I usually like to go through my questions and answers and edit them, and make sure that they are up to date. I was going through my skeptics.se questions, and in this question, Does a webpage with a black background save energy?, I noticed that the accepted answer had gone missing. What happened to it? It was a good answer, and it would be great if it could be brought back.
They were anecdotal answers: in other words, "I've personally measured it and...".
I know they were quite up voted. I know it will ruffle some feathers. The users will keep their reputation, but we can't have that kind of broken windows on the site.
Running your own pseudo-experiment and posting an answer is not proof of anything. It's anecdotal. It may convince yourself, but it won't (and shouldn't!) convince anybody else.
A real experiment has different characteristics:
- It's usually run on large samples.
- Its methodology is described accurately.
- It comes with sensible error estimations.
- Its data is disclosed to the public.
- It's peer-reviewed multiple times (before and after publication).
- It's generally replicated by other teams.
- It's based on peer-reviewed references
In other words - a far cry from a layman at home with a consumer watt-metre, no offense :-)
- How do you know that the answers are not fake?
- How accurate are the measurements?
- Are the monitors representative of the average in any form? Are they even meant to be representative?
- Have they tried to replicate the results?
The point here is, these answers are not good for the site - they are not looking at the science behind claims. They are doing amateur science. As such, I would completely welcome and encourage them on a PopSci site, but not here.
The whole point of this site is to look at evidence, facts, references, studies to answers some questions where other sites produce anecdotes and points of view.
If we allow such answers, then we produce clearly no better than the claims we want to investigate.
"The users will keep their reputation" - are you sure? I thought that rep vanishes with the answer! Mar 25, 2012 at 16:44
@HendrikVogt it's a new feature (exactly to allow these cleanups to happen). See: blog.stackoverflow.com/2012/03/… Mar 25, 2012 at 17:02
Ah, thanks! I didn't know this important detail of that recent change. (A bit strange, it seems: if the answer is alive 59 days, then you loose your rep, otherwise you keep it unless it had a score below 3 anyway.) Mar 25, 2012 at 17:24
1What about situations where there is no peer-reviewed, extensive literature? (like the answer in question). In this case, I clearly described my test methodology, and I am completely willing to respond to and even correct my answer if any further evidence come along. Mar 26, 2012 at 3:53
I tend to think that in the absence of any reasonable peer-reviewed literature, a transparently described, clearly noted as original research test is better then leaving the question unanswered. Mar 26, 2012 at 3:55
1I'm not claiming my answer is definitive, or even all-encompassing. It's merely a test of a few different LCDs, under a clearly-defined methodology. Mar 26, 2012 at 3:59
1I would agree the answers for questions about a topic where there is a lot of controversy must be held to a much higher level of testing rigor, but in a situation where there is little controversy, a simple test is often enough to confirm one versus the the other. Mar 26, 2012 at 4:02
1I mean, if my question is "Anecdotal", then the question linking to some random guy's blog seems far more anecdotal to me. Or does posting my answer to my website, and linking to it give it more weight, somehow? Mar 26, 2012 at 4:04
@FakeName, if there is no literature then don't answer. As simple as that. Alternatively, answer that "We don't know because no extensive studies have been performed". Both have worked for us elsewhere. Other users have found it possible to answer with references though, so the "absence of literature" angle doesn't hold. Mar 26, 2012 at 6:52
I argue that the evidence I presents is neither anecdotal or faith-based. I simply did some tests, explained my testing methodology, and posted the results. It should be trivially falsifiable by anyone with a LCD monitor and a watt-meter. Mar 26, 2012 at 7:25
The field of skepticism covers many things that never make it into peer-reviewed journals. Furthermore, there are many situations where a question may fall between explicitly researched ideas. Furthermore, it seems to me that a significant number of answers which do not get deleted are far more anecdotal or faith-based then my answer in this case. Mar 26, 2012 at 7:27
Basically... what? So if I published my results on a web-page somewhere, and referenced them, you would be ok with it? For instance, one of the answers you left links here, which does not have a single citation, and has vague hand-wavy references to "analysts note". If you're going to delete every post not citing exclusively peer-reviewed literature, at least be consistent. Mar 26, 2012 at 7:32
How do you know that the answers are not fake?- Well, there has to be some trust somewhere. I imagine that this is what the whole reputation system is supposed to address. This is really the only argument you make against my answer that really has any validity. Aside from referencing my Reputation scores on electronics.stackexchange.com, and pointing out that there are many pseudoscientific "peer-reviewed" journals, I really can't provide any way to demonstrate my trustworthyness. Mar 26, 2012 at 7:42
4@Sklivvz as per the meta question, the answer should have stayed. It is clearly noted that it is not authoritative peer reviewed research, however the answer was well worded and gives an indication in the absence of other evidence. With an appropriate disclaimer I don't see a problem. Mar 27, 2012 at 1:11