cells to kill the first virus infected cells. This would have been a first if it had worked.
So why don't we make antibodies which block HIV infection of cells ? One possibility is that there is no such antibody in our repertoire. That would mean we are out of luck. Another, however, is that we have B-cells which would make the antibody but don't have helper T-cells which would stimulate them. In that case, it might be possible to make a vaccine which will cause us to make an HIV blocking antibody by linking the key bit of the HIV surface protein (gp120) to another protein. A hapten. Such an approach is the basis of the H influenza vaccine which actually works (here it is H influenza polysaccharide linked to a protein).
Now it makes sense that helper T-cells which would help make exactly the right antibodies are missing. The reason is that we have proteins which bind to CD-4 just as HIV does. Therefore, it is possible that the helper T-Cells needed to make antibodies which bind to the HIV binding site are deleted, because they would help us make antibodies to our own proteins.
In fact, blocking antibodies exist in the human repertoire.
A fairly recent review is here.
Permanently promising perhaps, but not hopeless.
Previous googling sent me to the original now very old articles.
First Chimpanzees definitely do make antibodies which protect Monkeys from SIV (monkey HIV).
This article is 9 years old !
This seems much more promising and it is 13 years old !
B-12 a human antibody cloned from a long time HIV infected survivor blocks infection of human cells ! There must be something wrong.
They were still publishing about B-12 in 2003 and getting published in top journals.
HIV binding and infection has been understood in gruesome detail for over a decade.
OOh this is interesting. It is about how HIV evades blocking antibodies. The claim is that gp-120 is glycosylated (sugar stuck on it) when it reproduces in monocytes (not the helper t-cells which it kills but other cells in which it grows). If the virus made from helper T-Cells and monocytes are different, then an antibody which blocks one won't block the other. The authors don't say if the protein is glycoslyated when HIV grows in T-cells, but there must be some reason people were convinced that the antibody blocks infection.
I would say that
known and boring
In fact the evidence tends to suggest that the Merck experimental vaccine was worse than worthless
One of the possibilities is that the increase in the number of infections was related to the vaccine," meaning it could have made people more susceptible to HIV infection, said Dr. Keith Gottesdiener, vice president of clinical research at Merck Research Laboratories. He couldn't say how likely that was but said other factors, even coincidence, could be the explanation.
[snip]
"New data released Wednesday showed that to date, 49 of 914 vaccinated men became infected with HIV, compared with 33 of the 922 men who got dummy shots.
First what would the probability that as many as 49 or 82 infected men were vaccinated if the probability of infection were the same with and without the vaccine ?
The expected number would be 914/(914+922)= 40.821351
the variance 82*914*922/(914+922)^2 = 20.499611
the standard deviation (82*914*922)^0.5/(914+922)= 4.5276496
z score = (number - expected number)/standard deviation = 1.8063786
This should be distributed very very nearly like a unit normal under the null so the probability of a z score that high or higher is 0.035429612
Typically one considers a two tailed test, that is the probability that the absolute value of the z-score is that high that is 0.070859225.
I think that, during the interview with ABC, Dr. Gottesdiener was quite relieved that the standard confidence level for statistical significance is 5%.
However, the researchers found that volunteers with pre-existing immunity to this particular cold virus were much more likely to get infected with HIV if they got the AIDS vaccine than if they got the dummy shot.
To me that settles it.
Monday, March 24, 2008
Paul Krugman Doesn't Need My Help
(except to find a photograph of an obscure press conference) but I think that Jonathan Taplin is being unfair here and is also unfamiliar with the evidence on banking regulation and banking crises.
First, Krugman didn't mention the Glass-Steagall act in his column. Taplin must argue (as he does) that obviously the Glass-Steagall act was the key regulation which stabilized the banking system, so Krugman should have discussed it explicitly. It is possible that Krugman is not a hypocrite but that he considers other regulations to have been the key.
I don't know about Krugman, but I certainly think neither that the Glass-Steagall act was the key to relative banking stability while it was in force a, nor that the repeal was the key to the current crisis.
My reason is simple. The Glass-Steagal act was unique to the USA (as far as I know) but banking stability was common to developed countries while it was in force. I think that Taplin has decided that the experience of the rest of the world is irrelevant.
Now if he can explain how the Glass-Steagal act could have prevented our current problems, I'm all ears. I just don't see it.
The most relevant current new event is that investment banks are borrowing from the discount window. Thus the problem is not, say, that commercial banks which were allowed to own common stock gambled and lost. The way in which the Glass-Steagal act would have been relevant is if the investment banks had taken deposits. They didn't. The crisis is based on commercial banks being irrelevant, not on their being allowed to do things they couldn't do before. Repealing the Glass-Steagal act could have helped commercial banks compete in return and in reckless risk bearing with bond financed mortgage lenders. This is not what happened.
(except to find a photograph of an obscure press conference) but I think that Jonathan Taplin is being unfair here and is also unfamiliar with the evidence on banking regulation and banking crises.
He tells a tale of how after the Great Depression, Democrats worked to protect the banking system from runs, by enacting a split of Investment Banks and Commercial Banks.
[snip]
But never once in the whole article does he point out who yielded to the enticements of Wall Street--who was responsible for destroying the Glass-Steagall separation of Banks and Investment Banks--Bill Clinton.
First, Krugman didn't mention the Glass-Steagall act in his column. Taplin must argue (as he does) that obviously the Glass-Steagall act was the key regulation which stabilized the banking system, so Krugman should have discussed it explicitly. It is possible that Krugman is not a hypocrite but that he considers other regulations to have been the key.
I don't know about Krugman, but I certainly think neither that the Glass-Steagall act was the key to relative banking stability while it was in force a, nor that the repeal was the key to the current crisis.
My reason is simple. The Glass-Steagal act was unique to the USA (as far as I know) but banking stability was common to developed countries while it was in force. I think that Taplin has decided that the experience of the rest of the world is irrelevant.
Now if he can explain how the Glass-Steagal act could have prevented our current problems, I'm all ears. I just don't see it.
The most relevant current new event is that investment banks are borrowing from the discount window. Thus the problem is not, say, that commercial banks which were allowed to own common stock gambled and lost. The way in which the Glass-Steagal act would have been relevant is if the investment banks had taken deposits. They didn't. The crisis is based on commercial banks being irrelevant, not on their being allowed to do things they couldn't do before. Repealing the Glass-Steagal act could have helped commercial banks compete in return and in reckless risk bearing with bond financed mortgage lenders. This is not what happened.
Now we Know what McCain and Clinton have in Common.
They've Crossed the Comander in Chief Threshold
Oh sorry I miss typed. I meant to write "blatantly lied".
Come on, at least think of something plausible. For example, I claimed that I miss typed not that I "misspoke".
They've Crossed the Comander in Chief Threshold
Oh sorry I miss typed. I meant to write "blatantly lied".
The Clinton campaign's now putting it out there that Hillary Clinton "misspoke" when describing her fake death-defying landing at Tuzla, but it certainly seems to have occurred in her prepared remarks.
Come on, at least think of something plausible. For example, I claimed that I miss typed not that I "misspoke".
Don't Yank Away the Football America
I don't remember the last time I was obsessed with an election (my mom says I was totally impossible about Kennedy Vs Nixon when I was in the womb). I feel scared.
Not because I dread the possibility of a McCain victory, although I do, I always dread Republican victories. I think it is, because I almost hope that something rather good might happen.
Brownsox at DailyKos is very enthusiastic. So why did he (or she) prominently advise me to place my hopes in
Charlie Brown
CA-04 John Doolittle*# (Charlie Brown opposing)
AAAAARGGGGG
I don't remember the last time I was obsessed with an election (my mom says I was totally impossible about Kennedy Vs Nixon when I was in the womb). I feel scared.
Not because I dread the possibility of a McCain victory, although I do, I always dread Republican victories. I think it is, because I almost hope that something rather good might happen.
Brownsox at DailyKos is very enthusiastic. So why did he (or she) prominently advise me to place my hopes in
Charlie Brown
CA-04 John Doolittle*# (Charlie Brown opposing)
AAAAARGGGGG
Sunday, March 23, 2008
Banking Regulation Illustrated
Paul Krugman
"Hiding Behind the Invisible Hand"*
The scene
from the FDIC bulletin (warning *.pdf and the smugness might make your head explode).
I know the image is low quality, but it took me a while to find it. For one thing, not only did the American Banker's Association eagerly lobby for the deregulation which seriously threatens American Banking, but their magazine -- community banker -- had an article on the great event which claimed to be illustrated and was missing the digital photograph. No wonder they are going bankrupt.
* I am so used to reading "invisible hand" as "the Market" that I didn't get the Zenlike character of the title. What is the smell of one invisible hand clapping ?
** He's quoting himself. The NYT style guide is a mystery to the unenlightened.
Paul Krugman
"Hiding Behind the Invisible Hand"*
Actually, there was plenty of coordination — a coordinated effort to destroy effective regulation:
Consider the press conference held on June 3, 2003 — just about the time subprime lending was starting to go wild — to announce a new initiative aimed at reducing the regulatory burden on banks. Representatives of four of the five government agencies responsible for financial supervision used tree shears to attack a stack of paper representing bank regulations. The fifth representative, James Gilleran of the Office of Thrift Supervision, wielded a chainsaw.**
The lack of oversight, in short, was no oversight: it was part of the plan.
The scene
from the FDIC bulletin (warning *.pdf and the smugness might make your head explode).
I know the image is low quality, but it took me a while to find it. For one thing, not only did the American Banker's Association eagerly lobby for the deregulation which seriously threatens American Banking, but their magazine -- community banker -- had an article on the great event which claimed to be illustrated and was missing the digital photograph. No wonder they are going bankrupt.
* I am so used to reading "invisible hand" as "the Market" that I didn't get the Zenlike character of the title. What is the smell of one invisible hand clapping ?
** He's quoting himself. The NYT style guide is a mystery to the unenlightened.
Saturday, March 22, 2008
Hansen Vs Gore
This is interesting. Dr James Hansen, a world leading climatologist at NASA whose struggles against the Bush administration's effort to censor him are well known, also had some trouble with the Clinton administration.
From an interview at Democracy Now
h/t DarkSyde
Hmm what's with soot ? I used the google and got this simple summary of (some of) Hansen's research on the topic
[snip]
Oh.
Being dumb, I had thought of black smoke as a source of global cooling as, when it is up in the air, it absorbs light where there is little blocking the infrared it emits from escaping (nuclear winter and all that). I didn't think sooty snow was important.
However, I note that there isn't snow everywhere. Soot coming from, say Guangdong Province will not end up on snow. I would still guess that black smoke over tropical oceans, while unpleasant, would reduce global warming.
Anyway, Hansen and Nazarenko's research suggests a cost efficient way to fight global warming, but putting electrostatic dust precipitators in large fixed sources of soot. I would guess that the target for this effort would be Russia not China. Difficult to deal with, but it is much easier to reduce particulate emissions than to reduce C02 emissions.
I may be slandering Gore, but I suspect that he was afraid that the research would lead people in the USA to conclude that we have already done our part. My guess is that the USA stacks up much better in particulate emissions than in CO2 emissions.
