Control?: The Double Blind, Random Gontrol Group Method

For discussing anything related to physics, biology, chemistry, mathematics, and their practical applications.

Moderator: Flannel Jesus

Re: Control?: The Double Blind, Random Gontrol Group Method

Postby Flannel Jesus » Thu Apr 17, 2014 6:50 am

Considering the huge backlog of things I already have to read, I don't really have time to take suggestions on science literature from someone who lacks understanding to the degree that you do.

Choice quotes that show your level, curated by me:
the assumption of the DBRCGM is that the larger n, the more assurance we can derive from our comparison test. Yet, the larger n becomes, by my "complexity argument," not only does our hoped for control over hidden and confounding variables decrease, but in all likelihood we increase, and again in all likelihood exponentially, the amount of hidden and confounding variables present in what we are trying to control and measure

the phrase "random assignment" seems contradictory. If "control" is the end, how can randomization be the means? Randomness implies no (or little) controI. The same issue arises with the notion of assignment. If assigning is happening, then the alleged randomness is compromised, if not destroyed, again defeating the purpose of control.

And then to top it off you end with some nonsense about quantum physics!

I will need to be really bored before I'm interested in working my way through your reading list.
User avatar
Flannel Jesus
For Your Health
 
Posts: 5158
Joined: Thu Mar 31, 2011 11:32 pm

Re: Control?: The Double Blind, Random Gontrol Group Method

Postby landis » Fri Apr 18, 2014 12:42 am

Flannel Jesus: I'm genuinely interested in free exchange of ideas, but you seem to be more interested in painting a caricature of me. While you haven't outright violated your own rule (# 3) of this sub-forum, your last three or four responses come very close, IMHO, to insults. I think you own me an apology. How could you know so much about my "lack of understanding" when you barely even listened to me or hardly tried to hear where I'm coming from.
"Questions of reality are too important to be left to the scientists."
-Paul Feyerabend, The Tyranny of Science (p. 51; Polity: 2012).
"Arguments seldom make converts in matters philosophical."
-William James, Principles (p. 468, v. 1)
"Argument is propaganda for one observer, the essence of human discourse for another."
-Feyerabend, Against Method (p. 236; Verso: 2010)
Blog: http://thislandismylandis.wordpress.com/
landis
 
Posts: 36
Joined: Wed Mar 12, 2014 10:24 pm

Re: Control?: The Double Blind, Random Gontrol Group Method

Postby Flannel Jesus » Fri Apr 18, 2014 9:56 pm

When you start out by saying that random assignment is a bad thing, and large sample sizes are a bad thing...there's nothing to think but 'geeze, this guy really doesn't get it'.

You later back tracked and said that I misunderstood you, that you don't think they're a bad thing. But your language in the OP is pretty clear. It's clear to me that someone who says "First, the phrase "random assignment" seems contradictory. If "control" is the end, how can randomization be the means? Randomness implies no (or little) controI. The same issue arises with the notion of assignment. If assigning is happening, then the alleged randomness is compromised, if not destroyed, again defeating the purpose of control" doesn't understand even the basics of why double-blind studies are seen as valuable.

I mean, you completely miss the entire concept of 'assigning'; just completely. Assigning MUST happen in a study comparing eg a drug to eg a placebo. Some people MUST be assigned one and others assigned the other. You can't have a trial comparing the two without assigning different people one of the two. What in the world is the alternative? And how does it compromise the randomness? That is the very implementation of the randomness! 'Randomness' refers to how people are assigned to treatment or placebo.

In addition to completely misunderstanding what the word 'assign' means in this context, you've got 'control' pretty far off too. The control group in a randomized trial refers to the group that takes the placebo (usually; there are other potential control groups). Randomizing who gets put in the control group does not contradict the end of "control". You've just misunderstood what "control" means.

Your OP was just so so so far off the mark, and backtracking is sorta good in a way when you're that far off, but the sort of backtracking you need to do is much more extreme than what you have done; instead of just saying 'no no you misunderstood', I think it's really time you just erased everything from the chalkboard, said 'I've got something wrong here' and start with a clean slate.
User avatar
Flannel Jesus
For Your Health
 
Posts: 5158
Joined: Thu Mar 31, 2011 11:32 pm

Re: Control?: The Double Blind, Random Gontrol Group Method

Postby landis » Sun Apr 27, 2014 5:05 am

You're not going to apologize, are you? Oh well. I still think I deserve it.

