Wikipedia:Reference desk/Archives/Science/2018 February 23

Welcome to the Wikipedia Science Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

February 23

Olive oil

How much alpha-linolenic acid does 1 tablespoon of olive oil have in it? --User777123 (talk) 02:57, 23 February 2018 (UTC)

The answer is there in olive oil, under constituents. Matt Deres (talk) 04:22, 23 February 2018 (UTC)

WP is not a reliable source!

Try http://nutritiondata.self.com/facts/fats-and-oils/509/2 which has great resources. It is not loading for me pending a flash update, so I have no info, but this was my goto site when I went from a 39 to a 23 BMI. μηδείς (talk) 04:35, 23 February 2018 (UTC)

No one said it is. But it is supposed to cite them. And the article does cite a source for the info. If you feel it is not a RS per wikipedias definition then tag it as {{unreliablesource}}. If you have your own weird definition of reliable source, you should at least provide it before asking. Nil Einne (talk) 08:25, 23 February 2018 (UTC)

Medeis's link, which opened OK for me, does not actually list the different types of fatty acid. The source quoted in the article looks reasonable: it does seem to give a very thorough description of the chemistry, and quotes other sources for that data. In any case, the answer seems to come down to a range of 0% to 1.5% - which clearly suggests that the actual level will vary according to variety and location. The bottle in my kitchen doesn't include that information - it just lists saturated and unsaturated fatty acids. The implication may be that the identical bottle, bought somemonths apart, could have a slightly different chemical composition. Wymspen (talk) 12:02, 23 February 2018 (UTC)

An educated guess would be that it varies by commercial grade (virgin, extra-virgin, etc.) and probably also by cultivar used. Matt Deres (talk) 16:13, 23 February 2018 (UTC)

Olive oil seems to contain between very little and nothing at all of it. Maybe you are interested in this table of oils from seeds, with alpha-linolenic oil (ALA) content (average percent, first table from above): http://www.my-personaltrainer.it/nutrizione/alimenti-grassi-essenziali.html Language is not English but you have the Latin name of all plants. Olive is missing by the way. It is funny that many articles and ads about olive oil cite with much ballyhoo this ALA as an important constituent, but not the quantity contained. The only one somewhow doing so says "so much ALA as in mother milk", but again without figures. 194.174.76.21 (talk) 20:04, 23 February 2018 (UTC) Marco Pagliero Berlin

I just noticed there's some explanation in the source.

Regarding the polyunsaturated fatty acids (PUFAs), there is a wide range acceptable for extra virgin olive oil, however the linolenic acid has to be less than 0.9% per the International Olive Oil Council (IOOC) guidelines. Higher levels, e.g. 1.5%, do not present a nutritional problem, but the IOOC uses the linolenic acid level to establish the authenticity of the olive oil. Seed oils like canola oil have higher levels of linolenic acid.

So if you're getting quality olive oil, it should indeed be low. Nil Einne (talk) 00:02, 24 February 2018 (UTC)

Actually it does list different types of fatty acids. It just doesn't show them by default. (I noticed this while evaluating it.) If you click more details under 'Fats & Fatty Acids', it should show a list. That said, it only shows 18:03 which I assume to mean Linolenic acid. They don't list anything for 18:3 n-3, c,c,c or 18:3 n-6, c,c,c, so it may seem we don't know if it's alpha-Linolenic acid or gamma-Linolenic acid. Except that 'Total Omega-3 fatty acids', is the exact same as 18:03, strongly suggesting it's all alpha/n-3. I don't know why 18:03 rather than 18:3 as they do for all the other ones as well as the subtypes for 18:3. It seems we're not the first to wonder this [1] (who also come to a similar conclusion about alpha vs gamma). I did initially question that source, however their data is supposedly from USDA SR-21 so it may not be too bad even at a minimum their interpretation and presentation seems confusing. Anyway so their number is 0.761% which is within the range the source our article uses provides. (I initially thought it was around 1.6% but the default values are for 1 cup, not for 100g.) Nil Einne (talk) 00:00, 24 February 2018 (UTC)

See also here. Count Iblis (talk) 00:33, 24 February 2018 (UTC)

Where do you see the ALA content in that link? I only see EPA and DHA.--User777123 (talk) 01:44, 2 March 2018 (UTC)