This is interesting. Dr James Hansen, a world leading climatologist at NASA whose struggles against the Bush administration's effort to censor him are well known, also had some trouble with the Clinton administration.
From an interview at Democracy Now
AMY GOODMAN: So, before we go on to the Bush administration, where you did have the most trouble, can you talk about what happened during the Clinton years and how you were able to express or not your research?
DR. JAMES HANSEN: Well, the one particular event that stands out in my mind is when I wrote a paper called “Global Warming in the 21st Century: An Alternative Scenario,” in which I emphasized that it’s not only carbon dioxide, but other climate forcings—methane and black soot—and we need to address those also. And for some reason, the people in the White House didn’t like emphasis on the non-CO2 parts of the story, and I just—the press release just kept coming back, and I would try to change it, they would change it, and finally I gave up. I just couldn’t get a press release through the way I wanted it.
h/t DarkSyde
Hmm what's with soot ? I used the google and got this simple summary of (some of) Hansen's research on the topic
A new study found that emissions of soot, or black carbon, alters the way sunlight reflects off snow and may be responsible for as much as 25 percent of observed global warming over the past century.
[snip]
Hansen and Nazarenko used a leading worldwide-climate computer model to simulate the effects of greenhouse gases and other factors on world climate. The researchers incorporated into the model data from NASA spacecraft that monitor the Earth's surface, vegetation, oceans and atmosphere. The calculated global warming from soot in snow and ice, by itself, accounted for 25 percent of observed global warming in an 1880-2000 simulation. NASA's Terra and Aqua satellites allow daily monitoring of snow cover and reflectivity, helping scientists better understand the effects of soot on snow.
Oh.
Being dumb, I had thought of black smoke as a source of global cooling as, when it is up in the air, it absorbs light where there is little blocking the infrared it emits from escaping (nuclear winter and all that). I didn't think sooty snow was important.
However, I note that there isn't snow everywhere. Soot coming from, say Guangdong Province will not end up on snow. I would still guess that black smoke over tropical oceans, while unpleasant, would reduce global warming.
Anyway, Hansen and Nazarenko's research suggests a cost efficient way to fight global warming, but putting electrostatic dust precipitators in large fixed sources of soot. I would guess that the target for this effort would be Russia not China. Difficult to deal with, but it is much easier to reduce particulate emissions than to reduce C02 emissions.
I may be slandering Gore, but I suspect that he was afraid that the research would lead people in the USA to conclude that we have already done our part. My guess is that the USA stacks up much better in particulate emissions than in CO2 emissions.
Friday, March 21, 2008
Andrew Sullivan is very hard on Andrew Sullivan
He has a strong case, but I admire his courage and honesty.
John Cole too.
And he's funny.
He has a strong case, but I admire his courage and honesty.
John Cole too.
And he's funny.
Thursday, March 20, 2008
Fact Check Check
I have been ignoring the Washington Post Factcheck column written by Michael Dobbs.
My bad. He fact checked McCain on al Qaeda and Iran. He notes that McCain is full of it. He also notes that the McCain campaign (which seems to have switched from saying that McCain's claim was a slip of the tongue which he corrected immediately to saying it is true) edited out the key part of a document which they presented as proving McCain's claim (the key part was essentially saying the opposite of McCain's claim).
Good job. Still Dobbs could do better. For one things he refers to "the Sunni-Shiite ethnic divide" it is not an ethnic divide, it is a sectarian divide (there is an ethnic difference between McCain and Obama there is a sectarian difference between McCain and Lieberman. These are not obscure words).
More importantly Dobbs does not address McCain's claim that he mispoke and quickly corrected himself. He does provide the evidence which proves that it is a lie -- he notes that McCain made the same claim on two successive days. However, he doesn't mention at all the fact that McCain claimed that he misspoke. That is to say, he let a blatant demonstrable lie pass in his "fact check".
The rule appears to be that McCain is allowed to lie, then take the lie back, and no decent journalist will mention it.
I have been ignoring the Washington Post Factcheck column written by Michael Dobbs.
My bad. He fact checked McCain on al Qaeda and Iran. He notes that McCain is full of it. He also notes that the McCain campaign (which seems to have switched from saying that McCain's claim was a slip of the tongue which he corrected immediately to saying it is true) edited out the key part of a document which they presented as proving McCain's claim (the key part was essentially saying the opposite of McCain's claim).
Good job. Still Dobbs could do better. For one things he refers to "the Sunni-Shiite ethnic divide" it is not an ethnic divide, it is a sectarian divide (there is an ethnic difference between McCain and Obama there is a sectarian difference between McCain and Lieberman. These are not obscure words).
More importantly Dobbs does not address McCain's claim that he mispoke and quickly corrected himself. He does provide the evidence which proves that it is a lie -- he notes that McCain made the same claim on two successive days. However, he doesn't mention at all the fact that McCain claimed that he misspoke. That is to say, he let a blatant demonstrable lie pass in his "fact check".
The rule appears to be that McCain is allowed to lie, then take the lie back, and no decent journalist will mention it.
CNN Edits Video to Distort the Truth
This is journalistic fraud
Media Matters
via many bloggers.
OK How long did those "few moments" last ? McCain made the claim twice. The first time, as shown on CNN was separated from the correction by more than a few moments, by a reporter questioning McCain's claim and by McCain insisting that it was "common knowledge."
CNNs version is inconsistent with the facts, so they decided to suppress them by editing out the part of the video which proves that McCain's claim that he immediately corrected himself is a lie.
mediamatters has a great catch, but their embed video tag doesn't work.
This is journalistic fraud
Media Matters
Summary: Discussing Sen. John McCain's false claim that Iranian operatives are "taking Al Qaeda into Iran, training them and sending them back," CNN's Wolf Blitzer falsely asserted that McCain "quickly corrected [it] after some prodding from his Senate colleague, Joe Lieberman." Blitzer then aired a spliced video of McCain's misstatement immediately followed by his "correct[ion]." In fact, as The Washington Post reported, it was later in the press conference when McCain was "[p]ressed to elaborate" on his claim and after he reiterated that "Al Qaeda is going back into Iran and receiving training and are coming back into Iraq from Iran," that Lieberman "stepped forward and whispered" in his ear. McCain then corrected himself.
via many bloggers.
OK How long did those "few moments" last ? McCain made the claim twice. The first time, as shown on CNN was separated from the correction by more than a few moments, by a reporter questioning McCain's claim and by McCain insisting that it was "common knowledge."
CNNs version is inconsistent with the facts, so they decided to suppress them by editing out the part of the video which proves that McCain's claim that he immediately corrected himself is a lie.
mediamatters has a great catch, but their embed video tag doesn't work.
Caveat Lector II
More work on meta-analysis of the effects of anti-depressants. By now I am sure that I understand the notation "d" used by physicians when talking about statistics, since I can reproduced the results in Kirsch et al (2008). This, and repeated checking, convince me that I managed to read a *.gif and type numbers in correctly. Finally, I have reproduced the results calculated with a spread sheet using STATA 9. The following post might still contain boo-boos, but it is, I hope, getting close to a journal submission.
Kirsch et al (2008) estimate the average benefit of SSRI’s using a simple standard meta-analysis procedure “We conducted two types of data analysis …and another using each study's drug and placebo groups' arithmetic mean (weighted for the inverse of the variance) as the meta-analytic “effect size” [11]. They first calculated an estimate of the overall average change of the Hamilton Rating Scale of Depression (HRSD) for patients who received an SSRI as the precision weighted average of the changes reported for the treated subsample in each trial. Then they calculated the average change for patients who received the placebo in the same way and calculated the difference. This approach involves two methodological choices. First the use of precision weights (weights inverse to the reported variance of the estimate of the mean change in each study). Second the decision to calculate overall mean changes for treated and control groups and then take the difference rather than first calculating the difference in the HRSD then calculating a weighted average across trials of those differences. In each case, Kirsch chose a method which, under strong assumptions, gives an efficient and unbiased estimate of the true overall average benefit. In each case there are alternative approaches which are less efficient under those assumptions but which are unbiased not only when the Kirsch et al estimates are unbiased, but also for many cases in which the Kirsch et al estimates are biased. That is they are less efficient under the null but more robust. In each case the null hypothesis that the Kirsch et al estimator is unbiased has been tested and rejected. The available unbiased estimate of the overall average benefit of SSRI’s is equal to 2.65 HRSD units, which is considerably higher than Kirsch et al’s biased estimate
Bias due to precision weighting when disturbance terms are not normal.
First consider the estimation of the mean change in HRSD for patients who received an SSRI. A simple natural alternative to the precision weighted average is the simple average change in HRSD across all patients. This is equal to the average of the average change across trials weighted by the number of patients who received an SSRI (from now on Nsi). This is an unbiased estimate of the sample size (Nsi ) weighted average expected change across trials.
The precision weighted average is an unbiased and efficient estimate of the sample size (Nsi ) weighted average expected change across trials if, within each trial, the improvement of the average patient were equal to the expected improvement (itself a function of the characteristics of the trial) plus a disturbance term with a mean zero normal distribution. The assumption of normality has the critically important implication that for given population means and averages the sample mean and the sample average are distributed independently. This is a particular property of the normal distribution which does not hold for other distributions.
In general it is possible that sampling alone will create a positive correlation between the sample mean and the sample variance. This would mean that trials that showed a higher than expected improvement would have a lower than expected weight and, thus, that the precision weighted average is biased down.
In a first effort to see if this might be a problem, the 35 sample standard deviations (sds) were regressed on the sample mean improvements in RRSD (CHANGEs) giving a coefficient of 0.25 with a standard error of. 0.1 Under the null of zero correlation, the ratio of the coefficient and the standard error (2.51) definitely does not have a t-distribution, since the sample standard error is definitely not normally distributed.
A more reasonable second step was to regress sds on CHANGEs by weighted least squares with weights equal to Ns). This gives coefficient (0.39) with a standard error of (0.08). Cehbychev’s inequality is sufficient to reject the null of zero correlation at the 5% level without any distributional assumptions.
Since there is an estimate which is efficient and unbiased under the null and another estimate which is unbiased both under the null and the alternative, it is possible to test the null hypothesis that the precision weighted average is unbiased using a Wu-Hausman test. The square of the difference in the estimates divided by the difference in the estimated variances is distributed as a chi2(1) under the null.
The variance of the CHANGE within subsamples (Varsi and Varpi) are needed to calculate the precision weighted average of change and to calculate the variance of both the precision weighted average and the simple average. Varsi can be calculated from the data in Table 1 of Kirsch et al using the formula Varsi = (CHANGEsi /dsi)2 The variance of the precision weighted average is the inverse of the sum over i of (Nsi /Varsi ). The variance of the simple average is the sum over i of
Nsi Varsi divided by the square of the overall total number of patients who received SSRIs.
The calculated statistic is 133.15 overwhelmingly rejecting the null.
The huge Wu-Hausman test statistic is partly the result of one outlier protocol 62 subsample mild, which has a very low reported sds and CHANGEs. If the very low sds is replaced by the Ns weighted average sds the Wu Housman statistic drops to 20.57 which is still overwhelmingly significant.
Define Npi As the number of patients who received the placebo in trial i, sdpi as the standard deviation in the change in HRSD among patients who received the placebo in trial i and changepi as the average change in HRSD among patients who received the placebo in trial i.
