Kevin's Watch

participating in the evolution debate over in the tank some of my and others criticisms of science have been rebuffed by something like this:

I think most people don't really understand the methods and purpose of science

i have, in my reading come across multiple definitions of science and its objectives, many of them different, but it has to be said that much of my reading has been authored by thinkers making critiques of science. for various reasons i didn't have the pleasure of studying at university so much of my learning has been done within a vacum. i fully accept that my thoughts lack a certain objectivity at this stage so i wish to seek it(in part) from the Kevins watch community.

i dont expect a simple and easy definition but i would like to ask what exactly you find the methods and purposes of science to be?

hmmmm...i just posted in that other thread but now i'm thinking i should have posted here, Tazz.

I would say the purpose of science is to understand how the universe works. In many cases, we also want to go beyond learning for the sake and joy of learning, and put what we learn to practical use; to benefit ourselves. For example, we learn about the human body and micro-organisms, and use the knowledge to live longer and more healthy lives. We learn about fuels, gravity, and aerodynamics, and put communications and navigational satellites in orbit.

Kins has written a couple of things that could be a good beginning to the methods of science.

Kins wrote:The National Academy of Science regards scientific theory as a well substantiated explanation of some aspect of the natural world that can incoporate facts, laws, inferences, and tested hypotheses. No amount of validation changes a theory into a law.

Kins wrote:A theory is something that has not been disproven. Science sets hypotheses and experiments to disprove or find holes in the theories and start over again until the scientific community gets it 99.9% correct.

As I said in the other thread, in general, science tends to operate on the assumption that it is wrong, until it is proved otherwise.

--A

There have been many contributions to science that did not involve munitions. I thought the point of the thread was to discuss the process of scientific thought or the philosophy of science. Anything that I say in this post must be taken with the caveat that I am an experimentalist and not a theorist. In general scientists can be divided up into those two categories.

Science is always questioning. Until I, or a reputable source, have observed something it is unknown. One must contstantly be aware of the ease of making a post hoc or other causal fallacy. One must question the integrity of the data acquired and the integrity of the equipment used to acquire the data. One must make sure the data was treated properly after acquisition. It is a well know fact in scientific circles that data may be twisted and tortured to say things it really doesn't say. Usually the less mathematical treatment of data the better.

Unfortunately there is quite a bit of bad science out there and some of it really does look like good science. One must carefully scrutinize each source of information logically and thouroughly prior to deeming it valid or invalid.

Great post, Lorelei.

Indeed.

despite the splintering of the thread i am glad for the discussion that has taken place on the subject. thanks.

Lorelei wrote:Usually the less mathematical treatment of data the better.

Great post Lorelei. I wonder if you'd be so kind as to elaborate on the quote above...it sounds like a point I'd like to know. ( LoreMaster)

(Oh, and if you'd remind me what your field is again...) Thanks.

--A

Every time a set of data is collected there is a certain level of statisical variability and error from the method of data collection. This is known in most circles as "noise". If this noise level is high compared to the data of interest (signal), the data is noisy and hard to interpret. There are several methods of increasing your signal to noise ratio. Assuming that there is nothing wrong with the data collection device, the most common method of increasing this signal to noise ratio is to repeatedly record data on the sample under the same conditions. Sometimes it is not feasible to do so, or the test is destructive. In these cases there are some mathematical treatments of data that can be used to help artifically "clean" up the data. The most common tool for making data prettier is called smoothing. The process of smoothing the data actually removes data points from the data at set intervals to make it look, well smoother. Given enough smoothing cycles it is possible to make noisy data look really smooth and pretty. The problem with that is that it is no longer really useful data, much information has been lost. The trouble is you don't know whether the information lost was from a real phenomenon or it was noise. There have been cases; however, where data like this has been used to make conclusions.

Oh, I am a chemist btw.

Additional methods of statistical analysis work in similar ways: some remove all data at one or both extremes, considering it to be noise, for example. But all mathematical methods end up removing parts of data which could be actually viable, so it's always best to rely on reproducibility of the experiment.

Lorelei wrote:Every time a set of data is collected there is a certain level of statisical variability and error from the method of data collection. This is known in most circles as "noise". If this noise level is high compared to the data of interest (signal), the data is noisy and hard to interpret. There are several methods of increasing your signal to noise ratio. Assuming that there is nothing wrong with the data collection device, the most common method of increasing this signal to noise ratio is to repeatedly record data on the sample under the same conditions. Sometimes it is not feasible to do so, or the test is destructive. In these cases there are some mathematical treatments of data that can be used to help artifically "clean" up the data. The most common tool for making data prettier is called smoothing. The process of smoothing the data actually removes data points from the data at set intervals to make it look, well smoother. Given enough smoothing cycles it is possible to make noisy data look really smooth and pretty. The problem with that is that it is no longer really useful data, much information has been lost. The trouble is you don't know whether the information lost was from a real phenomenon or it was noise. There have been cases; however, where data like this has been used to make conclusions.

Oh, I am a chemist btw.

I know that does happen, but experiments that reduce noise variables help alot and you dont need to do 'data cleaning'.

I know that does happen, but experiments that reduce noise variables help alot and you dont need to do 'data cleaning'.

Yes, that would be good science, I was just explaining how over treating data mathematically is generally considered a bad idea.

Lorelei wrote:

I know that does happen, but experiments that reduce noise variables help alot and you dont need to do 'data cleaning'.

Yes, that would be good science, I was just explaining how over treating data mathematically is generally considered a bad idea.

I agree. I have always been against data cleaning because if you sampled from a normal population you are essentially damaging an (hopefully) unbiased grouping.

Fascinating. See, I knew it was a mistake to trust mathematicians.

Thanks for the info. I think that LoreMaster is right. Isn't the "noise" in itself part of the data that should be taken into account? Perhaps it forms or affects a trend that could be significant, but instead, ends up being discarded.

--A

Avatar wrote: I think that LoreMaster is right.

What the?! *Cue Twilight Zone music* Have I stumbled into a strange world where Avatar has agreed with me twice in one day? Oh man, I've been smoking some strong weed.

Without sharing it? Shame on you!

Yeah, I noticed the stars increasing...it's disconcerting to have a physical record of it. It might turn out to be a trend, rather than an anomoly...

--A