Hacker News Reader

Comment by awakeasleep | Hacker News Reader

I think the assumption that self-reported data can be trusted makes a mockery out of every field of study that relies on it.

Setting aside the fact that it's hard enough to design a double (triple?) blind study of mice, setting aside the fact that people will lie and joke around for pleasure, how can you ever get past the fact that 'who we are' and 'who we want to be' are inextricably linked?

View on HN

According to some accounts, Margaret Mead, the most celebrated anthropologist of her generation, was similarly trolled by Samoan girls: http://www.amazon.com/The-Trashing-Margaret-Mead-Anthropolog... At the very least, the controversy around that study should have led these researchers to be more circumspect with their data.

> double (triple?) blind study of mice

Would that be a study where the mice are not aware of if they are in the control or not?

Has the placebo effect been seen in mice?

You've never heard of the three blind mice? I believe the farmer's wife won a Nobel prize for her work in regeneration of severed limbs.

I'm a little ashamed to admit that you had me googling for that. ;)

Seriously though, the idea that mice might have a placebo effect is fascinating. I would have thought that the placebo effect would require some high-level form of self-awareness (which perhaps mice do have, I don't really know) and an understanding of at least the purpose of medicine. Maybe not though...

Mice are the most intelligent species on Earth. In fact, they are so smart that they are immune to placebo effects. However, they act like they aren't because our reactions provide them much entertainment - http://hitchhikers.wikia.com/wiki/Mice

Edited to add: On a more serious note, I don't know if mice are susceptible to a Hawthorne-like effect http://psychology.about.com/od/hindex/g/def_hawthorn.htm Even physical science researchers have to deal with observer effects http://en.wikipedia.org/wiki/Observer_effect_%28physics%29

Edited to clean up brain-drained errors.

I don't know what you or the OP are thinking of, but it's worth pointing out that the placebo effect is not just a phenomenon in the patients' minds; it can also be a phenomenon in doctors' minds.

In brief, any time you have a subjective measurement (for instance, the experimenter assigns a number from 1 to 5 based on how "distressed" the mice -- or human patients -- look), that subjective measurement can be subtly biased. So if the experimenter doesn't blind his experiments properly, you can see a placebo effect in animal studies.

It is appropriate to refer to these studies as double or triple blind because the single blind usually refers to the patient.

  Single blind: patients do not know which group they are in
  Double blind: experimenters do not know the group identities
  Triple blind: overseeing committees do not know the group identities

So animal studies can be assumed to be single blind by default, and become double-blind when the experimenters do not know the group identities.

My understanding is that the "placebo effect" doesn't refer to biased numbers or reporting, but exists when everything is measured correctly. It is objectively there, not a screwy result.

You're correct. What saalweachter was describing are ways to minimize threats to reliability and validity.

- http://www.socialresearchmethods.net/kb/relandval.php

- http://www.mathcs.duq.edu/~packer/Courses/Psy624/Validity.ht...

- https://en.wikipedia.org/wiki/Internal_validity

- https://en.wikipedia.org/wiki/External_validity

"Biased" is probably a loaded word. I think I may have implied intent without meaning to.

But basically, the placebo effect only exists for subjective measurements. (http://en.wikipedia.org/wiki/Placebo#Mechanism_of_the_effect) If something exists only in subjective measurements and not objective measurements... I'll let you draw your own inference.

Reply on news.ycombinator.com