It is possible to test the null that the precision weighted average change of patients who received the placebo is an unbiased estimate of the overall average expected change in HRSD when receiving a placebo against the alternative that it is biased but the simple average is unbiased. In this case the Wu-Hausman test statistic is 0.65 which is not at all significant.
Finally it is possible to test the null hypothesis that the estimate of the additional benefit of an SSRI over a placebo estimated by Kirsch et al (1.8) is biased against the alternative hypothesis that it is biased but the difference in the simple average changes in HRSD across patients (2.18) is an unbiased estimate of the additional benefit. First note that, as discussed below, both methods require the implicit assumption that the disturbances to individual patients’ change in HRSD are independent. This implies that the overall average CHANGE of SSRI treated patients is distributed independently of the overall average CHANGE of placebo treated patients, so the variance of the difference between these averages is equal to the sum of the variances of these averages.
In this case the Wu-Hausman test statistic is 38.07 which is overwhelmingly significant and provides strong evidence that Kirsch et al’s estimate is biased down.
Bias due to the Exclusion of Indicator Variables for Each Trial
A separate and, perhaps, more important issue is the difference between first calculating overall mean changes in HRSD for SSRI treated and placebo treated patients and taking the difference as Kirsch et all did, instead of first calculating the difference in HRSD improvement of SSRI and placebo treated patients within each trial and then calculating an overall average. A basic principle of clinical trials is that the only reliable evidence is differences between treated and controlled patients when neither the patients nor the researchers know who is treated. This suggests that the basic principles of clinical testing imply that the first step of valid analysis must be calculating these differences. The null hypotheses that other approaches are unbiased are tested and rejected below. The explanation of this statistical result should be familiar as concern about the potential biases are exactly the reason that controlled double blind studies are conducted.
Some notation will be useful in the discussion below. First Changepij is the change in HRSD of the jth patient in the ith trial. In general Changepij will be affected by factors specific to patient j and by factors common to all patients in trial i so
1) Changepij = c + dSSRI + i + ij .
If is correlated with dSSRI, OLS and weighted least squares estimates of will be biased. Even if is not correlated with dSSRI, such estimates will be inefficient, because they are based on the assumption that the variance of is zero.
In order to explain the hypothesis tests, it is helpful to explain how the relevant calculations can be interpreted as OLS or weighted least squares regressions. For simplicity, first the case of simple OLS regressions with data on individual patients from different trials is discussed. As discussed above, a very simple estimate of the effect of SSRI’s in addition to the placebo effect is the average change of HRSD of SSRI treated patients minus the average change in HRSD of placebo treated patients. This is exactly equal to the coefficient on an indicator variable (dSSRI) which is one if the patient took an SSRI and 0 otherwise in a simple OLS regression which also includes a constant term. This estimate is biased if there is any excluded variable which is correlated both with dSSRI and with the change in HRSD (CHANGE). Since different fractions of patients were given SSRI’s in different trials, indicator variables for the trials are correlated with dSSRI. An indicator variable for the trial implementing protocol 62 [mild] would be one for patients enrolled in that trial whether they received an SSRI or a placebo, and zero for other patients.
The coefficient of CHANGE on dSSRI in a regression which also contains 35 indicator variables for the 35 trials (or equivalently 34 indicators for 34 trials and a constant) is called the within estimator of the additional benefit the SSRIs, because it is based on differences within each group, in this case each clinical trial.
Now it is easy to calculate the within estimate using only the data reported in table 1 of Kirsch et al and in particular CHANGE and the number of patients (N) in each of the 70 subsamples. The OLS estimate with individual data is equal to the weighted least squares estimates using N as weights of the average change in HRSD on the average value of dSSRI (the fraction of the patients in the trial who received an SSRI) and the indicator for the trial.
The coefficient is also equal to the weighted average of the difference in CHANGE for the SSRI treated patients and the placebo treated patients in each trial. The weights to be used to obtain the within estimate are the harmonic mean of the number of patients who received the SSRI and the number of patients who received the placebo Nhi = (NsiNpi/(Nsi + Npi ). The variance of the within estimate of the coefficient on dSSRI is the sum over I of Nhi((Varsi/Nsi)+(Varpi/Npi)) divided by the square of the sum over I of Nhi.
The resulting estimate of the benefit of the SSRIs is 2.65.with a standard error of 0.255.
A very strong case can be made a priori that such indicator variables should be included in the regression. As noted above, the indicator variables are correlated with the variable of interest, since different fractions of patients received an SSRI in different trials. Different trials also have different average values of CHANGE for many reasons. First the average baseline HRSD is different for different trials. Second all trials include not only ingestion of an SSRI or a placebo but also clinical management which may differ across trials and cause different average changes in HRSD. Finally, application Hamilton rating scale requires judgment on the part of the clinician and the application of the scale might not be identical across trials.
Common factors affecting different patients in the same trial ( imply that OLS is not appropriate. At the very least, common factors imply that OLS is not efficient even if the common factors are not correlated with dSSRI. It also seems extremely unlikely that OLS will be unbiased.
To see this, note that the OLS estimate of the coefficient is the weighted average of the within estimate described above and the, so called, between estimate which is equal to a weighted least squares regression of each trial’s overall average of CHANGE on the fraction of patients in the trial who receive an SSRI (FSRRI) and a constant. The weights are equal to the total number of patients in the trial. For the data presented in Table 1 of Kirsch et al, the between estimate of the additional benefit of an SSRI over a placebo is actually negative (-.).
Assuming that the within estimate is unbiased, the OLS estimate is unbiased only if the sample size weighted between estimate is unbiased. It should be clear that the data used in calculating the between estimate are not the sort of data usually considered acceptable when evaluating a pharmaceutical. For example, if one were willing to trust between estimators, one could estimate the effect of a pharmaceutical using one trial in which all patients receive the pharmaceutical and another trial in which all patients received a placebo. The view that trials should be double blind is inconsistent with the view that between estimators should be used at all.
Even using the extremely limited data available in table 1 of Kirsch et al, it is possible to demonstrate in two different ways that the OLS estimate of the benefit of SSRIs is biased. First, the only other variable in the table – baseline, is correlated with ACHANGE and with FSSRI. ABASELINEi is the overall average initial HRSD of patients involved in trial i. Weighted least squares, where the total number of patients in each trial are used as weights, of ACHANGE and FSSRI on ABASELINE are reported below. They are sufficient to demonstrate that the OLS estimate of the average benefit of an SSRI is biased down.
Also it is possible to perform a Wu-Hausman test to test the null that the OLS estimate is unbiased and efficient. This null hypothesis is not extremely interesting, since, for the OLS estimate to be efficient, it would be necessary for the correlation within trials of CHANGE to be zero. However, it is easily performed and rejects the null that OLS is efficient and unbiased. The statistic, which has a Chi2(1) distribution under the null, is equal to 13.49 overwhelmingly rejecting the null that the OLS estimate is unbiased.
Similarly, it is possible to test whether the Kirsch et al estimator is unbiased by comparing the estimate to weighted least squares estimates of the model including a full set of indicator variables for the trials in which the weights are equal to the subsample size divided by the subsample variance. This last estimator is a precision weighted within estimator. The coefficient on dSSRI is equal to the weighted average difference in CHANGE with an SSRI and placebo for each trial with the weight on the difference in the ith trial equal to 1/((varsi/Nsi)+(varpi/Npi)). The precision weighted within estimate is 2.40.
The variance of the precision weighted within estimator is equal to the sum of these weights. Thus it is possible to perform a Wu-Hausman test of the null that both the Kirsch et al estimator and the precision weighted fixed effects estimator are unbiased against the alternative that only the precision weighted fixed effects estimator is unbiased. The statistic, which is distributed as a Chi2(1) under the null is equal to 55.37 overwhelmingly rejecting the null hypothesis.
Also the null hypothesis that the precision weighted within estimator and the subsample size weighted within estimator are both unbiased is rejected against the alternative that only the sample size weighted estimator is unbiased, since the relevant Wu-Hausman statistic is 13.49.
Finally, the null that the Kirsch et al estimator and the subsample size weighted within estimator are unbiased is rejected against the alternative that only the subsample size weighted estimator is unbiased since the Wu-Hausman test statistic is 65.00.
The Wu-Hausman test is only valid if one of the estimators is efficient under the null hypothesis. In out case, this requires the assumption that all the common terms are exactly zero. A more interesting hypothesis is that is iid across trials with a positive variance and is uncorrelated with the fraction of patients receiving an SSRI. In this case, the most efficient estimate is the random effects estimator which is a weighted average of the within and between estimates with weights equal to the inverse of the variances of the estimates.
Unfortunately, the correct weights for the between estimator are not the inverse of the number of patients in the trial. Rather they are the inverse of the sum of the variance of and the inverse of the number of patients in the trial times the variance of . The variances of and can be estimated by maximum likelihood (if one can explain weighted random effects to a statistical package). Alternatively they can be estimated by iterated weighted least squares starting with the weighted least squares between estimate for zero variance , estimating the variance of from the difference between the sum of squared residuals from that regression and the sum of Vari . Then repeating weighted OLS with the new weights and calculating a new estimate of the variance of and continuing until the procedure converges.
This gives a between estimate quite different from the first estimate based on the assumption that the variance of is zero. The estimated coefficient on dSSRI is -0.48 with estimated variance 2.43. Thus the random effects estimate of the coefficient on dSSRI is 2.57 with a variance of 0.0637. Note that the random effects estimate is very close to the within estimate. Nonetheless the Wu_hausman test statistic is of 3.93 implies rejection of the null hypothesis that the random effects estimate is unbiased (the 5% critical level is 3.84).
Conclusion
Of the estimates reported above, the only estimate of the overall average effect of the SSRI’s on HRSD which is not known to be biased is 2.65 -- the result of sample size weighted within estimation.
More work on meta-analysis of the effects of anti-depressants. By now I am sure that I understand the notation "d" used by physicians when talking about statistics, since I can reproduced the results in Kirsch et al (2008). This, and repeated checking, convince me that I managed to read a *.gif and type numbers in correctly. Finally, I have reproduced the results calculated with a spread sheet using STATA 9. The following post might still contain boo-boos, but it is, I hope, getting close to a journal submission.
Kirsch et al (2008) estimate the average benefit of SSRI’s using a simple standard meta-analysis procedure “We conducted two types of data analysis …and another using each study's drug and placebo groups' arithmetic mean (weighted for the inverse of the variance) as the meta-analytic “effect size” [11]. They first calculated an estimate of the overall average change of the Hamilton Rating Scale of Depression (HRSD) for patients who received an SSRI as the precision weighted average of the changes reported for the treated subsample in each trial. Then they calculated the average change for patients who received the placebo in the same way and calculated the difference. This approach involves two methodological choices. First the use of precision weights (weights inverse to the reported variance of the estimate of the mean change in each study). Second the decision to calculate overall mean changes for treated and control groups and then take the difference rather than first calculating the difference in the HRSD then calculating a weighted average across trials of those differences. In each case, Kirsch chose a method which, under strong assumptions, gives an efficient and unbiased estimate of the true overall average benefit. In each case there are alternative approaches which are less efficient under those assumptions but which are unbiased not only when the Kirsch et al estimates are unbiased, but also for many cases in which the Kirsch et al estimates are biased. That is they are less efficient under the null but more robust. In each case the null hypothesis that the Kirsch et al estimator is unbiased has been tested and rejected. The available unbiased estimate of the overall average benefit of SSRI’s is equal to 2.65 HRSD units, which is considerably higher than Kirsch et al’s biased estimate
Bias due to precision weighting when disturbance terms are not normal.