Flannel Jesus wrote:...random assignment is a bad thing...
Never said that.
Flannel Jesus wrote:...and large sample sizes are a bad
Never said that either.
Flannel Jesus wrote:there's nothing to think but 'geeze, this guy really doesn't get it'.
There are plenty of other things to think. I'm guessing you can't see that because you're more interested in caricaturizing my exchanges than understanding their poignancy.

Flannel Jesus wrote:You later back tracked and said that I misunderstood you, that you don't think they're a bad thing. But your language in the OP is pretty clear. It's clear to me that someone who says "First, the phrase "random assignment" seems contradictory. If "control" is the end, how can randomization be the means?
These are questions, not claims. Examining assumptions, clarification, and free idea exchanges are a far cry from "backtracking." For instance, I used the word "seem" to avoid exactly your misunderstanding.
Flannel Jesus wrote:Randomness implies no (or little) controI. The same issue arises with the notion of assignment. If assigning is happening, then the alleged randomness is compromised, if not destroyed, again defeating the purpose of control" doesn't understand even the basics of why double-blind studies are seen as valuable.
Your conflating the obvious limits of DBRCG studies with your caricature that I don't understand the basics. I couldn't even pose the OP without a basic understanding, and in previous posts I've explained how and why my understanding is advanced. You seem annoyed that I don't share your unexamined assumptions, which apparently has blinded you from the significance of my content.

Flannel Jesus wrote:I mean, you completely miss the entire concept of 'assigning'; just completely.
More substanceless rhetoric. I heard you the first time you said you disagreed, but you've yet to exchange an idea with me about it.
Flannel Jesus wrote:Assigning MUST happen in a study comparing eg a drug to eg a placebo. Some people MUST be assigned one and others assigned the other. You can't have a trial comparing the two without assigning different people one of the two.
Three unexamined assumptions in a row. There's no "MUSTs" or "cant's" about it, a reflection of your unwillingness to think creatively about scientific puzzles.
Flannel Jesus wrote:What in the world is the alternative?
This is precisely the problem. You're so indoctrinated you can't imagine alternatives. Are you one of those people who believe science is progressive?
Flannel Jesus wrote:And how does it compromise the randomness? That is the very implementation of the randomness! 'Randomness' refers to how people are assigned to treatment or placebo.
It's obvious that "assignment" and "randomness" are contradictory notions. What about this don't you understand?

Flannel Jesus wrote:In addition to completely misunderstanding what the word 'assign' means in this context, you've got 'control' pretty far off too. The control group in a randomized trial refers to the group that takes the placebo (usually; there are other potential control groups). Randomizing who gets put in the control group does not contradict the end of "control". You've just misunderstood what "control" means.
You've bought into the illusion. It's a common mistake, but forgivable. You don't need to keep restating the basics. I understand them thoroughly. Familiarize yourself with the "double pendulum" phenomenon and "the butterfly effect." Individuals are complex systems, groups even more so.

Flannel Jesus wrote:Your OP was just so so so far off the mark, and backtracking is sorta good in a way when you're that far off, but the sort of backtracking you need to do is much more extreme than what you have done; instead of just saying 'no no you misunderstood', I think it's really time you just erased everything from the chalkboard, said 'I've got something wrong here' and start with a clean slate.
You've confused my clarifications and good faith attempts at free exchanges of ideas with "backtracking." I guess you're too invested in your misunderstanding of me to see that.

I will engage with the others in this thread who have shown they can refrain from caricature and demonstrated they're willing to examine assumptions via a free exchange of ideas, but unless you apologize and start acting civil about it, I'm done interacting with you.
"Questions of reality are too important to be left to the scientists."
-Paul Feyerabend, The Tyranny of Science (p. 51; Polity: 2012).
"Arguments seldom make converts in matters philosophical."
-William James, Principles (p. 468, v. 1)
"Argument is propaganda for one observer, the essence of human discourse for another."
-Feyerabend, Against Method (p. 236; Verso: 2010)
Blog: http://thislandismylandis.wordpress.com/
landis
 
Posts: 36
Joined: Wed Mar 12, 2014 10:24 pm

Re: Control?: The Double Blind, Random Gontrol Group Method

Postby James S Saint » Sun Apr 27, 2014 6:12 am

Flannel Jesus wrote:When you start out by saying that random assignment is a bad thing, and large sample sizes are a bad thing...there's nothing to think but 'geeze, this guy really doesn't get it'.

That is the same impression that I got.

Perhaps it needed to be worded very differently if there was a different intention.
Clarify, Verify, Instill, and Reinforce the Perception of Hopes and Threats unto Anentropic Harmony :)
Else
From THIS age of sleep, Homo-sapien shall never awake.

The Wise gather together to help one another in EVERY aspect of living.