What link? Multiple have been provided. Your reply is at the bottom under CI's response, but is not indented at it. Nil Einne (talk) 16:04, 2 March 2018 (UTC)

Current Scientific opinion on Piaget and Kohlberg

I’ve been doing research on ethical theory from the standpoint of evolutionary theory, but I want to take in ideas from any discipline that have something scientifically rigorous to add. I’m very familiar with the philosophical literature on ethics but not with the psychological research. I recently came across Kohlberg’s work. I’ve just started to look at it but so far I’m very impressed and he seems to have some good empirical evidence. I’m wondering is there any modern scientific consensus on his work? I’m mostly interested in Kohlberg but I include Piaget because from what I understand their two approaches are closely related. Are there any major Issues with Kohlberg’s work? Are there competing theories I should focus on? I’m interested only in scientific approaches, and theories specifically about ethical development not general theories of personality or development. I’ve looked at Moral Foundation’s Theory. I looked at the Wikipedia page on Kohlberg’s ethical theory but was wondering if people had anything more to add that isn’t in the article, even non-encyclopedic POVs ;) Any feedback would be appreciated --MadScientistX11 (talk) 06:05, 23 February 2018 (UTC)

Are you talking about Jean Piaget and Lawrence Kohlberg? ←Baseball Bugs ^{What's up, Doc?} carrots→ 08:00, 23 February 2018 (UTC)

@Baseball Bugs: Sorry, I thought it would be obvious from the context but yes Jean Piaget and Lawrence Kohlberg. --MadScientistX11 (talk) 15:56, 23 February 2018 (UTC)

Since ethics is part of the field Philosophy you should move your question to Wikipedia:Reference desk/Humanities. --Kharon (talk) 18:33, 23 February 2018 (UTC)

Since Kohlberg's work was in the field of developmental psychology,which is" the scientific study of how and why human beings change over the course of their life," the question is located fine right here. That assumes that Kohlberg did experiments or some sort of science-based observation rather than just sitting in his office philosophizing. Edison (talk) 22:06, 23 February 2018 (UTC)

@Edison: Thanks, yes I agree, Kohlberg was a psychologist not a philosopher and I think my question belongs here. One of the things that surprised me was that he did indeed do experimental work where he tried to test his theory of moral development. For example, he had subjects who he followed at various points in their lives and had them respond to stereotypical scenarios (was what this person did wrong? if so why?) and then he had other researchers independently rate the answers based on his model and he achieved pretty strong consistency across evaluators and some good evidence that supports his model of how moral values evolve as people age. Similar to Piaget's theories about child development and my understanding is that Piaget was a strong influence on Kohlberg. I found it interesting because Kohlberg's work is pretty old and in the more modern work I'm reading now in evolutionary psychology he's not mentioned that much even though he seems to be as or more rigorous as much of the work I'm reading now. But I was also wondering, if there was some later work that pointed out major flaws in his methodology or otherwise. Sorry, I'm rambling on here, since no one has spoken up I'm going to follow up on my own and read some more of his work, but I'm still interested if people have any opinions either for or against his approach. --MadScientistX11 (talk) 18:37, 26 February 2018 (UTC)

Is Iodised salt broken in water?

If I put Iodised salt in water it's broken into salt (NaCl) and Iodine? 185.191.178.183 (talk) 16:40, 23 February 2018 (UTC)

There's a lot of chemistry to correct there. Let's see how much I can get through in the space of this forum. Iodised salt does not contain a mixture of sodium chloride and elemental iodine, it contains a mixture of sodium chloride and some other salt of iodine (i.e. an iodide salt or iodate salt such as potassium iodide or potassium iodate). Secondly, the word broken is imprecise here. All salts, when they dissolve in water, dissolve through a process called solvation, which includes a sort of "breaking"; ion-ion interactions in the solid salt are replaced with ion-solvent interactions in the solution. I suppose that involves "breaking" of a sort. Both the sodium chloride and potassium iodide undergo this process. --Jayron32 16:47, 23 February 2018 (UTC)