First consider the estimation of the mean change in HRSD for patients who received an SSRI. A simple natural alternative to the precision weighted average is the simple average change in HRSD across all patients. This is equal to the average of the average change across trials weighted by the number of patients who received an SSRI (from now on Nsi). This is an unbiased estimate of the sample size (Nsi ) weighted average expected change across trials.
The precision weighted average is an unbiased and efficient estimate of the sample size (Nsi ) weighted average expected change across trials if, within each trial, the improvement of the average patient were equal to the expected improvement (itself a function of the characteristics of the trial) plus a disturbance term with a mean zero normal distribution. The assumption of normality has the critically important implication that for given population means and averages the sample mean and the sample average are distributed independently. This is a particular property of the normal distribution which does not hold for other distributions.
In general it is possible that sampling alone will create a positive correlation between the sample mean and the sample variance. This would mean that trials that showed a higher than expected improvement would have a lower than expected weight and, thus, that the precision weighted average is biased down.
In a first effort to see if this might be a problem, the 35 sample standard deviations (sds) were regressed on the sample mean improvements in RRSD (CHANGEs) giving a coefficient of 0.25 with a standard error of. 0.1 Under the null of zero correlation, the ratio of the coefficient and the standard error (2.51) definitely does not have a t-distribution, since the sample standard error is definitely not normally distributed.
A more reasonable second step was to regress sds on CHANGEs by weighted least squares with weights equal to Ns). This gives coefficient (0.39) with a standard error of (0.08). Cehbychev’s inequality is sufficient to reject the null of zero correlation at the 5% level without any distributional assumptions.
Since there is an estimate which is efficient and unbiased under the null and another estimate which is unbiased both under the null and the alternative, it is possible to test the null hypothesis that the precision weighted average is unbiased using a Wu-Hausman test. The square of the difference in the estimates divided by the difference in the estimated variances is distributed as a chi2(1) under the null.
The variance of the CHANGE within subsamples (Varsi and Varpi) are needed to calculate the precision weighted average of change and to calculate the variance of both the precision weighted average and the simple average. Varsi can be calculated from the data in Table 1 of Kirsch et al using the formula Varsi = (CHANGEsi /dsi)2 The variance of the precision weighted average is the inverse of the sum over i of (Nsi /Varsi ). The variance of the simple average is the sum over i of
Nsi Varsi divided by the square of the overall total number of patients who received SSRIs.
The calculated statistic is 133.15 overwhelmingly rejecting the null.
The huge Wu-Hausman test statistic is partly the result of one outlier protocol 62 subsample mild, which has a very low reported sds and CHANGEs. If the very low sds is replaced by the Ns weighted average sds the Wu Housman statistic drops to 20.57 which is still overwhelmingly significant.
Define Npi As the number of patients who received the placebo in trial i, sdpi as the standard deviation in the change in HRSD among patients who received the placebo in trial i and changepi as the average change in HRSD among patients who received the placebo in trial i.
It is possible to test the null that the precision weighted average change of patients who received the placebo is an unbiased estimate of the overall average expected change in HRSD when receiving a placebo against the alternative that it is biased but the simple average is unbiased. In this case the Wu-Hausman test statistic is 0.65 which is not at all significant.
Finally it is possible to test the null hypothesis that the estimate of the additional benefit of an SSRI over a placebo estimated by Kirsch et al (1.8) is biased against the alternative hypothesis that it is biased but the difference in the simple average changes in HRSD across patients (2.18) is an unbiased estimate of the additional benefit. First note that, as discussed below, both methods require the implicit assumption that the disturbances to individual patients’ change in HRSD are independent. This implies that the overall average CHANGE of SSRI treated patients is distributed independently of the overall average CHANGE of placebo treated patients, so the variance of the difference between these averages is equal to the sum of the variances of these averages.
In this case the Wu-Hausman test statistic is 38.07 which is overwhelmingly significant and provides strong evidence that Kirsch et al’s estimate is biased down.
Bias due to the Exclusion of Indicator Variables for Each Trial
A separate and, perhaps, more important issue is the difference between first calculating overall mean changes in HRSD for SSRI treated and placebo treated patients and taking the difference as Kirsch et all did, instead of first calculating the difference in HRSD improvement of SSRI and placebo treated patients within each trial and then calculating an overall average. A basic principle of clinical trials is that the only reliable evidence is differences between treated and controlled patients when neither the patients nor the researchers know who is treated. This suggests that the basic principles of clinical testing imply that the first step of valid analysis must be calculating these differences. The null hypotheses that other approaches are unbiased are tested and rejected below. The explanation of this statistical result should be familiar as concern about the potential biases are exactly the reason that controlled double blind studies are conducted.
Some notation will be useful in the discussion below. First Changepij is the change in HRSD of the jth patient in the ith trial. In general Changepij will be affected by factors specific to patient j and by factors common to all patients in trial i so
1) Changepij = c + dSSRI + i + ij .
If is correlated with dSSRI, OLS and weighted least squares estimates of will be biased. Even if is not correlated with dSSRI, such estimates will be inefficient, because they are based on the assumption that the variance of is zero.
In order to explain the hypothesis tests, it is helpful to explain how the relevant calculations can be interpreted as OLS or weighted least squares regressions. For simplicity, first the case of simple OLS regressions with data on individual patients from different trials is discussed. As discussed above, a very simple estimate of the effect of SSRI’s in addition to the placebo effect is the average change of HRSD of SSRI treated patients minus the average change in HRSD of placebo treated patients. This is exactly equal to the coefficient on an indicator variable (dSSRI) which is one if the patient took an SSRI and 0 otherwise in a simple OLS regression which also includes a constant term. This estimate is biased if there is any excluded variable which is correlated both with dSSRI and with the change in HRSD (CHANGE). Since different fractions of patients were given SSRI’s in different trials, indicator variables for the trials are correlated with dSSRI. An indicator variable for the trial implementing protocol 62 [mild] would be one for patients enrolled in that trial whether they received an SSRI or a placebo, and zero for other patients.
The coefficient of CHANGE on dSSRI in a regression which also contains 35 indicator variables for the 35 trials (or equivalently 34 indicators for 34 trials and a constant) is called the within estimator of the additional benefit the SSRIs, because it is based on differences within each group, in this case each clinical trial.
Now it is easy to calculate the within estimate using only the data reported in table 1 of Kirsch et al and in particular CHANGE and the number of patients (N) in each of the 70 subsamples. The OLS estimate with individual data is equal to the weighted least squares estimates using N as weights of the average change in HRSD on the average value of dSSRI (the fraction of the patients in the trial who received an SSRI) and the indicator for the trial.
The coefficient is also equal to the weighted average of the difference in CHANGE for the SSRI treated patients and the placebo treated patients in each trial. The weights to be used to obtain the within estimate are the harmonic mean of the number of patients who received the SSRI and the number of patients who received the placebo Nhi = (NsiNpi/(Nsi + Npi ). The variance of the within estimate of the coefficient on dSSRI is the sum over I of Nhi((Varsi/Nsi)+(Varpi/Npi)) divided by the square of the sum over I of Nhi.
The resulting estimate of the benefit of the SSRIs is 2.65.with a standard error of 0.255.
A very strong case can be made a priori that such indicator variables should be included in the regression. As noted above, the indicator variables are correlated with the variable of interest, since different fractions of patients received an SSRI in different trials. Different trials also have different average values of CHANGE for many reasons. First the average baseline HRSD is different for different trials. Second all trials include not only ingestion of an SSRI or a placebo but also clinical management which may differ across trials and cause different average changes in HRSD. Finally, application Hamilton rating scale requires judgment on the part of the clinician and the application of the scale might not be identical across trials.
Common factors affecting different patients in the same trial ( imply that OLS is not appropriate. At the very least, common factors imply that OLS is not efficient even if the common factors are not correlated with dSSRI. It also seems extremely unlikely that OLS will be unbiased.
To see this, note that the OLS estimate of the coefficient is the weighted average of the within estimate described above and the, so called, between estimate which is equal to a weighted least squares regression of each trial’s overall average of CHANGE on the fraction of patients in the trial who receive an SSRI (FSRRI) and a constant. The weights are equal to the total number of patients in the trial. For the data presented in Table 1 of Kirsch et al, the between estimate of the additional benefit of an SSRI over a placebo is actually negative (-.).
Assuming that the within estimate is unbiased, the OLS estimate is unbiased only if the sample size weighted between estimate is unbiased. It should be clear that the data used in calculating the between estimate are not the sort of data usually considered acceptable when evaluating a pharmaceutical. For example, if one were willing to trust between estimators, one could estimate the effect of a pharmaceutical using one trial in which all patients receive the pharmaceutical and another trial in which all patients received a placebo. The view that trials should be double blind is inconsistent with the view that between estimators should be used at all.
Even using the extremely limited data available in table 1 of Kirsch et al, it is possible to demonstrate in two different ways that the OLS estimate of the benefit of SSRIs is biased. First, the only other variable in the table – baseline, is correlated with ACHANGE and with FSSRI. ABASELINEi is the overall average initial HRSD of patients involved in trial i. Weighted least squares, where the total number of patients in each trial are used as weights, of ACHANGE and FSSRI on ABASELINE are reported below. They are sufficient to demonstrate that the OLS estimate of the average benefit of an SSRI is biased down.
Also it is possible to perform a Wu-Hausman test to test the null that the OLS estimate is unbiased and efficient. This null hypothesis is not extremely interesting, since, for the OLS estimate to be efficient, it would be necessary for the correlation within trials of CHANGE to be zero. However, it is easily performed and rejects the null that OLS is efficient and unbiased. The statistic, which has a Chi2(1) distribution under the null, is equal to 13.49 overwhelmingly rejecting the null that the OLS estimate is unbiased.
Similarly, it is possible to test whether the Kirsch et al estimator is unbiased by comparing the estimate to weighted least squares estimates of the model including a full set of indicator variables for the trials in which the weights are equal to the subsample size divided by the subsample variance. This last estimator is a precision weighted within estimator. The coefficient on dSSRI is equal to the weighted average difference in CHANGE with an SSRI and placebo for each trial with the weight on the difference in the ith trial equal to 1/((varsi/Nsi)+(varpi/Npi)). The precision weighted within estimate is 2.40.
The variance of the precision weighted within estimator is equal to the sum of these weights. Thus it is possible to perform a Wu-Hausman test of the null that both the Kirsch et al estimator and the precision weighted fixed effects estimator are unbiased against the alternative that only the precision weighted fixed effects estimator is unbiased. The statistic, which is distributed as a Chi2(1) under the null is equal to 55.37 overwhelmingly rejecting the null hypothesis.
