Talk:Experiment

From Wikipedia, the free encyclopedia

WikiProject on Sociology This article is supported by the Sociology WikiProject, which gives a central approach to sociology and related subjects on Wikipedia. Please participate by editing the article Experiment, or visit the project page for more details on the projects.
Start This article has been rated as Start-Class on the quality scale.
High This article has been rated as High-importance on the importance scale.

Contents

[edit] Interent Resources

I found these to be interesting resources,bullass but not worthy of being on the main page:adam http://www.asp.ucar.edu/colloquium/1992/notes/part1/node86.html http://www.psychstat.smsu.edu/introbook/sbk02.htm

[edit] Natural experiment example

"For example, after Albert Einstein advanced his theory of general relativity according to which massive objects such as the sun should cause curving of spacetime, an observation was made of the light from a star which passed near the sun. This was done during a solar eclipse in order that light from the star could be detected. It was found that the light from the star was indeed deflected by the sun."

This was used as an example of a natural experiment. However, the discription is very minimal, and this experiment probably deserves its own page, which I couldn't find. Once that page is written up, we can reincorporate this experiment into the article.adam

If I recall correctly, this experiment (performed I think during a solar eclipse in the 1920s) was actually pretty bad - there was a lot of experimental error, and although it is commonly referenced in popular discussions, it wasn't a particulalry convincing result. The michelson morley experiment is a better example, IMHO. Chas zzz brown 23:43 Jan 12, 2003 (UTC)

Harris77 writes: I personally do not like the term natural experiment. I think the article should talk about naturalistic observation.

Also, i think that the heading should be "Studies" then talk about the different type of studies (correlational, clinical, case study and experimental). It would be a lot of work to change this and I don’t think I should

I am considering creating an article "Scientific Studies" and listing the studies.

[edit] hard science and soft science

Note on hard/soft science: Please do not delete this reference, as I believe it is an important aspect of the nature of science, though I may not have used the best words to identify this issue. I do not mean to imply that there is any such thing as an absolute ?hard science? or ?soft science?, or that a scientist working in a ?softer? field is any less of a scientist than one working in a ?harder? field. In fact, the soft sciences may require more competent and imaginative scientists because they are working with more constraints than those who work in the hard sciences. All that I wish to point out is that all sciences have their limits, and the study of each phenomena will be limited in different ways. To a large extent, fields may be arranged along a spectrum from soft (most limited) to hard (least limited), but in comparing some fields it is not possible to say that one has more limits than the other, only that they have different limits.

While this idea of hard/soft science is pretty common, I am unaware of any formal treatment of the issue. Perhaps it is covered in the philosophy of science section. Anyway, I believe that we should incorporate this into Wikipedia somewhere, and there should be a link to such a page from this page. If you can think of a better way of expressing this idea than ?hard/soft science?, please replace the term with the appropriate term. adam

This is a good start! Some other thoughts; maybe they should go in the experimental method article, or maybe here...

  • what makes a "good" experiment? one generally has a hypothesis to test; if the experiment is not well designed, it may neither support nor refute the hypothesis.
  • experimental error - what is it, and how does one attempt to reduce it?
  • replicability - easier to define for physics, harder to define for sociology.
  • the role of statistical analysis and meta-analysis in drawing conclusions from experiments.
  • a list of cool experiments that we can refer to (hopefully covered in Wikipedia...)

Other thoughts as they occur to me... Chas zzz brown 23:43 Jan 12, 2003 (UTC)

I had thought of a list of experiments. There are many experiments described in wikipedia...just do a search for "experiment". I read many of them when I wikified the term "experiment" to link to this page. Anyway, we should probably categorize them. One option is to categorize them by field (Physics, Economics, Astronomy, etc.) or we could categorize them as Controlled experiments, natural experiments, quasi-experiments, or thought experiments. adam

In my view, the present article doesn't start out well.
As for: Experiment: An experiment is [...] to identify a causal relationship between phenomena.

What's "causal" about the experimental determination of particle masses, for instance?
But that's not an experiment, it's a measurement! (sarcasm) Perhaps this just shows the bias that I have aquired from my studies of molecular biology, where we don't have any universal constants to measure. I made an effort to account for psycology and the social sciences, but it didn't occur to me that I would need to make an effort to think of things from a physicist's perspective. That's a strange type of experiment (to me). I vaguely remember the original experiment. Do measurement experiments even have a control? Should they be included as a different type of experiment? It sounds like a purely descriptive process. Are all measurements experiments? What about measurements of CO2 concentration or the measurement of the height of trees in a forest.
My dictionary (Webster's new world) defines experiment as "a test or trial untertaken to discover or demonstrate something." adam

I object to the word "verify" in the definition. I propose that the word "support" be substituted. Experiments can ONLY falsify hypotheses, at least in scientific method. Also, nowhere are mentioned other roles of experimentation. What about the role of experiment in discovery ? user:DCDuring



We might distinguish experimental determinations from statistical analysis and meta-analysis of experimental determinations ...
Is this the same as "measurement" and "analysis of measured data"? (I'm not sure what "meta-analysis" is) If so, I think it omits the idea of a controlled experiment. Controlled experiments require the comparison of two measurements, and to maximize confidence, these measurements should be made simultaneously...they shouldn't be separated and treated as independent measurements. Just make sure that isn't left out of the final article. adam

As for: An experiment is a set of actions and observations performed [...]

I prefer an experimental determination, or measurement, (in any one trial) to be characterized as:
The systematic acquisition of observational data, and application of a reproducible analysis procedure to the data, in order to derive one definite result value (within a range of confidence)
This sounds good. My only concern is that it seems to emphasize quantification, which isn't always essential (especially in biology).adam


Regards, Frank W ~@) R 02:52 Jan 13, 2003 (UTC).