Thank you for the info. I'd like to know please, if this solvation has to do something with temperature? (the higher temperature the dissolving faster - for example) I'm asking because I want to understand if I'm adding iodised salt to food while cooking, it's dissolved in the water to be potassium and iodain separately and goes to the blood via the digestive system. 185.191.178.183 (talk) 03:24, 24 February 2018 (UTC)

Yes, all ions are dissociated from each other when in solution. The ion channels responsible for transport of ions across membranes are highly selective, and your body will absorb and process potassium ions and iodide ions in different ways; your body does of course use iodine ions so it has ways of dealing with them, but yes, when they enter your blood stream, the potassium ions and the iodide ions are dissociated from each other in solution, the same way they would be in any water-based solution (blood being a water-based solution. This article describes specifically how iodide is absorbed. --Jayron32 03:58, 24 February 2018 (UTC)

Can someone reconcile the caldera and volcanic crater articles?

The former claims the two terms are distinct, while the latter claims the former is a subtype. I'm sure there are disagreements. 161.185.160.21 (talk) 17:23, 23 February 2018 (UTC)

I do not think that volcanic crater article claims that caldera is a subtype of volcanic crater. Ruslik_Zero 20:36, 23 February 2018 (UTC)

I've changed "crater" to "depression" in the volcanic crater article to help make the distinction. Does this help? Dbfirs 20:50, 23 February 2018 (UTC)

Clinical trial paradox

Not seeking medical advice (this originated as a math-related question rather than a medical one) but science desk seems better because I'm wondering how actual drug trials and the medical profession deal with the issue.

There's a syndrome S for which there are a bunch of prescription drugs D₁, D₂, etc. If you treat S with a placebo, about 25% of patients get better. Similarly if you treat S with D₁, about 25% get better, and the same with D₂, with D₃, and so on. The obvious reaction is "well those drugs are no better than placebos, they're being overmarketed by greedy pharma companies bla bla, the only cure is clean living so quit those filthy habits like editing Wikipedia!". In fact I think the FDA generally won't approve drugs that don't outperform placebos in clinical trials, which seems logical.

The above seems paradoxical for the following reason. Experiment might show that placebos are the same as a class: if placebo P₁ doesn't work then P₂ won't either, while the drug actions are much less correlated. So the medical strategy doctors observably use for treating S is: try drug D₁. If it doesn't work, try D₂, etc. If the drugs are independent, then after k of them, $1-\left({3/4}\right)^{k}$ of the patients should get better, and in the $k\rightarrow \infty$ limit, all of them should get better. So the drugs are (collectively) useful after all. In fact the same could be said even if the clinical trials showed each drug to be *less* likely to help than a placebo.

Drug trials afaict usually test one drug at a time, or there might be a multi-armed test where different sub-populations get different drugs. Do they ever test correlations between drugs? Is there a way to do that without testing multiple drugs on the same patients? Is this a known issue in the clinical trials field and something that gets reasonable study? It's obvious that doctors know about it informally which is why they try multiple drugs (this question is inspired by something I saw on a medical blog) but maybe there's a disconnect. Thanks. 173.228.123.121 (talk) 20:38, 23 February 2018 (UTC)

If a drug works even for some patients, it will still work better than placebo on average. The situation that you describe can only be possible if the drug actually makes the syndrome S worse in some patients. Then it will be up to researches to devise criteria when the drug should be used and when not. Ruslik_Zero 20:56, 23 February 2018 (UTC)

I don't see why a sometimes-working drug would be better than placebo on average (maybe it helps 20% of patients and leaves 80% with their condition unchanged or makes them worse, while placebo improves 25% and does nothing for the rest). But anyway that's a corner case. The controversy I see is that drugs get approved after (maybe only slightly) beating placebo, and then there's reproducability concerns or claims that the effect wasn't statistically significant enough, etc., yet doctors still consider the drug to be a useful part of their toolbox.