Also the null hypothesis that the precision weighted within estimator and the subsample size weighted within estimator are both unbiased is rejected against the alternative that only the sample size weighted estimator is unbiased, since the relevant Wu-Hausman statistic is 13.49.
Finally, the null that the Kirsch et al estimator and the subsample size weighted within estimator are unbiased is rejected against the alternative that only the subsample size weighted estimator is unbiased since the Wu-Hausman test statistic is 65.00.
The Wu-Hausman test is only valid if one of the estimators is efficient under the null hypothesis. In out case, this requires the assumption that all the common terms are exactly zero. A more interesting hypothesis is that is iid across trials with a positive variance and is uncorrelated with the fraction of patients receiving an SSRI. In this case, the most efficient estimate is the random effects estimator which is a weighted average of the within and between estimates with weights equal to the inverse of the variances of the estimates.
Unfortunately, the correct weights for the between estimator are not the inverse of the number of patients in the trial. Rather they are the inverse of the sum of the variance of and the inverse of the number of patients in the trial times the variance of . The variances of and can be estimated by maximum likelihood (if one can explain weighted random effects to a statistical package). Alternatively they can be estimated by iterated weighted least squares starting with the weighted least squares between estimate for zero variance , estimating the variance of from the difference between the sum of squared residuals from that regression and the sum of Vari . Then repeating weighted OLS with the new weights and calculating a new estimate of the variance of and continuing until the procedure converges.
This gives a between estimate quite different from the first estimate based on the assumption that the variance of is zero. The estimated coefficient on dSSRI is -0.48 with estimated variance 2.43. Thus the random effects estimate of the coefficient on dSSRI is 2.57 with a variance of 0.0637. Note that the random effects estimate is very close to the within estimate. Nonetheless the Wu_hausman test statistic is of 3.93 implies rejection of the null hypothesis that the random effects estimate is unbiased (the 5% critical level is 3.84).
Conclusion
Of the estimates reported above, the only estimate of the overall average effect of the SSRI’s on HRSD which is not known to be biased is 2.65 -- the result of sample size weighted within estimation.
Wednesday, March 19, 2008
My countrypeople are insane.
Pollingreport has results on opinions about the death penalty which prove this.
First, in different polls, about the same fraction of people (sadly around 30%) say they are opposed to the death penalty and say they are opposed to the death penalty for murder (get that not first degree murder just murder). To be opposed to the death penalty (as I am) means to be opposed to the death penalty for any crime including the rape torture and killing of someone one kidnapped (as I am). It is crazy that there are essentially no Americans who support the death penalty for such a horrible crime but oppose it for second degree murder.
Fortunately a similarly small fraction want to increase the number of executions. Do people have any idea how many people we would kill if we imposed the death penalty for plain ordinary murders ?
Pollingreport has results on opinions about the death penalty which prove this.
First, in different polls, about the same fraction of people (sadly around 30%) say they are opposed to the death penalty and say they are opposed to the death penalty for murder (get that not first degree murder just murder). To be opposed to the death penalty (as I am) means to be opposed to the death penalty for any crime including the rape torture and killing of someone one kidnapped (as I am). It is crazy that there are essentially no Americans who support the death penalty for such a horrible crime but oppose it for second degree murder.
Fortunately a similarly small fraction want to increase the number of executions. Do people have any idea how many people we would kill if we imposed the death penalty for plain ordinary murders ?
How do I agree with thee ? Let me count the ways.
You guys are having some trouble with this counting business.
Mark Kleiman has a suggestion for Hillary Clinton.
Kevin Drum is enthusiastic "It's a twofer! Count me in."
Mark Kleiman rejects such mild praise, claiming it's at least a threefer if not a fourfer. "That statement would have three advantages for HRC, and one for the country:" and then lists 1 though 4 without distinction.
One two three, what are we fighting for.
Anyway. I'm enthusiastic too and that makes ... three of us (five counting country Joe *and* the fish). Actually probably a few tens of thousand of use who have read the posts and millions to come.
You guys are having some trouble with this counting business.
Mark Kleiman has a suggestion for Hillary Clinton.
Given McCain's buffoonish performance in Jordan, wouldn't this be a good time for Hillary Clinton to say, "Gee, I thought he was ready to be Commander-in-Chief, but it sure doesn't sound like it. The least we should expect from the President is some basic knowledge about who our enemies are."
Kevin Drum is enthusiastic "It's a twofer! Count me in."
Mark Kleiman rejects such mild praise, claiming it's at least a threefer if not a fourfer. "That statement would have three advantages for HRC, and one for the country:" and then lists 1 though 4 without distinction.
One two three, what are we fighting for.
Anyway. I'm enthusiastic too and that makes ... three of us (five counting country Joe *and* the fish). Actually probably a few tens of thousand of use who have read the posts and millions to come.
The Distant Future ?
Doesn't Look Distant Enough
via Matt Yglesias.
Update: Matt has an update from the Onion News Network
Doesn't Look Distant Enough
via Matt Yglesias.
Update: Matt has an update from the Onion News Network
Tuesday, March 18, 2008
Marching Orders
Brad DeLong orders me to support the Frank-Dodd bill to guarantee mortgages through the FHA.
He modestly neglects to mention that, last week, he told The Dodd and Frank to do this
OK OK technically he told Hank Paulson to do this, but we know that was a rhetorical device. Saying the Bush administration should lead is like saying pigs should fly -- a sign of desperation not a proposal.
Brad DeLong orders me to support the Frank-Dodd bill to guarantee mortgages through the FHA.
He modestly neglects to mention that, last week, he told The Dodd and Frank to do this
If I were working for the Treasury right now, I would be saying: make this happen on Monday. There isn't time to set up a new bureaucrtacy--a HOLC, which is what Alan Blinder wanted to do as of three weeks ago. So use an existing bureaucracy: Fannie Mae. If I were Treasury Secretary Hank Paulson, I would spend the weekend building a legislative vehicle to introduce Monday morning on an emergency basis to give Fannie Mae the resources and the mission to undertake this mortgage rescue operation, and I think Fannie Mae is the right institution for the task: why does it have its government-sponsored status and guarantee if not to be used for purposes like these at times like these?
OK OK technically he told Hank Paulson to do this, but we know that was a rhetorical device. Saying the Bush administration should lead is like saying pigs should fly -- a sign of desperation not a proposal.
Monday, March 17, 2008
Cosmic Synchrony or Hmm Maybe there is a God
I was just thinking about how, if Jesus Christ were running for President, I would probably vote for the other candidate. The process of thought was that I was arguing in my mind with a friend who can't stand Barack Obama largely because Obama said he prays each night. Oddly, given the context, I imagined myself saying -- look are you going to wait until Jesus Christ returns and runs for President or ... welllll let me rephrase that (my friend is, as you have noticed, not a big fan of Jesus). Thinking further, I realised that I would probably vote against Jesus (in the Democratic primary I don't know about the general) because he was an absolute pacifist and I strongly disagree.
Then I surfed over to DailyKos and, by some miracle, Devilstower gave me a sign that I definitely wouldn't vote for Jesus (at least not in the Democratic primary)
To Jesus I say: No the rich don't deserve damnation, they* deserve progressive taxation. And what is wrong with laughing ?
Ok I admit I should have said "we" not "they" since by the standards of the last time you were on earth I am super rich.
To Devilstower I say: I much prefer those who follow the teachings of Christ to those who are sure they are Christians but invert most of them, but I prefer reasonable people to either.
To God I say: Ok you've given me another sign. I have no one to blame for my atheism but myself. If I appear in front of Peter, I won't whine or make excuses or claim I wasn't told.
* Actually I should say "we" not they as by the standards of 30 AD I am super rich.
I was just thinking about how, if Jesus Christ were running for President, I would probably vote for the other candidate. The process of thought was that I was arguing in my mind with a friend who can't stand Barack Obama largely because Obama said he prays each night. Oddly, given the context, I imagined myself saying -- look are you going to wait until Jesus Christ returns and runs for President or ... welllll let me rephrase that (my friend is, as you have noticed, not a big fan of Jesus). Thinking further, I realised that I would probably vote against Jesus (in the Democratic primary I don't know about the general) because he was an absolute pacifist and I strongly disagree.
Then I surfed over to DailyKos and, by some miracle, Devilstower gave me a sign that I definitely wouldn't vote for Jesus (at least not in the Democratic primary)
Damn You Rich!
by Devilstower
Sun Mar 16, 2008 at 02:57:01 PM PDT
Damn you rich! You already have your compensation.
Damn you who are well-fed! You will know hunger.
Damn you who laugh now! You will weep and grieve.
Damn you when everybody speaks well of you!
A rant from a radical preacher? Without a doubt. Someone on the Obama campaign? Well, Sen. Obama says so. That's the Scholars Translation of Luke 6:24-26, and the speaker is Jesus of Nazareth.
To Jesus I say: No the rich don't deserve damnation, they* deserve progressive taxation. And what is wrong with laughing ?
Ok I admit I should have said "we" not "they" since by the standards of the last time you were on earth I am super rich.
To Devilstower I say: I much prefer those who follow the teachings of Christ to those who are sure they are Christians but invert most of them, but I prefer reasonable people to either.
To God I say: Ok you've given me another sign. I have no one to blame for my atheism but myself. If I appear in front of Peter, I won't whine or make excuses or claim I wasn't told.
* Actually I should say "we" not they as by the standards of 30 AD I am super rich.
Sunday, March 16, 2008
George Orwell Is Immortal
Still on my language police kick, I note that I'm not the only one.
Paul Krugman is at it too
The swan song of the fascist Octopus is an example of a dying metaphor from "Politics and the English Language." I googled "fascist" and "octopus" and found many references to Orwell and none clearly about to fascism (or octopuses). One noted that there is no word press blog about fascist octopuses. Might have to set up "Swan Song of the Fascist Octopus" to replace the defunct "George Orwell Might Say." It's a pithy that the last link leads to a comment to brief to make it possible to understand what Dayv was talking about.
OK so googling "fascist octopus" (that is exact phrase) gives
"Results 1 - 10 of about 1,730 for "fascist octopus""
Hmmm how about
"George Orwell" Results 1 - 10 of about 4,220,000 for "george orwell"
Orwellian Results 1 - 10 of about 1,170,000 for orwellian
How does 1984 the book stack up against 1984 the year (1984 the plain ordinary number has no chance) ?
The book stomps the year
Although 2 of the top 10 hits are for the commercial introducing the MacIntosh.
Google suggests that googlers who aren't after the book try related searches
Searches related to: 1984
1984 movie the year 1984 1984 themes 1984 essays
1984 amazon 1984 text 1984 characters 1984 quiz
George Orwell owns the number 1984.
Results 1 - 10 of about 545,000 for newspeak
"Big Brother" number 10 is about the dictator in 1984. One through 9 are about reality TV shows. How many of the viewers know why they are called "Big Brother" ?
Results 1 - 10 of about 42,300 for minitrue.
Results 1 - 10 of about 10,700 for duckspeak.
I wonder what Newspeak word has the most google hits.
Results 1 - 10 of about 509,000 for "thought police"
although some don't seem clear on the concept as they refer to "Canada's thought police"
Still on my language police kick, I note that I'm not the only one.
Paul Krugman is at it too
Call the metaphor police!
