Talk:Blinded experiment

Untitled
Does anyone know who carried out the first true double blind trial and thus invented the methodology?

Noel jackson. LIFE.Newcastle


 * Suggest merging Blinding (medicine) into this topic. I also want to add references to the scientific method,  social sciences, psychology, and the physical sciences.  Blinding is not just important in medical clinical trials, its important in all sciences.  I had considered creating a topic Blinding (scientific method) and merging both into it, but Double-blind is already a pretty good topic and its the term most people would use when searching for information. Thatcher131 07:53, 11 February 2006 (UTC)


 * I did a rewrite incorporating the medical-specific material from Blinding (medicine) in preparation to an official merge. I added a bit about single-blinded studies which is redundant to the single blind entry but I think its useful to have it here in case someone comes from a link and doesn't know what blinded research is.  I also wrote it generically to cover all research as blinding is not important just in medical research.  I added one non-medical example and would love to see more non-medical examples. Thatcher131 04:45, 12 February 2006 (UTC)

controversy
Is there no controversy or questioning of the validity of blinded studies? What about human subjects being able to guess if they were in the active drug group, and that affecting the results? This is a big enough issue that the statistics of some studies are recalculated or seriously questioned... There must be some criticisms of blinded studies..? --159.178.248.49 (talk) 13:18, 15 March 2010 (UTC)

I agree. I intend to make a project of this issue and how it relates to the Active placebo effect, as even a small error in blinding can trigger a statistically significant result in a large enough study. This topic needs more attention. Wikiman2718 (talk) 08:25, 19 April 2019 (UTC)

Placebo & mice
For the risk of being laughed at. Is there any proof that mice experiments are not influenced by placebo effect. I can imagine the lab staff talking to the doomed mice "here you buggers, eat this, you will all die soon anyhow." and to the lucky mice "look over to you neighbors, they're gonna die soon, you are the lucky ones". And maybe the mice understand ? If not word for word, but maybe body language.

Is double blind method being used with mice experiments ?

-- Paparodo, dec 13th 2006 —The preceding unsigned comment was added by 217.10.60.85 (talk) 11:46, 13 December 2006 (UTC).

Double placebo numbers
Does anyone know if there was a double-blind study where BOTH groups were given a placebo, but both were told that one group was getting a med and one was not? Did anyone measure what is the % of success of placebo group that is always reported by a placebo group in a double-blind study.

What I am looking for is a percent of success reported from the placebo group?

Was there such study?

Can someone post that?

Thanks.

Atessitore05 13:50, 13 December 2006 (UTC)


 * I doubt there's any single number. Successful outcomes in either group (control and test) can potentially range anywhere from none to all. Given the variety of therapies being tested, and the conditions being studied, would you expect success to even be defined the same way, much less measured at a consistent rate? At any rate, the article on the placebo effect is probably a good place for you to look for more information. In fact I found this there: "...studies generally do not include an untreated group, so determining the actual size of the placebo effect, compared to totally untreated patients, is difficult...", and, again, not likely to yield similar results across dissimilar studies. 198.49.180.40 22:05, 5 April 2007 (UTC)

If BOTH groups were given placebo, the result should be 1:1 for both groups with an error within a statistical expectation...

The placebo group is group with no real treatment applied - so the possible success of this "treatment" is caused by placebo effect - either people do feel like the "treatment" helps them even if it does not help (important only if results are interpreted based on subjective reports - usual in psychology) or the believe that the treatment should help them really does help them (assuming psychological setup can really affect course of the studied disease). It is also possible the placebo effect does not apply and there is standard course on the placebo group...

Measuring of the placebo effect itself is not important in studies testing drugs etc.

The importancy of the blinded placebo group in a study is that placebo effect applies to both groups - to the so called placebo group and to the really treated group. Getting results, you can "decrease" possible placebo effect (does not matter how big it really is) from both groups getting "0" on placebo group and real added effect of the treatment on the really treated group. Sometimes you get "0" within statistical tolerance on both (means the studied treatment has no added effect). So if you create two placebo groups and will give them purple starch pills to help them from migraine, you will get results like: it helped 20% of the group for both... Then, if you give one of the group real purple ibuprofenum pill, you will get 20% on placebo group and 65% on ibuprofenum group giving you information ibuprofenum helps to at least 45% of the group (simply said ;)

Comparing two placebo groups, however, makes sense for balancing small groups, where inadmissibly high statisticall error is too probable. E.g. if you have group of 12 people as a statistical set and you divide it into two groups with 6 people, a placebo-placebo test could easily return something like 15% vs. 25% of positive results, though we definitely do expect equal results. This gives an important information about unbalances sets. In such case you can either exchange people in sets and try again or you have to count with higher statistical error. You could get 60% of positive results if the first group got the real treatment and 70% of positive results if the second group got it.