You are always more insecure than you think, just not by what you think.
The only absolute certainty is formed by the absolute lack of alternatives.
It is not merely "do what works", but "to accomplish what purpose in what time frame at what cost".
As long as the authority is secretive, the population will be subjugated.

Amid the lack of certainty, put faith in the wiser to believe.
Devil's Motto: Make it look good, safe, innocent, and wise.. until it is too late to choose otherwise.

The Real God ≡ The reason/cause for the Universe being what it is = "The situation cannot be what it is and also remain as it is".
.
James S Saint
ILP Legend
 
Posts: 25041
Joined: Sun Apr 18, 2010 8:05 pm

Re: Control?: The Double Blind, Random Gontrol Group Method

Postby landis » Fri May 02, 2014 9:38 pm

James S Saint wrote:
Flannel Jesus wrote:When you start out by saying that random assignment is a bad thing, and large sample sizes are a bad thing...there's nothing to think but 'geeze, this guy really doesn't get it'.

That is the same impression that I got.

Perhaps it needed to be worded very differently if there was a different intention.

Re-worded? Absolutely. After all this commotion (including the same post on another philosophy forum) I see that I've not explained myself very well at all. I've given the impression to you, Flannel Jesus, et al, that I want to throw the DBRCGM baby out with the DBRCGM bath water, which I don't. I regret that. Please accept my apology.
"Questions of reality are too important to be left to the scientists."
-Paul Feyerabend, The Tyranny of Science (p. 51; Polity: 2012).
"Arguments seldom make converts in matters philosophical."
-William James, Principles (p. 468, v. 1)
"Argument is propaganda for one observer, the essence of human discourse for another."
-Feyerabend, Against Method (p. 236; Verso: 2010)
Blog: http://thislandismylandis.wordpress.com/
landis
 
Posts: 36
Joined: Wed Mar 12, 2014 10:24 pm

Re: Control?: The Double Blind, Random Gontrol Group Method

Postby landis » Wed Jan 21, 2015 11:01 pm

Why There's No Cause to Randomize by John Worrall, professor of Philosophy, Logic & Scientific Method at the London School of Economics
"Questions of reality are too important to be left to the scientists."
-Paul Feyerabend, The Tyranny of Science (p. 51; Polity: 2012).
"Arguments seldom make converts in matters philosophical."
-William James, Principles (p. 468, v. 1)
"Argument is propaganda for one observer, the essence of human discourse for another."
-Feyerabend, Against Method (p. 236; Verso: 2010)
Blog: http://thislandismylandis.wordpress.com/
landis
 
Posts: 36
Joined: Wed Mar 12, 2014 10:24 pm

Re: Control?: The Double Blind, Random Gontrol Group Method

Postby landis » Fri Jun 26, 2015 12:06 am

1) What Evidence? In Evidence-Based Medicine by John Worral, LSE Professor of Philosophy, Logic & Scientific Method

2)
high quality supply of random numbers...It's surprisingly difficult to generate a truly pure series of random numbers. While it's easy enough to come up with an algorithm that produces a series of numbers that appear random, there are usually hidden flaws that can cause some digits or series to predictably recur. In addition, some sources of "random" numbers are not random at all, but have been deliberately massaged to ensure an artificial degree of homogeneity. All this is a big deal in the world of applied math.
-http://bleedingedge.pynchonwiki.com/wiki/index.php?title=Chapter_31, Commentary to Bleeding Edge, p. 341

3)
Between the wish and the thing the world lies waiting…. When I was in school I studied biology. I learned that in making their experiments scientists will take some group--bacteria, mice, people--and subject that group to certain conditions. They compare the results with a second group which has not been disturbed. This second group is called the control group. It is the control group which enables the scientist to gauge the effect of his experiment. To judge the significance of what has occurred. In history there are no control groups. There is no one tell us what might have been. We weep over what might have been, but there is no might have been. There never was. It is supposed to be true that those who do not know history are condemned to repeat it. I don’t believe knowing can save us. What is constant in history is greed and foolishness and a love of blood and this is a thing that even God--who knows all that can be known--seems powerless to change.