I’ve been worried for a while about the fact that, according to financial reporters, the freezing up of the credit markets is causing a financial meltdown. The world is ending in ice and fire, simultaneously. But this is true cause for alarm:
“The self-feeding downturn now in place shows signs of becoming deeply entrenched,” economists at Citigroup wrote Friday.
Uh oh. we’ve got a downturn that can feed itself and, at the same time, dig trenches.
The fascist octopus will sing its swan song any day now.
The swan song of the fascist Octopus is an example of a dying metaphor from "Politics and the English Language." I googled "fascist" and "octopus" and found many references to Orwell and none clearly about to fascism (or octopuses). One noted that there is no word press blog about fascist octopuses. Might have to set up "Swan Song of the Fascist Octopus" to replace the defunct "George Orwell Might Say." It's a pithy that the last link leads to a comment to brief to make it possible to understand what Dayv was talking about.
OK so googling "fascist octopus" (that is exact phrase) gives
"Results 1 - 10 of about 1,730 for "fascist octopus""
Hmmm how about
"George Orwell" Results 1 - 10 of about 4,220,000 for "george orwell"
Orwellian Results 1 - 10 of about 1,170,000 for orwellian
How does 1984 the book stack up against 1984 the year (1984 the plain ordinary number has no chance) ?
The book stomps the year
Although 2 of the top 10 hits are for the commercial introducing the MacIntosh.
Google suggests that googlers who aren't after the book try related searches
Searches related to: 1984
1984 movie the year 1984 1984 themes 1984 essays
1984 amazon 1984 text 1984 characters 1984 quiz
George Orwell owns the number 1984.
Results 1 - 10 of about 545,000 for newspeak
"Big Brother" number 10 is about the dictator in 1984. One through 9 are about reality TV shows. How many of the viewers know why they are called "Big Brother" ?
Results 1 - 10 of about 42,300 for minitrue.
Results 1 - 10 of about 10,700 for duckspeak.
I wonder what Newspeak word has the most google hits.
Results 1 - 10 of about 509,000 for "thought police"
although some don't seem clear on the concept as they refer to "Canada's thought police"
Mutant ness monster
Keeping with the ultra twityness of the parsing of "parsing" below, I note that Josh Marshall generated a very unusual ness monster* here
"Now, sometimes spinning campaigns come up with phrases that are so heavy-handedness and tendentious that it's just ridiculous"
It is unusual since the "ness" is clearly grammatically incorrect.
I think what happened is that Marshall generated a normal ness monster and then created the mutant nessie when trying to slay the normal one.
That is I suspect he wrote ""Now, sometimes spinning campaigns come up with phrases of such heavy-handedness and tendentiousness that it's just ridiculous"
Then thought something like "uh oh what if tendentiousness gets labled a ness monster. Ok so 'tendency' uhm no, even if that still has a second or third definition as the sort of thing that is by definition tendentious no one will understand and anyone who does will think I'm a pretentious twit. Better rewrite the sentence and lop off the ness."
Notice in his thoughts he forgot the other ness.
* I was just googling for the definition of "ness monster". I was delighted to find that the Monster was sighted in "The Decline and Fall of the Roman Empire"
I've done the experiment and Gibbon wins. To me "ferociousness" sounds much fiercer than "ferocity." No surprise. Kleiman might wish, but can hardly hope, that he writes better than Gibbon. or should it be "Kleiman might wish, but can hardly hope, that he write better than Gibbon" ?
Keeping with the ultra twityness of the parsing of "parsing" below, I note that Josh Marshall generated a very unusual ness monster* here
"Now, sometimes spinning campaigns come up with phrases that are so heavy-handedness and tendentious that it's just ridiculous"
It is unusual since the "ness" is clearly grammatically incorrect.
I think what happened is that Marshall generated a normal ness monster and then created the mutant nessie when trying to slay the normal one.
That is I suspect he wrote ""Now, sometimes spinning campaigns come up with phrases of such heavy-handedness and tendentiousness that it's just ridiculous"
Then thought something like "uh oh what if tendentiousness gets labled a ness monster. Ok so 'tendency' uhm no, even if that still has a second or third definition as the sort of thing that is by definition tendentious no one will understand and anyone who does will think I'm a pretentious twit. Better rewrite the sentence and lop off the ness."
Notice in his thoughts he forgot the other ness.
* I was just googling for the definition of "ness monster". I was delighted to find that the Monster was sighted in "The Decline and Fall of the Roman Empire"
In Chapter 2 of the Decline and Fall, Gibbon refers to the "ferociousness" of some of the barbarians whom Rome confronted. That still seems to me like a much inferior expression to "ferocity" -- say them both aloud, and decide for yourself which sounds scarier -- but if the test of a language is how it is written by those who write it best, I can't really deny that Gibbon has more authority than Kleiman.
I've done the experiment and Gibbon wins. To me "ferociousness" sounds much fiercer than "ferocity." No surprise. Kleiman might wish, but can hardly hope, that he writes better than Gibbon. or should it be "Kleiman might wish, but can hardly hope, that he write better than Gibbon" ?
This is an important analysis of the outrageous behavior of the SCLM*
Sadly I want to parse the new meaning of "parse." Eric Boehlert discusses the close analysis or "parsing" of Clinton's answers. He coins a word.
I think it can be best described as "parsing."
Even since Bill Clinton (accurately) said that the judgement of whether he was a perjury depends on what the definition of "is" is, people have used "parsing" to what he did during the deposition, that is careful phrasing and especially careful phrasing to mislead with technically true statement. It refers to what he did at that press conference (try to force reporters to admit that they understand plain English and so should admit that the story they and the Republicans have been flogging is a non-story). At least people don't think it refers to what he did in the Oval office (that is use his amazing work ethic, brain and knowledge to provide outstanding public service (plus a spot on a blue dress)).
The article is important.
A bit of the transcript after the question was asked (the first time)
The exchange took the longest time, because Kroft wouldn't take no for an answer.
Now are journalists malicious or ignorant ? Since the exchange is easily available, one would assume they were being malicious if one were Boehlert. However, it is possible that they didn't watch the video, that they didn't even read the transcript but that they wrote whole columns on a phrase taken out of context and fed into the rumor mill without checking. Being lazy, I wouldn't put it past them.
Also the dread "as far as I know" is clearly Clinton trying to be polite. If she had just repeated "no" and "of course not", it would suggest that Kroft was being stupid or crazy. It is rude to repeat an answer word for word when a question is rephrased. Of course Kroft was being stupid, crazy and/or malicious.
*So Clinton loses Media
Sadly I want to parse the new meaning of "parse." Eric Boehlert discusses the close analysis or "parsing" of Clinton's answers. He coins a word.
The 60 Minutes controversy -- specifically the intense media spin it sparked -- highlights a disturbing rise in a new form of campaign journalism, which might be best described as post-parsing.
I think it can be best described as "parsing."
Even since Bill Clinton (accurately) said that the judgement of whether he was a perjury depends on what the definition of "is" is, people have used "parsing" to what he did during the deposition, that is careful phrasing and especially careful phrasing to mislead with technically true statement. It refers to what he did at that press conference (try to force reporters to admit that they understand plain English and so should admit that the story they and the Republicans have been flogging is a non-story). At least people don't think it refers to what he did in the Oval office (that is use his amazing work ethic, brain and knowledge to provide outstanding public service (plus a spot on a blue dress)).
The article is important.
Less than one second. That's how long it took Sen. Hillary Rodham Clinton to answer, "Of course not," to Steve Kroft's question on 60 Minutes about whether she thought Sen. Barack Obama was a Muslim. You can time it yourself by watching the clip at YouTube.
Still, that didn't stop MSNBC's Chris Matthews from complaining on-air last week that it took Clinton "the longest time" to answer Kroft's question.
A bit of the transcript after the question was asked (the first time)
CLINTON: Of course not. I mean, that's--you know, there is not basis for that. You know, I take him on the basis of what he says. And, you know, there isn't any reason to doubt that.
KROFT: And you said you'd take Senator Obama at his word that he's not a Muslim.
CLINTON: Right. Right.
The exchange took the longest time, because Kroft wouldn't take no for an answer.
Now are journalists malicious or ignorant ? Since the exchange is easily available, one would assume they were being malicious if one were Boehlert. However, it is possible that they didn't watch the video, that they didn't even read the transcript but that they wrote whole columns on a phrase taken out of context and fed into the rumor mill without checking. Being lazy, I wouldn't put it past them.
Also the dread "as far as I know" is clearly Clinton trying to be polite. If she had just repeated "no" and "of course not", it would suggest that Kroft was being stupid or crazy. It is rude to repeat an answer word for word when a question is rephrased. Of course Kroft was being stupid, crazy and/or malicious.
*So Clinton loses Media
Saturday, March 15, 2008
I would like to object to a particular assertion made by Skidelsky
"On the other hand, it crowds out virtues that have no economic utility, like ... honor." I think that, so long as an apparent sense of honor can't be simulated by the dishonorable, a sense of honor is highly rewarded by the market system. Clearly if we can tell who is honorable, honorable people will enter into mutually beneficial agreements even if they are not effectively enforceable in courts. Thus they would succeed compared to dishonorable people.
Now, in theory, it is possible for someone with no sense of honor to simulate one until the gains from behaving dishonorable outweigh the reputational cost. Such an agent would do better than the genuinely honorable and the openly dishonorable.
Now, on the topic of economists and moral sentiments I invoke Adam Smith. I pretty sure that, if someone had told him about game theory, he would have argued in "The Theory of the Moral Sentiments" that, in the long run, rewarding both true honor and effectively simulated honor will promote true honor as a habit of mind.
Also, I think that the theoretical argument makes unreasonable demands on mere human efforts to be rational (as usual). I think that the vast majority of dishonorable people are not capable of effectively simulating honor (nor are honorable people able to hide that sometimes inconvenient characteristic). Psychologists say we have an amazing ability to detect people who are trying to cheat us. True honorability is favored in a market system full of people with that ability.
A deeper problem is that mores can evolve, so that the requirements of a sense of honor change. Once it was dishonerable in the extreme to charge interest on loans. The new mores under which made it is honorable (indeed which make interest free loans suspect) was selected. If Skidelsky objects to the market because it promotes behavior which he has promised himself he won't do, he objects because it does not favor his ideal of honor, not because it does not favor honor in general.
update: Welcome DeLongians. I must say that I am particularly flattered by a link from Brad to a post which praises "The Theory of the Moral Sentiments" and aims to get on Skidelsky's nerves as Brad excels in both of those fields.
I fear however, that the link also reflects the fact that, exactly because Brad excels in those fields, he has taken the post above to be proof that I am on his team*. Ah that is what I want to write about. Markets don't promote honor compared to what ?
Another alternative form of organization is a hierarchy or chain of command as in the military, the civil service (with less saluting) and large corporations (with executive wash rooms). Another is a traditional village or clan ("It Takes a Village" means "markets aren't enough"). Finally yet another is a team. A bit hierarchical with a coach and in American football a quarterback, but something else.
Why low and behold, we are told that team sports and the military promote honor. Oh really ? My sense is they promote treating people on your team very differently from the other guys. Within the team behave honorably or be cut (or at least shunned). With the other team get away with what you can. Work the refs.