However, real researches work with serious statistical analysis and with big enough statistical sets, and with calculated expected errors producing reliability intervals etc...

Also in real huge researches there is no time and money allocated for testing groups in placebo-placebo mode first. It is understandable if we know we can expect 1:1 result. --213.160.184.124 (talk) 10:10, 16 June 2009 (UTC)

What if there is no Subject?
I'd like clarification on the definition of a double blind study. What if there is an ignorant experimenter, but no subject at all? For instance, suppose we give three weapons to a ballistics expert and ask him which one fired a given bullet? Would that be a double blind test, even though there is only one person (the experimenter) unaware of the correct answer? —The preceding unsigned comment was added by 68.6.94.12 (talk) 16:33, 31 March 2007 (UTC).
 * Well, that isn't an example of a "study" at all. You're talking about an applied materials test with one sample, not a research experiment with a pool of subjects. Also, evaluating a therapeutic outcome is a really different problem than what you have described. Every patient won't have an identical outcome even after a well-tested therapy is provided, blindly or otherwise. Your example is basically the same though as what's described under "Forensic application". The only difference is that you're trying to have someone identify a bullet rather than a person. That person may or may not be biased, and the same is true of the person who's providing the samples to the tester. 198.49.180.40 21:14, 5 April 2007 (UTC)
 * Thank you for the reply. However, I'm still unclear on one thing: Is an experiment "double blind" whenever observer bias is mitigated by denying the observer knowledge of the expected outcome? Or does "double blind" have a narrower definition, and only applies to therapeutic studies with a pool of subjects who are also blinded? In effect, I'm asking whether the term means what it literally means, or whether its meaning has drifted -- preserving the valuable concept at its core, but dropping less significant details -- to the point where it applies to more situations than originally intended, including the ballistics example I cited above.


 * I ask because I've heard "double blind" applied to questionable situations, and I'm not sure if that is because of ignorance, or because the term is widely understood to have a meaning more flexible than the one described here in Wikipedia. Language is a slippery thing, and words tend to mean what people think they mean. I suppose I'm asking what people really think "double blind" means.


 * By the way, I can see how my ballistics example is similar to the photo lineup cited in the article, but that doesn't help answer my question. In the lineup example, the officer is conducting a single blind test of the witness' memory, but isn't the witness also conducting a test? And since the observer in the witness' test is different from the observer in the officer's test, might it be possible that the witness' test is double blind even though the officer's is not? Returning to my ballistics example, it seems reasonable to ask whether the ballistics expert is conducting a double blind test, regardless of what is known by the person providing the samples. —The preceding unsigned comment was added by 68.6.94.12 (talk) 18:22, 10 April 2007 (UTC).
 * In "single blind", the subjects are blinded but not the doctors. In "double blind", both the doctors and the subjects are blinded.
 * In your ballistics example, what would be the "single-blind" analogue? Presumably, the bullets themselves are not ignorant (blind). The same would be true in the photo lineup: the witness and the photo subjects are not blind, though the detective might be, or or might not. So, how are you imagining a case that would NOT be described as double-blind, in contrast to what you already described?
 * Lastly: I fully expect researchers to use research terms (like "double-blind", for example) differently from how the general public does. The meaning-drift you mention occurs thus, though I'd argue that intended meaning on the part of an ignorant utterer does not convey any actual meaning in the utterance. 198.49.180.40 22:13, 10 April 2007 (UTC)
 * I'm mostly interested in the meaning-drift issue, and you've addressed that. If I understand correctly, you are insisting that "double blind" applies only when there are experimenters and subjects, and both are blinded. I'm satisfied with that answer.


 * On the principle that no question should be left hanging, I'll answer the questions you asked in your last reply: When I've heard "double blind" used in situations similar to the ballistics example, there is no analog to "single blind"; there is only "double blind" and "unblinded". An unblinded test would be one in which the experimenter is aware of the expected outcome (e.g. "Here is the suspect's gun. Tell us if it fired the bullet."). I have no doubt that there is a valid distinction to be made between a blinded and unblinded experimenter in cases like these. What is at question is whether the term "double blind" can be used to make that distinction. You've given me the answer no.


 * By the way, I'm surprised at your claim that intended meaning conveys no actual meaning. I'd love to debate that with you, but perhaps we should do that in a linguistics forum somewhere, rather than this page. In any case, thanks for the time you spent responding to my questions here. —The preceding unsigned comment was added by 68.6.94.12 (talk) 03:13, 11 April 2007 (UTC).