I do think it's normal to dispense drugs that have a potential for making the illness worse (or even for causing illness, like vaccines given to healthy people). It's just a trade-off like anything else. 173.228.123.121 (talk) 00:12, 24 February 2018 (UTC)

You still seem to be conflating different things. A drug which doesn't help 80% of patients is a very big difference from a drug which makes the condition worse in 80% of patients. If the drug helps enough patients without making things worse in the rest, it's easily possible this improvement will be picked up even in a simple statistical test of all patients. Also I think you're overestimating the number of drugs which are approved which are known to make the condition worse in a subset of patients, especially if we're talking about something which is easily measured e.g. blood pressure, cholesterol levels, uric acid levels. More likely the drug has known side effects which sometimes could be worse than whatever you are trying to resolve. Also some drugs have a narrow therapeutic range and require careful monitoring because you may overcorrect what you're trying to adjust. E.g. warfarin. BTW, drugs are not always tested against placebos even in the initial approval stages. They may be tested against some existing drug for the same condition. Nil Einne (talk) 01:34, 24 February 2018 (UTC)

Patients don't know whether they're being given the placebo or not. So the same 25% that heal through the power of placebo will heal anyway, leaving us to assess the drug's usefulness on the basis of the other 75% (in the long run of course). 93.142.92.135 (talk) 01:33, 24 February 2018 (UTC)

Nil, I think you're focusing on about the least relevant part of the question, but obviously just as a math tautology, a drug can cause expected worsening of a condition and still be worth trying, depending on how bad the potential worsening is and whether it's reversable. E.g. if I'm nearsighted and I'm offered a drug that has 20% chance of making my nearsightedness better and 80% chance of making it worse (but it must taken every day to continue having any effect at all), then of course I'll want to try it. If it works, great. If not, I stop taking it and I'm back where I was. If there are 10 such drugs each with an independent 20% chance of helping, I try one after another and have a 90% chance of eventually finding one that helps, seems like a good deal. (That's obviously an artificial conceptual example rather than a common practical situation).

The question is simply about a line of reasoning that I see a lot, that goes: "Drug X was compared against a placebo (or against drug Y) in a clinical trial for treating disease D and it didn't do significantly better according to standard statistical tests. Therefore, drug X's claimed usefulness against disease D is unsupported by sound evidence, the actual experiences of patients getting better after taking it should be treated as meaningless anecdotes, and doctors should quit prescribing it". That reasoning is tempting (I believed it myself without thinking about it much) but on closer inspection it's clearly fallacious. So I'm wondering how recognized the fallacy is, whether it's taken into account in trial processes, etc. Of course I recognize that this may be far down in the weeds compared with other sources of bias in those experiments, usually in the other direction. 173.228.123.121 (talk) 03:17, 24 February 2018 (UTC)

I never said it's impossible. I'm simply said it's anywhere as common as your response seemed to suggest. (Not simply because of how many drugs work, but also because if a drug is like that, for a lot of conditions it's probably going to be abandoned maybe fairly early on. And that you need to differentiate between the far more common example where are drug does not appear to improve the condition in some patients, and the case where the drug appears to make the condition worse for some patients since you seem to conflate the two in your first 2 responses. You need to properly understand the basics before you try and understand more complicated things. To give an obvious example you said "I don't see why a sometimes-working drug would be better than placebo on average". But as Ruslik_Zero had already said, and I emphasised, it should be fairly obvious why a sometimes working drug is often statistically better than a placebo on average if you understand the basics i.e. well statistics and that most drugs do not make the condition worse, and the placebo effect still applies to the drug. (The other IP's point on the placebo effect was another thing I noticed your response seemed to be confused about, but I didn't comment on.) Or to put it a different way, as long as you are confused by the basics, you are liable to be even more confused if you try and analyse more complicated things, you can't simply dismiss understanding the basics as the "leave relevant part". Nil Einne (talk) 03:45, 24 February 2018 (UTC)

P.S. No part of my responses are intended to suggest a drug will always be statistically significantly better if it works in some people even if it doesn't generally work in others despite not making things worse. Isosorbide dinitrate/hydralazine is an obvious example here [2]. A number of cancer drugs are also approved dependent on some genetic indicator either in the cancerous cells or the patient. (Although they may not have actually bothered to test it for effectiveness on others initially, depending on their understanding of the mechanism of action etc. One of the points here of course is that with better understanding of what we're trying to achieve and why, we are able to better guess whether it's actually going to work. So in some ways you're still actually targeting a specific problem aiming to achieve a specific outcome. It's just more complicated than 'lowering high blood pressure'. And something like 'malignant breast cancer' is actually very non specific.) Nil Einne (talk) 04:05, 24 February 2018 (UTC)

(edit conflict) Ruslik0 wrote "If a drug works even for some patients, it will still work better than placebo on average" which is a non-sequitur for reasons I explained. It might often be true in practice, but it's not a valid logical inference. I interpreted an unqualified "if X then Y" as "if X then necessarily Y as a matter of pure deduction", rather than as "if X then often Y in the context of how people tend to test drugs". If Ruslik0 meant it the second way then oops, I misinterpreted their words, but that's different from being confused by the basics. If they meant it the first way, then it's a math error as far as I can tell.