That's not what I call honor. So how do markets promote a more honorable honor ? Well what was Marx writing about when he wrote "everything sacred is profaned. Everything solid melts into air" ? It was about how market relationships are brief and frequently disrupted.
Is that so bad ? That way we don't know which people we will have to ask to trust us in the future. That way, we can't count on saying "sure I lied to x but you are on my team( a fellow employee, a member of my clan ) so why do you care ?" when we ask someone to trust us.
The stable relationships do not promote adherence to universal principles of honorable conduct. One's reputation only matters within the group and only concerns whether one betrays the interests of the group or of its other individual members.
Loyalty to the team is not the sort of honor which I think should be promoted.
*I am.
"On the other hand, it crowds out virtues that have no economic utility, like ... honor." I think that, so long as an apparent sense of honor can't be simulated by the dishonorable, a sense of honor is highly rewarded by the market system. Clearly if we can tell who is honorable, honorable people will enter into mutually beneficial agreements even if they are not effectively enforceable in courts. Thus they would succeed compared to dishonorable people.
Now, in theory, it is possible for someone with no sense of honor to simulate one until the gains from behaving dishonorable outweigh the reputational cost. Such an agent would do better than the genuinely honorable and the openly dishonorable.
Now, on the topic of economists and moral sentiments I invoke Adam Smith. I pretty sure that, if someone had told him about game theory, he would have argued in "The Theory of the Moral Sentiments" that, in the long run, rewarding both true honor and effectively simulated honor will promote true honor as a habit of mind.
Also, I think that the theoretical argument makes unreasonable demands on mere human efforts to be rational (as usual). I think that the vast majority of dishonorable people are not capable of effectively simulating honor (nor are honorable people able to hide that sometimes inconvenient characteristic). Psychologists say we have an amazing ability to detect people who are trying to cheat us. True honorability is favored in a market system full of people with that ability.
A deeper problem is that mores can evolve, so that the requirements of a sense of honor change. Once it was dishonerable in the extreme to charge interest on loans. The new mores under which made it is honorable (indeed which make interest free loans suspect) was selected. If Skidelsky objects to the market because it promotes behavior which he has promised himself he won't do, he objects because it does not favor his ideal of honor, not because it does not favor honor in general.
update: Welcome DeLongians. I must say that I am particularly flattered by a link from Brad to a post which praises "The Theory of the Moral Sentiments" and aims to get on Skidelsky's nerves as Brad excels in both of those fields.
I fear however, that the link also reflects the fact that, exactly because Brad excels in those fields, he has taken the post above to be proof that I am on his team*. Ah that is what I want to write about. Markets don't promote honor compared to what ?
Another alternative form of organization is a hierarchy or chain of command as in the military, the civil service (with less saluting) and large corporations (with executive wash rooms). Another is a traditional village or clan ("It Takes a Village" means "markets aren't enough"). Finally yet another is a team. A bit hierarchical with a coach and in American football a quarterback, but something else.
Why low and behold, we are told that team sports and the military promote honor. Oh really ? My sense is they promote treating people on your team very differently from the other guys. Within the team behave honorably or be cut (or at least shunned). With the other team get away with what you can. Work the refs.
That's not what I call honor. So how do markets promote a more honorable honor ? Well what was Marx writing about when he wrote "everything sacred is profaned. Everything solid melts into air" ? It was about how market relationships are brief and frequently disrupted.
Is that so bad ? That way we don't know which people we will have to ask to trust us in the future. That way, we can't count on saying "sure I lied to x but you are on my team( a fellow employee, a member of my clan ) so why do you care ?" when we ask someone to trust us.
The stable relationships do not promote adherence to universal principles of honorable conduct. One's reputation only matters within the group and only concerns whether one betrays the interests of the group or of its other individual members.
Loyalty to the team is not the sort of honor which I think should be promoted.
*I am.
A Daring Footnote to a History
In Chapter 11 of "The Decline and Fall of the Roman Empire" I find the following footnote
Is anyone going to try to convince me that Gibbon was not amusing himself -- playing with fire by writing something which could be interpreted (if not of the parenthetical commment) as implying that he thinks the gospels are fables and that Jesus Christ may have been an impostor or a fanatic ?
In Chapter 11 of "The Decline and Fall of the Roman Empire" I find the following footnote
1. Apollonius of Tyana was born about the same time as Jesus Christ. His life (that of the former) is related in so fabulous a manner by his disciples, that we are at a loss to discover whether he was a sage, an impostor, or a fanatic.
Is anyone going to try to convince me that Gibbon was not amusing himself -- playing with fire by writing something which could be interpreted (if not of the parenthetical commment) as implying that he thinks the gospels are fables and that Jesus Christ may have been an impostor or a fanatic ?
The Last to Know
Paul Krugman discovers something "when I did learn about it, I also learned that some evil lurks in my heart."
Look, I'm a huge fan of Krugman, but I've known that some evil lurks in his heart since 1993. Notice how late I was. I worked in the same very small place NBER 1050 Mass av for years without finding out, but I was mainly there in the middle of the night. It's those big innocent eyes. How can evil lurk behind those eyes ?
Paul Krugman discovers something "when I did learn about it, I also learned that some evil lurks in my heart."
Look, I'm a huge fan of Krugman, but I've known that some evil lurks in his heart since 1993. Notice how late I was. I worked in the same very small place NBER 1050 Mass av for years without finding out, but I was mainly there in the middle of the night. It's those big innocent eyes. How can evil lurk behind those eyes ?
I don't find it easy to e-mail congresspeople who aren't Michael Capuano
First their web pages do not list an e-mail address. Instead there is a form to e-mail which takes a while to fill out. Second, and much worse, it requires a zip code. In the cases I checked I got an error message, because the zip code was not in the congressperson's district. Ahhh I get it. They want to hear from their constituents. They don't want spam from astroturf outfits (astrospam ?).
OK so how about a web page which gives me the zip codes in a given congressional district. The House web site has the inverse mapping
but when I google "congressional" AND "district" AND "zip" AND "code" I find many opportunities to pay good money for the data (astrospammer prices not concerned citizen prices).
I never understood that that data was valuable.
First their web pages do not list an e-mail address. Instead there is a form to e-mail which takes a while to fill out. Second, and much worse, it requires a zip code. In the cases I checked I got an error message, because the zip code was not in the congressperson's district. Ahhh I get it. They want to hear from their constituents. They don't want spam from astroturf outfits (astrospam ?).
OK so how about a web page which gives me the zip codes in a given congressional district. The House web site has the inverse mapping
but when I google "congressional" AND "district" AND "zip" AND "code" I find many opportunities to pay good money for the data (astrospammer prices not concerned citizen prices).
I never understood that that data was valuable.
Friday, March 14, 2008
Tales of the Weird II
When arguing that it is necessary to protect from lawsuits the telecommunications companies that allowed the government to eavesdrop on their customers without a court's permission Rep. Dan Lungren said
"There is not one iota of evidence that the companies acted inappropriately whatsoever,"
Look there is not one iota of evidence that I improperly wiretapped anyone either (I did't really). Does Congressman Lungren think it is necessary for Congress to grant me retroactive immunity ?
Typically if there is no evidence of a tort, Congress relies on the authority of courts to dismiss frivolous lawsuits and to require the plaintiffs to pay all costs and legal fees.
When arguing that it is necessary to protect from lawsuits the telecommunications companies that allowed the government to eavesdrop on their customers without a court's permission Rep. Dan Lungren said
"There is not one iota of evidence that the companies acted inappropriately whatsoever,"
Look there is not one iota of evidence that I improperly wiretapped anyone either (I did't really). Does Congressman Lungren think it is necessary for Congress to grant me retroactive immunity ?
Typically if there is no evidence of a tort, Congress relies on the authority of courts to dismiss frivolous lawsuits and to require the plaintiffs to pay all costs and legal fees.
Thursday, March 13, 2008
Wish I'd said that
-Tyler Cowen
via Kevin Drum
If I had to guess whether Wikipedia or the median refereed journal article on economics was more likely to be true, after a not so long think I would opt for Wikipedia. This comparison should give us pause.
-Tyler Cowen
via Kevin Drum
Tales of the Weird
I have read several articles by FBI agents about how they don't torture and how they know how to interrogate. These agents suggest that torture is not only wrong but stupid and a sign that other agencies aren't as qualified as they are.
All of that is, perhaps, somewhat called into question by the fact that FBI interrogators decided to duct tape Ali al-Marri's mouth shut while they interrogated him.
Look I understand how hard it is to get someone who can talk to talk, but what the hell is the point of interrogating someone whose mouth is duct taped shut ?
Was the idea to have Mr al-Marri write the answers ? Couldn't he just write the words of the chant ? I mean I know it's tough when you are interrogating and they guy you are interrogating won't let you get a word in edgewise, but, you know, it's not the words the interrogator is saying that are the point of the exercise.
I mean I'm not a superqualified expert on interrogation like the guys who interrogate for the FBI so maybe I'm confused, but I thought the point was what the interrogatee says.
I have read several articles by FBI agents about how they don't torture and how they know how to interrogate. These agents suggest that torture is not only wrong but stupid and a sign that other agencies aren't as qualified as they are.
All of that is, perhaps, somewhat called into question by the fact that FBI interrogators decided to duct tape Ali al-Marri's mouth shut while they interrogated him.
Two government officials said that the tape showed Mr. Marri being manhandled ... the interrogators dispensing the rough treatment on the tape were F.B.I. agents.
[snip]
He said that Mr. Marri was chanting loudly, disrupting his interrogation, and that interrogators used force to put duct tape on his mouth, while Mr. Marri resisted.
Look I understand how hard it is to get someone who can talk to talk, but what the hell is the point of interrogating someone whose mouth is duct taped shut ?
Was the idea to have Mr al-Marri write the answers ? Couldn't he just write the words of the chant ? I mean I know it's tough when you are interrogating and they guy you are interrogating won't let you get a word in edgewise, but, you know, it's not the words the interrogator is saying that are the point of the exercise.
I mean I'm not a superqualified expert on interrogation like the guys who interrogate for the FBI so maybe I'm confused, but I thought the point was what the interrogatee says.
Waldmann vs Waldman
and Glastris.
Paul Glastris wrote
OK time to name some names. Which secular liberals argue that most of the founding fathers were agnostics, or, at "best"* deists. I have never heard or read that claim.
First, all discussion on the issue by agnostic or atheist liberals (or radicals) concerned Franklin or Jefferson. The founding fathers numbered more than 3 so 2 of them were not most of them. When I read Waldman's second post on the topic, I learned that John Adams died a unitarian. I had assumed he lived and died a congregationalist (I don't count Waldman as an atheist or a liberal).
Second I have never heard or read the suggestion that Jefferson and Franklin were agnostic. I have heard the claim that they were deists. For example, I quote Waldman from the tpmcafé discussion to which Glastris linked "Actually I quote that passage in my book ... the same Franklin who said he was a Deist." Yes, note that Waldman agrees that Franklin said that he was a deist. Odd where those silly secular liberals get their crazy notions. Waldman goes on to note that, by the time of the constitutional convention he quite clearly stated that he had changed his mind. Thus Waldman's argument is that the founding father Franklin (unlike the younger Franklin) was not a deist.