Single blind
Is anyone against merging it here? For that matter, why not just have an article on blinding in general that covers all forms? Richard001 05:31, 24 April 2007 (UTC)


 * A very good idea. It was the first thing I thought when I googled for information on blind experiments and found two wikipedia articles covering the same thing.  As such I have done a rather crude merge and am more than happy for others to continue the process.  I agree that the article could be renamed to a more general blinding title.
 * On a completely unrelated idea, I would like to see further discussion of the ethical implications of double-blinds. I have been led to believe that here in Australia there are a number of situations where it is deemed unethical to conduct a double-blind and I would like to know more about this and see it included in the article. Fermion 10:38, 25 July 2007 (UTC)

What's it called...
when an experimenter tells the subject that something is X but then switches X with Y in order to observe the existence of the subject's bias? does it have a name? --AnY FOUR! 04:46, 4 August 2007 (UTC)

ABX testing
I'm an ex-subjective audio reviewer (Stereophile) who's had terrible arguments with Arny Krueger, et al, about problems with controlled testing -- namely, that listening under controlled conditions is not the same as listening casually (which is how we actually listen to music). This point -- that controlling the test conditions can also alter them -- needs discussion. WilliamSommerwerck (talk) 14:57, 15 February 2008 (UTC)

Not ubiquitous
Blinding is far from ubiquitous in the scientific method. It is relatively rare in chemistry, physics, geology, meteorology, astronomy, and forensic science. It is even relatively rare in many areas of biology and medicine. I recently reviewed a number of grant proposals related to traumatic brain injury; less than 10% included blinding in their proposed methods. Michael Courtney (talk) 11:45, 1 April 2008 (UTC)
 * You're right.  Blinded trials seem to be most used to test if a drug has a measurable effect against something else (placebo, another drug, etc.)   In this kind of testing situation it is ubiquitous, but otherwise not so much...  --159.178.248.49 (talk) 13:15, 15 March 2010 (UTC)

Male circumcision 'cuts' HIV risk - not a blind experiment at all
I think the reference:

"Male circumcision 'cuts' HIV risk" ... Retrieved on 2009-05-18.

is not related to blind experiment at all.

1. All the males knew about their circumcision and were equally informed about safe sexual behavior...

2. The experiment was a social behavior study with no intervention from the researcher's side concerning participants' conditions.

I agree male social behavior can be seen as a kind of blind activity if not under a supervision in general ;), but hey, in blind experiment we have to actively make a difference and at the same time we do hide the difference to participants or even researchers to avoid bias...

Even if researchers performed circumcision on half of the group it would not have been a blind experiment.

(btw. I do not think the experiment is correct concerning possible corellation between circumcision and safe sexual behavior - it is pretty logical to assume that people comming from society where it is usual to perform circumcision - i.e. society taking care about "these issues" - will take more care... If a father sends his boy to a tribal medicine man for circumcision, he will more probably also teach him how to avoid HIV)

I would rather remove all the note concerning ethical troubles. The note is valid for experiments in general - once we know the result and it can safely help we should use it for wide public if possible. Does not matter whether it is a blind experiment or not...

--213.160.184.124 (talk) 11:44, 16 June 2009 (UTC)

Triple-blind
The lead of the article mention triple-blind experiments and says that more will be discussed later in the article, but triple-blind experiments are never mentioned again in the article. Can somebody add information about what a triple-blind experiment is? --Bando26 (talk) 06:20, 30 August 2009 (UTC)

ortography
I believe this is "double blind," and not "double-blind." The latter refers to the adjectival form only, as, by WordNet, "a test procedure in which the identity of those receiving the intervention is concealed from both the administrators and the subjects until after the test is completed; designed to reduce or eliminate bias in the results." —Preceding unsigned comment added by Twipley (talk • contribs) 23:59, 3 November 2009 (UTC)

When blinding fails
Recently I have seen sites question the claims of pharma companies that their nicotine replacements products(patches, gums, etc) double the chance of quitting. These claims are all based on double blind studies, but critics argue that the control group of these studies were inproperly blinded.

The problem is that a smoker is intimately familiar with the effects on nicotine and will know if they have been administered a dose or a placebo. In some tests undisclosed amounts of nicotine has been added to the placebos, which could have an adverse effect on quitting. Here is a study that calls for researchers to follow up on how the test subjects perceived the treatment. 141.16.91.106 (talk) 16:23, 29 April 2011 (UTC)


 * You may enjoy the article I am writing on unblinding. You will have to wait until I finish drafting the article for the link to be good.
 * Wikiman2718 (talk) 09:53, 20 April 2019 (UTC)

Blinding, Experiments and other uses
This same technique is used in elections and IT (crypto particularly), perhaps the uses section could be expanded to include those.

I'd suggest that blind experiments are one application of a more general technique. — Preceding unsigned comment added by 77.86.117.208 (talk) 09:50, 8 August 2018 (UTC)