As for: it seems to emphasize quantification, which isn't always essential (especially in biology) [...] Controlled experiments require the comparison of two measurements

Therefore it does seem essential, including in biology, that any one of these (two or more) rather simple and quantitative measurements can be obtained separately (albeit perhaps in simultaneous trials), and that their result values are commensurate between trials.
This is the sort of situation that I'm thinking of: We do an experiment with two samples. At the end of the experiment, we report that "sample A has a signal that is 60x as strong as sample B". We repeat the experiment the next day, and we find that the signal from sample A (day 2) is 1/5 the signal from sample A (day 1), however when comparing the samples from day 2, sample A is still 60x as strong as sample B. This could happen for several reasons; perhaps one of our reagents degraded (a radioactive tracer?), or the cells were growing at 37.2 degrees instead of 37 degrees. I suppose it would be possible to perform the experiment such that we get some sort of absolute value, but it would be extremely tedious and that value would apply just to one very narrow set of conditions. We avoid this issue by focusing on the comparison between the control and the experimental sample, and we avoid all those little complications by functionally treating them in the same way. By "functionally", I mean that we grow cells in the same incubator at the same time, trusting that there is a minimal difference in conditions within the incubator, rather than trying perfectly control the conditions in an absolute, measured sense. Perhaps this all beside the point, and the experiments as I describe them are just a way of working around the fact that we cannot make perfect measurements and have perfect control over the conditions of the experiment.
Regarding quantification: A biologist might report "the treatment caused an increase of signal from the nucleus. Therefore, our modifications caused the signal producing molecule to be targeted to the nucleus." We can draw firm conclusions and increase our knowledge of how the cell works without making any reference to numbers. Of course, numbers are nice when we can get them, but we

can't always get them with a reasonable amount of work.


Finally, the word "value": My first response to this word is "numbers!", which is what set me off. However, I suppose that "True" is also a value, and "nucleus" could also be a value. But what we need to think of is how the reader will interpret this, not what the complete, "real" meaning of the word is. adam

As for: My dictionary (Webster's new world) defines experiment as "a test or trial untertaken to discover or demonstrate something."

Along with this, I'd prefer to emphasize that "something" needs to be definite, and to be derived from observations.


As for: Are all measurements experiments? What about measurements of CO2 concentration or the measurement of the height of trees in a forest.

Well, on one hand, it seems straightforward to patch up a measurement to obtain a statement of a test or demonstration, e.g.
It has been tested whether, and demonstrated that, the CO2 concentration in these trials was less than 10 %.
On the other hand, I agree, considering correlations between commensurate measurements is different (and indeed more advanced) than the notion of one measurement itself.

As for: Just make sure that isn't left out of the final article.

I certainly prefer that, too; if the present article on Experiment emphasized the role of measurements in order to obtain Controlled experiments; and vice versa, that (sets of) individual quantitative Measurements provide the basis for more sophisticated and derivative experimental procedures.
Besides, being a relative newbee Wikipedian, I'm still trying to determine whether and how expressions of such preferences are supposed to be honored in the resulting articles ...
Try to do the best that you can.  :^) In general, try not to cut things out of the article (without moving them somewhere or such). If you make a change and I think it is stupid and much worse than the original, I can always revert to the original...but that is essentially declaring war, so it is to be avoided. Anyway, you have a good point, so I wouldn't do that. I guess if you made a change and I thought that it left out an important point, then the ball would be in my court and I would have to try to rewrite it to incorporate the things that both of us think are important. As it is right now, I'd rather not make the modifications myself because I'm not sure I understand exactly what you are proposing, so I'd rather that you make the modifications that you see fit and then I'll judge that against my understanding of the issue. Good luck! adam

As for: I'm not sure what "meta-analysis" is

As far as I understand the Wikipedian who introduced this notion in this page, it seems to be concerned with how to obtain advanced experimental procedures (such as controlled experiments) from commensurate result values, obtained by sets of individual measurements.
Okay, that's pretty much what I had guessed. I'd rather not use that term in the article unless it has a good description. adam
Best regards, Frank W ~@) R 20:05 Jan 13, 2003 (UTC).

[edit] Dependent and independent variables

I'm concerned that this article doesn't mention dependent and independent variables by name (they may be implied in some places). It seems to me this is integral to the definition of an "experiment". I will return when I have time and attempt to provide this terminology in the article. I also think, as much as possible, the article should reflect what is generally true of experimentation without making too much of "hard science" and "soft science" differences. These may be interesting philosophical issues for somewhere else, but they are not central to the concept of experimentation. Any experiment - be it in physics, psychology, or trying to figure out why your entertainment center in your living room isn't working, shares essential features. These, I think, should be the focus of a general article on Experiment. Feedback before I take a stab? SJS1971 00:17, 20 January 2007 (UTC)

[edit] Quotes

The following were removed from the article to this talk page, pending proper citation to the source. The last one may be a hoax. --Bejnar 18:58, 28 May 2007 (UTC)

"We have to learn again that science without contact with experiments is an enterprise which is likely to go completely astray into imaginary conjecture." — Hannes Alfven
"Today's scientists have substituted mathematics for experiments, and they wander off through equation after equation, and eventually build a structure which has no relation to reality." — Nikola Tesla
"To experiment is the most unique of human attributes" Rossafar Gerium

Quotes belong on Wikiquote anyway, not in the article (unless there's some context they belong in). Also, both "Rossafar Gerium" and his quote get zero ghits; I'm not going to say that the quote is a hoax, but it appears entirely unattributable. Zetawoof(ζ) 19:30, 28 May 2007 (UTC)