Now, before criticizing a guest at Kevin "always click the link" Drum, I should have clicked alll the links. Still I think that your summary creates a false balance between the clearly false claims of the religious right and a claim by a tiny minority of secular liberals and, indeed, tiny minorities of secular radicals and of atheist liberals.
I am now going to quote and challenge Waldman from the tpmcafè and then, if I get to it, his first post on the topic at ???
"But – and I know I’ll be accused of unmanly centrism on this – there’s a non-sequitor sometimes offered by progressives, too. Some argue that because many of the Founders wanted separation of church and state, they therefore must have been either irreligious or wanted less role for religion in American life."
I don't accuse him of "unmanly"* centrism. For one thing, I don't consider "unmanly" to be a pejorative. I accuse him of dishonest use of the rhetoric of centrism. He is linking his clearly accurate characterization of named conservative Christians who claim that the founders (without distinction) were conservative Christians with his claim that "some"* say that people who believe in separation of church and state must be "irreligious"*. If people fell for rhetorical tricks, the overwhelming evidence for the first claim would trick them into accepting the second different claim. From the comments I have read at the café it doesn't seem to have worked.
I note, in passing, that Waldman's accusations against secular liberals vary over time. I recall the post which I have read criticizing the claim that the founders were deists. Here he criticizes the claim that the founders were irreligious. These are very different claims. I haven't actually heard or read any claim that any founder was irreligious. Of course Waldman has studied the question at length and I'm sure he cites such claims in his book. I note my ignorance only to argue that they are rare. I think that the some who claim that the founders were irreligious are few and clearly flaky. As to the argument that a belief in separation of church in state implies irreligion, I find it hard to imagine any supporter of separation of church and state who has that idea. I'm sure such people exist. I don't think I have met one. I think they are very very rare. I am going to check how long I have to read Waldman, before I get to the name of someone who argues that the founders were irreligious or that most of them were deists. I will also look for the name of a supporter of separation of church and state who thinks that if one supports separation of church and state, then one must be irreligious, but I don't expect to find it.
Oh now I'm pissed. Commenter redshift wrote "However, I've seen no evidence that Mr. Waldmann bothered to find out anything more about the beliefs of 18th-century Unitarians than he did about 18th-century Deists before he started throwing around assertions about them.
Posted by Redshift." Yeah well RReddshiftt how about you readd before you question someone else's scholarshipp ? "Mr Waldmann" is named Waldman nott Waldmann !
OK here we are
My apologies for not responding in the comment area earlier. I actually had the mistaken understanding that I wasnt supposed to go into the comment area. My mistake.
OK didn't take long
That's a name. And not someone ranting to himself. It is not clear to me if Allen is saying that the founders were irreligious or that they weren't devout. So one name so far and it is still very early.
Here is a specific claim "It's a fair question. I know that on a personal level I heard "the Founders were all Deists" from my liberal friends about as much as I hear "the Founders were all Christians" from my conservative friends." No names here, but if he is talking about personal conversations with friends he shouldn't name names. His experience differs from mine (to put it mildly). I have heard neither argument from my friends.
There is also direct evidence from the comments thread. Redshift again "Progressives aren't trying to prove all the founders weren't religious (though that's certainly true for some of them)" If the them refers to founders not progressives, then Redshift has made a claim which I read there for the first time.
On the other hand, all of the evidence on the frequency of the view that "because many of the Founders wanted separation of church and state, they therefore must have been either irreligious or ..." is negative. Commenter after commenter stresses that the religious beliefs of the founders are irrelevant to any question about separation of church and state.
Brilliant commenter nindid suspects that Waldman is making "an attempt to mollify conservatives who hate nothing more than the liberal bias of facts. If they hear that progressives got it equally wrong - reality be damned - they might listen a bit more." I think that might be exactly right (although I can read Waldman's mind no more than nindid can).
OK so much for the other hand. amcarey writes "many of the Founders WERE Deists and that this fact DOES strengthen the case for separation of church and state, particularly from the perspective of an originalist. (I'm not one," evidently even though he is not an originalist, he thinks that the deism of many of the founders strengthens the case for spearation at least a little although not as much as it would if original intent were a valid interperative tool. I don't see how it could matter at all to a non originalist. I deduce a sloppy use of "particularly". Plust TOO MANY CAPS.
In the unlikely case that anyone has read this far, this is the end of Waldmann vs Waldman. I liked the title, but there is no need for someone else to pile on Waldman. He is getting torn to pieces over there. His few defenders don't make arguments about the content of his claim but just insult his critics or, in one case, suggest that maybe he will write quite different things in the near future so people should wait before criticizing what he wrote so far.
*The quotation marks are honest to God quotation marks as I am quoting Glastris. Heaven forfend that I use "scare quotes."
and Glastris.
Paul Glastris wrote
WERE THE FOUNDING FATHERS REALLY CHRISTIANS?....Religious conservatives have long insisted that the Framers were deeply and traditionally Christian, an assertion central to their contention that America was founded as a "Christian nation." Secular liberals, by contrast, have long argued that most of the Founders were agnostics or, at best, Deists who believed that reason, not scripture, is the true path to understanding the Almighty.
OK time to name some names. Which secular liberals argue that most of the founding fathers were agnostics, or, at "best"* deists. I have never heard or read that claim.
First, all discussion on the issue by agnostic or atheist liberals (or radicals) concerned Franklin or Jefferson. The founding fathers numbered more than 3 so 2 of them were not most of them. When I read Waldman's second post on the topic, I learned that John Adams died a unitarian. I had assumed he lived and died a congregationalist (I don't count Waldman as an atheist or a liberal).
Second I have never heard or read the suggestion that Jefferson and Franklin were agnostic. I have heard the claim that they were deists. For example, I quote Waldman from the tpmcafé discussion to which Glastris linked "Actually I quote that passage in my book ... the same Franklin who said he was a Deist." Yes, note that Waldman agrees that Franklin said that he was a deist. Odd where those silly secular liberals get their crazy notions. Waldman goes on to note that, by the time of the constitutional convention he quite clearly stated that he had changed his mind. Thus Waldman's argument is that the founding father Franklin (unlike the younger Franklin) was not a deist.
Now, before criticizing a guest at Kevin "always click the link" Drum, I should have clicked alll the links. Still I think that your summary creates a false balance between the clearly false claims of the religious right and a claim by a tiny minority of secular liberals and, indeed, tiny minorities of secular radicals and of atheist liberals.
I am now going to quote and challenge Waldman from the tpmcafè and then, if I get to it, his first post on the topic at ???
"But – and I know I’ll be accused of unmanly centrism on this – there’s a non-sequitor sometimes offered by progressives, too. Some argue that because many of the Founders wanted separation of church and state, they therefore must have been either irreligious or wanted less role for religion in American life."
I don't accuse him of "unmanly"* centrism. For one thing, I don't consider "unmanly" to be a pejorative. I accuse him of dishonest use of the rhetoric of centrism. He is linking his clearly accurate characterization of named conservative Christians who claim that the founders (without distinction) were conservative Christians with his claim that "some"* say that people who believe in separation of church and state must be "irreligious"*. If people fell for rhetorical tricks, the overwhelming evidence for the first claim would trick them into accepting the second different claim. From the comments I have read at the café it doesn't seem to have worked.
I note, in passing, that Waldman's accusations against secular liberals vary over time. I recall the post which I have read criticizing the claim that the founders were deists. Here he criticizes the claim that the founders were irreligious. These are very different claims. I haven't actually heard or read any claim that any founder was irreligious. Of course Waldman has studied the question at length and I'm sure he cites such claims in his book. I note my ignorance only to argue that they are rare. I think that the some who claim that the founders were irreligious are few and clearly flaky. As to the argument that a belief in separation of church in state implies irreligion, I find it hard to imagine any supporter of separation of church and state who has that idea. I'm sure such people exist. I don't think I have met one. I think they are very very rare. I am going to check how long I have to read Waldman, before I get to the name of someone who argues that the founders were irreligious or that most of them were deists. I will also look for the name of a supporter of separation of church and state who thinks that if one supports separation of church and state, then one must be irreligious, but I don't expect to find it.
Oh now I'm pissed. Commenter redshift wrote "However, I've seen no evidence that Mr. Waldmann bothered to find out anything more about the beliefs of 18th-century Unitarians than he did about 18th-century Deists before he started throwing around assertions about them.
Posted by Redshift." Yeah well RReddshiftt how about you readd before you question someone else's scholarshipp ? "Mr Waldmann" is named Waldman nott Waldmann !
OK here we are
My apologies for not responding in the comment area earlier. I actually had the mistaken understanding that I wasnt supposed to go into the comment area. My mistake.
OK didn't take long
Obviously there's a lot more detail in the book, but here's just one example: In The Nation magazine, Brooke Allen declares that "the Founding Fathers were not religious men." As for Washington, she writes, "Religion seems to have played a remarkably small role in his own life."
I believe both of those statements are wrong. But I also dont think their inaccuracy undermines the fundamental case /for/ separation of church and state.
That's a name. And not someone ranting to himself. It is not clear to me if Allen is saying that the founders were irreligious or that they weren't devout. So one name so far and it is still very early.
Here is a specific claim "It's a fair question. I know that on a personal level I heard "the Founders were all Deists" from my liberal friends about as much as I hear "the Founders were all Christians" from my conservative friends." No names here, but if he is talking about personal conversations with friends he shouldn't name names. His experience differs from mine (to put it mildly). I have heard neither argument from my friends.
There is also direct evidence from the comments thread. Redshift again "Progressives aren't trying to prove all the founders weren't religious (though that's certainly true for some of them)" If the them refers to founders not progressives, then Redshift has made a claim which I read there for the first time.
On the other hand, all of the evidence on the frequency of the view that "because many of the Founders wanted separation of church and state, they therefore must have been either irreligious or ..." is negative. Commenter after commenter stresses that the religious beliefs of the founders are irrelevant to any question about separation of church and state.
Brilliant commenter nindid suspects that Waldman is making "an attempt to mollify conservatives who hate nothing more than the liberal bias of facts. If they hear that progressives got it equally wrong - reality be damned - they might listen a bit more." I think that might be exactly right (although I can read Waldman's mind no more than nindid can).
OK so much for the other hand. amcarey writes "many of the Founders WERE Deists and that this fact DOES strengthen the case for separation of church and state, particularly from the perspective of an originalist. (I'm not one," evidently even though he is not an originalist, he thinks that the deism of many of the founders strengthens the case for spearation at least a little although not as much as it would if original intent were a valid interperative tool. I don't see how it could matter at all to a non originalist. I deduce a sloppy use of "particularly". Plust TOO MANY CAPS.
In the unlikely case that anyone has read this far, this is the end of Waldmann vs Waldman. I liked the title, but there is no need for someone else to pile on Waldman. He is getting torn to pieces over there. His few defenders don't make arguments about the content of his claim but just insult his critics or, in one case, suggest that maybe he will write quite different things in the near future so people should wait before criticizing what he wrote so far.
*The quotation marks are honest to God quotation marks as I am quoting Glastris. Heaven forfend that I use "scare quotes."