Talk:mean absolute deviation

Deletion discussion

The following information passed a request for deletion.

This discussion is no longer live and is left here as an archive. Please do not modify this conversation, but feel free to discuss its conclusions.

---

mean absolute deviation

From the same IP as data value above. Evidently sum of parts. —Mr. Granger (talk • contribs) 12:57, 25 September 2015 (UTC)

• Looks like sum-of-parts to me - so delete. SemperBlotto (talk) 06:49, 26 September 2015 (UTC)

I have noticed that anything mathematical easily gets doomed SOP here, whereas terms like "embalming fluid" (right above) provoke lengthy discussions which often result in keeping the term or definition due to lack of consensus. I'd say this is statistical term which on one hand has a clear definition but which on the other hand can refer to a number of concepts as both the averaging method and the definition of the central point may vary. --Hekaheka (talk) 23:57, 27 September 2015 (UTC)

No, this really is just SOP. Delete. —Μετάknowledge^{discuss/deeds} 03:06, 28 September 2015 (UTC)

Keep. It cannot be sum of parts because the special meaning of absolute is not in Wiktionary. I've clarified the definition and believe that it is now no longer a sum of parts. Dbfirs 07:34, 28 September 2015 (UTC)

Could you explain why the new definition is not sum of parts? To me it looks like mean + absolute deviation. —Mr. Granger (talk • contribs) 12:38, 28 September 2015 (UTC)

Yes, I see what you mean. The definition was incorrectly linked, so I didn't notice that we have absolute deviation. I still think it's worth an entry because the meaning is not easily found from the separate words. Dbfirs 20:53, 30 September 2015 (UTC)

@Dbfirs I've added absolute. We should've had it.​—msh210℠ (talk) 18:23, 12 October 2015 (UTC)

Thanks, the meaning can now be inferred from the separate words, though possibly with some difficulty. I'm now neutral on the keep or delete scale. (We've kept more obvious sums of parts.) Dbfirs 09:51, 16 October 2015 (UTC)

IMO, if one knows no statistics, the meaning of this term is not transparent from its components. If one does, it is. So, Delete.

IMO it is analogous to: If one knows no German, many German expressions are not transparent because one would not necessarily know how to construct their meaning from their components. If one does know German, many expressions, and even single compound words, are transparent from their components.

@Dbfirs: "It cannot be sum of parts because the special meaning of absolute is not in Wiktionary." If you were to mean by this that we have to keep expressions that would be transparent because we have failed to provide a proper definition of component terms, I would have to disagree. We simply have an obligation to add any justifiable omitted definition, as you did here with absolute. DCDuring TALK 10:35, 16 October 2015 (UTC)

I would agree with you were it not that we keep lots of other sums of parts on tenuous grounds. Dbfirs 10:59, 16 October 2015 (UTC)

• Keep. It only became clear to me what that is supposed to mean when I read the definition provided by Dbfirs ("In a set of data, the average of the absolute values of the deviations from a chosen central point"), and after I checked multiple sites that provided even better clarity. mathworld.wolfram.com[1] states that this thing is also called "mean deviation", and the site only includes the data's mean as the central point. The original definition "The amount of variation or dispersion of a set of data values" was so vague as to be nearly worthless. Now as to whether this is a sum of parts, I would ask, deviation from what? The sum of parts term would be, the mean absolute deviation from the mean. If I take the data set minimum as the reference point, why is not the mean absolute deviation from that reference point also mean absolute deviation? That could be answered by pointing to absolute deviation, but that states that the deviation is from "a given value" without making it clear that the given value is not e.g. the minimum; it further states "(such as mean or expected value)" but that still does require that the given value has to be some kind of central point; it merely hints. And even then, the user would have to know to look up "mean" + "absolute deviation" rather than "mean" + "absolute" + "deviation". And we would have no way of stating that "mean deviation" is a synonym in our lexicographical database; we should have mean deviation since "absolute" is merely implied. Thus, I do not think this is really a sum of parts, and even if it is, I think the user is better off with our having the entry. --Dan Polansky (talk) 08:31, 25 October 2015 (UTC)

  Even with absolute now expanded with "As measured using an absolute value", the reader still has to figure out that it is this sense of "absolute" that applies, and not a sense that is the opposite of relative. And the reader has to click to absolute value. We can do a much better service by providing a clear definition at mean absolute deviation, possibly even with an example in a usage note.

  As a question for attestation, it should be clarified whether anyone really uses the term "mean absolute deviation" to refer to "median absolute deviation" and whether it is "median absolute deviation from the median" or "median absolute deviation from the arithmetic average". Thus, we should look at actual lexicographical data to find what central points are actually plugged into the term. --Dan Polansky (talk) 08:40, 25 October 2015 (UTC)

• Keep. The term appears in the 2011 Dictionary of Statistics & Methodology: A Nontechnical Guide for the Social Sciences, in a 1918 Dictionary of Military Terms, and the 2014 Concise Oxford Dictionary of Mathematics, among others. bd2412 T 14:02, 28 October 2015 (UTC)

Kept. bd2412 T 17:45, 1 November 2015 (UTC)