-Duena Alfosno in Cormac McCarthy's All the Pretty Horses (pp. 238-9. Everyman: '99. My italics.)
"Questions of reality are too important to be left to the scientists."
-Paul Feyerabend, The Tyranny of Science (p. 51; Polity: 2012).
"Arguments seldom make converts in matters philosophical."
-William James, Principles (p. 468, v. 1)
"Argument is propaganda for one observer, the essence of human discourse for another."
-Feyerabend, Against Method (p. 236; Verso: 2010)
Blog: http://thislandismylandis.wordpress.com/
landis
 
Posts: 36
Joined: Wed Mar 12, 2014 10:24 pm

Re: Control?: The Double Blind, Random Gontrol Group Method

Postby landis » Thu Jul 30, 2015 5:11 am

See also: The Crisis of Confidence in Medical Research by Allen Frances, Professor Emeritus, Duke University (he was also the DSM-IV Task Force Chair and famously also said, "There is no definition of a mental disorder. It’s bullshit. I mean, you just can’t define it" [http://www.wired.com/2010/12/ff_dsmv/]).
"Questions of reality are too important to be left to the scientists."
-Paul Feyerabend, The Tyranny of Science (p. 51; Polity: 2012).
"Arguments seldom make converts in matters philosophical."
-William James, Principles (p. 468, v. 1)
"Argument is propaganda for one observer, the essence of human discourse for another."
-Feyerabend, Against Method (p. 236; Verso: 2010)
Blog: http://thislandismylandis.wordpress.com/
landis
 
Posts: 36
Joined: Wed Mar 12, 2014 10:24 pm

Re: Control?: The Double Blind, Random Gontrol Group Method

Postby landis » Wed Nov 25, 2015 3:56 am

Listen to philosopher Nigel Warburton interview John Worrall, professor of philosophy of science at the London School of Economics, on RTCs and Evidence-Based Medicine at Philosophy Bites.

Worrall, Philosophy Bites
"Questions of reality are too important to be left to the scientists."
-Paul Feyerabend, The Tyranny of Science (p. 51; Polity: 2012).
"Arguments seldom make converts in matters philosophical."
-William James, Principles (p. 468, v. 1)
"Argument is propaganda for one observer, the essence of human discourse for another."
-Feyerabend, Against Method (p. 236; Verso: 2010)
Blog: http://thislandismylandis.wordpress.com/
landis
 
Posts: 36
Joined: Wed Mar 12, 2014 10:24 pm

Re: Control?: The Double Blind, Random Gontrol Group Method

Postby Moreno » Wed Nov 25, 2015 11:00 am

landis wrote:See also: The Crisis of Confidence in Medical Research by Allen Frances, Professor Emeritus, Duke University (he was also the DSM-IV Task Force Chair and famously also said, "There is no definition of a mental disorder. It’s bullshit. I mean, you just can’t define it" [http://www.wired.com/2010/12/ff_dsmv/]).

Though the problem here is not double blind tests, but more on the ontological level and in terms of cause, effect, etc.
User avatar
Moreno
ILP Legend
 
Posts: 10305
Joined: Sat Nov 24, 2007 5:46 pm

Re: Control?: The Double Blind, Random Gontrol Group Method

Postby landis » Wed Nov 09, 2016 10:34 pm

Moreno wrote:
landis wrote:See also: The Crisis of Confidence in Medical Research by Allen Frances, Professor Emeritus, Duke University (he was also the DSM-IV Task Force Chair and famously also said, "There is no definition of a mental disorder. It’s bullshit. I mean, you just can’t define it" [http://www.wired.com/2010/12/ff_dsmv/]).

Though the problem here is not double blind tests, but more on the ontological level and in terms of cause, effect, etc.
True, but too much has been made of the correlations versus causes distinction (although it still has its uses). Correlations are probabilistic indicators of causality (positive correlations) or non-causality (negative correlations). In other words, causality and correlation are not mutually exclusive.

I also disagree with Frances that all anecdotal evidence is untrustworthy. Indeed, most studies are designed to answer questions that arise from anecdotal observations; anecdotes are the incubators for research questions and as such are a necessary part of the investigational process.
"Questions of reality are too important to be left to the scientists."
-Paul Feyerabend, The Tyranny of Science (p. 51; Polity: 2012).
"Arguments seldom make converts in matters philosophical."
-William James, Principles (p. 468, v. 1)
"Argument is propaganda for one observer, the essence of human discourse for another."
-Feyerabend, Against Method (p. 236; Verso: 2010)
Blog: http://thislandismylandis.wordpress.com/
landis
 
Posts: 36
Joined: Wed Mar 12, 2014 10:24 pm

Re: Control?: The Double Blind, Random Gontrol Group Method

Postby landis » Wed Nov 09, 2016 11:36 pm

The link to the article by John Worral, LSE Professor of Philosophy, Logic & Scientific Method titled "What Evidence? In Evidence-Based Medicine" in my previous post in this thread is now wrongly directing to a cocaine article. This will take you to right the right place: What Evidence? In Evidence-Based Medicine.
This is especially important to my illusion of control point. As Worral puts it:

There are...indefinitely many possible alternative causal factors to the characteristic features of the intervention under test. But “background knowledge” indicates which of these are plausible alternatives. It is difficult to see how we can do better than control—whether in advance or post hoc—for all plausible alternatives. The idea that randomization controls all at once for known and unknown factors (or even that it “tends” to do so) is a will-o’-the-wisp. The only solid argument for randomization appears to be that standard means of implementing it have the side-effect of blinding the clinical experimenter and hence controlling for a known factor—selection bias. But if selection bias can be eliminated or controlled for in some other way then randomization is inessential (p. 18; my bolds).
"Questions of reality are too important to be left to the scientists."
-Paul Feyerabend, The Tyranny of Science (p. 51; Polity: 2012).
"Arguments seldom make converts in matters philosophical."
-William James, Principles (p. 468, v. 1)
"Argument is propaganda for one observer, the essence of human discourse for another."
-Feyerabend, Against Method (p. 236; Verso: 2010)
Blog: http://thislandismylandis.wordpress.com/
landis
 
Posts: 36
Joined: Wed Mar 12, 2014 10:24 pm

Re: Control?: The Double Blind, Random Gontrol Group Method

Postby landis » Thu Nov 10, 2016 12:08 am

Dr. H. Gilbert Welch--physician, renowned epidemiologist, professor at Dartmouth Medical School, nationally recognized expert on the effects of medical testing and author of the highly acclaimed book Overdiagnosed: Making People Sick in the Pursuit of Health (Beacon, 2012)--is a big fan of the DBRCGM. Yet in his latest book Less Medicine, More Health: 7 Assumptions That Drive Too Much Medical Care (Beacon, 2015), he admits it has great limitations (pp. 46-8).

First he cites what he calls a "wonderful parody of evidence based medicine" from the British Medical Journal titled, "Parachute use to prevent death and major trauma related to gravitational challenge: systematic review of randomised controlled trials" (summary here; full report here). Then he says:

The evidence-based critics have a point: randomized trials are not always possible. We can't randomize people to receive what we believe might be harmful exposure, like jumping out of airplane without a parachute or taking up the habit of cigarette smoking" [recall/cf. the citation my last post from by John Worral, LSE Professor of Philosophy, Logic & Scientific Method, who contends that, "Ethical issues are often inextricably intertwined with methodological issues..." (p. 19)]. And the more sophisticated critics would be quick to point out three other limitations of randomized trials.

First, they may focus on the wrong comparison. There have been a number of randomized trials of catheter ablation in atrial fibrillation. The comparison group typically consists of patients taking antiarrythmic drugs. Why would I be interested in the comparison? We already know that antiarrythmic drugs don't work that well in atrial fibrillation--and they have loads of side effects (including death). What I want to know is how ablation compares to the standard treatment of rate control and anticoagulation.

Second, they may focus on the wrong outcome. The typical outcome measure for randomized trials of catheter ablation in atrial fibrillation is whether the patients still have arrhythmia or not--as determined by EKG.... But patients don't care about what their EKG looks like, they care about how they feel....

Third, they may focus on the wrong patients. One of the pharmaceutical industry's favorite strategies is to study the effect of a drug on the few patients who have severe disease, find some benefit, and then hope the doctors extrapolate the benefit to the many patients with less severe forms of the disease. It's a clever strategy: it's like testing parachutes on the few people who jump out of airplanes and then selling them as protection against falls to the many people who walk downstairs. Severely ill patients always stand to benefit more from intervention than those who are less severely ill (it's all about baseline risk....). Yet the harms of intervention are roughly equivalent in the two groups. So the net effect of intervention regularly looks better in the severely ill. To be sure, the problem can work in the reverse direction: if a trial focuses on patients who are only moderately ill, and then finds no benefit, it may have missed an important effect among the severely ill.

Finally, the results of randomized trials reflect the typical patient that is enrolled in the study..., the average effect of intervention. Some individuals may do much better; others may do much worse (pp. 47-9; my bolds).
"Questions of reality are too important to be left to the scientists."
-Paul Feyerabend, The Tyranny of Science (p. 51; Polity: 2012).
"Arguments seldom make converts in matters philosophical."
-William James, Principles (p. 468, v. 1)
"Argument is propaganda for one observer, the essence of human discourse for another."
-Feyerabend, Against Method (p. 236; Verso: 2010)
Blog: http://thislandismylandis.wordpress.com/
landis
 
Posts: 36
Joined: Wed Mar 12, 2014 10:24 pm

Previous

Return to Science, Technology, and Math



Who is online

Users browsing this forum: No registered users