That placebos sometimes have a real effect is widely reported (Wikipedia's placebo article describes various mechanisms for this, though some studies claim the opposite;[3] I should update the article) and I've heard this is sometimes true even when the person knows they're getting a placebo.

The disconnect between reported experience and clinical trial results is substantial (the whole field of alternative medicine could be described as people taking stuff that failed clinical trials and saying that it works anyway). Some of it is obviously bogus but some could possibly be explained by the effect under discussion. I don't know how important the issue is, but I do get the impression that it's overlooked. Anyway I appreciate the responses though at this point we've probably run out of meaningful ideas. Thanks! 173.228.123.121 (talk) 05:46, 24 February 2018 (UTC) (Added: I see now that by "on average" Ruslik0 may have meant averaged over drugs, or averaged over trials, or whatever. I had read it as averaged over patients participating in a specific trial). 173.228.123.121 (talk) 05:53, 24 February 2018 (UTC)

No they were simply wrong since it might affect some people badly. Many drugs can have some bad side effects and it is a question of balance of the benefits against the possible side effects. Unfortunately people react a bit differently to things. You'll find a list of possible side effects on a leaflet with any packet of medicine. The assesment has been that the risk of these is acceptable compared to the benefits. The balance is normally tipped to the side of avoiding bad side effects and that will cause some drugs to not be approved unless a good criterion can be found for spotting when the bad effects might occur and avoiding them. It is a bad idea to use such drugs and risk being unlucky. Dmcq (talk) 10:23, 24 February 2018 (UTC)

I think that you misunderstand the statistics involved. If the probability of placebo effect in a given patient is $p_{1}=1/4$ (as you said) and the probability that a drug will help this patient is $p_{2}$ then the probability that the drug will make a difference is $P=1-(1-p_{1})(1-p_{2})=p_{1}+p_{2}(1-p_{1})\geq p_{1}$. The equality will mean that $p_{2}=0$, which implies that the drug is ineffective. So, in your hypothetical example when all drugs have the same efficiency as placebo, the efficiency of all those drugs is zero. So, regardless of how you combine them the efficiency that you will get is the efficiency of the placebo. Ruslik_Zero 18:54, 24 February 2018 (UTC)

Agreed. In a properly controlled trial, where the control patients are influenced to the same level of conviction that they're taking the real drug as the actual users of the drug, the placebo effect works equally on both sides. So, even for a terribly ineffective drug with 5% success rate, 25% of the patients (in your example) will heal from the placebo effect, and 5% of the other 75% will heal from the drug, giving a total of 28.75%. Technically, it's possible to have a drug's success rate lower than the placebo effect, for example if 15% of the patients heal and the other 85% experience horrible side effects negating the placebo effect. In that case though, this might create and ethical dilemma since giving the patients the drug appears to be worse than doing nothing. 93.142.90.67 (talk) 19:26, 24 February 2018 (UTC)

Ruslik0, sure, if you model it that way then that formula is valid. But IMHO it's an unwarranted assumption about what the drug does. I'd start with the experimental results. If you see 25% of the placebo patients and 20% of the drug patients get better (and it's a big enough sample that that's unlikely to be a chance result) you have to take the numbers as they come, rather than saying there's something wrong with them because they don't fit your model. You similarly can't rigorously infer that the drug didn't help any patients. There could be a confounder that shuts off the placebo effect for the drug patients, so the 20% that got better that really did get helped by the drug. I agree that this doesn't sound likely with real drugs, but it's ok mathematically as far as I can tell. 93.142.90.67 yes I think we're saying similar things, more or less. 173.228.123.121 (talk) 01:51, 25 February 2018 (UTC)